Table 1.
Details of predictors selected for spatially explicit modelling of SOC.
| Scorpan factors | Predictors | Description | Relationship to SOC |
|---|---|---|---|
| s | Dominant soil | 21 types, soil and terrain database (SOTER) for Nepal53 | SOC varies with soil type. E.g. clay soils have a much higher SOC than sandy loam54 |
| c, r, a | Elevation | Digital elevation model, 30 m spatial resolution, ASTER55 | Related to mean annual air temperature; higher elevation areas have a lower temperature |
| o | NDVI | Based on cloud-free Landsat 7 median composite image for ca. 2000, computed from cloud-free imagery acquired in the period 1999–201256, version 1.6 | NDVI is a proxy of landscape-scale photosynthetic activity forest57, and thus, long-term NDVI can be used as a proxy for net primary productivity58 and inputs of organic matter into the soil |
| o, a | Cost surface | Cost surface calculated using GRASS r.walk59. It represents the time in seconds needed to reach each grid cell from the road network based on national road data60. The slope (derived from DEM) was used as a friction surface | In a human-dominated landscape, the proximity of forest increases the likelihood of disturbance |
| o, a | Protected Status | Binary protected/non-protected area mask. Polygon layer from Dept of National Parks and Wildlife Conservation, Nepal, rasterised to 30 m pixel | The sites under the protection have lower disturbance and higher SOC stocks than comparable sites. For e.g. sites outside the protected area can lose twice the amount of SOC compared to protected sites in the humid tropical forest61 |
| o | Distance from edge | Distance from the edge of the forest boundary towards the core using a forest mask62. This raster proximity map was derived by applying the gdal:proximity function63 | The likelihood of disturbance is higher near the edge of the forest patch than near the core |
| r | Slope | Slope gradient derived from the digital elevation model | The erosion potential increases with slope gradient, and thus steep slopes are likely to have shallow soils. The shallow soils have a small water storage capacity and generate runoff more frequently than deeper soils |
| r, c | Wind exposure | The average ‘Wind Effect Index’ for all directions using an angular step. A dimensionless index. Values below 1 indicate wind shadowed areas, whereas values above 1 indicate wind-exposed areas. Derived using the SAGA-GIS package64 | Considers aspect. Sites North-facing slopes can have three times higher SOC levels than South-facing slopes65 |
| r, c | TWI | Topographic wetness index (TWI) indicates the potential for water to accumulate. A high index value indicates the high water accumulation potential. Unitless. Computed using the SAGA-GIS package64 | Higher soil moisture availability favours plant productivity and thus more carbon inputs to soil |
| r | TPI | Topographic position index (TPI) compares the elevation of each cell in a DEM to the mean elevation of a specified neighbourhood around that cell. A 100 m × 100 m neighbourhood was used. Positive TPI values represent locations that are higher than the average of their neighbourhood window (e.g. ridges), negative values are lower (e.g. valleys), and flat areas are close to66 | Curvature controls the water redistribution and substrate thickness and is an important determinant of SOC65 |
| p, a | Parent materials | Soil and terrain database (SOTER) for Nepal53 provides a parent material type map representing eleven types based on lithology | Parent materials determine the soil properties and determine SOC distribution67 |
The spatial resolution of all predictors is 30 m.