Proximal remote sensing of dissolved organic matter in aqua-culture ponds via multi-temporal spectral correction

Wenxu Lv¹Yancang Wang^1*Huiqiong Cao¹Peng Cheng¹Xiaohe Gu²Zhuoran Ma³Mengjie Li¹Ruiyin Tang¹Qichao Zhao¹Xuqing Li¹Lan Zhang¹Shuaifei Liu¹

• ¹North China Institute of Aerospace Engineering, Langfang, China
• ²Beijing Academy of Agriculture and Forestry Sciences Information Technology Research Center, Beijing, China
• ³Monash University, Melbourne, Australia

Dissolved organic matter (DOM) is a critical indicator of aquatic environmental quality, and its concentration affects the quality of aquaculture products. Integrating unmanned aerial vehicle (UAV)-based multispectral data with machine learning algorithms enables accurate estimation of DOM. However, the stability of models in different periods—such as those affected by seasonal variations and environmental condition changes—is the key factor affecting their application. This study employed a spectral correction method to unify multi-temporal datasets. Estimation models were constructed using the 2023 dataset with Light Gradient Boosting Machine (LightGBM)、Extreme Gradient Boosting (XGBoost) and Random Forest (RF) algorithms, and their cross-year performance was validated on the 2024 dataset through transfer learning. Results demonstrated that models constructed with corrected data performed better, with the average R² of the test set increased by 15.67%, the RMSE decreased by 10.27%, and the MAE decreased by 6.44%. After transfer learning optimization, the model using the corrected spectrum still exhibited superior performance in 2024. Compared with the original spectrum, an average R² improvement of 30.67%, along with reductions of 17% in RMSE and 11.67% in MAE. The RF model yielded the best performance, with an R² of 0.82, on the test set. The approach proposed in this study effectively mitigates the temporal impact on model performance and enhances the temporal generalization capability of DOM estimation models.