Ground-based evaluation of MODIS snow cover product V6 across China: Implications for the selection of NDSI threshold

Impacts of snow cover seasonality on spring land surface phenology of forests in Changbai mountains of Northeast China

Snow cover phenology (SCP) strongly affects forest spring phenology (the start of growing season, SOS), but the underlying mechanism of the relationship varies. In this study, we aimed to analyze the relationship between forest SOS and SCP, and investigate the mechanisms about how changes of SCP affect forest SOS. To do so, we extracted forest SOS and SCP from multiple remote sensing datasets and analyzed the spatio-temporal patterns of both in Changbai Mountains (2001-2020). We assessed the relationships between SCP and forest SOS using partial least squares regression analysis and investigated the potential mechanism of SCP changes affecting on forest SOS using path analysis. We found earlier forest SOS (-0.5 days/year), prolonged snow cover duration (SCD, 0.43 day/year), and earlier snow cover end day (SCED, -0.1 days/year) in the region. The results indicated that SCD showed negative influence while SCED showed positive influence on forest SOS in most of the region. Results revealed that the influence of SCP on forest SOS was mainly through altering spring temperature and the dominant path of SCP influencing forest SOS followed hydrothermal gradients. Our study reveals new insights into the influence of changing SCP on forest SOS, which provides the theoretical basis for including SCP in the phenological models.

Spatio-temporal analysis of climate and irrigated vegetation cover changes and their role in lake water level depletion using a pixel-based approach and canonical correlation analysis

Lake Urmia, located in northwest Iran, was among the world's largest hypersaline lakes but has now experienced a 7 m decrease in water level, from 1278 m to 1271 over 1996 to 2019. There is doubt as to whether the pixel-based analysis (PBA) approach's answer to the lake's drying is a natural process or a result of human intervention. Here, a non-parametric Mann-Kendall trend test was applied to a 21-year record (2000-2020) of satellite data products, i.e., temperature, precipitation, snow cover, and irrigated vegetation cover (IVC). The Google Earth Engine (GEE) cloud-computing platform utilized over 10 sub-basins in three provinces surrounding Lake Urmia to obtain and calculate pixel-based monthly and seasonal scales for the products. Canonical correlation analysis was employed in order to understand the correlation between variables and lake water level (LWL). The trend analysis results show significant increases in temperature (from 1 to 2 °C during 2000-2020) over May-September, i.e., in 87 %-25 % of the basin. However, precipitation has seen an insignificant decrease (from 3 to 9 mm during 2000-2019) in the rainy months (April and May). Snow cover has also decreased and, when compared with precipitation, shows a change in precipitation patterns from snow to rain. IVC has increased significantly in all sub-basins, especially the southern parts of the lake, with the West province making the largest contribution to the development of IVC. According to the PBA, this analysis underpins the very high contribution of IVC to the drying of the lake in more detail, although the contribution of climate change in this matter is also apparent. The development of IVC leads to increased water consumption through evapotranspiration and excess evaporation caused by the storage of water for irrigation. Due to the decreased runoff caused by consumption exceeding the basin's capacity, the lake cannot be fed sufficiently.

Assessing Snow Phenology and Its Environmental Driving Factors in Northeast China

Snow cover is an important water source and even an Essential Climate Variable (ECV) as defined by the World Meteorological Organization (WMO). Assessing snow phenology and its driving factors in Northeast China will help with comprehensively understanding the role of snow cover in regional water cycle and climate change. This study presents spatiotemporal variations in snow phenology and the relative importance of potential drivers, including climate, geography, and the normalized difference vegetation index (NDVI), based on the MODIS snow products across Northeast China from 2001 to 2018. The results indicated that the snow cover days (SCD), snow cover onset dates (SCOD) and snow cover end dates (SCED) all showed obvious latitudinal distribution characteristics. As the latitude gradually increases, SCD becomes longer, SCOD advances and SCED delays. Overall, there is a growing tendency in SCD and a delayed trend in SCED across time. The variations in snow phenology were driven by mean temperature, followed by latitude, while precipitation, aspect and slope all had little effect on the SCD, SCOD and SCED. With decreasing temperature, the SCD and SCED showed upward trends. The mean temperature has negatively correlation with SCD and SCED and positively correlation with SCOD. With increasing latitude, the change rate of the SCD, SCOD and SCED in the whole Northeast China were 10.20 d/degree, −3.82 d/degree and 5.41 d/degree, respectively, and the change rate of snow phenology in forested areas was lower than that in nonforested areas. At the same latitude, the snow phenology for different underlying surfaces varied greatly. The correlations between the snow phenology and NDVI were mainly positive, but weak correlations accounted for a large proportion.

Glacier Area and Snow Cover Changes in the Range System Surrounding Tarim from 2000 to 2020 Using Google Earth Engine

Glacier and snow are sensitive indicators of regional climate variability. In the early 21st century, glaciers in the West Kunlun and Pamir regions showed stable or even slightly positive mass budgets, and this is anomalous in a worldwide context of glacier recession. We studied the evolution of snow cover to understand whether it could explain the evolution of glacier area. In this study, we used the thresholding of the NDSI (Normalized Difference Snow Index) retrieved with MODIS data to extract annual glacier area and snow cover. We evaluated how the glacier trends related to snow cover area in five subregions in the Tarim Basin. The uncertainty in our retrievals was assessed by comparing MODIS results with the Landsat-5 TM in 2000 and Landsat-8 OLI in 2020 glacier delineation in five subregions. The glacier area in the Tarim Basin decreased by 1.32%/a during 2000–2020. The fastest reductions were in the East Tien Shan region, while the slowest relative reduction rate was observed in the West Tien Shan and Pamir, i.e., 0.69%/a and 1.08%/a, respectively, during 2000–2020. The relative glacier stability in Pamir may be related to the westerlies weather system, which dominates climate in this region. We studied the temporal variability of snow cover on different temporal scales. The analysis of the monthly snow cover showed that permanent snow can be reliably delineated in the months from July to September. During the summer months, the sequence of multiple snowfall and snowmelt events leads to intermittent snow cover, which was the key feature applied to discriminate snow and glacier.

An Automated Snow Mapper Powered by Machine Learning

Snow preserves fresh water and impacts regional climate and the environment. Enabled by modern satellite Earth observations, fast and accurate automated snow mapping is now possible. In this study, we developed the Automated Snow Mapper Powered by Machine Learning (AutoSMILE), which is the first machine learning-based open-source system for snow mapping. It is built in a Python environment based on object-based analysis. AutoSMILE was first applied in a mountainous area of 1002 km2 in Bome County, eastern Tibetan Plateau. A multispectral image from Sentinel-2B, a digital elevation model, and machine learning algorithms such as random forest and convolutional neural network, were utilized. Taking only 5% of the study area as the training zone, AutoSMILE yielded an extraordinarily satisfactory result over the rest of the study area: the producer’s accuracy, user’s accuracy, intersection over union and overall accuracy reached 99.42%, 98.78%, 98.21% and 98.76%, respectively, at object level, corresponding to 98.84%, 98.35%, 97.23% and 98.07%, respectively, at pixel level. The model trained in Bome County was subsequently used to map snow at the Qimantag Mountain region in the northern Tibetan Plateau, and a high overall accuracy of 97.22% was achieved. AutoSMILE outperformed threshold-based methods at both sites and exhibited superior performance especially in handling complex land covers. The outstanding performance and robustness of AutoSMILE in the case studies suggest that AutoSMILE is a fast and reliable tool for large-scale high-accuracy snow mapping and monitoring.

Evaluation and Intercomparison of Topographic Correction Methods Based on Landsat Images and Simulated Data

Topographic effects in medium and high spatial resolution remote sensing images greatly limit the application of quantitative parameter retrieval and analysis in mountainous areas. Many topographic correction methods have been proposed to reduce such effects. Comparative analyses on topographic correction algorithms have been carried out, some of which drew different or even contradictory conclusions. Performances of these algorithms over different terrain and surface cover conditions remain largely unknown. In this paper, we intercompared ten widely used topographic correction algorithms by adopting multi-criteria evaluation methods using Landsat images under various terrain and surface cover conditions as well as images simulated by a 3D radiative transfer model. Based on comprehensive analysis, we found that the Teillet regression-based models had the overall best performance in terms of topographic effects’ reduction and overcorrection; however, correction bias may be introduced by Teillet regression models when surface reflectance in the uncorrected images do not follow a normal distribution. We recommend including more simulated images for a more in-depth evaluation. We also recommend that the pros and cons of topographic correction methods reported in this paper should be carefully considered for surface parameters retrieval and applications in mountain regions.

Snow Cover and Snow Persistence Changes in the Mocho-Choshuenco Volcano (Southern Chile) Derived From 35 Years of Landsat Satellite Images

Mountain regions have experienced above-average warming in the 20th century and this trend is likely to continue. These accelerated temperature changes in alpine areas are causing reduced snowfall and changes in the timing of snowfall and melt. Snow is a critical component of alpine areas - it drives hibernation of animals, determines the length of the growing season for plants and the soil microbial composition. Thus, changes in snow patterns in mountain areas can have serious ecological consequences. Here we use 35 years of Landsat satellite images to study snow changes in the Mocho-Choshuenco Volcano in the Southern Andes of Chile. Landsat images have 30 m pixel resolution and a revisit period of 16 days. We calculated the total snow area in cloud-free Landsat scenes and the snow frequency per pixel, here called “snow persistence” for different periods and seasons. Permanent snow cover in summer was stable over a period of 30 years and decreased below 20 km2 from 2014 onward at middle elevations (1,530–2,000 m a.s.l.). This is confirmed by negative changes in snow persistence detected at the pixel level, concentrated in this altitudinal belt in summer and also in autumn. In winter and spring, negative changes in snow persistence are concentrated at lower elevations (1,200–1,530 m a.s.l.). Considering the snow persistence of the 1984–1990 period as a reference, the last period (2015–2019) is experiencing a −5.75 km2 reduction of permanent snow area (snow persistence > 95%) in summer, −8.75 km2 in autumn, −42.40 km2 in winter, and −18.23 km2 in spring. While permanent snow at the high elevational belt (>2,000 m a.s.l.) has not changed through the years, snow that used to be permanent in the middle elevational belt has become seasonal. In this study, we use a probabilistic snow persistence approach for identifying areas of snow reduction and potential changes in alpine vegetation. This approach permits a more efficient use of remote sensing data, increasing by three times the amount of usable scenes by including images with spatial gaps. Furthermore, we explore some ecological questions regarding alpine ecosystems that this method may help address in a global warming scenario.

Remote Sensing of Snow Cover Variability and Its Influence on the Runoff of Sápmi’s Rivers

The boreal winter 2019/2020 was very irregular in Europe. While there was very little snow in Central Europe, the opposite was the case in northern Fenno-Scandia, particularly in the Arctic. The snow cover was more persistent here and its rapid melting led to flooding in many places. Since the last severe spring floods occurred in the region in 2018, this raises the question of whether more frequent occurrences can be expected in the future. To assess the variability of snowmelt related flooding we used snow cover maps (derived from the DLR’s Global SnowPack MODIS snow product) and freely available data on runoff, precipitation, and air temperature in eight unregulated river catchment areas. A trend analysis (Mann-Kendall test) was carried out to assess the development of the parameters, and the interdependencies of the parameters were examined with a correlation analysis. Finally, a simple snowmelt runoff model was tested for its applicability to this region. We noticed an extraordinary variability in the duration of snow cover. If this extends well into spring, rapid air temperature increases leads to enhanced thawing. According to the last flood years 2005, 2010, 2018, and 2020, we were able to differentiate between four synoptic flood types based on their special hydrometeorological and snow situation and simulate them with the snowmelt runoff model (SRM).

Observing Snow Cover and Water Resource Changes in the High Mountain Asia Region in Comparison with Global Mountain Trends over 2000–2018

The quantification of snow cover changes and of the related water resources in mountain areas has a key role for understanding the impact on several sectors such as ecosystem services, tourism and energy production. By using NASA-Moderate Resolution Imaging Spectroradiometer (MODIS) images from 2000 to 2018, this study analyzes changes in snow cover in the High Mountain Asia region and compares them with global mountain areas. Globally, snow cover extent and duration are declining with significant trends in around 78% of mountain areas, and the High Mountain Asia region follows similar trends in around 86% of the areas. As an example, Shaluli Shan area in China shows significant negative trends for both snow cover extent and duration, with −11.4% (confidence interval: −17.7%, −5.5%) and −47.3 days (confidence interval: −70.4 days, −24.4 days) at elevations >5500 m a.s.l. respectively. In spring, an earlier snowmelt of −13.5 days (confidence interval: −24.3 days, −2.0 days) in 4000–5500 m a.s.l. is detected. On the other side, Tien Shan area shows an earlier snow onset of −28.8 days (confidence interval: −44.3 days, −8.2 days) between 2500 and 4000 m a.s.l., governed by decreasing temperature and increasing snowfall. In the current analysis, the Tibetan Plateau shows no significant changes. Regarding water resources, by using Gravity Recovery and Climate Experiment (GRACE) data it was found that around 50% of areas in the High Mountain Asia region and 30% at global level are suffering from significant negative temporal trends of total water storage (including groundwater, soil moisture, surface water, snow, and ice) in the period 2002–2015. In the High Mountain Asia region, this negative trend involves around 54% of the areas during spring period, while at a global level this percentage lies between 25% and 30% for all seasons. Positive trends for water storage are detected in a maximum 10% of the areas in High Mountain Asia region and in around 20% of the areas at global level. Overall snow mass changes determine a significant contribution to the total water storage changes up to 30% of the areas in winter and spring time over 2002–2015.

A Conditional Probability Interpolation Method Based on a Space-Time Cube for MODIS Snow Cover Products Gap Filling

Seasonal snow cover is closely related to regional climate and hydrological processes. In this study, Moderate Resolution Imaging Spectroradiometer (MODIS) daily snow cover products from 2001 to 2018 were applied to analyze the snow cover variation in northern Xinjiang, China. As cloud obscuration causes significant spatiotemporal discontinuities in the binary snow cover extent (SCE), we propose a conditional probability interpolation method based on a space-time cube (STCPI) to remove clouds completely after combining Terra and Aqua data. First, the conditional probability that the central pixel and every neighboring pixel in a space-time cube of 5 × 5 × 5 with the same snow condition is counted. Then the snow probability of the cloud pixels reclassified as snow is calculated based on the space-time cube. Finally, the snow condition of the cloud pixels can be recovered by snow probability. The validation experiments with the cloud assumption indicate that STCPI can remove clouds completely and achieve an overall accuracy of 97.44% under different cloud fractions. The generated daily cloud-free MODIS SCE products have a high agreement with the Landsat–8 OLI image, for which the overall accuracy is 90.34%. The snow cover variation in northern Xinjiang, China, from 2001 to 2018 was investigated based on the snow cover area (SCA) and snow cover days (SCD). The results show that the interannual change of SCA gradually decreases as the elevation increases, and the SCD and elevation have a positive correlation. Furthermore, the interannual SCD variation shows that the area of increase is higher than that of decrease during the 18 years.

Changes in Snow Cover Dynamics over the Indus Basin: Evidences from 2008 to 2018 MODIS NDSI Trends Analysis

The frozen water reserves on the Earth are not only very dynamic in their nature, but also have significant effects on hydrological response of complex and dynamic river basins. The Indus basin is one of the most complex river basins in the world and receives most of its share from the Asian Water Tower (Himalayas). In such a huge river basin with high-altitude mountains, the regular quantification of snow cover is a great challenge to researchers for the management of downstream ecosystems. In this study, Moderate Resolution Imaging Spectroradiometer (MODIS) daily (MOD09GA) and 8-day (MOD09A1) products were used for the spatiotemporal quantification of snow cover over the Indus basin and the western rivers’ catchments from 2008 to 2018. The high-resolution Landsat Enhanced Thematic Mapper Plus (ETM+) was used as a standard product with a minimum Normalized Difference Snow Index (NDSI) threshold (0.4) to delineate the snow cover for 120 scenes over the Indus basin on different days. All types of errors of commission/omission were masked out using water, sand, cloud, and forest masks at different spatiotemporal resolutions. The snow cover comparison of MODIS products with Landsat ETM+, in situ snow data and Google Earth imagery indicated that the minimum NDSI threshold of 0.34 fits well compared to the globally accepted threshold of 0.4 due to the coarser resolution of MODIS products. The intercomparison of the time series snow cover area of MODIS products indicated R2 values of 0.96, 0.95, 0.97, 0.96 and 0.98, for the Chenab, Jhelum, Indus and eastern rivers’ catchments and Indus basin, respectively. A linear least squares regression analysis of the snow cover area of the Indus basin indicated a declining trend of about 3358 and 2459 km2 per year for MOD09A1 and MOD09GA products, respectively. The results also revealed a decrease in snow cover area over all the parts of the Indus basin and its sub-catchments. Our results suggest that MODIS time series NDSI analysis is a useful technique to estimate snow cover over the mountainous areas of complex river basins.

Comparative evaluation of VIIRS daily snow cover product with MODIS for snow detection in China based on ground observations

Accurate spatiotemporal information of snow cover not only is important for investigating the mechanisms of climate change but also greatly contributes to hydrological modelling in mountainous regions. The Suomi-National Polar-orbiting Partnership (S-NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) (referred to as VNP) daily snow cover product is recently released and expected to take place of Moderate Resolution Imaging Spectroradiometer (MODIS) snow cover products in near future. As an important addition to the widely used MODIS products, there is also an urgent need for a reliable accuracy evaluation and comparison of VNP for future large-scale daily snow cover mapping. This study for the first time evaluates the accuracy of VNP daily snow cover data in China using daily snow depth observations from 330 stations. The accuracy of VNP data is generally good with the averaged CK (Cohen's Kappa) and FS (F-Score) as high as 0.72 and 0.75, respectively, but considerably decreases to 0.50 and 0.52 for the Tibetan Plateau. VNP shows slightly better accuracy than MODIS TERRA for stations outside the Tibetan Plateau owing to its higher spatial resolution, but its accuracy is lower than TERRA for those within the Tibetan Plateau possibly due to its longer time interval between ground observation and satellite overpass time. By contrast, VNP shows much better accuracy than MODIS AQUA in China including both outside and within the Tibetan Plateau. This study provides important implications for optimal use of VNP and MODIS daily snow cover products in China, which may further contribute to more accurate snow variation information for climate analysis and cryospheric hydrological modelling.

Improved Landsat-Based Water and Snow Indices for Extracting Lake and Snow Cover/Glacier in the Tibetan Plateau

Identifying water and snow cover/glaciers (SCG) accurately is of great importance for monitoring different water resources in the Tibetan Plateau. However, discriminating between water and SCG remains a difficult task because of their similar spectral characteristic according to the physical principles of remote sensing. To efficiently distinguish different kinds of water resources automatically, here we proposed two new indices including: (i) the normalized difference water index with no SCG information (NDWIns) to extract lake water and suppress SCG: and (ii) the normalized difference snow index with no water information (NDSInw) to extract SCG and suppress lake water. Both new water and snow indices were tested in the Tibetan Plateau using Landsat series, showing that the overall accuracies of NDWIns and NDSInw were in the range of 94.6–97.0% and 94.9–97.0% in mapping the lake water from SCG and mapping the SCG from lake water, respectively. Further comparisons suggest that these new two indices improved upon the previous normalized difference snow index/modified normalized difference water index (NDSI/MNDWI) in mapping the water body and SCG. While the present study only focuses on the validation over certain areas in Tibetan Plateau, the newly proposed NDWIns and NDSInw have the potential for better monitoring the lake water and snow/glacier areas over other cold regions around the globe.

Gap-Filling of a MODIS Normalized Difference Snow Index Product Based on the Similar Pixel Selecting Algorithm: A Case Study on the Qinghai–Tibetan Plateau

Cloud contamination has largely limited the application of the Moderate Resolution Imaging Spectroradiometer(MODIS) normalized difference snow index (NDSI). Here, a novel gap-filling method based on spatial-temporal similar pixel interpolation was proposed to remove cloud occlusions in MODIS NDSI products. First, the widely used Terra and Aqua combination and three-day temporal filter methods were applied. The remaining missing NDSI information was estimated by using similar eligible pixels in the remaining cloud-free portion of a target image through a spatial-temporal similar pixel selecting algorithm (SPSA). The MODIS daily NDSI product data from 2003 to 2018 in the Qinghai–Tibetan Plateau (China) was used as a case study. The results demonstrate that the three-step methodology can generate almost completely cloud-free, daily MODIS NDSI images, reducing the cloud-gap fraction from >45% to less than 1.5% on average. The validation results of the SPSA method exhibited a high accuracy, with a high R2 exceeding 0.78, a low mean absolute error of 2.77%, a root mean square error of 3.78%, and a 96.92% overall accuracy. The proposed method can fill cloud gaps without a significant loss of accuracy, especially during snow cover transition periods (autumn and spring), which may provide more accurate cloud-free NDSI data for climate change and energy balance studies.

Assessing the Snow Disaster and Disaster Resistance Capability for Spring 2019 in China’s Three-River Headwaters Region

Frequent snowfall and low temperatures led to a considerable snow disaster in some areas of China’s Three-River Headwaters Region (TRHR) in Qinghai province in the spring of 2019, exerting a considerably negative influence on animal husbandry production in local grasslands. Based on a model of snow disaster classification and quantitative estimations of disaster-stricken animal husbandry, we propose a comprehensive disaster resistance capability index (CDRCI) using remote sensing, ground monitoring, and statistical investigations. With a comprehensive assessment of the space distribution and the magnitude of snow disasters, combined with a quantitative determination of disaster-stricken animal husbandry, the proposed CDRCI calculates how grassland animal husbandry is affected by snow disasters in different counties of the TRHR. The results indicate that approximately 2.31 million sheep and yaks were affected by moderate to severe snow disasters in the TRHR, accounting for 78.3% of the total livestock in the affected region. Of these affected livestock, approximately 1.54 million sheep and yaks were specifically affected by severe snow disasters, accounting for 52.1% of the total number of livestock. The CDRCIs for grassland animal husbandry in both Yushu and were moderate, being higher for the former than for the latter. We confirmed that the proposed CDRCI can accurately evaluate the magnitude of snow disasters in terms of how they affect grassland animal husbandry. The CDRCI is a way of relating the number of animal deaths to spatial disaster prevention and resistance. We expect that this research will provide important theoretical support for formulating snow disaster resistance policy, for example for increasing the construction of grassland animal husbandry infrastructure as well as providing greater stored forage material.

Snow Cover Evolution in the Gran Paradiso National Park, Italian Alps, Using the Earth Observation Data Cube

Mountainous regions are particularly vulnerable to climate change, and the impacts are already extensive and observable, the implications of which go far beyond mountain boundaries and the environmental sectors. Monitoring and understanding climate and environmental changes in mountain regions is, therefore, needed. One of the key variables to study is snow cover, since it represents an essential driver of many ecological, hydrological and socioeconomic processes in mountains. As remotely sensed data can contribute to filling the gap of sparse in-situ stations in high-altitude environments, a methodology for snow cover detection through time series analyses using Landsat satellite observations stored in an Open Data Cube is described in this paper, and applied to a case study on the Gran Paradiso National Park, in the western Italian Alps. In particular, this study presents a proof of concept of the preliminary version of the snow observation from space algorithm applied to Landsat data stored in the Swiss Data Cube. Implemented in an Earth Observation Data Cube environment, the algorithm can process a large amount of remote sensing data ready for analysis and can compile all Landsat series since 1984 into one single multi-sensor dataset. Temporal filtering methodology and multi-sensors analysis allows one to considerably reduce the uncertainty in the estimation of snow cover area using high-resolution sensors. The study highlights that, despite this methodology, the lack of available cloud-free images still represents a big issue for snow cover mapping from satellite data. Though accurate mapping of snow extent below cloud cover with optical sensors still represents a challenge, spatial and temporal filtering techniques and radar imagery for future time series analyses will likely allow one to reduce the current cloud cover issue.

A Two-Stage Fusion Framework to Generate a Spatio–Temporally Continuous MODIS NDSI Product over the Tibetan Plateau

The Tibetan Plateau (TP) is an important component of the global environmental system, on which the snow cover greatly affects the regional climate and ecology. Moderate resolution imaging spectroradiometer (MODIS) snow cover products have been demonstrated to be appropriate for investigating the snow cover over the TP. However, they are subject to cloud obscuration, and the TP’s extremely complex terrain makes the snow monitoring difficult. Therefore, in this paper, we propose a two-stage spatio–temporal fusion framework for the cloud removal of MODIS C6 snow products, including an adjusted Terra and Aqua combination (TAC) and a spatio–temporal fusion based on Gaussian kernel function and error correction (STF-GKF-EC). To the best of our knowledge, this is the first time that a spatio–temporally continuous daily 500-m MODIS normalized difference snow index (NDSI) product has been generated for the TP, which greatly improves the spatial and temporal resolutions of the current snow cover products. The main stage, STF-GKF-EC, adaptively weights the spatial and temporal correlations by the Gaussian kernel function, and further takes the rapid changes of snow cover into consideration through the error correction. The experiments indicated that STF-GKF-EC removes clouds completely, achieving an overall accuracy (OA) and mean absolute error (MAE) of 91.48% and 3.88, respectively. Based on the cloud-removed results, during 2001–2017, as far as the intra-annual variation is concerned, a large proportion of the snow cover appears between October and May, with a peak in February/March, and the variation is mainly controlled by temperature. For the inter-annual variation, an obvious increasing trend of 0.68/year for NDSI is observed before 2005, followed by a slight decreasing trend of 0.16/year, in which precipitation is a better explanation factor than temperature.

Cloud Masking for Landsat 8 and MODIS Terra Over Snow-Covered Terrain: Error Analysis and Spectral Similarity Between Snow and Cloud

Automated, reliable cloud masks over snow-covered terrain would improve the retrieval of snow properties from multispectral satellite sensors. The U.S. Geological Survey and NASA chose the currently operational cloud masks based on global performance across diverse land cover types. This study assesses errors in these cloud masks over snow-covered, midlatitude mountains. We use 26 Landsat 8 scenes with manually delineated cloud, snow, and land cover to assess the performance of two cloud masks: CFMask for the Landsat 8 OLI and the cloud mask that ships with Moderate-Resolution Imaging Spectroradiometer (MODIS) surface reflectance products MOD09GA and MYD09GA. The overall precision and recall of CFMask over snow-covered terrain are 0.70 and 0.86; the MOD09GA cloud mask precision and recall are 0.17 and 0.72. A plausible reason for poorer performance of cloud masks over snow lies in the potential similarity between multispectral signatures of snow and cloud pixels in three situations: (1) Snow at high elevation is bright enough in the "cirrus" bands (Landsat band 9 or MODIS band 26) to be classified as cirrus. (2) Reflectances of "dark" clouds in shortwave infrared (SWIR) bands are bracketed by snow spectra in those wavelengths. (3) Snow as part of a fractional mixture in a pixel with soils sometimes produces "bright SWIR" pixels that look like clouds. Improvement in snow-cloud discrimination in these cases will require more information than just the spectrum of the sensor's bands or will require deployment of a spaceborne imaging spectrometer, which could discriminate between snow and cloud under conditions where a multispectral sensor might not.