Uncertainties in land use data may have substantial effects on environmental planning recommendations: A plea for careful consideration

Felix Neuendorf; Julia Thiele; Christian Albert; Christina von Haaren

doi:10.1371/journal.pone.0260302

. 2021 Nov 24;16(11):e0260302. doi: 10.1371/journal.pone.0260302

Uncertainties in land use data may have substantial effects on environmental planning recommendations: A plea for careful consideration

Felix Neuendorf ^1,^*, Julia Thiele ¹, Christian Albert ², Christina von Haaren ¹

Editor: Gerald Forkuor³

PMCID: PMC8612552 PMID: 34818356

Abstract

A key challenge of environmental planning is to craft recommendations for future sustainable spatial development amid ubiquitous uncertainties. This paper aims to explore how different data uncertainties, usually unknown to the planner, may influence environmental planning recommendations. We apply a case study-based approach, in which we provide three illustrative examples of how data with different kinds and levels of uncertainty affect environmental assessments and, by that, the decision-support provided by environmental planning. The cases stem from different spatial levels in Germany and consider ‘Regional soil-based climate change mitigation’ in the region of Hannover, ‘State-wide habitat conservation siting’ in the federal state of Saxony-Anhalt, and ‘National renewable energy planning’. Based on the three examples, we discuss implications for planning practice and derive recommendations for further research. The three cases studies illustrate the substantial effects of data uncertainty on environmental assessments and planning recommendations derived from those results. We identify four problem constellations of dealing with data uncertainty in environmental planning that relate to the severeness of uncertainty impacts, the responsibility of the decision-maker, and the kinds of impacts that wrong decisions may have. We close with recommendations for further research, among others to develop robust and pragmatic methods for identifying the uncertainty levels in environmental data and assessment results.

Introduction

A key challenge of environmental planning is to craft recommendations for future sustainable spatial development amid ubiquitous uncertainties due to incomplete understandings of the ecological, social and economic conditions, trends, and interrelations [1, 2]. A recent review revealed that the main planning relevant categories of uncertainties in environmental planning can be associated with four categories [3]: Data uncertainties, which are most commonly associated with geometric uncertainties that arise mainly from resolution, thematic uncertainties that occur when values are attributed to objects, and temporal mismatches between data sources [4]. Model uncertainties, the second category, relate to challenges in choosing appropriate indicators and variables for environmental assessments [5]. Models applying only a small number of explanatory variables are generally easier to use for planners, but are often inferior to more complex and dynamic models in understanding environmental system change. The most obvious trade-off in model development for planning purposes is between precision and usefulness, for example when a complex model would become more demanding in data requirements [6]. In addition, a very high level of detail can be impractical and a source of uncertainty in planning applications when the interpretation of complex model results puts high demands on user [7]. The third category, projection uncertainties, is linked to projecting future states of the environment—atypical challenge for planning as a strong forward looking action [8]. Plans often suggest that problems are of a static nature, but in reality they are highly dynamic [9] and can often not be predicted with the understanding of future behaviour of various components [10]. Finally, valuation uncertainties are introduced when norms and laws need to be interpreted to form scales for valuation [3]. While data, model and valuation uncertainties are prominent at multiple stages of the planning cycle, projection uncertainties come into play in particular when scenarios for future developments are created and potential scenario impacts are being explored.

Uncertainty of spatial and statistical data presents one of the most important uncertainties, as such input data inaccuracies propagates through the entire environmental assessment and planning process. Relevant concerns of environmental planning in terms of data uncertainties include, among others, mistakes in categorising (for example, if particular land parcels are incorrectly attributed to land use types), inconsistencies between the different datasets (mainly the location and spatial delineation of features) and impacts of data resolution (for instance if the resolution it too coarse to divide site-specific management recommendations to farmers). Some of these uncertainties may be obvious, but not quantified, others may be hidden in the data set. Both types of uncertainties are usually not explicitly communicated with the data provision. Nevertheless, the planner has to decide which data detail will suffice for which specific planning task—risking misjudgement. As planners are striving for efficiency in the planning process, they tend to use the data set that is least detailed while still fulfilling addressing the planning question with sufficient accuracy and robustness. Also, not uncommon, data has to be used which is obviously not optimal but available, and often planning recommendations necessarily reflect an unquantified or unquantifiable uncertainty.

Over the last decades, environmental planners devised some strategies for providing decision-support amid such unknown or suspected data uncertainties. One is the participation strategy according to which the involvement of diverse knowledge holders in the environmental assessment and planning process can integrate complementary insights and minimize potential bias. However, this strategy seems to be effective only at local level. Most common in theory but costly and thus rarely used in practice is the adaptive management strategy [11, 12] that proposes to frame interventions as hypotheses, to monitor impacts of implementation, and to continuously revise decisions in face of enhanced understandings [13]. In Europe, applying such strategies is also guided by the legally enshrined precautionary principle according to which environmental planners should use caution and additional safeguards in situations where scientific understanding and data certainty is lacking and decisions could potentially cause serious harm to people and nature [14]. Finally, environmental planners often presume that an intuitively felt uncertainty may be present in their basic data, but that this will not relevantly affect the response measures. The logic behind this is that planners often strive for multifunctional measures which intentionally target a larger area than monofunctional measures that usually address only smaller areas, for example to safeguard a certain species. Along this line, planners generally presume that the inaccuracies of land use data will not affect the validity of recommended measures which address multiple purposes in an aggregated spatial unit. Furthermore, in case of precautionary measures, the uncertainties of proposals of future land uses seem to be much more relevant than uncertainties stemming from inaccuracies of cartographic land use data.

The problem remains, however, that environmental planners usually miss comprehensive understanding of the kind and degree of severity of the uncertainty embedded in their data, and how this might affect their assessment results which constitute the basis for their recommendations. In consequence, planners remain unsure if the above outlined uncertainty tackling strategies are needed, if they work or if opportunities for better targeted or efficient recommendations are missed. We hypothesize that a lack of uncertainty information or awareness may lead to substantial mismatches between actions proposed and needed, and that providing more uncertainty information about geodata would allow for better-adapted planning recommendations.

This paper aims to explore how different data uncertainties, usually unknown to the planner, may influence environmental planning recommendations. We apply a case study-based approach, in which we provide three illustrative examples of how data with different kinds and levels of uncertainty affect environmental assessments and, by that, the decision-support provided by environmental planning. Based on the three examples, we discuss implications for planning practice and derive recommendations for further research.

Methodological approach

Our research method follows a typical qualitative and comparative case study analysis [15]. In order to test our hypothesis, we selected three case studies from different levels of environmental planning decision making in Germany (Table 1). The first case study, ‘Regional soil-based climate change mitigation’, concerns environmental planning for Greenhouse Gas (GHG) emission reduction from soils in the region of Hannover. The case study focuses on identifying priority areas for climate protection and management. Potential users of such information would be the state or the regional and local governments. The second case study, ‘State-wide habitat conservation siting’, addresses the identification of valuable habitats for protection planning in the German federal state of Saxony-Anhalt. Case study number three, ‘National renewable energy planning’, concerns an environmental planning effort aiming to identify areas for the siting of renewable energies, which should also guide decision-making at the regional level.

Table 1. Data applied in the case studies.

Case study	Case study location / model run	Land use / Land cover classification	Date of mapping / release date	Reference size / spatial accuracy	Number of different LC classes
First case study: ‘Regional soil-based climate change mitigation’	Location: Lower Saxony	Biotope-mapping	2007–2011	-	557^*
		ATKIS (ni-LBM)	2009	1ha	44
		CORINELand Cover	2012	25ha	30
Second case study: ‘State-wide habitat conservation siting’	Location: Saxony-Anhalt	Biotope-mapping	starting 2009	-	184^*
	Location: Saxony-Anhalt	ATKIS (Basis-DLM)	Starting2011^**	3–5 m	50
Third case study: ‘National renewable energy planning’	First model run	• DLM250	2015	100 m	36
	First model run	• DEM200	2014	Grid cell size 200 m; Accuracy: Planimetry:± 1–3 m; Altimetry: ± 3–10 m	-
	Second model run	• ATKIS (Basis-DLM)	2018	3–5 m	50
	Second model run	• DEM50	2017	Grid cell size 50 m; Accuracy: Planimetry: ± 3 m; Altimetry: ± 2–4 m	-

Open in a new tab

* Federal states use different classification manuals which result in different number of classes on a comparable level.

** Very high actuality on classes that are listed as “up-to date”, mainly infrastructure or demarcations with legal consequences.

For each case study, we apply two or three data sets and compare the assessment results in which the data is used, and the planning recommendations deduced from the assessment results. We compare resulting maps also in terms of spatial consistency, meaning areas where there is a spatial match of class labels between different outputs. One of the data sets applied in each case represent data easily available at the respective planning level that might thus be chosen by planners as the basis for assessments despite of having no detailed information about inherent uncertainties—which is a realistic assumption (easy data variant). The second or third data set applied represented the best available data and was used as a reference (reference variant). The juxtaposition of the planning outputs based on data of different levels of uncertainty, and an exploration whether the uncertainty of the less detailed data would be acceptable for the planning purposes, will provide us input for the reflection on implications of data uncertainties on environmental planning recommendations in the discussion section. In addition, we explored whether knowledge about structural uncertainties or errors in the dataset could lead to better adapted planning recommendations even if the easily available data set is used.

Data sources and properties in the three case studies

Our first two case studies, ‘Regional soil-based climate change mitigation’ and ‘State-wide habitat conservation siting’, draw on two or three kinds of spatial datasets commonly used in different sectors of spatial planning in Germany (see Table 1). As the first and coarsest dataset, we used CORINE Land Cover 25 (CLC25) data. It represents the European standard for land cover classifications and was introduced in 1985 as part of the Copernicus programme. It is being updated every six years using the latest remote sensing technologies. We used the 2012 dataset that has been generated using IRS P6 LISS and rapid eye data and has a minimum mapping unit of 25 hectares. The CLC 25 data of 2012 has a proclaimed thematic accuracy of > = 85% with the addition of the phrase “probably achieved” [16].

The cadastral data, called ATKIS (Digital landscape model—base data), is the second data source used in our case studies. ATKIS data currency differs between federal states with a maximum time span of five years between complete updates. In Saxony-Anhalt, the base data set is updated every three years to ensure a higher degree of topicality. Selected land use/land cover classes such as roads, water bodies and protected areas are updated every three to twelve months. Thus, in most cases, the ATKIS dataset represents the most up-to-date information on land use and land cover. The specified positional accuracy is three to five meters. For Lower Saxony, we used the data of Lower Saxony land cover model, which resembles the ATKIS base data but has less thematic classes and a minimum mapping unit of one hectare.

Biotope mapping data, the third dataset, is used as a reference, as it is the most detailed type of land-use data used for environmental planning in Germany and we assume that it depicts the real circumstances best. Biotope mappings are generated using a combination of remote sensing, mainly interpretation of aerial images, ATKIS data and field surveys. There is no minimum mapping unit for biotope mappings, other than that the biotopes should be easily distinguishable in the field [17]. However, also this basic data set cannot be considered completely reliable, as it is not updated as often as ATKIS data. Furthermore, while spatial accuracy of the biotope delimitation can be considered relatively similar across federal states, the assessments differ in their class hierarchies and in specific thematic information given within each class [17, 18]. This highlights the importance of the uncertainty analysis of different data sources, and of describing the sources of uncertainty to be expected in each data set.

For the purpose of the assessment of inconsistencies between the data sets, the classes have been aggregated. We opted to aggregate classes with high relevance for landscape planning into groups that best reflect important areas, the baseline being higher aggregation levels in the biotope mapping. A different aggregation scheme might be used when addressing inconsistencies in another branch of planning. In difficult cases, experts were consulted to ensure accurate classifications.

Assessment methods applied in the case studies

The case study ‘Regional soil-based climate change mitigation’ illustrates the impacts of applying two sets of commonly used land use data with different levels of consistency on recommendations for minimizing GHG emissions from soils in the Region of Hannover. The land use data sources used included ATKIS data, and CLC 25 data and for reference, biotope mapping data. For each dataset, we applied the same method for estimating potential CO₂-emission and retention potentials [19]. The applied method combines data about soil types and land-use/land-cover which are classified as to climate relevance, and should be available for a coherent time span of 5 years. The investigation was conducted for the reference year 2015, i.e., the results show the potential CO₂ retention function for this year. For this purpose, the biotope type mapping was extended with information from the LEA portal (SLA) on the agricultural impacts of the years 2010–2015. For the investigation of the ATKIS data, it had to be assumed that both grassland and arable land have existed for more than 5 years due to a non-optimal data situation.

The case study example ‘State-wide habitat conservation siting’ focussed on the impacts of using data sources of different uncertainty levels when identifying valuable habitats for species protection planning in the German federal state of Saxony-Anhalt. We used ATKIS data, and biotope mapping data as reference. We identified habitats of very high conservation value in the biotope mapping dataset, and compared the consistency with respective classes in the ATKIS data. The considered land use classes include ‘fen, bog and marshlands’, ‘heathlands’, and ‘small natural standing waters’. We expected a higher inconsistency in the latter class, because ATKIS data only has standing water as a delimitable class and no distinctive features regarding the area size. We applied descriptive statistics to explore commonalities and differences, an analysis method that can be easily communicated and comprehended by different addressees.

The case study ‘National renewable energy planning’ explored impacts of using datasets of different uncertainty levels on the potentially suitable sites for renewable energies, taking into account both human and nature-conservation needs [20]. We used the data and results of two model runs, the first one was conducted by Walter et al. 2018 and the second one by Thiele et al. 2021, for an uncertainty analysis in order to quantify the influence of data resolution on identifying opportunity spaces for generating renewable energy.

In the first run of modelling the area potentials for renewable energy, data was used which are easily available on national scale [21], namely the digital landscape model 1:250,000 (DLM250) and the digital elevation model 200 (DEM200). These spatial data were considered sufficient for generating a satisfactorily accurate estimation of opportunity spaces for renewable energy generation that are compatible with human and nature conservation considerations. It was also assumed that the data accuracy was sufficient to scale down the spatial results to the level of federal states. The DLM250 was used to select, for example, settlements, infrastructures, water bodies or arable land. Areas with slopes greater than or equal to 30° were determined as not suitable sites for wind turbines [21]. These areas were identified in the first run with the digital elevation model (DEM200), which has a grid cell size of 200 m. The height accuracy of this low-resolution model is +/- 20 m.

In order to define the inaccuracies of the results due to the data resolution, a second model run with more precise data was conducted [20, 22], namely ATKIS Basis-DLM and the DEM50 digital elevation data. For the selection of land use types such as settlements and water bodies, the ATKIS Basis-DLM was applied, which provides a much higher surface sharpness with a resolution of 1:25,000 than the lower-resolution DLM250 used in the first run. The second run also applied the DEM50 elevation data with grid cell sizes of 50 m and a height accuracy of +/- 2.5 m to 10 m, instead of the coarser one from the first run. We compared the spatial results by quantifying differences in the spatial input data, and by subtracting the result dataset of the first model run from the result dataset of the second model run. In order to compare the two digital elevation models (input data), the DEM200 was converted to a grid with a grid cell sizes of 50 m. Since the potential areas for onshore wind energy of the two model runs were also provided as grids with a grid cell size of 50 m, they could be compared without further transformations.

Results

The results of the assessment performed in the case study ‘Regional soil-based climate change mitigation’ (Fig 1, Table 2) show dramatic differences between the modelled quantities of CO₂depending on the data sets used [23]. Differences in the spatial delineation of peatland and bog soil areas between ATKIS or CLC 25 data are most important here, as the quantitative CO₂ emission risk and potential of those soil types are much higher than that of other hydromorphic soils (>1700 versus >70t CO₂). While the overall consistency of ATKIS and CLC 25 data is relatively high with 76% and 80%, respectively, substantial differences in the CO₂assessment results emerge depending on which data source and associated uncertainties are applied (see Table 2). The large discrepancies between the datasets highlight the importance of a good representation of sites of importance for climate protection. The consistency of areas of highest level of climate protection (level 5) is substantially lower, with ATKIS data only achieving 31% and CLC 25 data achieving 44% of consistency compared with the reference data. The decline in correctly represented area of around 2,267 hectares when using ATKIS data corresponds to about 4.2 million t CO₂ retention potential at risk to not be considered in planning proposals. Surprisingly, using coarser CLC 25 data leads to better results, with a smaller area deficit of around 1,472 hectares that accounts for around 2.8 million t CO₂ retention potential untapped.

Table 2. Quantitative results of the case study ‘regional soil-based climate change mitigation’.

Climate protection function according to Saathoff et al. 2013		Level 1: Potential CO₂ emission of 1700–2600 t/ha	Level 2: Potential CO₂ emission of 70–160 t/ha:	Level 3: No emission/retention	Level 4: Potential CO₂retention of 70–160 t/ha	Level 5: Potential CO₂ retention of 1700–2600 t/ha	Overall (ha) / CO₂ Balance (t) [- = Retention. + = Emission]
Data set		LU/LC: Arable land and Grassland Soiltypes: Peat and bog soils	LU/LC: Arable land (<5 years old) Soil types: all other hydromorphic soils	LU/LC: Arable land (>5 years old) and Grassland (<5 years old) Soil types: all other soils	LU/LC: Grassland (>5 years old) Soil types: all other hydromorphic soils	LU/LC: Peatland and bogs Soil types: Peat and bog soils
Biotope Mapping *Reference dataset*	Surface area in ha	7,842	733	96,154	10,518	4,155	119,401
Biotope Mapping *Reference dataset*	Potential CO2 emission or retention in t	14,798,341	67,905	0	986,309	7,418,954	6,460,982
ATKIS	Surface area in ha	4,817	0	98,993	21,897	1,887	127,594
	Difference to reference in ha	-3,025	-733	+2,839	+11,380	-2,267	+8,193
	Consistent area in ha (same lvl as in reference)	3,455	-	83,344	6,470	1,279	91,093
	Consistency in %	44.0	0	86.7	61.5	30.8	76.3
	Potential CO2 emission or retention in t	8,941,704	0	0	2,384,509	3,261,472	3,295,723
	Deviation from reference regarding the emission/retention potential in t	-5,856,637	-67,905	0	+1,398,199	-4,157,482	-3,165,259
CLC 25	Surface area in ha	2,262	370	118,127	12,487	2,683	135,928
	Deviation from reference regarding the in ha	-5,580	-363	+21,973	+1,969	-1,472	+16,527
	Consistent area in ha (same lvl as in reference)	1,403	3	88,735	5,043	1,830	95,608
	Consistency in %	17.9	0.4	92.3	47.9	44.0	80.1
	Potential CO2 emission or retention in t	5,070,617	66,242	0	1,387,255	4,579,275	**-829,671**
	Deviation from reference regarding the emission/retention potential in t	-9,727,724	-1,663	0	+400,946	-2,839,680	-7,290,653

Open in a new tab

The use of ATKIS data results in an underestimation of about 5.8 million tons of potential CO₂ emissions compared to the reference data. This is because a total of 3,025 fewer hectare were mapped as Level 1 which corresponds to around 44% of the features labelled consistently in comparison to the reference data. In contrast to the highest retention level, the consistency levels further decrease, for example for CLC 25 data set with only 18%. This results in a loss of around 5,800 hectares in contrast to the 7,840 hectares in climate protection level 1 at the reference scale, which accounts for a potential underestimation of the CO₂ emission potentials of around 9.7 million tons.

These observations are very important when looking at the overall climate balance for the region. The use of our reference dataset suggests a climate balance would trend more towards a highlighted potential emission of around 6.5 million t CO₂. When using ATKIS data this would be reduce by about 50% to an emission potential of about 3.3 million t CO₂. Using CLC 25 data the statement would switch completely, in that the planners would assume a retention potential of about 830,000 t CO2.

Results of the case study on ‘State-wide habitat conservation siting’ (Fig 2, Table 3) illustrate that about 413 hectares of area with a need for conservation protection according to Biotope mapping data are not correctly represented when using ATKIS data (II + IV). However, about 58 hectares of those incorrectly mapped areas are at least attributed to another type of area with conservation protection needs. This would lead to a total of 355 ha of areas important to conservation not recognized in planning recommendations due to the lower detail of the data used. As expected, a large share of about 3,570 hectares of areas are wrongly identified as areas that require protection status, mainly due to larger standing water bodies represented in the ATKIS data.

Table 3. Quantitative results of the case study ‘state-wide habitat conservation siting’.

	Type of area considered	Fens, moor and marshlands (ha)	Heathland (ha)	Small natural standing waters (ha)	Total area (ha)
Biotope mapping (very important areas)	Reference data (most accurate data on biotopes)	247.45	388.72	103.48	739.65
ATKIS Basis-DLM	Comparison data	313.72	633.84	3003.84	3981.40
Assessment ofinconsistencies	I. Areas with consistent labels	74.90	179.87	71.58	326.35
	II. Areas with semi consistent labels (matching other very important areas)	5.64		52.10	57.74
	III. Excess areas(areas that are wrongly associated as important in ATKIS)	233.18	453.97	2880.15	3567.30
	IV. Missed areas (areas identified as important areas in the reference data, but not identified as such in ATKIS data)	120.82	204.62	30.12	355.56

Open in a new tab

Results for the case study ‘National renewable energy planning’ show that the area potentials for generating renewable energy are reduced when applying higher-resolution data: According to the first model run, 8,300 km² (2.3% of the area of Germany) are usable for onshore wind energy in a way that is compatible with humans and nature. The second model run shows a human- and nature-friendly usable area potential of 5,320 km² (1.5% of the area of Germany) [24]- a reduction of 36%, 2,980 km² [22] (Fig 3).

Compared to the first model run, industrial and commercial areas, for example, increase by 241% (4,616 km²) in the second model run. Recreation and leisure areas even increase by 3,561%. This can be explained by the fact that smaller industrial or recreational areas are represented in the ATKIS Basis-DLM, but not in the DLM250. The selected built-up areas of the Basis-DLM are slightly smaller (6%, 1,687 km²) than in the selection from the DLM250, but the selection of the Basis-DLM is considerably more exact and detailed. For this reason, after a 750-m-bufferfor noise protection was added to the built-up areas, the area in the second model run (Basis-DLM) increases by 123% (151,414 km²) compared to the first model run (DLM250) (Fig 4).

Infrastructures (e.g. roads, railroad lines), which are available as line features in the DLM250 as well as in the ATKIS Basis-DLM, could also be represented in a more detailed manner in the second model run due to the fact that the ATKIS Basis-DLM offers a variety of attributes compared to the DLM250. Road axes could thus be buffered according to width classes and railroad lines were buffered according to their number of tracks (2.5 m single-track railroad line, 4.5 m double-track railroad line) [20, 22].

Areas with slopes greater than or equal to 30° record an increase of 1,150%(6,570 km²) in the second model run (DEM50) compared to the first model run (DEM200). The difference between the selections of the two digital elevation models with identical SQL expression is particularly visible when zooming in on low mountain ranges: according to the DEM200, for example, almost no areas with a slope of ≥ 30° are found in the Harz Mountains [22] (Fig 5).

We can summarize for the case ‘National renewable energy planning’, that we observed the same effect on planning-relevant information as in the case of ‘Regional soil-based climate change mitigation’: The seemingly small error for instance in the spatial data for built up area for Germany has a much bigger effect when included in an assessment. The area excluded for wind turbines increases, as also a small hamlet (not considered in theDLM250 data set) has to be surrounded by a 750m buffer zone for noise protection.

Discussion and conclusion

This paper has provided three case studies from landscape and environmental planning in the German regional, federal state and national levels that illustrate implications of using data of different kinds and levels of uncertainty on environmental assessments and planning recommendations. All findings are discussed in the light of the integration of uncertainty assessments in planning practise still lacking [25, 26].

The case study on ‘regional soil-based climate change mitigation’ showed that in landscape planning the use of coarse data like CLC or data with fewer classes such as in ATKIS goes hand in hand with a serious loss of valuable information which can lead to a completely different planning statement. Areas with a high relevance for environmental planning are increasingly affected by inconsistencies which translate to uncertainties in potential planning proposals. The case study ‘State-wide habitat conservation siting’ showed that even the use of high resolution data might lead to erroneous results if aggregated thematic information does not match the classes of interest, as the increase in area with conservation need is mainly due to a lack in specific information in the ATKIS data rendering the distinction between different features as guesswork. Unfortunately, respective areas are of high importance for landscape planning and environmental planning in general. The third case study, ‘National renewable energy planning’, illustrated that applying coarser vs. more detailed spatial data in efforts to identify area potentials for renewable energy can yield differences by about 2980 km² in falsely allocated areas for onshore wind energy. The potential areas for onshore wind energy have decreased by 34% when using more detailed data. Taken together, the results re-emphasize the importance of carefully selecting datasets for landscape and environmental analysis for planning, as the first impression of sufficient data certainty might be misleading, and error-propagation through data application in models can result in substantially misleading results. Also more information is needed about characteristics of data sets. We found that data sets may be selective and inhomogeneous in their level of detail which may have considerable influence when used in assessment procedures.

The results of the three illustrative case studies also provide opportunities for reflecting on the implications for planning recommendations. Based on the findings, we distinguish four problem constellations in which the data uncertainty should be considered inappropriate as a basis for decision-support in landscape and environmental planning. The first constellation is when the potential errors in planning recommendations may lead to a pronounced different decision of a relevant dimension which counteracts legitimate planning objectives. This applies, for example, in the case study ‘National renewable energy planning’ where the results using easily obtained data could not been used for downscaling development limits for renewable energy because false allocations would harm environment and people.

A second problem constellation of inappropriate data uncertainty emerges if potential decisions based on this data lead to consequences which are destructive in a dimension for which the actual decision maker cannot take the responsibility. This applies, for example, if the decisions affect a longer time span than the decision-makers regular phase in office, or if it affects a larger area than that of which she or he has legitimate decision power. This problem constellation is demonstrated in both the case of ‘regional soil-based climate change mitigation’, where regional decisions about GHG mitigation affect, if so only marginally, the global climate conditions. In this case, the certainty of the assessment results based on ATKIS or even more so CLC 25 data does not provide sufficient certainty to find the areas where climate protection measures would be most needed and efficient and justify mandatory protection and management measures. Many areas of special importance would be overlooked others would be protected unnecessarily. The problem constellation also occurs in the case of ‘State-wide habitat conservation siting’, when, for example areas of European value are not identified and not appropriately protected by federal state-level decision-makers.

The third problem constellation is represented in situations where consequences of a decision may affect the rights of private subjects in a disproportionate way, in particular to use ones’ own property. Such impacts are usually forbidden, for example by the German constitution. This problem constellation also applies in the case studies of ‘regional soil-based climate change mitigation’, when the uncertain data would lead to recommendations that force land users to adopt particular GHG retention measures that are not actually helpful based on the real land use of the area. Vice-versa, it applies in the case of ‘State wide habitat siting’ when farmers are obliged to mow their grasslands only late in the season to protect particular breeding bird species, which in fact do not occur on those conditions.

Finally, the fourth problem constellation is when a large effort or expensive response would be required to address the issues raised by the model outcome and the data uncertainty is also large. Even if no harm would be caused by misallocating the response action, the resulting costs would violate the rules of sound financial management and budgetary rigour that should be applied at all political levels in the European Union. The case study ‘regional soil-based climate change mitigation’ shows that the certainty of assessment results based on ATKIS or even more so CLC 25 data (easily available data) are not suitable for an appropriate identification of areas where climate protection measures would be most efficient. In all of those problem constellations, planners need to be careful in making decisions and should apply the strategies of caution as outlined in the introduction, for example to try to use more precise data, or to apply the participation, adaptive management or precautionary principle.

We are aware that data uncertainty is only one component of the uncertainty spectrum and we posit that future research is also needed to incorporate considerations of all types of uncertainties in practical planning. In conjunction with our findings we propose that opportunities for future research are to develop robust and pragmatic methods for identifying the uncertainty levels in environmental data and assessment results, and to devise approaches for enhancing data certainty wherever possible. Developments in the remote sensing community on data quality assessment and communication [27, 28] can act as a suitable starting point. Established guidelines on data quality, as used for earth observation data [29], can be easily adapted to be used in data generation in general. In addition, planners should pay much more attention to distortions in modeling results due to data limitations. Uncertainties in assessment results and deduced planning measures have to be interpreted as to their relevance for the issues at stake, leading to more informed decision-making in the face of uncertainty. In many cases, a fast and preliminary decision within bounded uncertainty that addresses the problem constellations above would be better than an ill-informed decision or no decision at all. The introduced problem constellations already cover a broad range of potential planning and decision dilemmas but can only act as a starting point and should be extended when more experiences will be compiled and analyzed. Future research towards this goal should also cover other known types of uncertainties in planning which would only help in tackling these complex problems.

The realistic dilemma of using easily available data instead of better but hard to obtain data with uncertain added value in terms of relevance for answering the planning question will not vanish in the near future. In this situation it is important that uncertainties have to be communicated to the users of planning recommendations. The analysis of uncertainties though, still proposes a multitude of challenges for the practical planner [30]. In order to support practical environmental planning in dealing with uncertainties in a reasonable way, uncertainty analyses with big data sets should be performed for all standard geodata types which are used in environmental planning. This will produce uncertainty markers which can readily be used in practical planning as a basis for calculating the uncertainty in modelling or ecosystem service assessment.

If the uncertainty of the readily available data is known, the uncertainty of the assessment results as well as consequences for deduced respective measures can be judged as to their acceptability. Furthermore, planning recommendation can be adapted to the uncertain nature of the assessment results. For instance, the comparably little difference between ATKIS and CLC in the overall uncertainty of the state data, can be interpreted as a possibility for using CLC as very actual and readily available data when summarized results are expected for defining general risks or deciding about need for action. However, in most landscape planning applications this cannot be accepted for example for monitoring the development of habitat area or even for roughly calculating the land use induced CO₂ emissions on state level. In particular when area specific delimitations are needed and consequences for private property anticipated even ground truth mapping may be required in order to avoid unwanted consequences. Finally, in order to enhance the acceptability of uncertainty communication to decision makers, future research should explore how behavioral interventions such as splitting complex, uncertain decisions into smaller decisions with regard to space and time [31] would increase acceptability.

In a nutshell, the differences detected in the uncertainty of different land use classes were surprisingly high as well as the effect this had on the uncertainty of thematic ES-assessment results and thus on planning recommendations. This emphasizes our plea towards practical planning to pay more attention to data uncertainties as well as to distortions in their propagation in modelling and assessment.

Acknowledgments

The Authors would like to thank Jordan Siebel for his help in preparing the manuscript. We also thank the editor and two reviewers for their constructive comments on an earlier version of this manuscript.

Data Availability

Data for case study 1 and 2 is published on the Zenodo Data repository (DOI:10.5281/zenodo.5602765) and can be accessed via the URL: https://zenodo.org/record/5602766. Data for Case study 3 (DOI: 10.25835/0072778) is published and available via the URL: https://data.uni-hannover.de/dataset/dataset-areas-with-low-and-medium-spatial-vulnerability-to-a-prototype-wind-turbine.

Funding Statement

This study was funded by the German Research Foundation (HA 3301/8-1, 250763334), https://www.dfg.de/en/index.jsp - FN, and the German Federal Agency for Nature Conservation (FKZ 3515 82 4300), https://www.bfn.de/en.html, in cooperation with the German Ministry for the Environment, Nature Conservation and Nuclear Safety, - JT. The publication of this article was funded by the Open Access Fund of the Leibniz Universität Hannover. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Marcot B.G. The Science and Management of Uncertainty—Dealing with Doubt in Natural Resource Management. 2020: 286. doi: 10.1201/9780429299575 [DOI] [Google Scholar]
2.von Haaren C, Galler C, Albert C. (Eds.). Landschaftsplanung. 2021. ISBN: 978-3-8252-8579-1. German.
3.Neuendorf F, von Haaren C, Albert C. Assessing and coping with uncertainties in landscape planning: an overview. Landscape Ecology. 2018;33:861–878. doi: 10.1007/s10980-018-0643-y [DOI] [Google Scholar]
4.Schulp CJE, Alkemade R. Consequences of Uncertainty in Global-Scale Land Cover Maps for Mapping Ecosystem Functions: An Analysis of Pollination Efficiency. Remote Sensing. 2011;3(9):2057–2075. doi: 10.3390/rs3092057 [DOI] [Google Scholar]
5.Schulp CJE, Burkhard B, Maes J, van Vliet J, Verburg PH. Uncertainties in Ecosystem Service Maps: A Comparison on the European Scale. PLoS ONE. 2014;9(10): e109643. doi: 10.1371/journal.pone.0109643 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.McVittie A, Norton L, Martin-Ortega J, Siameti I, Glenk K, Aalders I. Operationalizing an ecosystem services-based approach using Bayesian Belief Networks: An application to riparian buffer strips. Ecological Economics, 2015;110:15–27. doi: 10.1016/j.ecolecon.2014.12.004 [DOI] [Google Scholar]
7.van der Biest K, Vrebos D, Staes J, Boerema A, Bodi MB, Fransen E, et al. Evaluation of the accuracy of land-use based ecosystem service assessments for different thematic resolutions. Journal of Environmental Management, 2015;156:41–51. doi: 10.1016/j.jenvman.2015.03.018 [DOI] [PubMed] [Google Scholar]
8.Gret-Regamey A, Brunner SH, Altwegg J, Bebi P. Facing uncertainty in ecosystem services-based resource management. Journal of Environmental Management, 2013;127:145–154. doi: 10.1016/j.jenvman.2012.07.028 [DOI] [PubMed] [Google Scholar]
9.Kubiszewski I, Costanza R, Paquet P, Halimi S. Hydropower development in the lower Mekong basin: alternative approaches to deal with uncertainty. Regional Environmental Change, 2013;13(1):3–15. doi: 10.1007/s10113-012-0303-8 [DOI] [Google Scholar]
10.Foley AM. Uncertainty in regional climate modelling: A review. Progress in Physical Geography, 2010;34(5):647–670. doi: 10.1177/0309133310375654 [DOI] [Google Scholar]
11.Ahern J. Theories, methods and strategies for sustainable landscape planning. In: Tress B., Tress G., Fry G., Opdam P. (eds) From landscape research to landscape planning: Aspects of integration, education and application. 2006;119–131. ISBN: 978-1-4020-3979-9 [Google Scholar]
12.Kato S, Ahern J. ‘Learning by doing’: adaptive planning as a strategy to address uncertainty in planning. Journal of Environmental Planning and Management, 2008;51(4):543–559. doi: 10.1080/09640560802117028 [DOI] [Google Scholar]
13.Walters CJ, Holling CS. Large-Scale Management Experiments and Learning by Doing. Ecology, 1990;71(6):2060–2068. doi: 10.2307/1938620 [DOI] [Google Scholar]
14.Read R, O’Riordan T. The Precautionary Principle Under Fire. Environment: Science and Policy for Sustainable Development, 2017;59(5): 4–15. doi: 10.1080/00139157.2017.1350005 [DOI] [Google Scholar]
15.Yin R. K. Case Study Research Design and Methods (5th ed.). 2014. Sage. ISBN-13: 978–1452242569 [Google Scholar]
16.Jaffrain G, Sannier C. GMES Initial Operations / Copernicus Land monitoring services—Validation of products—(Second Specific Contract). Corine Land Cover 2012. Final Validation Report. Villeneuve d’Ascq (FR): Systèmes d’Information à Référence Spatiale; IGN France International; 2017 [cited 2021 Mar 20]. https://land.copernicus.eu/user-corner/technical-library/clc-2012-validation-report-1.
17.von Drachenfels O. Kartierschlüssel für Biotoptypen in Niedersachsen unter besonderer Berücksichtigung der gesetzlich geschützten Biotope sowie der Lebensraumtypen von Anhang I der FFH-Richtlinie, Stand Februar 2020. 11th ed. Hannover: NLWKN; 2020. German.
18.Landesamt für Umweltschutz Sachsen-Anhalt. Katalog der Biotoptypen und Nutzungstypen für die CIR-luftbildgestützte Biotoptypen- und Nutzungstypenkartierung im Land Sachsen-Anhalt. Halle: Landesamt für Umweltschutz Sachsen-Anhalt; 1992. German.
19.Saathoff W, von Haaren C, Dechow R, Lovett AA. Farm-level assessment of CO2 and N2O emissions in Lower Saxony and comparison of implementation potentials for mitigation measures in Germany and England. Regional Environmental Change. 2013;13:825–841. doi: 10.1007/s10113-012-0364-8 [DOI] [Google Scholar]
20.Thiele J, Wiehe J, Gauglitz P, Pape C, Lohr C, Bensmann A, et al. Konkretisierung von Ansatzpunkten einer naturverträglichen Ausgestaltung der Energiewende, mit Blick auf strategische Stellschrauben. BfN-Skripten. 2021;614. doi: 10.19217/skr614 German. [DOI] [Google Scholar]
21.Walter A, Wiehe J, Schlömer G, Hashemifarzad A, Wenzel T, Albert I, et al. Naturverträgliche Energieversorgung aus 100% erneuerbaren Energien 2050. BfN-Skripten. 2018;501. doi: 10.19217/skr501 German. [DOI] [Google Scholar]
22.Wiehe J, Thiele J, Walter A, Hashemifarzad A, zumHingst J, von Haaren C. Nothing to regret: Reconciling renewable energies with human wellbeing and nature in the German Energy Transition. International Journal of Energy Research. 2020;45(1):745–758. doi: 10.1002/er.5870 [DOI] [Google Scholar]
23.Neuendorf F, Dataset: Exemplary landscape planning results using different qualities of LU/LC data to derive implications for practical planning. 2021.
24.Thiele J, Wiehe J, von Haaren C. Dataset: Areas with low and medium spatial vulnerability to a prototype wind turbine. 2021.
25.Lees J, Jaeger JAG, Gunn JAE, Noble BF. Analysis of uncertainty consideration in environmental assessment: an empirical study of Canadian EA practice. J Environ Plan Manag. 2016;59(11):2024–2044. doi: 10.1080/09640568.2015.1116980 [DOI] [Google Scholar]
26.Larsen SV, Kørnøv L, Driscoll P. Avoiding climate change uncertainties in Strategic Environmental Assessment. Environmental Impact Assessment Review. 2013;43:144–150. doi: 10.1016/j.eiar.2013.07.003 [DOI] [Google Scholar]
27.Batini C, Blaschke T, Lang S, Albrecht F, Abdulmutalib HM, Barsi A, et al. Data quality in remote sensing. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences—ISPRS Archives. 2017;42(2W7):447–53. doi: 10.5194/isprs-archives-XLII-2-W7-447-2017 [DOI] [Google Scholar]
28.Stritih A, Bebi P, Grêt-Regamey A. Quantifying uncertainties in earth-observation-based ecosystem service assessments. Environmental Modelling & Software. 2019;111:300–310. doi: 10.1016/j.envsoft.2018.09.005 [DOI] [Google Scholar]
29.Group on Earth observations. GEOSS Data Quality Guidelines. 2013. https://www.earthobservations.org/documents/dsp/GEOSS_Data_Quality_Guidelines.pdf
30.Hamel P, Bryant BP. Uncertainty assessment in ecosystem services analyses: Seven challenges and practical responses. Ecosystem Services, 2017;24:1–15. doi: 10.1016/j.ecoser.2016.12.008 [DOI] [Google Scholar]
31.vonHaaren C. & Othengrafen F. The Babel Fish Toolkit: Understanding and using behavioural mechanisms and interventions in landscape planning. disP—The Planning Review. 2019;55(2): 22–35. doi: 10.1080/02513625.2019.1630185 [DOI] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0260302.r001

Decision Letter 0

Gerald Forkuor

23 May 2021

PONE-D-21-14058

Data uncertainty in LU/LC data is unevenly distributed and has disproportionately high impacts on assessment results in environmental planning

PLOS ONE

Dear Dr. Neuendorf,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The study seem to have overly concentrated on spatial resolution as the main source of uncertainty in environmental modelling to the neglect of equally important sources/factors, including from the modelling process itself. Please, provide a justification for this or explore other sources in line with recommendations from the reviewers. In addition, kindly explore possibilities of improving the clarity of the structure and text, as well as strengthening the discussion based on recent works. Please, pay attention to comments on terminologies and ensure consistency. Further, ensure that you have provided and/or corrected the relevant information in the supporting data.

Please submit your revised manuscript by Jul 07 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gerald Forkuor

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: General comments:

In this manuscript, the authors assess the implications of data uncertainty on assessments in environmental planning based on two practical case studies. Uncertainties are important but often ignored in environmental assessments, so this manuscript addresses a relevant gap in the literature. However, I find that the clarity of both the structure and text could be improved, and the discussion could be strengthened with reference to existing work on uncertainties in general and more specific data uncertainties in spatial data. In addition, I think a stronger focus on the criteria for when the uncertainty is too high for specific applications would highlight the novelty of this paper. I would also suggest a language edit, as there are several typos and unclear formulations (see details in specific comments below).

1) In this manuscript, the authors mainly address data uncertainty. In particular, they focus on thematic accuracy and spatial resolution. While these are important sources of uncertainty, it would be helpful to put them in context with other types of uncertainties that affect assessments and decisions in environmental planning (such as model uncertainties, the inherent variability of environmental systems, or ambiguity among views of different stakeholders). How important are data uncertainties in relation to other sources of uncertainty? Addressing this question would add some depth to the discussion.

4) There has been a lot of work in the remote sensing community about various dimensions and metrics of data quality, and their effect on the usability of geospatial data. I think that referring to this literature would help strengthen the content of this paper, see e.g.:

• GEOSS Data Quality guidelines (https://www.earthobservations.org/documents/dsp/GEOSS_Data_Quality_Guidelines.pdf)

Specific comments

5) Lines 44-46: “Often…” this sentence is overly complex and difficult to understand, please consider rephrasing.

6) Lines 54-55: It would be helpful to define what authors mean by terms such as “data quality” and “preciseness”, as data quality includes both accuracy and precision, as well as some other attributes (e.g. appropriate temporal, spatial, and thematic resolution).

7) Line 100: An “error matrix” is commonly called a “confusion matrix” in the field of remote sensing classifications, which are a common source of LU/LC data, so it might be useful to mention both terms here.

8) Table 1: To make it easier to connect this table to the case studies, please consider adding a column or label to indicate which datasets are used for which case study.

9) Table 1 footnote: Typo “statesuse”. What do you mean by topicality in this context?

10) Lines 139-141: I don’t understand this sentence – why are biotope mapping useless because landscape planners are obliged to use appropriate data? Please clarify.

11) Lines 170-171: This sentence seems like it would fit more into the introduction, as it stresses the general relevance of choosing the appropriate spatial and thematic resolution.

12) Line 209: By accuracy, do you mean producer’s or users’ accuracy? Please specify.

13) Line 243: Phrases such as “dramatic effect” would be more appropriate in the discussion.

14) Lines 264-268: These results highlight the need to choose an appropriate spatial resolution, but do not really inform about uncertainty.

15) Lines 281-282: “When looking at…” this sentence is strangely formulated, please consider rephrasing.

16) Line 284: Typo “compatbible” should be “compatible”.

17) Lines 294-295: Here, you might want to mention that certain land-cover classifications (e.g. random forest-based classifications) can in fact provide spatially explicit information about uncertainties.

19) Lines 322-325: These questions should be presented when the case study in introduced in the methods section.

Reviewer #2: This paper addresses an important topic of data spatial scale and its influence on landscape modeling. The paper shows that using data of different spatial scales (and from different sources) can result in different results when used in modeling. The general conclusion that uncertainty in results is often overlooked in modeling studies is an important one.

Below I list some major concerns. I think a simple solution to addressing my concerns would be to change the theme of the paper away from providing a method of characterizing uncertainty (which the paper does not explore), and instead focus on a comparison of the dependence of modeling results on the spatial resolution and other properties of the input data. I think the latter approach will allow you to still achieve your main aim of highlighting the importance of considering how the data used affects the resulting model.

Major concerns

1. The paper makes broad and sweeping claims, including that it provides a simple and practical method for characterizing uncertainty, that are not supported by the analysis or the results. There is no investigation of the uncertainty of any of the data sets, or in the modeling itself. Also, as a proposed method for characterizing data suitability, by comparing the results with higher spatial resolution data, it does not seem to be useful to me, because why not use the higher spatial resolution data in the first place?

Minor issues

2. Avoid “error matrix” for the above reason and instead use a term like “confusion matrix”.

4. I suggest reserving “significant” for “statistically significant”. Instead, for general use, use terms like notable.

5. L28. I don’t see an analysis of the propagation of inaccuracy.

6. L30. I don’t see any “differences in uncertainty”, but instead differences in results.

8. Use a subscript for the 2 in CO2 throughout.

9. L92 – I can’t understand this sentence – perhaps rephrase?

10. L115 – “no significant missing data” – does this mean there is some missing data? If so, please explain.

11. Figure 1. Please add (a) scale bar (in m, or km, as may be appropriate) and (b) a legend.

13. L189, 190 and 195/196. The details of the grid cell size should be in Table 1. (I suggest using “grid cell size” or “grid cell dimension”, or even “pixel size”, not “grid width”)

15. L198 – “user’s accuracy”, not “user accuracy” (but seem my comment above about not using the term accuracy).

16. L199 – what is “combined accuracy” – do you just mean just “the range in the user’s accuracy”? – but that doesn’t make sense since the lowest value seems to be 0%, not 82%? Do you perhaps mean the “overall accuracy” (which is defined as the sum of the correctly classified areas in each class, divided by the total area of the map)?

17. Figure 2. The y-axis should be “Agreement”, not “Spatial accuracy” (the latter would be a measure of the spatial registration of the two data sets, which I don’t believe is what you mean). Is it possible to use color – I struggled to tell the different grays apart.

18. L219 – “correlation” – do you perhaps mean “confusion”?

19. Figure 3. Are the legends correct? For example, for Figure 3 (a), my interpretation from the legend (“misclassified GRASSLAND”) is that these are areas that are in reality GRASSLAND but have been misclassified. But I think you actually might mean “Proportion of pixels misclassified as GRASSLAND”?

20. L260 “rundue” should be two words.

21. L274 – replace “geometric” with “spatial resolution”.

22. L294/295 – I don’t understand the sentence (or the subsequent discussion) and my best interpretation doesn’t seem to be correct. The confusion matrix is a spatial overlay and produces a summary of agreement. It is true however that it does not consider the contiguity of the distribution of pixels in any one class, but I don’t see why that should matter.

23. L300 – yes, it has uncertainty. Since “to a certain degree” are not defined, they can be removed.

25. L321 – I don’t understand why you are saying “the uncertainty is too high” when the “effect is small”?

26. L332 I don’t understand why small land holders would be affected more than large ones. Wouldn’t small ones be too small to show up in the analysis, and therefore be ignored? Perhaps you could elaborate.

27. L345. I suggest not using a term like “better” (which is not defined), but instead simply state the properties that you mean by “better” – e.g. data with a spatial resolution that matches the modeling requirements, or whatever you mean.

28. L359. I don’t think you have analyzed the uncertainty in the products you have generated, as there are many other sources of uncertainty, including in the model itself, not to mention uncertainty in the “ground truth” data. I think instead you have shown that different data sets can produce different results when applied to a model, and therefore it is important to consider what spatial scale (as well as other data characteristics, such as classes mapped, etc) the model assumes for input data.

29. Figure 4. Could you use color here, too, to make it easier to interpret?

30. Supporting data. What are the units of measurement in the supporting confusion matrix data tables? Ha, pixels, etc? (if pixels, what size?). The tables are very hard to read because of the decimals (it is hard to get a visual idea of where the confusions lie) and the size of the numbers. Units like square km would make the table a lot more visually understandable, and perhaps make your overall points easier to make.

32. Supporting data – what does “balance" mean? What is "sum/balance"?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Ana Stritih

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Nov 24;16(11):e0260302. doi: 10.1371/journal.pone.0260302.r002

Author response to Decision Letter 0

4 Aug 2021

Editor comments

Response: Dear Editor, Dear Dr. Forkuor, thank you for your in-depth review and the critical remark. We understand that there, we obviously did not manage to express ourselves clear enough and have now revised the manuscript accordingly.

We aknowledge the fact that there are multiple sources of uncertainty. For that we added a whole section to the introduction highlighting different sources of uncertainties in planning (Lines 37-56)

To highlight the importance of data uncertainty we write:

“Uncertainty of topographical data presents one of the most important uncertainties, as such input data inaccuracies propagates through the entire environmental assessment and planning process. Relevant concerns of environmental planning in terms of data uncertainties include, among others, mistakes in categorising (for example, if particular land parcels are incorrectly attributed to land use types), inconsistencies between the different datasets (mainly the location and spatial delineation of features) and impacts of data resolution (for instance if the resolution it too coarse to divide site-specific management recommendations to farmers). Some of these uncertainties may be obvious, but not quantified, others may be hidden in the data set. Both types of uncertainties are usually not explicitly communicated with the data provision. Nevertheless, the planner has to decide which data detail will suffice for which specific planning task – risking misjudgement.” (Lines 57-66)

In addition, we write in the discussion section that we are fully aware of the fact that other, additional sources of uncertainty also play a role, but that it was beyond the scope of this paper to explore them in greater detail. We write:

“We are aware that data uncertainty is only one component of the uncertainty spectrum and we posit that future research is also needed to incorporate considerations of all types of uncertainties in practical planning. In conjunction with our findings we propose that opportunities for future research are to develop robust and pragmatic methods for identifying the uncertainty levels in environmental data and assessment results, and to devise approaches for enhancing data certainty wherever possible.” (Lines 366-371)

In addition, kindly explore possibilities of improving the clarity of the structure and text, as well as strengthening the discussion based on recent works. Response: Thanks. We have have done our best to improve the clarity of the manuscript structure by revisiting the writing of the entire manuscript and adapting it to increase the flow of argumentation and enhance the style of writing. We have done so by critically re-reading and carefully editing the entire manuscript ourselves, and we have asked a language consultant to review and improve the manuscript for us.

In addition, we have strengthened the discussion by better weaving our arguments into the latest state of knowledge as reflected in recent publications.

Please, pay attention to comments on terminologies and ensure consistency. Further, ensure that you have provided and/or corrected the relevant information in the supporting data. Thanks for flagging this. We now consistently refer to data uncertainty in the manuscript when referring to the kind of uncertaintywe have considered. The provided supporting data is no longer relevant for the manuscript and has been removed

When submitting your revision, we need you to address these additional requirements.

1) Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Response: Thanks – yes, we have done as suggested.

2) We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide. Response: Thank you for noting this. We are in the process of making the data publicly available on our own repository. The access will be provided via DOI numbers.

The data for the third case study is already available, the data for case study one and two will be available in short time.

Reviewer #1

GENERAL COMMENTS:

In this manuscript, the authors assess the implications of data uncertainty on assessments in environmental planning based on two practical case studies. Uncertainties are important but often ignored in environmental assessments, so this manuscript addresses a relevant gap in the literature. Response: Thank you very much for the insightful review and the many comments for further improvement. Please excuse the short handling of some of your specific comments as these have been obsolete after the complete revision of the manuscript. Nonetheless we would like to thank you for each comment as these greatly helped us in improving the whole manuscript.

However, I find that the clarity of both the structure and text could be improved, and the discussion could be strengthened with reference to existing work on uncertainties in general and more specific data uncertainties in spatial data. Thanks for noting. We agree and have substantially and comprehensively revised the entire manuscript. In fact, in order to comply with your suggestions, we needed to revise everything as you will see in the track-changes version. We hope this meets your expectations.

In addition, I think a stronger focus on the criteria for when the uncertainty is too high for specific applications would highlight the novelty of this paper. Response: Thank you for pointing this out. We had a long discussion within the team of authors of how to best address this. We have considered moving the criteira to the introduction or the methods section and tried our best to revise the manuscript accordingly. In the end, however, we realized that the best place to put them is in the discussion where we critically reflect on the case studies. We hope you agree and appreciate the changes. In any way, we are grateful for your advice and suggestions which has spurred our further reflection, and we are confident that this change has substantially enhanced the quality of our manuscript.

I would also suggest a language edit, as there are several typos and unclear formulations (see details in specific comments below). Response: Thank you for the suggestion. A comprehensive language edit has been performed, which we believe has resulted in a major improvement of the quality of our manuscript.

“A recent review revealed that the main planning relevant categories of uncertainties in environmental planning can be associated with four categories [3]: Data uncertainties, which are most commonly associated with geometric uncertainties that arise mainly from resolution, thematic uncertainties that occur when values are attributed to objects, and temporal mismatches between data sources [4]. Model uncertainties, the second category, relate to challenges in choosing appropriate indicators and variables for environmental assessments [5]. Models applying only a small number of explanatory variables are generally easier to use for planners, but are often inferior to more complex and dynamic models in understanding environmental system change. The most obvious trade-off in model development for planning purposes is between precision and usefulness, for example when a complex model would become more demanding in data requirements [6]. In addition, a very high level of detail can be impractical and a source of uncertainty in planning applications when the interpretation of complex model results puts high demands on user [7]. The third category, projection uncertainties, is linked to projecting future states of the environment � atypical challenge for planning as a strong forward looking action [8]. Plans often suggest that problems are of a static nature, but in reality they are highly dynamic [9] and can often not be predicted with the understanding of future behaviour of various components [10]. Finally, valuation uncertainties are introduced when norms and laws need to be interpreted to form scales for valuation [3]. While data, model and valuation uncertainties are prominent at multiple stages of the planning cycle, projection uncertainties come into play in particular when scenarios for future developments are created and potential scenario impacts are being explored.” (Lines 37-56

We added a new part that highlights the importance of data uncertainty:

“Uncertainty of topographical data presents one of the most important uncertainties, as such input data inaccuracies propagate through the entire environmental assessment and planning process. Relevant concerns of environmental planning in terms of data uncertainties include, among others, mistakes in categorising (for example, if particular land parcels are incorrectly attributed to land use types), inconsistencies between the different datasets (mainly the location and spatial delineation of features) and impacts of data resolution (for instance if the resolution it too coarse to divide site-specific management recommendations to farmers). Some of these uncertainties may be obvious, but not quantified, others may be hidden in the data set. Both types of uncertainties are usually not explicitly communicated with the data provision. Nevertheless, the planner has to decide which data detail will suffice for which specific planning task – risking misjudgement.” (Lines57-66)

We pick up on that briefly in the discussion but focused on our findings and possible implications:

2) A clearer structure would help to make the paper easier to follow. In particular, I think it would be important to clearly define the criteria of when uncertainty is “too high” at an early stage in the manuscript, as this is a particularly novel and relevant aspect. Then, the results could be presented in relation to these criteria. Furthermore, I would suggest organizing the Methods section by case studies, and clearly describing the specific questions (as currently described in the discussion), data, and type of uncertainty addressed in each case study. It might also be helpful to have a short reference for each case study, e.g. Case study A and B or “carbon mapping” and “renewable energies”, which would be used consistently throughout the paper. Response: Dear reviewer, thank you for pointing this out. We fully agree that the structure of our manuscript could be improved. We have invested substantial time and efforts over the last weeks to address this issue, including several rounds of discussion and reflection within the author teams, and the elaboration of several versions and improvements of the manuscript.

In this process, we have come to the conclusion that the criteria for deciding about the level at which uncertainties are too high are not so easy to draw generically, but that this rather requires a case-specific contemplation and consideration. As mentioned, explained and justified above, we have decided to move the criteria into the discussion.

Our new manuscript outline takes up your kind advance to structure the methods section by case studies. As suggested by you, we now clearly descibre each specific question, we state the kind of data we use and our estimation of the inherent data uncertainty. We also describe that we define two scenarios for each case study, with the use of different levels of data uncertainties, and we clarify how we derive planning recommendations for these scenarios.

We also follow your advice about the naming of the case studies: They are now referred to as “Regional soil-based climate change mitigation”, “State-wide habitat conservation siting”, and “National renewable energy planning”

3) It seems to me that especially the results of the second case study (choice of suitable areas for RE) mainly highlight the need to select the appropriate spatial and thematic resolution of data, rather than the uncertainty of the data itself. The importance of choosing the appropriate resolution is well known in the spatial modelling/assessment community, but its implications for practical decisions in environmental planning are still interesting. I think it would therefore be useful to explicitly state the types of data uncertainty that are addressed in each case study (e.g. misclassification, loss of information due to spatial or thematic aggregation and insufficient resolution, etc.). Response: We agree with you that with the second (now third) case study we mainly considered the resolution of the spatial input data in a nationwide assessment of human- and nature-compatible usablearea potentials for renewable energies (RE) and not the uncertainty (e.g. misclassification) of the spatial input data itself.With this study, we were able to emphasize that high-resolution input data should be used, even for studies on a nationwide scale, in order to increase the quality of the resulting statementsor recommendations which is more in line with the updated focus of the paper. Future studies in the field of potential area assessmentsof RE can now refer to this study.

In order to clarify this, we have adjusted the manuscript:

“and impacts of data resolution” (Addition in the introduction) (Line 62)

“Case study number three, ‘National renewable energy planning’, concerns an environmental planning effort aiming to identify areas for the siting of renewable energies, which should also guide decision-making at the regional level.” (Lines 110-112)

“Two model runs were performed with input data of different resolutions to quantify the influence of data resolution to identify opportunity spaces for generating renewable energy, taking into account criteria for human and nature conservation compatibility.” (Lines 189-191)

• GEOSS Data Quality guidelines (https://www.earthobservations.org/documents/dsp/GEOSS_Data_Quality_Guidelines.pdf)

• Batini C, Blaschke T, Lang S, Albrecht F, Abdulmutalib HM, Barsi, et al. Data quality in remote sensing. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives. 2017;42(2W7):447–53. Response: Thank you for this important comment. We agree with you that these publications and the work in the remote sensing community would have strengthend the core aspect of the first version of the paper. Since the focus is different now, we opted to include these references in the discussion as a starting point for the planner community to build upon. The lines read as follows:

‘Good starting points are developments in the remote sensing community on data quality assessment and communication (25, 26). Established guidelines on data quality, as used for earth observation data [27], have the potential to be used in data generation in general.’ (Lines 368-371)

SPECIFIC COMMENTS:

5) Lines 44-46: “Often…” this sentence is overly complex and difficult to understand, please consider rephrasing. Response: This part of the manuscript has been completely revised.

8) Table 1: To make it easier to connect this table to the case studies, please consider adding a column or label to indicate which datasets are used for which case study. Response: Thank you for this comment, we added a column with information to the respective case study.

9) Table 1 footnote: Typo “statesuse”. What do you mean by topicality in this context? Response: Topicality means actuality in this context. We aknowledge that it might be a misleading term, so we changed it to actuality instead.

10) Lines 139-141: I don’t understand this sentence – why are biotope mapping useless because landscape planners are obliged to use appropriate data? Please clarify. Response: This part of the manuscript has been completely revised.

11) Lines 170-171: This sentence seems like it would fit more into the introduction, as it stresses the general relevance of choosing the appropriate spatial and thematic resolution. Response: This part of the manuscript has been completely revised.

12) Line 209: By accuracy, do you mean producer’s or users’ accuracy? Please specify. Response: This part of the manuscript has been completely revised.

13) Line 243: Phrases such as “dramatic effect” would be more appropriate in the discussion. Response: Thank you for this remark, we carefully checked the new manuscript for such phrases and made sure they were located correctly.

14) Lines 264-268: These results highlight the need to choose an appropriate spatial resolution, but do not really inform about uncertainty. Response: Absolutely, therefore we have included some adjustments into the manuscript to clarify this. We have listed the adjustment in the response to your third major concern ("It seems to me that..."); thank you very much for your comment.

15) Lines 281-282: “When looking at…” this sentence is strangely formulated, please consider rephrasing. Response: This part of the manuscript has been completely revised.

16) Line 284: Typo “compatbible” should be “compatible”. Response: Thank you, we thoroughly checked the manuscript for spelling errors.

18) Lines 315-321: These criteria are very relevant. For a clearer structure, I would suggest that these criteria already be mentioned earlier in the manuscript (e.g. in the introduction or methods section), so that the reader can keep them in mind when presented with the results. Response: We discussed about reframing the manuscript and putting the criterias at the front but opted against it. The criterias have been reformulated into problem constellations and discussed thoroughly in the discussion and conclusion section.

19) Lines 322-325: These questions should be presented when the case study in introduced in the methods section. Response: This part of the manuscript has been completely revised.

Reviewer #2

GENERAL COMMENTS:

This paper addresses an important topic of data spatial scale and its influence on landscape modeling. The paper shows that using data of different spatial scales (and from different sources) can result in different results when used in modeling. The general conclusion that uncertainty in results is often overlooked in modeling studies is an important one. Response: Dear Reviewer, thank you very much for these important remarks. The revised manuscript now picks up on this and goes much more in depth.

Please excuse the short handling of the majority of your minor remarks, but most of them are reffering to text bodies and subjects that no longer are a focal point in the manuscript.

We now exlusively focus on a case study-based exploration of the impacts of data uncertainty on planning recommendations. We descreibt the methods applied in each of the three case studies in detail in the methods section, and we clearly present the results for each case study in the results section. We thereby distinguish two scenarios in each case study – one scenario applying data of very high uncertaintiy levels, and one scenario applying data with lower uncertainty. The comparision of recommendations deriverd based on results in both scenarios then allow us in the discussion section to critically reflect upon the implications of data uncertainties on environemtnal planning recommendations

MAJOR CONCERNS

Regarding your question about why not using better data in the first place, we would like to argue that the main idea of the paper is now to illustrate the implications of applying data of difererent levels of uncertaintiy on planning recommendations. We follow this aim because environmental planners always need to make planning recommendations amid uncertainties and have devised some simple heuristics, i.e. planning principles, for intuitively addressing uncertainties. However, we find that this can cause serious harm, as planners often cannot appropriately estimate the degree of data uncertainty truly involved in their assessments. Thus, we wrote this paper with the intention to shed light on the implications of applying data with different uncertainty levels in decision making. In fact, planners often lack the appropriate certain data, so our cases are just for illustration of what differences using this sub-optimal data can make.

2. There are multiple sources of uncertainty in comparing data sets – spatial resolution is simply one of the sources of differences. Even something as subtle as the difference in how the classes are defined can be important. Other issues include spatial location accuracy, thematic classification accuracy, date of acquisition, etc. The mismatch between the classes required by the model and the classes in the geospatial data may be even greater than the issue of the spatial resolution. Response: We aknowledge the fact that there are multiple layers to data uncertainty and uncertainty in general, thank you for this important remark. We added some parts to the introduction and the discussion to refer to other sources of uncertainty in planning. To cover other data related uncertainties such as temporal factors we chose the data that best matched the other data we used, for example on crops, which resembles the work actual planners should have done.

3. After making the correct statement, that “There is no spatial environmental planning without uncertainties”, the paper then goes on to treat the biotype mapping as “ground truth”, and to call differences between the biotype data and other data sets as uncertainty in the other data, not in the biotype data. I think it is a fundamental flaw to suggest the biotype data are the truth and by implication so are the model results using the biotype data. Response: Thank you for this very important comment. We aknowledge that Ground truth is a misleading term in this case. We opted to substitute the term “ground truth” with “reference” and changed this wording for the manuscript. Nevertheless we still believe that the biotope mapping is the best data available in our case and that there is no inherent flaw in using it as reference data. This data simply has the highest thematic and spatial accuracy by a great margin. We agree that there are also uncertainties inherent to this data, and added this to the data description section. This also correlates with your next important comment.

4. It would be useful to provide a great deal more information on how each data set was generated, and to explain the methods of analysis in more detail. In particular, information on the biotype data seems important – what scale is it produced at and more detail on how it is mapped? How were the DEM products generated (e.g. photogrammetry, lidar, etc), and what is their source? In Table 1, you could separate out the minimum mapping unit area and the pixel size for raster data sets. What is the spatial uncertainty in the location of the data? For the data overlay for generating the confusion matrix, it is useful to state the resampling method used (presumably nearest neighbor). What is the cross-walk for the conversion of the data legends between the different data sets? The concept of the emission level classes (figure 4) and their definitions should be explained in the methods. Response: Thank you for this important comment. We reworked the section about the different data sets and included more information on how the data was collected and what the minimum mapping unit where. It should now give a better understanding of the different data sets. We also added information to the table that gives an overview of the different data sets and hope that this is now more comprehensible.

We also added a sentence regarding the emission level classes and added respective information to the new results table for case study one. We aknowledge the fact that the model chosen is important in general but for our case it was only important to use a method that takes all assessed data sets as input. We expect that all methods that address emission and retention would be affected by changes in extremely important classes, like moors and heathlands.

“For each dataset, we applied the same method for estimating potential CO2-emission and retention potentials [19]. The applied method combines data about soil types and land-use/land-cover which are classified as to climate relevance, and should be available for a coherent time span of 5 years.” (Lines 168-171; Table 2)

5. There appear to be some errors in the supporting document’s calculation of the confusion matrices user’s accuracies. For example, for Saxony Anhalt, I get different values for both Mining and landfill sites (I get 0.0) as well as Areas of sparse vegetation (I get 0.06) user’s accuracies. I suggest checking all the numbers. Response: Thank you for pointing this out, there has indeed been an error in the calculation of several classes.The supporting information in question though is no longer relevant for the paper and has been removed.

6. I suggest presenting both user’s and producer’s accuracies in Figure 2 and the text, and not just user’s accuracies, since producer’s accuracies are equally important. (Though see comments below about the term “accuracy”.) Response: We agree that it would be beneficial to present both accuracies, but due to a new focus and structure the confusion matrix is no longer part of the results.

MINOR ISSUES:

1. As explained above, I don’t think it is conceptually correct to refer to the differences in the maps as “accuracy” or even “uncertainty”. A better term would be agreement or consistency. For example, you could use the terms “user’s consistency” and “producer’s consistency” instead of “user’s accuracy” and “producer’s accuracy” – though you’d need to define the terms since they are not commonly used. Using agreement or consistency would not preclude making an argument that one data set more closely matches the assumed spatial scale (or other properties) assumed by the model. For example, if the model has wetlands as a key class, and wetlands are generally smaller than the 25 ha MMU of the Corine landcover, then the use of the finer spatial scale data will allow the model to incorporate these smaller wetlands. Response: Thank you for this insightful comment, we fully agree that definitions of terms are of great importance. Our new manuscript version now includes the addition of various kinds of uncertainty which are common in environmental planning and and explanation that we focus exclusively on data uncertainty and how it depicts in planning results. Since the whole focus has been shifted, the confusion matrix is no longer a focal point of the analysis, albeit the basic principle of overlaying data still remains. In these cases we use the term consistency for the manuscript. We also introduced that term in the introduction. The section reads as follows:

“Relevant concerns of environmental planning in terms of data uncertainties include, among others, mistakes in categorising (for example, if particular land parcels are incorrectly attributed to land use types), inconsistencies between the different datasets (mainly the location and spatial delineation of features) and impacts of data resolution (for instance if the resolution it too coarse to divide site-specific management recommendations to farmers).” (Lines 58-63)

2. Avoid “error matrix” for the above reason and instead use a term like “confusion matrix”. Response: Thank you for this comment, as already mentioned, the confusion matrix is no longer part of the manuscript.

3. Avoid the term “ground truth” – this is generally thought not to be a good term in remote sensing, and in a paper on uncertainty in data layers it seems strange to imply that any data could be the “truth”. Response: Thank you for this remark, we changed the wording to “reference” throughout the manuscript to avoid any confusion.

4. I suggest reserving “significant” for “statistically significant”. Instead, for general use, use terms like notable. Response: Thank you for this comment, we changed the wording accordingly.

5. L28. I don’t see an analysis of the propagation of inaccuracy. Response: Thank you for this comment. This statement was not fitting in the abstract and has been completely removed during the rephrasing of that section. We now phrased the aim of this paper as follows:

“This paper aims to explore how different data uncertainties, usually unknown to the planner, may influence environmental planning recommendations.” (Lines 18-20)

6. L30. I don’t see any “differences in uncertainty”, but instead differences in results. Response: This part of the manuscript has been completely revised.

7. L31. I think you are promising too much in claiming to have demonstrated uneven distribution in uncertainties. However, you have shown that the magnitude of the differences in results between using different data sets can vary, depending on the application and even location. Response: Thank you. We agree and have completely revised the respective section. The important sentence regarding this comment now reads as follows:

“We apply a case study-based approach, in which we provide three illustrative examples of how data with different kinds and levels of uncertainty affect environmental assessments and, by that, the decision-support provided by environmental planning.” (Lines 20-22)

8. Use a subscript for the 2 in CO2 throughout. Response: This has been changed throughout the manuscript.

9. L92 – I can’t understand this sentence – perhaps rephrase? Response: Thank you for pointing this out; this is now case study number three. The phrasing is now as follows: has been reworded:

10. L115 – “no significant missing data” – does this mean there is some missing data? If so, please explain. Response: This part of the manuscript has been completely revised. The data this referred to is no longer a part of the analysis that the manuscript is based on.

11. Figure 1. Please add (a) scale bar (in m, or km, as may be appropriate) and (b) a legend. This figure has been removed in the process of changing the focus of the manuscript.

12. I think it would be useful to point out to the reader that the issue is not just the pixel size of raster data, but the minimum mapping unit area – it is convenient that the CORINE data specifies this so clearly. Response: Thank you for this comment. Although the focus is still on the inherent uncertainties and implications of using different data it is no longer the analysis of the uncertaitnty itself that is the focal point of the manuscript.

13. L189, 190 and 195/196. The details of the grid cell size should be in Table 1. (I suggest using “grid cell size” or “grid cell dimension”, or even “pixel size”, not “grid width”) Response: We agree with you and have adjusted the table 1 and the text accordingly.For this we have used "grid cell size", thank you.

14. L195 – In my experience, even 50 m is still very coarse for this type of modeling. I think this reinforces the idea that often in modeling we have to use data that don’t entirely fit our purpose. Response: Digital elevation models with a grid cell size of 10m and 25m can also be requested from the Federal Agency for Cartography and Geodesy. We have decided in this investigation for the DEM50 due to computing capacities and agree with you completely that in modeling also data are integrated which are faster to process or whose provision is realized faster. We added a part to the discussion highlighting the need to carefully select appropriate data sets for environmental anaylsis. This part reads as follows:

“Taken together, the results re-emphasize the importance of carefully selecting datasets for landscape and environmental analysis for planning, as the first impression of sufficient data certainty might be misleading, and error-propagation through data application in models can result in substantially misleading results.” (Lines 322-325)

15. L198 – “user’s accuracy”, not “user accuracy” (but seem my comment above about not using the term accuracy). Response: This part of the manuscript has been completely revised.

18. L219 – “correlation” – do you perhaps mean “confusion”? Response: This part of the manuscript has been completely revised.

20. L260 “rundue” should be two words. Response: This has been fixed, thank you.

21. L274 – replace “geometric” with “spatial resolution”. Response: This part of the manuscript has been completely revised.

23. L300 – yes, it has uncertainty. Since “to a certain degree” are not defined, they can be removed. Response: This part of the manuscript has been completely revised.

24. L302 - The term “geodata uncertainty marker” is not defined and it is not clear any case that one can generalize for other applications – your work shows that even for different areas using the same dataset, the agreement varies. I suggest dropping this. Response: Thank you for this comment. In the process of redesigning the manuscript, this part has been changed and is no longer a focal point of the paper.

25. L321 – I don’t understand why you are saying “the uncertainty is too high” when the “effect is small”? Response: This part of the manuscript has been completely revised.

29. Figure 4. Could you use color here, too, to make it easier to interpret? Response: This figure is no longer included in the manuscript

31. Supporting data. Very important – all the user’s and producer’s accuracies don’t seem to be %, though they are labeled as such. I think they are just straight proportions (on a 0-1 scale). Response: After the revision of the manuscript this supporting data has been removed as it is no longer relevant.

32. Supporting data – what does “balance" mean? What is "sum/balance"? Response: After the revision of the manuscript this supporting data has been removed as it is no longer relevant.

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(69.2KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0260302.r003

Decision Letter 1

Gerald Forkuor

31 Aug 2021

PONE-D-21-14058R1

Uncertainties in land use data may have substantial effects on environmental planning recommendations: a plea for careful consideration

PLOS ONE

Dear Dr. Neuendorf,

The revisions required are minor. Please, make the necessary changes or provide the required clarifications. If a particular recommendation is not feasible, please provide cogent explanations as to why.

Please submit your revised manuscript by Oct 15 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

Gerald Forkuor

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

Reviewer #2: No

**********

6. Review Comments to the Author

Reviewer #1: The revised version of the manuscript is greatly improved in terms of clarity and provides an important message for the planning community.

Reviewer #2: The authors have made a good-faith effort to take into account my concerns. My suggestions below are mainly minor and mainly merely a matter of clarifying wording or terms that I find confusing, as unfortunately the revised paper is often a bit confusing.

1. In a number of locations “topographical” is used where it appears “land cover” would be a better term (topographical means the form of the land – e.g. hill, valley, etc). See L57, L137

2. L81-85 – The meaning of this is not clear – can you rephrase?

3. L107 – Adressees – not sure what this means – do you mean “Potential users”?

4. L135 – Actuality – do you mean “currency” (as in the data is up to date)? Or something else?

5. L145 – Pictures – replace with “images”

6. Methods – do you explain how you handle your spatial overlay operations evaluating “consistency” where the pixel size differ? Specifically, how do you resize the pixels?

7. L191 onwards – in the results you refer to “After Walter” and “After Thiele” – I suggest making it clear by including those names in the text here and also explaining explicitly how your methods relate to those sources – e.g. are you following their procedures or using their output or what?

8. L215, Figure 1 and Table 1. The word “consistent” alone does not explicitly explain the meaning that I think you are applying to this term. If I understand it correctly, you do two comparisons: an aspatial comparison of total areas mapped as a certain class, and a spatial overlay operation where you look for agreement in class labels of pixels. You appear to use “consistent” to mean the latter only. I think to make this usage clearer it would help a lot to say “consistent labels” or “consistent labeling” or “consistency in labeling of pixels” or “spatial agreement”. I also think you should state in the methods that you have these two ways of evaluating agreement.

9. L221 and also figure 2. In some cases you use the term “area” where I think you mean “class” or “label”. For example, in L221 – I think the point you are making is that errors in the delineation of certain classes can have a large effect on the model outcome.

10. L227 – “this case” – do you mean “this class”? I’m also confused about “deficits” and “untapped” – it would be clearer to simply talk about “disagreements” and “differences” or “not mapped”.

11. L231 – perhaps rephrase along the lines of “In contrast, the use of ATKIS data results in an underestimation of about 5.8 million tons of potential C02 emissions compared to the reference data.”

12. In the next sentence replace with something like “This is because a total of 3,025 fewer ha were mapped as Level 1…”

13. Table 2. This is a tricky table to read and absorb. I think it would help a lot to eliminate the use of decimals for the t and ha values. This level of precision (1/100 of 1 ha!) is of little purpose when the numbers differ as much as they do (thousands to millions of ha). Important: the “Potential CO2 emission or retention” cell label for the reference Biotope needs units (presumably t).

14. L248 – add a caveat “if we assume the Biotope mapping is correct.”

15. L251 – It wouldn’t be potentially disregarded, but would be disregarded – i.e. rephrase something like “This would lead to a total of 255 ha of areas important to conservation not recognized in the planning recommendations due to the lower detail of the data used.”

16. Table 3 – indicate in the table that Biotope is the reference mapping

17. Table 3 – need to clarify in the headings that it is “class label” that matches, or that the “class is not mapped” – rather than the area that is missed or consistent (which would mean something else).

18. L302 – thematic attributes - do you mean fewer classes?

19. L307- thematic information is lacking – do you mean that if the thematic classes do not match the classes of interest?

20. L312 – 2980 km2 – the significance of the area is hard to understand – I recommend also specifying this as a percentage.

21. L323- easy data – do you mean “easily obtained data??

22. L344 – albeit this would not be necessary – do you mean “that are not actually helpful based on the real land use of the area.”

23. L347 – I don’t understand what this fourth constellation is – do you mean something like “when the potential effort or response required to address the issues raised by the model outcome is large but the data uncertainty is also large?

24. L375-6 – I can’t understand this sentence.

25. L394 – I don’t understand the sentence (“in how far”?) and what does “behavioral mechanisms” mean? Can you rephrase this?

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Ana Stritih

Reviewer #2: No

PLoS One. 2021 Nov 24;16(11):e0260302. doi: 10.1371/journal.pone.0260302.r004

Author response to Decision Letter 1

27 Oct 2021

E: The revisions required are minor. Please, make the necessary changes or provide the required clarifications. If a particular recommendation is not feasible, please provide cogent explanations as to why.

- Thank you very much for the advice. We have done as suggested

R1: The revised version of the manuscript is greatly improved in terms of clarity and provides an important message for the planning community.

- Thank you for the kind words.

R2: The authors have made a good-faith effort to take into account my concerns. My suggestions below are mainly minor and mainly merely a matter of clarifying wording or terms that I find confusing, as unfortunately the revised paper is often a bit confusing.

- Thank you for the kind words and the detailed re-read of the manuscript.

R2: 1. In a number of locations “topographical” is used where it appears “land cover” would be a better term (topographical means the form of the land – e.g. hill, valley, etc). See L57, L137

- Thank you for this important remark. We changed the wording to “spatial and statistical data” in the first case and to “land use / land cover” data in the second case as we think these are the best fitting options.

R2: 2. L81-85 – The meaning of this is not clear – can you rephrase?

- Thank you for flagging this. We have reformulated this section as follows:

“Finally, environmental planners often presume that an intuitively felt uncertainty may be present in their basic data, but that this will not relevantly affect the response measures. The logic behind this is that planners often strive for multifunctional measures which intentionally target a larger area than monofunctional measures that usually address only smaller areas, for example to safeguard a certain species. Along this line, planners generally presume that the inaccuracies of land use data will not affect the validity of recommended measures which address multiple purposes in an aggregated spatial unit. Furthermore, in case of precautionary measures, the uncertainties of proposals of future land uses seem to be much more relevant than uncertainties stemming from inaccuracies of cartographic land use data.”

R2: 3. L107 – Adressees – not sure what this means – do you mean “Potential users”?

- We have changed this to ‚potential users‘

R2: 4. L135 – Actuality – do you mean “currency” (as in the data is up to date)? Or something else?

- Thanks. We changed the term to ‚currency‘ as kindly suggested.

R2: 5. L145 – Pictures – replace with “images”

- Changed as suggested

R2: 6. Methods – do you explain how you handle your spatial overlay operations evaluating “consistency” where the pixel size differ? Specifically, how do you resize the pixels?

- We added a section to explain how the raster data was resized. This reads as follows:

„In order to compare the two Digital Elevation Models (input data), the DEM200 was converted to a grid with a grid cell sizes of 50 m. Since the potential areas of the two model runs were also provided as grids with a grid cell size of 50 m, they could be compared without further transformations.”

R2: 7. L191 onwards – in the results you refer to “After Walter” and “After Thiele” – I suggest making it clear by including those names in the text here and also explaining explicitly how your methods relate to those sources – e.g. are you following their procedures or using their output or what?

- Thank you for flaggining this. We have rephrased the whole section, beginning with: “The case study ‘National renewable energy planning’ explored impacts of using datasets of different uncertainty levels on the potentially suitable sites for renewable energies, taking into account both human and nature-conservation needs [20]. We used the data and results of two model runs, the first one was conducted by Walter et al. 2018 and the second one by Thiele et al. 2021, for an uncertainty analysis in order to quantify the influence of data resolution on identifying opportunity spaces for generating renewable energy...”

R2: 8. L215, Figure 1 and Table 1. The word “consistent” alone does not explicitly explain the meaning that I think you are applying to this term. If I understand it correctly, you do two comparisons: an aspatial comparison of total areas mapped as a certain class, and a spatial overlay operation where you look for agreement in class labels of pixels. You appear to use “consistent” to mean the latter only. I think to make this usage clearer it would help a lot to say “consistent labels” or “consistent labeling” or “consistency in labeling of pixels” or “spatial agreement”. I also think you should state in the methods that you have these two ways of evaluating agreement.

- Thank you for this comment. We added an explanatory sentence to the method section to specify what is meant with consistent in this case.It reads as follows: “We compare resulting maps also in terms of spatial consistency, meaning areas where there is a spatial match of the class labels between the different outputs.”

R2: 9. L221 and also figure 2. In some cases you use the term “area” where I think you mean “class” or “label”. For example, in L221 – I think the point you are making is that errors in the delineation of certain classes can have a large effect on the model outcome.

- We understand the confusion and have rewritten the sentence as follows: “… highlight the importance of a good representation of sites of importance for climate protection. “

R2: 10. L227 – “this case” – do you mean “this class”? I’m also confused about “deficits” and “untapped” – it would be clearer to simply talk about “disagreements” and “differences” or “not mapped”.

- Sorry for the confusion, we have revised the sentence for clarity but did not change the exact wording as it would alter the intended message.It now reads: Surprisingly, using coarser CLC 25 data leads to better results, with a smaller area deficit of around 1,472 hectares that accounts for around 2.8 million t CO2 retention potential untapped.

R2: 11. L231 – perhaps rephrase along the lines of “In contrast, the use of ATKIS data results in an underestimation of about 5.8 million tons of potential C02 emissions compared to the reference data.”

- Thank you, we have rephrased the section

R2: 12. In the next sentence replace with something like “This is because a total of 3,025 fewer ha were mapped as Level 1…”

- Thank you, we have rephrased as suggested.

R2: 13. Table 2. This is a tricky table to read and absorb. I think it would help a lot to eliminate the use of decimals for the t and ha values. This level of precision (1/100 of 1 ha!) is of little purpose when the numbers differ as much as they do (thousands to millions of ha). Important: the “Potential CO2 emission or retention” cell label for the reference Biotope needs units (presumably t).

- Yes, you are absolutely right. We have implemented both changes as suggested.

R2: 14. L248 – add a caveat “if we assume the Biotope mapping is correct.”

- We have added a more detailed explanation in the data description section where we state the following: “…as it is the most detailed type of land-use data used for environmental planning in Germany and we assume that it depicts the real circumstances best.”

R2: 15. L251 – It wouldn’t be potentially disregarded, but would be disregarded – i.e. rephrase something like “This would lead to a total of 255 ha of areas important to conservation not recognized in the planning recommendations due to the lower detail of the data used.”

- Yes, we have done as suggested. Thanks so much for the very useful recommendations.

R2: 16. Table 3 – indicate in the table that Biotope is the reference mapping

- Thanks, we have done accordingly.

R2: 17. Table 3 – need to clarify in the headings that it is “class label” that matches, or that the “class is not mapped” – rather than the area that is missed or consistent (which would mean something else).

- Thank you. We have added a title for column 2 to clarify what is meant. The column title reads”Type of area considered”. We also added some specification in the cells below where needed for further clarification.

R2: 18. L302 – thematic attributes - do you mean fewer classes?

- Yes, this is what we mean. We have corrected it accordingly.

R2: 19. L307- thematic information is lacking – do you mean that if the thematic classes do not match the classes of interest?

- Yes. We revised the sentence to be more comprehensible.

R2: 20. L312 – 2980 km2 – the significance of the area is hard to understand – I recommend also specifying this as a percentage.

- Thanks for the important comment, we have integrated the following: “Thus, the potential area for onshore wind energy in the second model run is 34% smaller than in the first model run.”

R2: 21. L323- easy data – do you mean “easily obtained data??

- We changed to easily obtained data.

R2: 22. L344 – albeit this would not be necessary – do you mean “that are not actually helpful based on the real land use of the area.”

- Yes, this is what we mean. We changed the sentence as suggested.

R2: 23. L347 – I don’t understand what this fourth constellation is – do you mean something like “when the potential effort or response required to address the issues raised by the model outcome is large but the data uncertainty is also large?

- Yes, this is what we mean. We now write, as suggested: “… the fourth problem constellation is when the potential effort or response required to address the issues raised by the model outcome is large but the data uncertainty is also large. This applies, for example, even if the effect would be harmless, as… “

R2: 24. L375-6 – I can’t understand this sentence.

- Thank you for pointing this out. Indeed, this sentence is distorted. We have revised it and now write: “The realistic dilemma of using easily available data instead of better but hard to obtain data with uncertain added value in terms of relevance for answering the planning question will not vanish in the near future.”

R2: 25. L394 – I don’t understand the sentence (“in how far”?) and what does “behavioral mechanisms” mean? Can you rephrase this?

- Thank you for flagging this. We have revised it as follows: “Finally, in order to enhance the acceptability of uncertainty communication to decision makers, future research should explore how behavioral interventions such as splitting complex, uncertain decisions into smaller decisions with regard to space and time (29) would increase acceptability.”

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(75.7KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0260302.r005

Decision Letter 2

Gerald Forkuor

8 Nov 2021

Uncertainties in land use data may have substantial effects on environmental planning recommendations: a plea for careful consideration

PONE-D-21-14058R2

Dear Dr. Neuendorf,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Gerald Forkuor, Ph.D

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0260302.r006

Acceptance letter

Gerald Forkuor

16 Nov 2021

PONE-D-21-14058R2

Uncertainties in land use data may have substantial effects on environmental planning recommendations: a plea for careful consideration

Dear Dr. Neuendorf:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Gerald Forkuor

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(69.2KB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(75.7KB, docx)}

Data Availability Statement

[pone.0260302.ref001] 1.Marcot B.G. The Science and Management of Uncertainty—Dealing with Doubt in Natural Resource Management. 2020: 286. doi: 10.1201/9780429299575 [DOI] [Google Scholar]

[pone.0260302.ref002] 2.von Haaren C, Galler C, Albert C. (Eds.). Landschaftsplanung. 2021. ISBN: 978-3-8252-8579-1. German.

[pone.0260302.ref003] 3.Neuendorf F, von Haaren C, Albert C. Assessing and coping with uncertainties in landscape planning: an overview. Landscape Ecology. 2018;33:861–878. doi: 10.1007/s10980-018-0643-y [DOI] [Google Scholar]

[pone.0260302.ref004] 4.Schulp CJE, Alkemade R. Consequences of Uncertainty in Global-Scale Land Cover Maps for Mapping Ecosystem Functions: An Analysis of Pollination Efficiency. Remote Sensing. 2011;3(9):2057–2075. doi: 10.3390/rs3092057 [DOI] [Google Scholar]

[pone.0260302.ref005] 5.Schulp CJE, Burkhard B, Maes J, van Vliet J, Verburg PH. Uncertainties in Ecosystem Service Maps: A Comparison on the European Scale. PLoS ONE. 2014;9(10): e109643. doi: 10.1371/journal.pone.0109643 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0260302.ref006] 6.McVittie A, Norton L, Martin-Ortega J, Siameti I, Glenk K, Aalders I. Operationalizing an ecosystem services-based approach using Bayesian Belief Networks: An application to riparian buffer strips. Ecological Economics, 2015;110:15–27. doi: 10.1016/j.ecolecon.2014.12.004 [DOI] [Google Scholar]

[pone.0260302.ref007] 7.van der Biest K, Vrebos D, Staes J, Boerema A, Bodi MB, Fransen E, et al. Evaluation of the accuracy of land-use based ecosystem service assessments for different thematic resolutions. Journal of Environmental Management, 2015;156:41–51. doi: 10.1016/j.jenvman.2015.03.018 [DOI] [PubMed] [Google Scholar]

[pone.0260302.ref008] 8.Gret-Regamey A, Brunner SH, Altwegg J, Bebi P. Facing uncertainty in ecosystem services-based resource management. Journal of Environmental Management, 2013;127:145–154. doi: 10.1016/j.jenvman.2012.07.028 [DOI] [PubMed] [Google Scholar]

[pone.0260302.ref009] 9.Kubiszewski I, Costanza R, Paquet P, Halimi S. Hydropower development in the lower Mekong basin: alternative approaches to deal with uncertainty. Regional Environmental Change, 2013;13(1):3–15. doi: 10.1007/s10113-012-0303-8 [DOI] [Google Scholar]

[pone.0260302.ref010] 10.Foley AM. Uncertainty in regional climate modelling: A review. Progress in Physical Geography, 2010;34(5):647–670. doi: 10.1177/0309133310375654 [DOI] [Google Scholar]

[pone.0260302.ref011] 11.Ahern J. Theories, methods and strategies for sustainable landscape planning. In: Tress B., Tress G., Fry G., Opdam P. (eds) From landscape research to landscape planning: Aspects of integration, education and application. 2006;119–131. ISBN: 978-1-4020-3979-9 [Google Scholar]

[pone.0260302.ref012] 12.Kato S, Ahern J. ‘Learning by doing’: adaptive planning as a strategy to address uncertainty in planning. Journal of Environmental Planning and Management, 2008;51(4):543–559. doi: 10.1080/09640560802117028 [DOI] [Google Scholar]

[pone.0260302.ref013] 13.Walters CJ, Holling CS. Large-Scale Management Experiments and Learning by Doing. Ecology, 1990;71(6):2060–2068. doi: 10.2307/1938620 [DOI] [Google Scholar]

[pone.0260302.ref014] 14.Read R, O’Riordan T. The Precautionary Principle Under Fire. Environment: Science and Policy for Sustainable Development, 2017;59(5): 4–15. doi: 10.1080/00139157.2017.1350005 [DOI] [Google Scholar]

[pone.0260302.ref015] 15.Yin R. K. Case Study Research Design and Methods (5th ed.). 2014. Sage. ISBN-13: 978–1452242569 [Google Scholar]

[pone.0260302.ref016] 16.Jaffrain G, Sannier C. GMES Initial Operations / Copernicus Land monitoring services—Validation of products—(Second Specific Contract). Corine Land Cover 2012. Final Validation Report. Villeneuve d’Ascq (FR): Systèmes d’Information à Référence Spatiale; IGN France International; 2017 [cited 2021 Mar 20]. https://land.copernicus.eu/user-corner/technical-library/clc-2012-validation-report-1.

[pone.0260302.ref017] 17.von Drachenfels O. Kartierschlüssel für Biotoptypen in Niedersachsen unter besonderer Berücksichtigung der gesetzlich geschützten Biotope sowie der Lebensraumtypen von Anhang I der FFH-Richtlinie, Stand Februar 2020. 11th ed. Hannover: NLWKN; 2020. German.

[pone.0260302.ref018] 18.Landesamt für Umweltschutz Sachsen-Anhalt. Katalog der Biotoptypen und Nutzungstypen für die CIR-luftbildgestützte Biotoptypen- und Nutzungstypenkartierung im Land Sachsen-Anhalt. Halle: Landesamt für Umweltschutz Sachsen-Anhalt; 1992. German.

[pone.0260302.ref019] 19.Saathoff W, von Haaren C, Dechow R, Lovett AA. Farm-level assessment of CO2 and N2O emissions in Lower Saxony and comparison of implementation potentials for mitigation measures in Germany and England. Regional Environmental Change. 2013;13:825–841. doi: 10.1007/s10113-012-0364-8 [DOI] [Google Scholar]

[pone.0260302.ref020] 20.Thiele J, Wiehe J, Gauglitz P, Pape C, Lohr C, Bensmann A, et al. Konkretisierung von Ansatzpunkten einer naturverträglichen Ausgestaltung der Energiewende, mit Blick auf strategische Stellschrauben. BfN-Skripten. 2021;614. doi: 10.19217/skr614 German. [DOI] [Google Scholar]

[pone.0260302.ref021] 21.Walter A, Wiehe J, Schlömer G, Hashemifarzad A, Wenzel T, Albert I, et al. Naturverträgliche Energieversorgung aus 100% erneuerbaren Energien 2050. BfN-Skripten. 2018;501. doi: 10.19217/skr501 German. [DOI] [Google Scholar]

[pone.0260302.ref022] 22.Wiehe J, Thiele J, Walter A, Hashemifarzad A, zumHingst J, von Haaren C. Nothing to regret: Reconciling renewable energies with human wellbeing and nature in the German Energy Transition. International Journal of Energy Research. 2020;45(1):745–758. doi: 10.1002/er.5870 [DOI] [Google Scholar]

[pone.0260302.ref023] 23.Neuendorf F, Dataset: Exemplary landscape planning results using different qualities of LU/LC data to derive implications for practical planning. 2021.

[pone.0260302.ref024] 24.Thiele J, Wiehe J, von Haaren C. Dataset: Areas with low and medium spatial vulnerability to a prototype wind turbine. 2021.

[pone.0260302.ref025] 25.Lees J, Jaeger JAG, Gunn JAE, Noble BF. Analysis of uncertainty consideration in environmental assessment: an empirical study of Canadian EA practice. J Environ Plan Manag. 2016;59(11):2024–2044. doi: 10.1080/09640568.2015.1116980 [DOI] [Google Scholar]

[pone.0260302.ref026] 26.Larsen SV, Kørnøv L, Driscoll P. Avoiding climate change uncertainties in Strategic Environmental Assessment. Environmental Impact Assessment Review. 2013;43:144–150. doi: 10.1016/j.eiar.2013.07.003 [DOI] [Google Scholar]

[pone.0260302.ref027] 27.Batini C, Blaschke T, Lang S, Albrecht F, Abdulmutalib HM, Barsi A, et al. Data quality in remote sensing. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences—ISPRS Archives. 2017;42(2W7):447–53. doi: 10.5194/isprs-archives-XLII-2-W7-447-2017 [DOI] [Google Scholar]

[pone.0260302.ref028] 28.Stritih A, Bebi P, Grêt-Regamey A. Quantifying uncertainties in earth-observation-based ecosystem service assessments. Environmental Modelling & Software. 2019;111:300–310. doi: 10.1016/j.envsoft.2018.09.005 [DOI] [Google Scholar]

[pone.0260302.ref029] 29.Group on Earth observations. GEOSS Data Quality Guidelines. 2013. https://www.earthobservations.org/documents/dsp/GEOSS_Data_Quality_Guidelines.pdf

[pone.0260302.ref030] 30.Hamel P, Bryant BP. Uncertainty assessment in ecosystem services analyses: Seven challenges and practical responses. Ecosystem Services, 2017;24:1–15. doi: 10.1016/j.ecoser.2016.12.008 [DOI] [Google Scholar]

[pone.0260302.ref031] 31.vonHaaren C. & Othengrafen F. The Babel Fish Toolkit: Understanding and using behavioural mechanisms and interventions in landscape planning. disP—The Planning Review. 2019;55(2): 22–35. doi: 10.1080/02513625.2019.1630185 [DOI] [Google Scholar]

PERMALINK

Uncertainties in land use data may have substantial effects on environmental planning recommendations: A plea for careful consideration

Felix Neuendorf

Julia Thiele

Christian Albert

Christina von Haaren

Roles

Abstract

Introduction

Methodological approach

Table 1. Data applied in the case studies.

Data sources and properties in the three case studies

Assessment methods applied in the case studies

Results

Fig 1. Spatial representation of the results of the case study ‘regional soil-based climate change mitigation’.

Table 2. Quantitative results of the case study ‘regional soil-based climate change mitigation’.

Fig 2. Spatial representation of selected results of the case study ‘state-wide habitat conservation siting’.

Table 3. Quantitative results of the case study ‘state-wide habitat conservation siting’.

Fig 3. Comparison of the potentially usable areas for onshore wind energy (left: First model run; right: Second model run [24]).

Fig 4. Comparison of the selection of built-up areas from the Basis-DLM (left) and DLM250 (right) with a noise buffer of 750 m exemplified by the municipality of Straßlach-Dingharting south of Munich.

Fig 5. Comparison of the two elevation models DEM50 and DEM200 with a slope ≥ 30° illustrated by the example of a Bavarian Alps section (left) and a Harz section (right).

Discussion and conclusion

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Gerald Forkuor

Roles

Author response to Decision Letter 0

Decision Letter 1

Gerald Forkuor

Roles

Author response to Decision Letter 1

Decision Letter 2

Gerald Forkuor

Roles

Acceptance letter

Gerald Forkuor

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases