A Remote Sensing Methodology Based on Elastic Net and XGBoost for Tidal Flat Photovoltaic Extraction Using GF-1 Satellite Imagery

Satellite remote sensing technology serves as the primary methodology for monitoring tidal flat photovoltaics, but the research on photovoltaic remote sensing mainly focuses on land-based systems. While the characteristics of tidal flat photovoltaic are obviously different from that of land photovoltaic, and its background environment is complex, so it is difficult to extract accurately. To solve this problem, based on the remote sensing data of Gaofen-1 (GF-1) Satellite, this paper uses elastic net to select the relevant spatial spectrum response characteristics of tidal flat photovoltaic, and selects seven features that can characterize tidal flat photovoltaic, namely blue band reflectance, green band reflectance, the mean value, Normalized Difference Water Index (NDWI), information entropy, the difference and ratio of red and blue bands, and brightness. On this basis, this paper develops the tidal flat photovoltaic remote sensing extraction method based on extreme gradient boosting algorithm. The experimental findings demonstrate that the proposed method exhibits robust extraction performance across diverse tidal flat photovoltaic regions, achieving an overall recall of 86.28% and an F1-Score of 0.91. Compared to Support Vector Machine (SVM) and Random Forest algorithms (RF), the recall and F1-Score are improved by more than 10% and 7%, respectively. Additionally, the method exhibits the advantages of low false extraction rate, fine edge extraction and fast training speed, making it an effective technical solution for tidal flat photovoltaic remote sensing monitoring and extraction.