Field study for the impact of rain-on-snow events on snow cover during snow accumulation period

Summary

The impact of rain-on-snow (ROS) events on snow cover is a key to uncovering disaster mechanisms of ROS events, but related research is currently lacking. Therefore, we conducted a field study to investigate the effects of ROS events on snow cover during snow accumulation period. The results showed that ROS events accelerated the decrease rate of snow albedo and altered the metamorphic processes of snow layer. Additionally, ROS events were more likely to form ice layers on snow surface during snow accumulation period, and once formed, these ice layers did not melt for several days, affecting the survival of foragers and hoofed animals. Changes in snow surface also altered heat transfer processes within snow cover, affecting the energy balance of permafrost. More interestingly, ROS events reduced the decrease rate of snow depth during snow accumulation period. These findings provided scientific basis for further understanding of ROS events and their disaster mechanisms.

Graphical abstract

Highlights

• •ROS events accelerated the decrease rate of snow albedo in snow accumulation period
• •ROS events were more likely to form ice layers on snow surface
• •Changes in snow surface altered heat transfer processes within snow cover
• •ROS events reduced the decrease rate of snow depth during snow accumulation period

Natural sciences; Earth sciences; Environmental science

Introduction

Rain-on-snow (ROS), which specifically refers to rain that falls on snow surface, is prevalent at high latitudes and altitudes around the globe,¹ and occurs mainly in late autumn and early spring, but also occasionally in winter.²

ROS events are extremely harmful, which not only induce avalanche disasters,³ but also are highly susceptible to the more catastrophic ROS flooding.^4,5,6 However, scholars have different knowledge about the mechanism of flooding triggered by ROS events. For example, Singh et al., 1997⁷ argued that ROS flooding occurred primarily due to the formation of preferential flow within the snow cover and the input of rainfall, rather than rain-induced snowmelt. Juras et al., 2016⁸ suggested that rainfall first squeezes out non-rainwater from the snow cover before contributing to runoff when a ROS event occurs. Haleakala et al., 2023,⁹ on the other hand, thought that the acceleration of snow ablation by ROS events is the key to the formation of ROS flooding.

To further reveal the triggering mechanism of ROS flooding, we conducted a field ROS experiment (rainfall intensity 30 mm/h, cumulative rainfall 30 mm) during snow ablation period, and observed and analyzed the changes in snow albedo, snow water content, and snow depth during the ROS event and the following 7 days.¹⁰ Moreover, to minimize the limitations of the study results, snow plots with similar initial parameters of the snow cover were selected nearby as a no-rainfall control experiment. The final data showed that the snow albedo decay rate increased by a factor of 2 on the day of the ROS event. Within 7 days of the rainfall event, the average rate of snow albedo decay increased by 55.6% and the average rate of snow depth reduction increased by a factor of one. At the same time, the ROS event changed the mode of water transport within snow cover, saturating snow cover rapidly. Therefore, our results suggested that the main reason for ROS flood triggering was the rapid melting of high moisture content snow within a short period of time due to ROS events.

Nevertheless, ROS events also occur frequently in late fall and winter,^11,12,13 but their effects on snow cover after occurring in late fall and winter are currently unknown and often overlooked. Therefore, based on the ROS experiment during snow ablation period, this study intends to preliminarily recognize the influence pattern of ROS events on snow cover during snow accumulation period by conducting the ROS experiment during snow accumulation period in the same experimental region. Furthermore, to facilitate comparison with the results of the snow ablation period, the dimensions of the experimental snow plots, rainfall intensity, and the cumulative rainfall were the same as those of the previous field experiment, but optimized in terms of observation time and parameters. In this study, we observed the changes of snow albedo, snow water content, snow density, snow depth, and temperatures of snow surface and snow bottom before, during, and 5 days after the ROS event. The study is aimed at understanding the effects of ROS events on snow cover during the accumulation period, supplementing and perfecting the mechanism of ROS events, and providing a scientific basis for dealing with the disasters caused by ROS events.

Results

Changes in snow surface structure

On 22 October, when observations were made at 15:59 after the rainfall experiment, it was found that an ice layer about 5 cm thick had formed on the ROS experiment snow plot surface, and that this ice layer persisted for several days. However, there was little change in the surface structure of comparison snow plot (Figure 1). This difference was not accidental. Such changes in the snow surface after ROS events did not occur during the snow ablation period (Yang et al., 2023). This also indicated that although short-term air temperature increase caused ROS events to occur during the snow accumulation period, the low air temperature after the events inevitably caused rainwater to freeze on the snow surface.

Figure 1.

Differences in surface structure between experiment and comparison snow plots after the ROS event

Surface structure of experiment (A) and comparison (B) snow plots at 15:59 on 22 October.

Changes in snow albedo

Combining Tables 1 and S1, it can be seen that on 22 October, before the ROS event (i.e., before 13:12), there was not much difference in the albedo decay rate between two snow plots; however, after the ROS event (i.e., after 13:12), the albedo decay rate of the ROS experiment snow plot was significantly increased compared to the comparison snow plot. It should be noted, however, that due to the drop in air temperature, an ice layer formed on the ROS experiment snow plot surface at 15:59 (Figure 1), resulting in an increase in snow albedo. On 23 October (i.e., the day after the ROS event), after 2.5 h, the albedo of the ROS experiment snow plot decreased by 0.156 and that of the comparison snow plot decreased by 0.180, with the albedo decay rate of the comparison snow plot being higher than that of the ROS experiment snow plot. On 24 October (i.e., the third day after the ROS event), after 2.5 h, the albedo of the ROS experiment snow plot decreased by 0.235 and that of the comparison snow plot decreased by 0.214, which, compared to 23 October, showed a higher rate of albedo decay for the ROS experiment snow plot than for the comparison snow plot. On 25 October (i.e., the fourth day after the ROS event), after 2.5 h, the albedo of the ROS experiment snow plot decreased by 0.065, and that of the comparison snow plot decreased by 0.067, which, compared to 24 October, when the albedo decay rate of the comparison snow plot was in turn higher than that of the ROS experiment snow plot. On 26 October (i.e., the fifth day after the ROS event), after 2.5 h, the albedo of the ROS experiment snow plot decreased by 0.110, and that of the comparison snow plot decreased by 0.076, demonstrating that the albedo decay rate of the ROS experiment snow plot was, in turn, higher than that of the comparison snow plot.

Table 1.

Snow albedo at different moments in time

Measured time		Snow albedo		Measured time		Snow albedo
		Experiment snow plot	Comparison snow plot			Experiment snow plot	Comparison snow plot
October 22	11:23	0.783	0.798	October 23	11:30	0.853	0.933
	11:33	0.779	0.793		12:00	0.831	0.896
	11:43	0.776	0.791		12:30	0.809	0.885
	11:56	0.773	0.790		13:00	0.785	0.881
	13:42	0.693	0.787		13:30	0.735	0.811
	14:22	0.689	0.785		14:00	0.697	0.753
	15:05	0.650	0.784
	15:59	0.807	0.784
October 24	11:30	0.795	0.890	October 25	11:30	0.809	0.876
	12:00	0.772	0.884		12:00	0.795	0.862
	12:30	0.766	0.838		12:30	0.768	0.856
	13:00	0.718	0.807		13:00	0.764	0.834
	13:30	0.685	0.787		13:30	0.758	0.810
	14:00	0.560	0.676		14:00	0.744	0.809
October 26	11:30	0.817	0.887	–	–	–	–
	12:00	0.809	0.877	–	–	–	–
	12:30	0.778	0.861	–	–	–	–
	13:00	0.753	0.853	–	–	–	–
	13:30	0.730	0.830	–	–	–	–
	14:00	0.707	0.811	–	–	–	–

The above analysis shows that from 22 October to 26 October, there were higher and lower albedo decay rates for the ROS experiment snow plot compared to that of the comparison snow plot. However, in general, before the ROS event, the albedo and its decay rate of the ROS experiment snow plot and the comparison snow plot did not differ much, and after the ROS event, the albedo decay rate of the ROS experiment snow plot was significantly faster than that of the comparison snow plot, which indicated that ROS events accelerated the decay rate of snow albedo during the accumulation period, but the effect was not as significant as that of the melting period.¹⁰

Changes in snow water content

The entry of rain into snow cover increases snow water content, and this increase in water content affects snow density and altering the heat exchange within snow cover. Therefore, to know the effect of ROS events on snow water content during the accumulation period, it is first necessary to understand how snow water content varies in the natural state. For this purpose, the variation of water content inside the comparison snow plot (i.e., natural state) from 22 October to 26 October was analyzed, and the results were shown in Figure 2. As can be seen from the figure, on 22 October, the water content of the 0–5 cm snow layer was low and fluctuated very little, the water content of the 10 cm and lower snow layers was high, and the water content of the 10–15 cm snow layer fluctuated greatly. Compared with 22 October, the water content of each snow layer changed on 23 October, as evidenced by a lower and less fluctuating water content in the 0–10 cm snow layer, and a higher water content in the 15 cm and lower snow layers than in the surface layer. On 24 October, the water content of the snow surface was almost unchanged, and the snow layer with the largest fluctuation in water content was that at a snow depth of 5 cm. On 25 October, the water content of the 5–10 cm snow layer fluctuated more, which was the result of the slow infiltration of liquid water under the effect of gravity. On 26 October, liquid water reached the 15 cm snow layer by gravity. Overall, liquid water inside the contrasting snow plot infiltrated slowly downward by gravity.

Figure 2.

Changes in liquid water content inside snow cover in the natural state from 22 October to 26 October

In the natural state, snow water content during the melting period was significantly higher than that during the accumulation period¹⁰ (Figure 2); moreover, the water content of the natural snow surface was highest during the melting period, but was instead lowest during the accumulation period. This difference in the snow cover itself during different periods may lead to different effects of ROS events on snow cover during different periods.

Figure 3 showed the changes in snow water content before and after ROS events. As can be seen from the figure, before ROS events (i.e., before 13:12 on 22 October), the water content of all snow layers was less than 1%, especially the snow surface, which had the lowest water content; however, with the occurrence of ROS events, the water content of snow surface increased rapidly. However, it is interesting to note that the large amount of liquid water increased in snow surface did not rapidly infiltrate into snow interior, but rather remained in the snow up to 5 cm down from the surface (Figure 3). There were two possible reasons for this phenomenon: (1) Due to long-term slow metamorphism, a dense layer was formed on snow surface (Figures 4 and 5), resulting in the inability of rainfall strikes to penetrate the dense layer to form a preferential flow path. (2) Although a dense layer is formed on snow surface, snow cover was a porous medium and its water-holding capacity was enhanced when the water content was very low.

Figure 3.

Changes in liquid water content inside snow cover before and after the occurrence of ROS from 22 October to 26 October

Figure 4.

Changes in snow density in the natural state from 22 October to 26 October

Figure 5.

Changes in snow density before and after the occurrence of ROS from 22 October to 26 October

Changes in snow density

Snow density is one of the important parameters of snow accumulation, and the change of snow density affects the heat transfer, radiation characteristics and snow water content. Therefore, to understand the effect of ROS events on snow density, the characteristics and changes of snow density during the accumulation period in the natural state were first analyzed (Figure 4). The results showed that in the natural state, the density of snow surface during the accumulation period was greater than that of other layers, which indicates that a dense layer was formed in snow surface during the accumulation period after a long period of slow metamorphism. In addition, the variation of snow density in each layer from 22 October to 24 October was small, even in some snow layers with large fluctuations in water content (Figures 2 and 4). Interestingly, however, the magnitude of change in snow density began to increase from 25 October onwards, especially in the 10–15 cm snow layer. This change seems to be explained by the fact that with the fluctuation of snow water content and air temperature in the previous period (Figures 2 and S3), snow cover went through repeated melting and freezing processes, and metamorphism began to increase, and snow cover began to slowly densify.

Figure 1 and Table 1 showed that the occurrence of ROS events changed the energy balance of snow cover and snow surface structure, which may affect the natural variation of snow density. For this reason, the changes in snow density before and after ROS events from 22 October to 26 October were analyzed, and the results were shown in Figure 5. As can be seen from the figure, before the ice layer on snow surface was formed (i.e., before 15:59 on 22 October), the density of snow surface was above 0.25 g/cm³, which was greater than that of other snow layers, indicating that a dense layer was formed on snow surface, which was consistent with what was shown in Figure 4, and can be further used to explain the experimental phenomenon in Figure 2. Furthermore, from the experimental data, there was little change in snow density of each snow layer from 22 October to 24 October, but from 25 October onwards, snow density of each snow layer began to increase, which was similar to what is shown in Figure 4, and may be related to changes in snow water content and air temperature (Figures 2 and S3).

Further comparative analysis of Figures 4 and 5 showed that from 22 October to 24 October, except for the formation of an ice layer on the ROS experiment snow plot surface, the difference in snow density between the comparison snow plot and the ROS experiment snow plot was not significant and the changes were small. However, from 25 October onwards, snow density of the comparison snow plot fluctuated most in the snow layer at a depth of 10 cm, while snow density of the ROS experiment snow plot fluctuated most in the snow layer at a depth of 15 cm. Combined with the thickness of the ice layer, it can be assumed that the formation of the ice layer on snow surface caused snow layer, which was the most actively metamorphosed layer inside snow cover, to increase its thickness by one ice layer (5 cm) downwards. In other words, the occurrence of ROS events not only changed snow surface structure, but also altered the metamorphism of the snow layer inside snow cover. This alteration may affect the heat transfer within snow cover, which ultimately affected the energy balance of the permafrost at the base of snow cover.^14,15

Changes in snow depth

The effect of ROS events on snow depth was significant, but this significant effect was unexpected (Figure 6). On 22 October, the day of the ROS event, the reduction rate of snow depth in the ROS experiment plot was 8.86 times that of the comparison snow plot, however, on the second to fifth days after the ROS event (i.e., from 23 October to 26 October), none of the reduction rates of snow depth in the ROS experiment snow plot were higher than that of the that of the comparison snow plot, suggesting that the ROS event accelerated snow depth reduction on the day of the event, but decreased the reduction rate of snow depth later in the day. This seemed to be contradictory because, as can be seen from Table 1, snow albedo of the ROS experiment snow plot was generally lower than that of the comparison snow plot, which resulted in an increase in the amount of energy going into the interior of the ROS experiment snow plot compared to the comparison snow plot, and this increase in energy should accelerate snow melting, but the opposite was true. A reasonable explanation would be that the reduction in snow depth in two snow plots was not primarily due to ablation, but that the densification of snow covers itself, and sublimation, played a dominant role. This was because, as can be seen from Figure S3, the air temperature was below −5 °C, while the snow melting point was normally around 0 °C. Moreover, based on Figures 4 and 5, it was clear that the snow cover was fresh snow, so the decrease in snow depth in the experiment snow plot on the day of the ROS event was mainly caused by rainfall strikes, and the reason why the decrease rate of snow depth in the experiment snow plot was lower than that in the comparison snow plot in the later period was mainly because the formation of the ice layer on the experiment snow plot surface impeded the sublimation of snow surface.

Figure 6.

Changes in snow depth in the experiment snow plot and the comparison snow plot from 22 October to 26 October

Blue solid dots indicate the experiment snow plot and red solid dots indicate the comparison snow plot.

To further clarify the effect of ROS events on snow depth, the first data of snow depth observations for each day from 22 October to 26 October was selected for analysis, and the results were shown in Figure 7. As can be seen from the figure, the reduction rate of snow depth for the ROS experiment snow plot was lower than that for the comparison snow plot for the five-day period from 22 October to 26 October. This further suggested that the occurrence of ROS events had the potential to reduce the reduction rate of snow depth during the accumulation period.

Figure 7.

Changes in snow depth in the experiment snow plot and the comparison snow plot over a 5-day period

Blue solid dots indicate the experiment snow plot and red solid dots indicate the comparison snow plot.

Changes in temperatures of snow surface and snow bottom

The data recorded from 22 October to 26 October for the surface and bottom temperatures of the ROS experiment snow plot and comparison snow plot (Figure 8) indicated that there was not much difference in the temperatures at the surface and at the bottom of the ROS experiment snow plot and the comparison snow plot prior to the ROS event (i.e., before 13:12 on 22 October). However, at the time of the ROS event and on the day after the event (i.e., after 13:12 on 22 October), the surface temperature of the ROS experiment snow plot was higher than that of the comparison snow plot during the day and lower at night, and the bottom temperature was lower than that of the comparison snow plot during the day and higher than that of the comparison snow plot at night. On the day following the ROS event (i.e., 23 October), the surface temperatures of the experiment snow plot and the comparison snow plot were not significantly different before noon, but after noon and until the evening of 23 October, the surface temperatures of the experiment snow plot were higher than those of the comparison snow plot; the bottom temperatures were higher than those of the comparison snow plot from the morning of 23 October until about 18:00 p.m. on 23 October. Starting from the third day after the ROS event (i.e., from 24 October), the difference between the surface temperatures of the experiment snow plot and the comparison snow plot stabilized, generally showing that the surface temperature of the experiment snow plot was higher than that of the comparison snow plot when solar radiation was strongest, and that there was not a significant difference between the surface temperatures of the experiment snow plot and that of the comparison snow plot when the solar radiation was weakening or at zero; whereas the bottom temperatures had always been shown to be higher in the experiment snow plot than in the comparison snow plot. The above analyses indicated that the occurrence of ROS events did alter the heat exchange and energy balance of snow surface and interior, and that this effect did not disappear within a few days. This further suggested that the occurrence of ROS events affected the energy balance of permafrost,^14,15 resulting in a range of ecohydrological responses.

Figure 8.

Changes in temperatures of snow surface and snow bottom in the experiment snow plot and the comparison snow plot from 22 October to 26 October

The blue dashed line indicates the bottom temperature of the comparison snow plot, the blue solid line indicates the bottom temperature of the experiment snow plot, the red dashed line indicates the surface temperature of the comparison snow plot, and the red solid line indicates the surface temperature of the experiment snow plot.

Discussion

The impact of ROS events on snow cover during snow accumulation period

This study shows that when a ROS event occurs during the snow accumulation period, a layer of ice forms on the snow surface and the ice does not melt for several days. This phenomenon does not occur by chance. Air temperature is low during the snow accumulation period, and even if a short-term warming of air temperature leads to the occurrence of ROS events, the cooling that follows the rains will inevitably lead to the formation of an ice crust on the snow surface or at the ground surface. This is in general agreement with the findings of Putkonen and Roe, 2003,¹⁴ Putkonen et al., 2009,¹⁶ and Hansen et al., 2014,¹⁷ but is markedly different from the effect of ROS events on snow cover during the snow ablation period.¹⁰ For the snow albedo, the formation of the ice crust causes fluctuations in the snow albedo, but as the energy inside the snow accumulates, it causes an increase in the liquid water on the ice crust surface, which reduces the albedo. Compared to snow cover in its natural state, ROS events generally reduced the snow albedo during the snow accumulation period, which is the same as, but not as significant as, the effect of ROS events on snow cover in the snow ablation period.^10,18,19 During the snow accumulation period, long-term metamorphism formed a dense layer on the snow surface, making it impossible for rainfall strikes to form preferential flow inside the snow, which, together with the large amount of rainwater that collected on the snow surface to form an ice crust, resulted in a smaller effect of ROS events on snow water content. This differs greatly from the effect of ROS events on snow water content during the snow ablation period.¹⁰

Many studies have shown that ROS events accelerate snow ablation,^10,20,21 but the results of this study suggest that the occurrence of ROS events reduces the rate of snow depth reduction during the snow accumulation period. The main reasons for this difference may be: (1) the effect of air temperature. The lower air temperature during the snow accumulation period is not conducive to snow melt. (2) Formation of ice crust. The occurrence of ROS events led to the forming of an ice crust on the snow surface, and the snow depth began to decrease only after the ice on the snow surface had melted. Moreover, the changes in the temperatures of the snow surface and snow bottom in the experiment and comparison snow plots suggest that the ROS events altered the energy transfer processes within the snow during the snow accumulation period, which may affect the energy balance of the permafrost. This is in general agreement with the research results of Putkonen and Roe, 2003¹⁴ and Westermann et al., 2011.²²

Although the study by Eiriksson et al. (2013) showed that different snow thicknesses may lead to differences in the effect of ROS events on the snow cover during the snow ablation period, the snow thickness may mainly affect the energy transfer processes within the snow after a ROS event during the snow accumulation period, with less impact on the changes to other parameters. This is because: (1) the formation of ice crust on the snow surface after a ROS event is mainly related to air temperature, which is generally low during the snow accumulation period. (2) The formation of a dense layer on the snow surface is the result of long-term metamorphism of the snow, and is less related to the snow thickness; therefore, during the snow accumulation period, the snow thickness has less influence on the changes in the snow water content and snow density after a ROS event. (3) The thicker the snow cover is, the less the changes in the snow surface layer affect the snow bottom, and the more difficult it is for rainwater to reach the snow bottom. Therefore, the snow thickness has a greater influence on the energy transfer process within the snow after a ROS event.

The changes in snow water content before and after ROS events indicated that ROS events during the snow accumulation period were less likely to trigger floods. This was because the formation of floods required large amounts of liquid water, but during the snow accumulation period, rain was likely to freeze due to low air temperature and then be released slowly, having little effect on runoff. Of course, the extreme warming that continues after ROS events during the snow accumulation period is likely to trigger floods, which has become possible in the context of climate change. Furthermore, compared to snow in its natural state, changes in the surface structure and density of snow caused by ROS events were likely to increase the probability of avalanches. The formation of ice layers on the snow surface increased the load on snow, while increased metamorphism deeper within the snow led to an increase in the amount of unstable snow. These factors create more favorable conditions for avalanches.

This study showed that ROS events slowed down the rate of snow depth decrease during the snow accumulation period. However, the persistence of this impact was closely related to changes in air temperature in the later stages. If air temperature rose significantly one month or three months after the ROS event, the reduction in snow depth will accelerate. This was inevitable because ROS events caused a decrease in snow albedo, resulting in more energy being absorbed. Even so, the ice cover on the snow surface caused by ROS events during the snow accumulation period can last for a week or even several weeks. Long-term ice cover on snow surface not only affects the walking of hoofed animals, but also restricts the foraging of some animals, thereby affecting their survival.

Such changes in the snow bottom temperature may be significantly influenced by snow depth. As snow depth increased, heat was less able to rapidly conduct to snow bottom. Under these conditions, the effects of ROS events could not reach snow bottom in a short time, and may even be hindered by internal ice layers within the snow, making it difficult to reach the base. In such cases, the snow bottom temperature was not significantly affected by ROS events. However, this was only temporary. The fact that ROS events influence the heat exchange process within the snow was well established. Snow depth only affected the response time of deep snow to ROS events. Snow depth had little effect on changes in the snow surface temperature. With global air temperature continuing to rise and extreme events becoming more frequent, ROS events were likely to become increasingly common. This study may have some limitations, but it provided a data foundation for understanding the impact of ROS events on snow and permafrost. It also demonstrated that the impact of ROS events was significant and required attention. The emergence of new phenomena warned us that new challenges might arise in cold region hydrological research.

Conclusions

In this study, a ROS experiment during snow accumulation period was carried out at an altitude of 4592 m near Bayi Glacier in the Qilian Mountains, and a no-rainfall experiment was set up as a comparison. By analyzing the changes in parameters such as snow albedo, snow water content, snow density, snow depth and snow surface and bottom temperatures of the ROS experiment snow plot and the comparison snow plot before, during, and 5 days after the ROS event, the effects of the ROS event on snow cover during the accumulation period were understood. The main conclusions are as follows:

ROS events accelerated the decay rate of snow albedo during the accumulation period, altering the metamorphism of snow cover. At the same time, the occurrence of ROS events may lead to the formation of ice on snow surface during the accumulation period, and once formed, the ice cannot melt within a few days, disrupting the energy balance on snow surface and within it. This may affect the energy balance of permafrost, resulting in a range of ecohydrological responses. Moreover, ROS events did not significantly increase snow water content during the accumulation period, nor did they alter the liquid water transport processes within snow cover. It was even interesting to note that the occurrence of ROS events rather reduced the reduction rate of snow depth during the accumulation period. The results of the study revealed the effects of ROS events on snow cover, improved the disaster-causing mechanism of ROS events, and provided a reference for water resources management and disaster prevention and mitigation in China.

Limitations of the study

In this study, to improve the reliability of the results, a control experiment without rainfall was specifically set up. This is also something that other studies^7,8,23,24 have overlooked. However, the limited sample size resulting from only five consecutive days of observation and a single control experiment may restrict the generalizability of the study. Therefore, future studies should further conduct ROS experiments under different snow periods, meteorological conditions, and altitudes to supplement and refine the law of the effect of ROS events on snow cover.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, RenshengChen (crs2008@lzb.ac.cn).

Materials availability

This study did not generate new materials.

Data and code availability

• •All data reported in this paper will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We are very thankful to the anonymous reviewers whose comments helped improve the manuscript. This study was funded by the National Natural Science Foundation of China (42171145), the Gansu Provincial Science and Technology Program (25JRRA409), the Gansu Provincial Science and Technology Program (22ZD6FA005), the National Natural Science Foundation of China (42171147), the Gansu Provincial Major Science and Technology Program (24ZD13FA004), the Gansu Provincial Science and Technology Program (25JRRA864), and the Scientific Research Fund of Hunan Provincial Education Department (22A0497).

Author contributions

Conceptualization: Y.Z.; methodology: Y.Z. and C.R.; investigation: Y.Z.; writing ‒ original draft: Y.Z.; formal analysis: Y.Z., C.R., L.Z., and W.X.; writing ‒ review & editing: C.R., L.Z., and L.G.; data curation: Y.Z., B.S., and L.X.; funding acquisition: C.R., Y.Z., L.Z., B.S., and L.G.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Raw and analyzed data	This paper	–
Software and algorithms
Matlab 2017a	Open-source software	https://ww2.mathworks.cn/products/matlab.html

Method details

Field experiment station

At lower altitudes and latitudes, snow cannot remain on ground surface for long in autumn and winter, even on some shady back slopes where it can remain for long periods of time but is inaccessible to vehicles loaded with instruments. However, the Bayi Glacier in the Qilian Mountains has better transport conditions, and more importantly, in autumn in the region, snow starts to accumulate and the daytime air temperature are suitable for conducting rainfall experiments. In addition, in the central Qilian Mountains, there is the Upper Heihe River Ecohydrology Experimental Research Station of Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, which can provide logistical support for field experiments. Therefore, by combining the above conditions, a location at an altitude of 4592 m (39.0095°N, 98.8716°E), which is 2 km away from the Bayi Glacier, was finally selected as the field experiment station (Figure S1).

Experimental devices

The field conditions are complex and varied, so a portable, fully automated artificial rainfall simulator (Figure S2) was designed to adapt to the different field experiment conditions. The rainfall simulator consists of an adjustment bracket, a water-storage bucket, nozzles, a pressure sensor, a control system, and a water pipe. The adjustment bracket can be adapted to different slopes, and the water-storage bucket is placed on the adjustment bracket to steady the whole device. Three rainfall nozzles are installed at a distance of about 4 m from the ground in order to simulate natural rainfall as much as possible and to avoid the influence of wind on rainfall uniformity. The different combinations of nozzles guarantee a rainfall uniformity of more than 0.9 in a circle with a radius of 1 m and a rainfall intensity between 10 and 180 mm/h. The rainfall simulator is powered by a generator, which can be easily controlled by a control system to deliver water from a water-storage bucket to spray nozzles by a water pipe through a pump, and to achieve different rainfall intensities by adjusting the pressure and selecting different spray nozzles. The rainfall intensity of rainfall simulator was calibrated before the experiment was carried out. Furthermore, the rainfall simulator was equipped with a portable weather station to record meteorological conditions during the experiment in real time.

Experimental snow plots

Currently, the snow plot size for ROS experiment varies widely, and the choice of snow plot size was closely related to experiment’s purpose. Specifically, for rainwater infiltration in snow cover, snow plot sizes were chosen to be 1 m × 1 m or 1.5 m × 1 m (Conway & Benedict, 1994). Snow plot sizes of 1.5 m × 0.5 m and 0.5 m × 0.5 m were chosen for lateral flow of snow cover and for the process of generating runoff in snow cover and rainwater percentage problem during ROS occurrence (Eiriksson et al., 2013; Juras et al., 2016, 2017), respectively. For the water flow generation process and solute transport within snow cover during ROS events, larger snow plots were needed to illustrate the problem, so snow plot sizes of 2 m × 2 m and 6 m × 3 m are chosen, respectively (Singh et al., 1997; Feng et al., 2001; Lee et al., 2008, 2010). Moreover, the purpose of this study was to investigate the effect of ROS events on snow properties during the accumulation period. Meanwhile, it was compared with our completed experiment on the effect of ROS events on snow melting process (Yang et al., 2023), so as to complement and improve the disaster-causing mechanism of ROS events. Therefore, the snow plot size of 1 m × 1 m was chosen for this study. Considering the significant spatial variability in snow characteristics, we selected two snow plots with identical or similar parameters for comparative experiments. One snow plot was used for the ROS experiment, while the other served as a control experiment without rainfall. Comparing the results with a snow plot in its natural state can better illustrate the impact of the ROS experiment on snow, and this was overlooked in previous field ROS experiments.

Measurement of snow parameters

Snow albedo and snow depth

Snow albedo is the ratio of the total radiant energy emitted from each direction to the total radiant energy incident on snow surface per unit time, per unit area. In this study, the total radiant energy outgoing and incoming is measured using MP-200 Total Radiation Meter. The measuring range of this radiometer is 0∼1999 W/m² with an accuracy of ±5%. Snow depth is measured using an iron ruler with an accuracy of 1 mm. In addition, the ruler is held upright while measuring snow depth.

Snow density and snow water content

Snow water content and snow density are measured using Snow fork, which can accurately measure snow water content in the range of 0-10% and snow density in the range of 0-0.6 g/cm³. However, the spatial heterogeneity of snow cover is large, so three consecutive measurements are taken at each point, and the final results are averaged over three measurements.

Temperature at the bottom and surface of the snow

Snow temperature is recorded using the HOBO TidbiT MX Temperature 400′ Data Logger. The instrument has a measurement accuracy of 0.01 °C. There are four such instruments in total. Two of them are used to measure the snow surface temperature of the ROS experiment snow plot and comparison snow plot, respectively, while the remaining two are used to measure the bottom temperature of two snow plots. However, when measuring snow temperature, to ensure data accuracy, the instrument must be shielded from direct sunlight and air contact. Therefore, when measuring snow surface temperature, the instrument is placed 5 cm below the snow surface and completely covered by snow. Additionally, depending on snow melting conditions, the instrument’s position must be adjusted as needed to maintain a consistent 5 cm depth below the snow surface. When measuring the bottom temperature of a snow plot, the instrument is placed at the bottom of the snow plot and completely covered by snow. It is important to note that when placing the instrument, care should be taken to avoid disrupting the natural structure of the snow plot.

Experimental parameters

Rainfall parameters

To understand the effect of ROS events on snow cover during the accumulation period, a ROS experiment was carried out at the field experiment station on 22 October 2023 using the experiment device shown in Figure S2. In addition, to better reflect the effect of ROS events on snow cover during the accumulation period, a snow plot without rainfall on snow surface was also set up as a comparison experiment. The rainfall parameters are shown in Table S1. At 13:12, rainfall began with an intensity of 30 mm/h and lasted for 30 minutes before stopping. Measurement of various parameters of the two snow blocks then commenced. After the parameter measurements were completed, rainfall began again at 13:52 with the same intensity and lasted for 30 minutes before stopping. The two rainfall events were considered as a single ROS event. The reason for dividing the event into two rainfall processes was to understand the changes in snow parameters during the occurrence of the ROS event.

Snow plot initial parameters

Although setting up a comparison experiment can better reflect the effect of ROS events on snow cover during the accumulation period, there are more uncontrollable factors in the field experiment, especially the spatial heterogeneity of snow cover. Therefore, two snow plots with little difference in parameters such as snow depth, snow albedo, snow density, and water content were selected as experimental snow plots as much as possible at the experiment station. For two snow plots of ROS experiment and comparison, the initial snow depths were 24.7 cm and 25.3 cm, and snow albedos were 0.783 and 0798, respectively. The other initial parameters of two snow plots are shown in Table S2. Also, to prepare experiment snow plots, a larger block (1.5 m×1.5 m) was first prepared by digging snow cover from all four sides. The snow plots were then carefully further reduced to 1 m×1 m to maintain the natural structure of the snow plots as much as possible.

Meteorological data

Meteorological data help to understand the weather conditions during the field experiment and help to explain experimental phenomena. Therefore, a ZL6 mini meteorological monitoring system was used to record data on air temperature, relative humidity, solar radiation and wind speed at 2 m above ground level at the field experiment station, as shown in Figure S3. No natural rainfall occurred during the experiment (not shown in the figure). As can be seen in Figure S3, solar radiation fluctuated on 22 October and 24 October due to cloud cover, and the weather is clear on 23 October, 25 October and 26 October. In addition, from 22 October to 26 October, the maximum air temperature at the field experiment station was -3.5 °C and the minimum air temperature was -16.2 °C; the relative humidity ranged from 30% to 70%; and the wind speed ranged from 0.22 to 8.41 m/s.

Quantification and statistical analysis

The statistical analysis in this study was relatively simple, involving only linear regression. All statistical analyses and data visualizations were performed using Matlab R2017a and Origin 2018.

Published: September 23, 2025