Articles
PFG - Journal of Photogrammetry, Remote Sensing and Geoinformation Science (25122819)93(3)pp. 231-250
Snow depth is a critical parameter for meteorological, climatological, and hydrological models, enhancing regional water resource information and weather prediction accuracy. AMSR2 passive microwave data with 10-kilometer spatial resolution results in high error and uncertainty in snow depth estimation. This study attempted to develop machine learning algorithms to downscale AMSR2 snow depth data to 1 kilometer spatial resolution. One of the primary factors inflicting errors in snow depth downscaling is the absence of proper identification of snow-covered area boundaries (mixed pixels), leading to extensional uncertainty. In the present study, a novel method based on the theory of random sets has been developed to resolve the issue of mixed pixels in the boundaries of snow-covered areas for the European Alps. In addition to snow cover data acquired from random sets, auxiliary data for developing and training machine learning algorithms comprises of the Digital Elevation Model (DEM), Land Surface Temperature (LST), and Land Cover Type data were processed. The machine learning algorithms employed includes Support Vector Regression (SVR), Multi-Layer Perceptron (MLP), and Random Forest (RF). The RF algorithm, when considering snow cover data derived from random sets method, confirmed the excellent performance among the SVR and MLP algorithms. Using RF, the RMSE decreased from 31.67–16.34 cm, the correlation coefficient increased from 0.312–0.81, and the relative error decreased from 9.85–0.52% (reduced error and increased accuracy). The results based on evaluation metrics indicates an outstanding performance of the proposed innovative method for downscaling of AMSR2 snow depth. © Deutsche Gesellschaft für Photogrammetrie, Fernerkundung und Geoinformation (DGPF) e.V. 2025.
Iranian Journal of Remote Sensing and GIS (25886185)17(1)pp. 61-78
Atmospheric water vapor is a key parameter in modeling the energy balance on the earth's surface and plays a major role in keeping the temperature of the earth's atmosphere balanced. Retrieving of this parameter, as the most influential atmospheric parameter on the sensors received radiance, is of great importance. Since the atmospheric water vapor content in the near of surface is more and its temporal and spatial changes are more intense, the measurements of ground meteorological stations, despite their high accuracy, are not generalizable due to temporal and spatial limitations and point measurements. Therefore, it seems necessary to provide practical satellite-based methods to accurate and continuous retrieval of this parameter with appropriate spatial distribution. The aim of this research is to present four innovative and accurate methods to estimate the near surface atmospheric water vapor of Isfahan province in 2020 with a resolution of 1 km, through the integration of meteorological station data, sensor data and finally validating and comparing their performance. For this purpose, correcting the bias error of water vapor sensor data during the co-scaling stage and correcting the interpolation error of ground station observations was put on the agenda. Material and Methods: Different sensors measure water vapor with different sensitivities and spatial resolution. Therefore, it is necessary to provide methods based on the simultaneous use of diffferent sensor data and their integration to ground station observations, in order to simultaneously improve the accuracy and spatial resolution (1 km) of retrieved near surface water vapor. In the first method used in this research, the near surface water vapor is retrieved using the water vapor absorbing and non-absorbing bands of the MODIS, through the band ratio method and using ground observations. In the second method, first, observations of near surface water vapor of ground stations are converted to 1 km grid using the inverse distance interpolation (IDW) method. Then, during the steps of the proposed method and using the water vapor values estimated by the first method, the interpolation error in each pixel is removed. In the third method, the resolution of AIRS-derieved water vapor product is reduced to 1 km by combining MODIS data during an operation similar to the steps of the second method, with the difference that the AIRS sensor product is used instead of ground station observations. It is necessary to eliminate the bias error of near surface water vapor product of the AIRS during the co-scaling stage by first. Estimation of near surface water vapor using MODIS column water vapor product is the fourth method. Of course, due to the difference in content, it is necessary to unite the two sets and equate them with an approprite method. Results and Discussion: In order to model and validate the estimation of atmospheric near surface water vapor at a spatial resolution of 1 km using the different mentioned methods, 66.6% of the data were randomly used for training and the remaining 33.3% were used to evaluate the accuracy and validation. Finally, the implementation results of the methods have been compared with each other. The validation results of proposed methods show that the second method, which is based on the generalization of accurate observations of ground stations and removing their interpolation error, during integration with the water vapor values retrieved from first method, has the best performance (R2=0.55, RMSE=1.05 Gr/Kr). Conclusion: Considering the better performance of the second method in retrieving the mixing ratio of near surface water vapor with high accuracy and resolution of 1 km, and with the aim of using the capabilities of satellite-based products and data, it is recommended to combine them with each other and also with ground observations. © 2025, Shahid Beheshti University. All rights reserved.
