Extended Abstract
Introduction
At the local scale, snow cover is influenced by land features and affects water availability in some arid and semi-arid regions such as Iran. Since snowmelt plays an important role in total runoff (Azizi & Akhtar, 2022), understanding the extent and pattern of SCA melt in mountainous regions, including the Zagros highlands (the source of the Karun River), is of great importance. At the regional scale, snow accumulation and melt processes are mainly dominated by latitude and air temperature (Lopez-Moreno et al., 2014) and local topography factors (Schmidt et al, 2009). Therefore, a more comprehensive study is necessary to clarify the relationship between SCA and land features. Monitoring snow cover characteristics based on ground observations is a difficult and costly task that is limited in area (Liu et al., 2-17). Recent advances in remote sensing techniques and satellite capabilities make it possible to monitor snow in data-sparse areas, such as complex mountainous areas and cold regions (Frei et al., 2012). Traditional methods using satellite images to determine snow cover used snow indices such as NDSI and S3. it is difficult to determine the correct threshold value. Therefore, it is sometimes easier to let the classifier decide this problem (Nurzyńska et al, 2013). Therefore, the aim was to investigate the performance of the system with more complex classifiers. Since the distribution of snow particle features is unknown, classifiers that do not require this information, such as the support vector machine (SVM) algorithm, are suitable (Nurzyńska et al, 2013). This paper presents an analysis of the distribution and duration of snow cover in the North Karun River watershed using a topographic effect approach.The innovation of this research lies in combining advanced remote sensing methods with multiparameter topographic analysis in one of the most important snow basins of the Zagros.
Methodology
The present study used Lansat-8 images for a 5-year period (2019 to 2024) to quantify the spatial heterogeneity of snow cover due to topographic effects in the North Karun watershed. Here, the SVM machine learning approach was used to classify the images. Two performance metrics, including kappa coefficient and overall classification accuracy, were used to evaluate the performance of the SVM model..
Results and Discussion
The results of the assessment of the accuracy of snow cover mapping in the 5-year period using the SVM method are acceptable with an overall accuracy and kappa coefficient higher than 0.9. The estimated average annual SCA was calculated with a maximum area of ​​140,597.37 hectares (13.25%) and a minimum area of ​​63,957.82 hectares (6.03%) in the years 1398-1399 and 1399-1400, respectively. The seasonal variation analysis showed that the maximum average SCA was (70.44%) in winter and the minimum (1.34%) in summer. The results of snow cover in relation to altitude above sea level showed that the average SCA% below 2000 m above sea level was less than 2%, while above 2500 m it reached 70%. The lowest SCA was below 2500 m above sea level in summer. The results of the slope effect analysis showed that the maximum and minimum SCA% were found in the slope classes 20-40 and 77.33-60, respectively. Also, the snow cover analysis showed that the north-facing (33.75%) and east-facing (27.09%) directions had the largest snow cover extent, and the south-facing (19.98%) and west-facing (19.18%) directions had the lowest snow cover extent.
Conclusion
The results of the snow cover survey in the North Karun watershed located in Chaharmahal Bakhtiari province showed The average monthly SCA was maximum in Bahman (34.28%) and minimum in Mehr (0.05%). From Mehr to Bahman, positive changes were observed in the region. From Esfand to Shahrivar, the change in snow cover was negative, which means that the melting of snow cover accelerates in Farvardin and Ordibehesht, especially at low altitudes, and continues throughout the summer season. Factors such as elevation differences from sea level have a significant impact on the temperature distribution and snow accumulation/melt in the region, and the higher SCA corresponds well with the high Zardkouh mountain range. The slope and direction of the mountain also affect the distribution of snow cover through changes in radiation and energy balance in mountainous areas. The dataset and results of this study can be used in developing a model for runoff prediction and in assessing the impacts of climate change, and also provide useful insights into regional snow cover variance and can be a guide for managing snowmelt water resources with different terrain characteristics.