Enhancing Surface Soil Moisture Estimation through Integration of Artificial Neural Networks Machine Learning and Fusion of Meteorological, Sentinel-1A and Sentinel-2A Satellite Data  

作　　者：Jephter Ondieki Giovanni Laneve Maria Marsella Collins Mito Jephter Ondieki;Giovanni Laneve;Maria Marsella;Collins Mito(DIMA-Department of Mechanical and Aerospace Engineering, Sapienza University of Rome, Roma, Italy;SIA-School of Aerospace Engineering, Sapienza University of Rome, Roma, Italy;Department of Physics, University of Nairobi, Nairobi, Kenya)

机构地区：[1]DIMA-Department of Mechanical and Aerospace Engineering, Sapienza University of Rome, Roma, Italy [2]SIA-School of Aerospace Engineering, Sapienza University of Rome, Roma, Italy [3]Department of Physics, University of Nairobi, Nairobi, Kenya

出　　处：《Advances in Remote Sensing》2023年第4期99-122,共24页遥感技术进展（英文）

摘　　要：For many environmental and agricultural applications, an accurate estimation of surface soil moisture is essential. This study sought to determine whether combining Sentinel-1A, Sentinel-2A, and meteorological data with artificial neural networks (ANN) could improve soil moisture estimation in various land cover types. To train and evaluate the model’s performance, we used field data (provided by La Tuscia University) on the study area collected during time periods between October 2022, and December 2022. Surface soil moisture was measured at 29 locations. The performance of the model was trained, validated, and tested using input features in a 60:10:30 ratio, using the feed-forward ANN model. It was found that the ANN model exhibited high precision in predicting soil moisture. The model achieved a coefficient of determination (R²) of 0.71 and correlation coefficient (R) of 0.84. Furthermore, the incorporation of Random Forest (RF) algorithms for soil moisture prediction resulted in an improved R² of 0.89. The unique combination of active microwave, meteorological data and multispectral data provides an opportunity to exploit the complementary nature of the datasets. Through preprocessing, fusion, and ANN modeling, this research contributes to advancing soil moisture estimation techniques and providing valuable insights for water resource management and agricultural planning in the study area.For many environmental and agricultural applications, an accurate estimation of surface soil moisture is essential. This study sought to determine whether combining Sentinel-1A, Sentinel-2A, and meteorological data with artificial neural networks (ANN) could improve soil moisture estimation in various land cover types. To train and evaluate the model’s performance, we used field data (provided by La Tuscia University) on the study area collected during time periods between October 2022, and December 2022. Surface soil moisture was measured at 29 locations. The performance of the model was trained, validated, and tested using input features in a 60:10:30 ratio, using the feed-forward ANN model. It was found that the ANN model exhibited high precision in predicting soil moisture. The model achieved a coefficient of determination (R²) of 0.71 and correlation coefficient (R) of 0.84. Furthermore, the incorporation of Random Forest (RF) algorithms for soil moisture prediction resulted in an improved R² of 0.89. The unique combination of active microwave, meteorological data and multispectral data provides an opportunity to exploit the complementary nature of the datasets. Through preprocessing, fusion, and ANN modeling, this research contributes to advancing soil moisture estimation techniques and providing valuable insights for water resource management and agricultural planning in the study area.

关 键 词：Soil Moisture Estimation Techniques Fusion Active Microwave Multispectral Data Agricultural Planning 

分 类 号：S15[农业科学—土壤学]