Document Type : علمی - پژوهشی
Authors
1
PhD Student in Agricultural Meteorology, Water Engineering Department, Sari University of Agricultural Sciences and Natural Resources, Sari, Iran
2
Department of Water Engineering, Faculty of Agricultural Engineering, University of Sari Agricultural Sciences and Natural Resources, Sari, Iran
3
Associate Professor, Department of Water Engineering, Sari University of Agricultural Sciences and Natural Resources, Sari, Iran
4
Associate Professor, Department of Water Engineering, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran
5
Professor, Department of Watershed Science and Engineering, Sari University of Agricultural Sciences and Natural Resources, Sari, Iran
Abstract
Introduction: Estimation of spatiotemporal patterns of precipitation is difficult in Mazandaran province due to several factors, including the lack of ground observations, diverse climate zones, and extreme mountain slopes. Satellite precipitation products can provide a suitable solution for measuring the amount of precipitation, especially in areas with scattered ground stations, and provide a new approach to observe precipitation globally with remote sensing. However, despite the wide use of these products in various fields of study, the quantitative evaluation of these products is a fundamental challenge due to their inherent error and uncertainty, which should be considered in different temporal and spatial scales before their direct use.
Materials and methods: In the first step, CHIRPS, CMORPH, SM2RAIN, PERSIANN-CDR and IMERG gridded precipitation products were extracted from the database of each product in NetCDF format at the global level. Then, precipitation data for each product was selected for grids located in Mazandaran province by coding in the R programming environment and Geographic Information System (GIS). In the next step, the evaluation and comparison of these products against 15 synoptic stations in the Mazandaran province at the station-grid and regional spatial scales and monthly and annual time scales using the statistical evaluation criteria of Spearman's correlation coefficient (CC), Root Mean Square Error (RMSE), Mean Absolute Error (MAD), Kling-Gupta efficiency (KGE) and Nash-Sutcliffe efficiency (NSE), as well as drawing statistical diagrams (Taylor, etc.) were performed. In order to evaluate the grid to the station, the data of the closest grid to each synoptic station was extracted for each precipitation product and compared with the precipitation data of station in different time scales.
Results and discussion: The results of the regional evaluation of satellite precipitation products on a monthly and annual scale have shown that the IMERG and CMORPH products are more compatible, while the CHIRPS product performed well only in the dry months of the year. However, the accuracy of the products was higher on a monthly scale than on an annual scale. The Taylor diagram results indicated relatively high accuracy of IMERG, CHIRPS and CMORPH products (correlation 0.8-0.7) at stations located in coastal areas and relatively low accuracy at high altitudes (correlation 0.35). While for the PERSIAN product, the data accuracy was low for all regions, with a negative correlation in coastal areas. However, the results of the evaluation of precipitation products at the station level showed the better performance of the IMERG product and then CMORPH (CC=0.7-0.8 and RMSE=2-4 mm) in estimating monthly precipitation mainly in the eastern half of the province. While the lowest accuracy and the weakest performance was for the PERSIANN-CDR product. The highest CC value of precipitation products was equal to 0.8, which was mainly in the eastern and coastal areas of the province, but the lowest value was related to the PERSIANN-CDR product (CC=0.05). The RMSE values are mainly between 2 and 15 mm in the mountainous and eastern half of the province, respectively, and the lowest values are for the IMERG and CMORPH products. The values of KGE and NSE were also closer to the optimal value mainly in the coastal and eastern areas (NSE=0.5 and KGE=0-1), which was better for the CMORPH product. However, the BIAS values for all products varied between (-4)-2 mm, which was underestimated in coastal and low-altitude areas, but overestimated in high-altitude areas. Investigating the effect of the distance between the synoptic station and grid precipitation on the accuracy of the precipitation product has also shown that for the CMORPH, IMERG (in high areas) and SM2RAIN products, a low (higher) distance between the station and the grid has reduced (increased) the uncertainty. However, for most of the precipitation products, the significant increase in elevation has increased the uncertainty in satellite precipitation estimation.
Conclusion: The most accurate precipitation product in Mazandaran province includes the IMERG product. One of the main problems in the accuracy of precipitation products in this region is the complex topography (large height difference) and the proximity to the Caspian Sea. However, the use of modern methods such as remote sensing systems as a solution for estimating precipitation represents scientific advances in this field, and this can be used as a guide for decision-makers in climate and hydrological studies.
Keywords
)