Introduction

Understanding the spatial distribution and temporal evolution of seasonal snow depth, defined as the vertical distance from the snow surface to the snow base, is essential for water resource managers, climate researchers, local governments, and avalanche forecasters. However, quantifying snow depth is challenging, especially in mountainous regions, due to the multiscale nature of the physical processes governing the snow depth distribution. The spatio-temporal patterns of snow depth variation in steep and complex mountainous terrains have not been fully explored, thus, an attempt has been made to determine the spatial statistical patterns governing the annual average as well as the performance of the fifth generation of ECMWF/ERA5 atmospheric reanalysis data to estimate the monthly average snow depth in five provinces located in the northwestern region of Iran (1982-2022) should be evaluated during different decades along with estimating the values affected by altitude, snowfall, temperature and slope direction. The results of this research, through quantification in high spatial and temporal resolutions, for snow depth variation and snow pack changes of relatively inaccessible mountain ranges, have a high potential in reducing the uncertainty of analysis related to snow and snow water resources.

Methodology

The northwest of Iran is located between 34 degrees and 44 minutes to 39 degrees and 25 minutes of north latitude and 44 degrees and 3 minutes to 49 degrees and 52 minutes of east longitude. In order to investigate the spatial autocorrelation changes of the average snow depth in northwest Iran during the years 1982-2022, from the data obtained from the ECMWF/ERA5 database based on daily data, and to identify and understand the spatial patterns of snow depth, from the statistical and graphic models based in the information system environment used geographically. In the study of temporal-spatial changes of the average snow depth of the region in different time periods including 4 decades ((1991-1982), (2001-1992), (2011-2002), (2012-2022)) and the whole period of 41 years (2022- 1982)), Moran's general spatial autocorrelation indices (Moran's I) and Getis-Ord Gi* statistic were used. In order to investigate the effect of snowfall and temperature on snow depth, using inferential statistics methods such as Man-Kendall, Sen's slope and linear regression, the trends of changes in average minimum temperature and monthly snowfall were investigated.

Discuss

The results of the present research showed that in the studied area, despite the decrease in the average monthly snow depth in autumn to 0.022 and its increase in winter to 0.402 meters of water equivalent (especially in February equal to 0.107 meters of water equivalent), the first decade of each of the six investigated months has the highest amount of snow depth, but with the passage of time, the extent (number of pixels) and the amount of snow depth have decreased. Also, snowfall, especially in November to March, has more effect on snow depth than temperature. During recent years, especially in the fourth decade, the occurrence of freezing rains justifies the noticeable increase in snow depth. In the changes of snow depth within decades, the effect of altitude, including the direction of the northern slope, was clearly visible, and the highest values of snow depth in each of the decades were concentrated in high-altitude areas. However, in general, the amount of snow depth in the entire northwest area has decreased significantly during the last four decades. It was also found that the snow depth of the region has a spatial autocorrelation and a strong cluster pattern. In other words, in the pattern governing the annual average snow depth of the region, there is more tendency towards the formation of cold clusters containing low values.

Conclusion

The evaluation of temporal changes of snow depth also strengthened the hypothesis of the occurrence of snow drifts leading to an increase in snow cover and reciprocally increasing the depth of snow in the months of the cold seasons during recent years, especially in the fourth decade. The results of the analysis of the changes in precipitation in the 99th percentile showed that the amount of this type of precipitation has increased significantly during the last decade of the study, and this led to a relative increase in the depth of snow in the last decade compared to previous decades, although in general, the amount of snow depth Snow in the entire northwest has decreased significantly during the studied period. The results obtained from Moran's index and zoning of hot spots show that in all the studied periods, the cluster pattern and hot spots of snow depth are almost concentrated on the high mountain areas. Cold clusters (including low values of snow depth) are also observed more often in the low-altitude areas of the northern part of the region and especially in the areas around Lake Urmia, and often the extent of negative spatial autocorrelation at the 95% confidence level is greater than positive spatial autocorrelation, in other words, the probability of occurrence shallow snow values (cold clusters) were higher.