Improving precipitable water vapor estimations of the Moderate Resolution Imaging Spectroradiometer (MODIS) using metaheuristic algorithms

Abstract

Extracting Total Precipitable Water Vapor (TPW) and its variations from the Moderate Resolution Imaging Spectroradiometer (MODIS) data on a local scale is challenging. Meta-heuristic algorithms have a high ability to solve complex engineering problems of hydrology and climatic studies. Hence, the present study extracts TPW values using meta-heuristic algorithms for MODIS surface reflectance data. In this regard, radiosonde observations of six stations in the western part of Iran during 2019 and 2020 were used as the ground truth data for training algorithms and evaluating the results. The results were also validated with Global Positioning System (GPS) TPW data of Tehran-Mehrabad station. Moreover, MODIS surface reflectance (MOD021) and cloud mask (MOD35) products were prepared to distinguish cloudy days from non-cloudy, along with MODIS TPW (MOD05) for corresponding dates. Meta-heuristic algorithms, including Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Harmony Search (HS), Invasive Weed Optimization (IWO), Gray Wolf Optimization (GWO), and Grasshopper Optimization Algorithm (GOA), were applied to improve the performance of MOD05 algorithm. The results showed the proper performance of all the selected algorithms. Nevertheless, the GOA algorithm acquired the best result. The Root Mean Square Error (RMSE) and R values obtained from comparing GOA results with radiosonde data were 2.96 mm and 0.82. Those values for comparing GOA results with GPS data were 2.94 mm and 0.83, respectively. However, MODIS TPW, RMSE, and R showed 6.53 mm and 0.69, respectively, compared to radiosonde data and 6.55 mm and 0.68 compared to GPS data. Overall, the results proved the proper performance of the used meta-heuristic algorithms in extracting TPW from MODIS measurements on a local scale.