What are snow reserves and how important is it for water storage?
The phenomenon of climate change in recent years has affected hydrological processes such as precipitation regime, evaporation from the water surface, groundwater recharge, runoff rate, and snowmelt (Ezzati et al., 2016). Snow-covered areas, as an important component with a wide distribution on the earth's surface, have revealed seasonal changes and shown great sensitivity to climate change. It is worth noting that snow, due to its high albedo, heat, and water storage properties, plays an important role in the world's energy and water cycles and can indicate global climate changes in recent decades due to climate change that we have witnessed in snow-covered areas (Groisman et al., 1994; Wang and Li, 2006). Snowfall has occurred (Fattahi and Moghimi, 2018) and has also caused changes in the resulting runoff regime at different scales (Al et Kang, 2021). Snow cover, in addition to having a significant impact on climate change, is considered an important source of water for Asian rivers and affects hydrological and biological processes (Immerzeel, 2008a; Zhang et al., 2012; Tang et al., 2013 It is noted that more than one-sixth of the world's population relies on water from mountain snowmelt (Barnett et al., 2005). Since Iran is located in the arid and semi-arid region of the Earth and faces frequent water stress, snowfall is considered the most important factor affecting the amount of water reserves in this country and is one of the important factors controlling the hydroclimate of watersheds (Miryaghoubi and Qanbarpour, 2010). Snow cover in many regions of the world has a direct impact on human life, including engineering, irrigation, travel, recreation, and hydrology (Tsai et al., 2019). As mentioned, snow is one of the important forms of precipitation in the hydrological cycle of mountainous areas, which is used to provide drinking water resources and agriculture in the form of delayed flows in High water seasons and minimum flows play a valuable role in dry seasons (Goodinson et al., 2000al). So that the water equivalent of snow cover provides about one-third of the water needed by the agricultural sector, soil moisture, groundwater storage, and water resources of lakes and rivers around the world (Zhai and Gray, 2002; Raispour, 2018). In addition to the above, another important aspect of snow is its interaction with vegetation. The melting of this water source helps the survival and fertility of vegetation (Broxton et al., 2015; Adam et al., 2009). In this regard, studying snow cover changes is of great importance in monitoring and maintaining water management for ecosystem processes and irrigation methods (Dahe et al, 2006). According to meteorology and climate experts, snow monitoring is a necessity, because the physical properties within the snow affect daily and even long-term climate changes in the region (Ghulam and Bashir, 2010). Monitoring snow cover is the best solution to investigate spatial changes in this phenomenon and its regional precipitation distribution.
Snow cover can be measured using measurement stations, modeling, remote sensing technology, and applications (Li et al., 2015al et Lindsay). Although ground stations provide experts with accurate information from the measurement location, unfortunately, due to the lack and in many cases the absence of meteorological stations, sufficient information cannot be provided to produce long-term snow data at a spatial scale. Therefore, in terms of spatial scale, there are always limited encounters (Li et al., 2018; Zhou et al., 2013al). Although monitoring the spatial and temporal characteristics of snow using modeling is also possible, the lack of information reduces the accuracy and precision of modeling results (Marks and Garen, 2005). Over the past decades, advances in remote sensing have partially overcome some of the problems associated with snow measurement, such as poor station distribution, lack of access to high-altitude areas, and missing data. Recent advances in remote sensing technology have enabled satellite data to extract snow cover information in inaccessible areas.Rugged terrain and harsh climates have made
Jain et al., 2009; Immerzeel et al., 2009; Rittger et (
. )al., 2013; The high reflectivity of snow compared to most surfaces has made satellites a suitable tool for measuring snow cover (Yang et al., 2015), In recent years, the use of machine learning (ML) and artificial intelligence (AI) models in remote sensing and GIS has been very useful for forecasting and generating probability maps (Arabameri et al., 2021), and has been used with great speed and variety (Gholamnia et al., 2020). Among the existing models, artificial neural network models are devices or software that are organized based on the neural structure of the human brain (Garg et al., 2021) and exhibit behaviors that are similar to those in the functioning of the human brain or that can be interpreted as one of the human behaviors (Massaudi et al., 2021). Studies show that this network They have the ability to learn, remember, forget, infer, recognize patterns, classify information, and many other skills of the human brain (Kalogirou & Mellit, 2021) and have the ability to use this model to predict and simulate climate data (Moghanlo et al., 2021). Studies have been conducted in Iran and the world in the field of snow level monitoring and forecasting, some of which are mentioned below: Joha and Khalsa (2007), in a study, assessed seasonal snow cover for 8 basins in Central Asia using MODIS sensor images and concluded that the greatest decrease in snow cover belongs to the Shantian region and the reason for the variation in snow cover is due to variable rotational patterns during the winter (Khalsa, 2008). Markan et al. (2015), in a study, assessed daily snow cover products snow in the Atlas Mountains of Morocco, the results of the studies showed that the daily MODIS product can be used with reasonable confidence for snow cover in the southern Mediterranean region despite difficult detection conditions (Al et Marchane, 2015). Darian et al. (2017), in a study they conducted for the central Alborz mountainous region in northern Iran, compared three changes in snow levels between 2002 and 2015 and, by developing a sequential cloud removal algorithm, estimated the snow level more accurately using MODIS satellite images. Their research results showed that the snow cover of the central Alborz mountainous region decreased sharply in a short period (13 years) (Al et Dariane, 2017). Mir Mousavi and Sabour (2012) conducted a study to monitor snow cover changes using MODIS imagery in northwest Iran. The data used in this study are MODIS satellite imagery of northwest Iran from 2000 to 2009. The method used in this study is the NDSI index, unsupervised and supervised classifications. An examination of the maps related to snow cover changes in April showed that during the study period, the lowest snow cover amount was in 2008 and the highest area was in 2007. This indicates a 1000 percent change over a decade in The extent of snow cover in the northwestern region of Iran indicates the vulnerability of water resources dependent on snowmelt in this region.The results also showed that in years when the average cold season temperature is lower, the snow cover level in spring of that year is higher than other years. Saboor and Mirmousavi (2012). Dadashi et al. (2014) in a study calculated the snow level using MODIS sensor images in Tehran province. They used MODIS sensor images from 2014 and the NDSI method to advance the research goals. The results of this study showed that the snow cover index along with spectral thresholding on bands 4 and 6 provides a stable relationship in extracting the snow cover map. In this study, the extracted snow area and its results were compared with the area obtained from NASA images. Dadashi et al. (2014) Jafari (2016) in a study using effective variables such as latitude, temperature, amount and direction of the slope of the land surfaces in The country's radiometric stations tried to estimate the height of the permanent snow line. Given that the amount of energy received from the sun varies if the Earth is flat,There are not many 16 climate stations used for the whole of Iran. Then, using the temperature situation and the angle of the sun's elevation at the stations and the changes in the angle of the sun's elevation on sloping surfaces with different directions and the resulting temperatures, they established a relationship and used these relationships to estimate the height of the permanent snow boundary or the water and ice balance line. The initial results indicate that in the presented method, only by using the latitude of the location, direction, slope and longitude (to identify the climatic station of the region), it is easy to estimate the height of the permanent snow boundary, and there is no longer a need for long Wright methods and topographic maps (Jafari, 2016). Karbou et al. (2021), in a study in the French Alps region, used Sentinel 2 satellite images to identify the snow boundary in the period 2017 to 18 and identified the snow boundary and then used Sentinel 1 radar images to identify the snow depth and the water flow produced from it (Karbou et al., 2021). In this study, the aim is to investigate the trend of changes in snow levels in terms of time and space in Mazandaran province over an 18-year period using MODIS satellite images. And its coverage area was extracted, then the areas that caused changes in snow boundary during the studied time period were identified using the change detection technique and introduced as dependent variables in the MLP neural network model. In the following work, the climatic, environmental and geomorphological variables were evaluated and the changes in this trend were predicted for the 1429 time period, and then its accuracy was examined using the ROC curve method.