What is snow cover climatology? Can it be obtained through ERA5 Land reanalysis data and snow cover?

Understanding the distribution, patterns, and characteristics of snowfall and snow cover within a given region over extended periods is important. Snow climatology provides valuable insights into the seasonal and long-term variations in snowfall, helping researchers and meteorologists understand the impacts of climate change on snow accumulation, melt rates, and snowmelt runoff. In this study, in order to understand the spatial and temporal variation in snow cover in Turkey, the temporal and spatial dynamics of snow cover in the country were analyzed during the latest and longest period from 1970 to 2022 using ERA5-Land reanalysis product. It is aimed (1) to show snow-covered area (SCA), snow duration, and snow depth trends over the country; (2) to examine the altitudinal difference of snow phenology response to climate change; and (3) to evaluate the Snow Cover Frequency Maps from MODIS Snow Cover Products with the reanalysis snow depth data. It is found that the “false snow” mapping problem still exists in the MOD10C1_CGF Snow Cover Frequency maps over Turkey, especially in the melting period. We found that an increasing trend of 0.4 ◦C/decade and snow duration have a decreasing trend due to the early melting between 1970 and 2022. This trend is even more noticeable at elevations below 2000 m. Another important finding is the decreasing trend in snow duration at altitudes below 500 m, indicating a shift from snow to rain for precipitation types. Approximately 98% of the seasonal snow cover on the Earth is located in the Northern Hemisphere [1]. Seasonal snow cover, highly responsive to precipitation and temperature fluctuations, exhibits significant spatial and temporal variations in snow mass distribution, influencing freshwater resource recharge, runoff patterns, and near-surface temperature cooling [2]. With climate change driving non-uniform shifts in snowmelt and precipitation, it impacts ecosystems, populations, and economic activities, even in regions devoid of snow. Furthermore, seasonal snow significantly influences the carbon cycle by insulating soil, affecting vegetation growth, and altering carbon fluxes in frozen soil and permafrost areas [3]. The presence of seasonal snow is a crucial element in the global climate system, affecting surface energy, hydrology, heat exchange, and ecosystems. Its high albedo and low thermal conductivity directly impact land surface energy balance, soil temperature, and atmospheric circulation. The Northern Hemisphere heavily relies on snowpack as a natural water storage, making accurate assessment of its spatial and temporal changes essential for climate monitoring, model evaluation, and water resource management [2,3].In the hydrological cycle, snow represents seasonal water storage from where water is rapidly released during the melting period. Seasonal snow cover in the mountainous areas accumulates much of the water that becomes streamflow, fills water supply reservoirs, and recharges critical groundwater aquifers in the spring and summer months. Snowmeltdriven water is not only demanded by nearly one sixth of the world population as fresh drinking water supply, but it also supports other usages such as industrial, hydro-power, and irrigation applications [4]. Therefore, snow-covered area (SCA) is an important hydrologic variable for streamflow prediction [5], and observations of areal extent have been used in some hydrologic forecasts for decades [6]. Spatiotemporal alterations in the extent of snow cover due to the changing climate are considered to have significant socio-economic impacts in the near future [7,8]. Globally, temperature and precipitation patterns are predicted to change markedly as a result of climate change. Particularly, the regions with a cold or hot semi-arid climate and the Mediterranean climate zone are expected to be strongly affected. A 25–30% decrease in precipitation and increased evaporation are expected by the end of the 21st century in the Mediterranean region, to be accompanied by an even stronger reduction in runoff of up to 30–40%. Climate change influences the snow-dominated basins’ hydrology in two ways. First, climate change causes decrease in runoff amounts, which leads to water insufficiency. Second, it results in shifts in melting times, leading to early floods and/or summer droughts. Owing to increasing air temperatures, snow has begun to ablate 8 days earlier in northern Alaska compared to the mid-1960s [9]. In addition, declines and earlier occurrences of the maximum snow water equivalent (SWE) have been observed in the Cordillera of western North America [10–13]. Wang et al. [14] evaluated the spatial and temporal dynamic variations in snow in the Northern Hemisphere from 2000 to 2015. They found out that there was a downward fluctuating trend in the variation in snow cover in the Northern Hemisphere. The snow depth and snow cover duration days have decreased across the Northern Hemisphere over the past few decades [15–19]. Reanalysis products are of great importance to improve our understanding of the cryosphere and its interaction with the climate in places where ground observations are limited or do not exist. ERA5-Land reanalysis product [20] is an integral and operational component of the Copernicus Climate Change Service. Snow depth estimates from ERA5- Interim, ERA5, and ERA5-Land are compared with two sets of observations where the sites are distributed among North America, Europe, and Japan in Muñoz-Sabater et al. [20].In that study, it is presented that ERA5-Land shows lower RMSE over the sites with moderate altitude, i.e., between 1300 and 2500 m, and, with its higher horizontal resolution, it provides a better orographic representation. In the same study, it is also revealed that ERA5 performs better compared to ERA5-Land for heights above 3300 m. In another study, Varga and Breuer [21] found that the ERA5 product represents snow depth remarkably well, with correlation coefficients above 0.9 over a low altitude Central European region. They showed that ERA5-Land overestimates daily mean SD by 2–3 cm for some stations and displays lower correlations (0.7–0.9) during the 26-year time span. Kouki et al. [22] analyzed time series of snow water equivalent (SWE) and snow extent (SCE) in ERA5 and ERA5-Land reanalysis data to compare the time series with several satellite-based datasets in the northern hemisphere in spring 1982–2018. They found that SCE is accurately described in ERA5-Land, whereas ERA5 shows notably larger SCE than the satellite-based datasets. Snow depth retrieved from Sentinel-1 and simulated snow depths of ERA5 and ERA5-Land are compared over the Tibetan Plateau (TP) by Lei et al. [23]. They found out that ERA5-Land matches in situ observations better than ERA5 over the TP. Sahu and Gupta [24] utilized both atmospheric reanalysis and satellite-based snow products to reveal the relation between climate change and long-term snow cover trends in Chandra Basin, Western Himalaya region. Annual and seasonal variations in snow cover extent over the area between 2001–2017 were realized using Moderate Resolution Imaging Spectroradiometer (MODIS) 8-day snow cover product MOD/MYD10A2 (~500 m) together with precipitation and temperature data from ECMWF’s ERA5 reanalysis dataset (~0.25◦ ).The results indicated a decreasing trend in SCA with increasing temperature, showing a higher rate of correlation as compared to precipitation. ERA5 was also considered as a reliable alternative in long-term trend analysis in case of insufficient in situ observations. Haag et al. [25] also investigated the scope of local climate change and its impacts on the subsistence-based communities over a High Asia region in the Pamir Mountains of Tajikistan, namely, the Savnob and Roshorv villages. They used a holistic perspective by analyzing the statistical trends on temperature and snow cover using downscaled ERA5 temperature data and the MODIS snow cover product MOD10A1 together with the community observations. Monthly temperature trends between 1979 and 2018 were derived from the elevation-corrected [26] and downscaled ERA5 gridded temperature product. The accuracy of the newly created ERA5 temperature dataset was assessed with in situ observations from five ground stations. In order to follow the local snow cover trends over the study region, temporal gap-filled MOD10A1 snow cover data from 2001 to 2018 were employed. To analyze the local community’s observations and opinions on local livelihoods, ecological events, and environmental and climatological conditions such as temperature, precipitation, and snow cover, synchronous research activities composed of workshops and interviews were also held. The elevation-corrected ERA5 temperature dataset indicated a statistically significant increase in summer temperatures in both Savnob and Roshorv with 0.32 ◦C per decade and 0.38 ◦C per decade, respectively. Even though it was not statistically significant, MOD10A1 snow cover data over the region revealed an increasing trend in the timing of snow onset and a decreasing trend in the timing of snow offset, indicating a shortening of the snow season. However, in Savnob Village between 2001 and 2018, a statistically significant shortening of the snow period was observed, i.e., 5.4 weeks. The results of the workshops and interviews also pointed to an earlier time of harvest, earlier start of fieldwork, decreasing demand for heating resources, and earlier melting of snow for both villages. Ma et al. [27] explored the spatiotemporal variation and driving mechanisms of snow depth and snow phenology on the TP from daily passive microwave remote sensing snow depth dataset and meteorological data for the period 1979–2020. The findings indicated that the TP exhibited significant elevation-dependent warming (EDW) at 0.04 ◦C per decade per kilometer, along with decreasing maximum snow depth by 0.12 cm and snow cover duration by 4.1 days per decade per kilometer. EDW correlated with reduced maximum snow depth and snow cover duration. This feedback could exacerbate EDW and earlier snowmelt, potentially increasing droughts and floods. In Li et al. [28], snow depth and snow cover trends over the Tianshan Mountains in the High Asia region were evaluated by employing and comparing four snow depth datasets from ERA5, ERA5-Land, passive microwave (PMW) (i.e., Special Sensor Microwave Imager (SSM/I) and SSSMI Sounder (SSMI/S)), and dynamically downscaled simulation by Weather Research and Forecasting model (TSS) during the period from 1981 to 2018. The study assessed the snow depth and snow cover from these datasets against in situ meteorological data and interactive multisensor snow and ice mapping system (IMS) snow cover datasets over the region. Then, the spatial patterns and temporal variations in snow-related metrics (i.e., annual mean snow depth and snow cover days) based on the four datasets were compared. The results revealed that although there existed some discrepancies among the four datasets, they were all able to catch the related spatial patterns of snow-related metrics over the study area, indicating a decrease in annual mean snow depth and snow cover days from the north to the east across the Tianshan Mountains. While the ERA5 and ERA5-Land were the only datasets that revealed a significant reduction in annual mean snow depth over the entire study, ERA5, ERA5-Land, and TSS indicated a significant decreasing trend in snow cover days. Alonso-González et al. [29] studied the snowpack dynamics over Lebanese mountains through the assimilation of MODIS fractional snow-covered area (fSCA) product with ERA5 reanalysis covering the period 2010–2017. The generated 1 km regional-scale snow reanalysis (ICAR_assim) was realized using an ensemble-based data assimilation of MODIS fSCA through an energy and mass snow balance model (i.e., the Flexible Snow Model,FSM2). The necessary boundary and initial conditions required by the regional atmospheric simulation (i.e., Intermediate Complexity Atmospheric Research model, ICAR) were provided by the ERA5 atmospheric reanalysis product. The resultant SWE products exhibited a high degree of agreement with MODIS gap-filled snow cover data, with R = 0.98 and RMSE = 3.0%. The results also highlighted the high temporal variability in the snowpack over the Lebanese mountain ranges, where the most important snow freshwater reservoir resides in the middle elevations (2200–2500 m). The usefulness of data assimilation in snow extent studies in data-scarce regions was also emphasized. As the Mediterranean basin is considered being one of the most adversely affected regions in the near future due to climate change, according to the simulations regarding the future climatic conditions [30,31], several studies also focused on the spatiotemporal trends in the hydroclimatic variables over Anatolia and the Near East Region by assimilating atmospheric reanalysis products and satellite-based observations. The trends in near-surface air temperature, precipitation, SWE, runoff, and evapotranspiration between 1979 and 2010 over Turkey were evaluated in Gokmen [32] using ERA-Interim and ERA-Interim/Land reanalysis products. The temperature trend analysis from 1979 to 2010 indicated an average warming of 1.26 ◦C in Turkey, where the largest warming was observed in the western coastal areas next to the Aegean Sea and in the southeastern regions. These trends in air temperature were supported also by the ground measurements from about 100 weather stations, yet they exhibited slightly higher trends ranging from 1 to 2.5 ◦C. Regarding the precipitation, SWE and runoff trends during the same period, resulting from ERA-Interim and ERA-Interim/Land, were quite different. While the ERA-Interim dataset indicated significant decreasing trends on these parameters in some parts of inner/southeastern Anatolia, ERA-Interim/Land showed no or minor trends over the same regions. Extensive comparisons with precipitation and SWE gauge data suggested that hydrological trends shown by the ERA-Interim/Land dataset were relatively closer to the observations. The overall results indicated no widespread and strong hydrological trends throughout the country from 1979 to 2010, despite the strong warming trends observed for the same period.Another recent study that investigated the long-term behavior of seasonal snowpack in the eastern parts of Anatolia comprehensively was performed by Yılmaz et al. [33]. The study geographically covered the land area confined between the Mediterranean Sea, Black Sea, Caspian Sea, Arabian Sea, and Red Sea, i.e., the Near East. The study employed several remote sensing products (GRACE: liquid water equivalent thickness anomaly for, MODIS Aqua/Terra: snow-covered area, AMSR-E: SWE, AMSR2: SWE and snow depth, SRTM30: elevation) in combination with atmospheric reanalysis datasets (ERA5: 2 m air temperature, precipitation, snow energy, and mass balance) to reveal the relationship between the terrestrial water storage anomalies and the mountain snowpack over the four important snow-fed river basins of the Near East region, i.e., Euphrates and Tigris basin, Kura-Araks basin, Çoruh basin, and the Van Lake basin. The GRACE dataset covered the period between 2002 and 2017, whereas MODIS Aqua and Terra snow-related datasets spanned 2002–2018 and 2000–2018, respectively. SWE data from AMSR-E between 2002 and 2011, and from AMSR2 between 2012 and 2018 were used. The parameters retrieved from the ERA5 product were from 2002 to 2018. Monthly water storage anomalies derived from the GRACE dataset indicated an increasing trend in the water loss in the whole Near East region, especially after the severe drought in 2007. Results of the analysis on the terrestrial water storage anomalies over these four basins showed relatively higher negative water storage trends at higher altitude ranges (above 1000–1500 m). Snow-covered day analyses from MODIS showed negative trends over high mountainous areas in all basins with an average rate of ~4 weeks/decade. No direct relationship was observed between the seasonal temperature climatology and rainfall/snowfall events. Instead, the rate of change in snowfall events showed seasonal and regional variations, indicating a difference in the response patterns of the snowpack in the Mediterranean mountains to the changing climate.There are also several studies that put emphasis on similar declining trends on the snow cover extend over the US. Pederson et al. [34] conducted an analysis of the snow water equivalent (SWE) in the western United States. Their investigation revealed that, commencing from 1980, there has been a synchronous and overall diminishing trend in both the northern and southern Rocky Mountains, which can be attributed to the increasing springtime temperatures. In the period preceding 1980, the snowpack had typically displayed a dipole pattern, characterized by opposing anomalies in the northern and southern regions. Lute and Abatzoglou [35] investigated the role of specific extreme snowfall events in influencing the seasonal SWE in the western United States. They characterized snowfall events as the cumulative accumulation of SWE over 3-day periods. Their research revealed that the highest decile of these events accounted for a substantial 69% of the inter-annual variation in snowfall water equivalent. In the study conducted by Harpold et al. [36], an analysis of the SNOTEL SWE data for the central and southern Rocky Mountains spanning the years 1984 to 2009 was undertaken. The findings of this study unveiled extensive reductions in both the maximum SWE values and the duration of snow cover in the region. Taken together, these studies consistently demonstrate a decline in snow cover on the ground, with some of the most extreme reductions observed within the past decade to 15 years. This aligns with the anticipated outcomes resulting from the direct impact of rising temperatures. The MODIS snow cover product is a vital component of remote sensing technology designed to monitor and analyze snow cover extent and the changes on the Earth’s surface. Onboard NASA’s Terra and Aqua satellites, MODIS captures high-quality imagery with moderate spatial resolution, allowing for the accurate assessment of snow-covered regions. By utilizing different spectral bands, the MODIS snow cover product can distinguish between snow, ice, and other land cover types, enabling researchers and decision-makers to track the dynamic variations in snow cover over time. Recently developed MODIS snow cover frequency maps at ~5 km spatial resolution for the 20-year period (i.e., 2001–2020) provide a unique source of information on snow climatology [37]. In this study, in order to understand the spatial and temporal variation in snow cover in Turkey, the temporal and spatial dynamics of snow cover in the country was analyzed during the latest and longest period from 1970 to 2022 from ERA5-Land reanalysis products It is aimed (1) to show SCA, snow duration, and snow depth trends over the country; (2) to examine the altitudinal difference of snow phenology response to climate change; and (3) to evaluate the Snow Cover Frequency Maps from MODIS Snow Cover Products with the reanalysis snow depth data. 2. Materials and Methods 2.1. Study Area Most of the country is situated on the Anatolian plateau, and it is bordered by the Mediterranean Sea to the south, the Aegean Sea to the west and the Black Sea to the north. The country has a total of 783,562 km2 land surface area and lies between 36◦–42◦ N latitudes and 26◦–45◦ E longitudes. In addition to its typical mild mid-latitude Mediterranean climate, dramatically different climate conditions (dry mid-latitude steppe, temperate continental, and oceanic) prevail due to the presence of complex morphology and highmountain ridges. The mean elevation is 1132 m, whereas the maximum elevation is 5137 m (Figure 1). In general, elevation is high, and altitude differences are large in the north-east, east, and south Anatolia. The west and south-east have lower altitudes. The coastal areas possess milder climates, but North Anatolian and Taurus mountains prevent marine effects penetrating to the inland areas by positioning parallel to the sea [38]. Therefore, the Anatolian plateau receives limited precipitation and has a continental climate with hot and dry summers and cold and snowy winters [39]. The annual average temperature and precipitation values of Turkey for 1970–2022 are 13.3 ◦C and 618.9 mm, respectively [40,41]. The Köppen climate classification map and the main land use/cover maps are presented in Figure 1b,c. The majority of the land use/cover is shrubland, grassland, and woodland covering the mountainous regions. The high altitudes of eastern part of country, Anatolian plateau is covered with grassland. At lower altitudes, cropland is seen.e the mean values). 4. Conclusions This study has shown that MOD10C1_CGF Snow Cover Frequency maps, one for each day of the year, can be used to retrieve the snow climatology over a complex terrain. The validation of the maps was performed with the ERA5-Land reanalysis dataset, which covers the latest and longest period from 1970 to 2022. Although MOD10C1_CGF Snow Cover Frequency maps were derived from the coarse resolution ~5 km daily MODIS CMG cloud-gap-filled and “false snow” filtered product, with an additional spatial filter, the “false snow” mapping problem is still persistent. After deleting the outliers, the maps present the snow climatology well. Knowing the discrepancies in the reanalysis products, namely, ERA5 and ERA5-Land with the satellite-based datasets, both are able to capture the interannual variability quite accurately. ERA5-Land reanalysis product is found to be valuable to understand the spatial and temporal variation in snow cover in Turkey. The temporal and spatial dynamics of snow cover in the country was analyzed during the latest and longest period from 1970 to 2022.