Remote sensing technology applications for water quality inversion in large rivers are common. However, their application to medium/small-sized water bodies within rural areas is limited due to the low spatial resolution of remote sensing images. In this work, a typical small rural river was selected, and high-resolution unmanned aerial vehicle (UAV) multispectral images and ground monitoring data of the river were obtained. Then, a comparative analysis of three univariate regression models and nine machine learning models (Ridge Regression (RR), Support Vector Regression (SVR), Grid Search Support Vector Regression (GS-SVR), Random Forest (RF), Grid Search Random Forest (GS-RF), eXtreme Gradient Boosting (XGBoost), Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), and Catboost Regression (CBR)) for their accuracy in the prediction of turbidity (TUB), total nitrogen (TN), and total phosphorus (TP) was performed. TUB can be achieved by simple statistical regression models. The CBR model exhibited the best performance for the three index inversions on the test set evaluation metrics: R² (0.90~0.92), RMSE (7.57 × 10⁻³~1.59 mg/L), MAE (0.01~1.30 mg/L), RPD (3.21~3.56), and NSE (0.84~0.92). The water pollution of the study area was closely related to its land-use pattern, excessive and irrational fertilizer application, and distribution of pollutant outlets.

Accurate prediction of spatial variation in water quality in small microwaters remains a challenging task due to the complexity and inherent limitations of the optical properties of small microwaters. In this paper, based on unmanned aerial vehicles (UAV) multispectral images and a small amount of measured water quality data, the performance of seven intelligent algorithm-optimized SVR models in predicting the concentration of chlorophyll (Chla), total phosphorus (TP), ammonia nitrogen (NH₃-N), and turbidity (TUB) in small and micro water bodies were compared and analyzed. The results show that the Gray Wolf optimized SVR model (GWO-SVR) has the highest comprehensive performance, with R² of 0.915, 0.827, 0.838, and 0.800, respectively. In addition, even when dealing with limited training samples and different data in different periods, the GWO-SVR model also shows remarkable stability and portability. Finally, according to the forecast results, the influencing factors of water pollution were discussed. This method has practical significance in improving the intelligence level of small and micro water body monitoring.

The continuous and effective monitoring of the water quality of small rural rivers is crucial for rural sustainable development. In this work, machine learning models were established to predict the water quality of a typical small rural river based on a small quantity of measured water quality data and UAV hyperspectral images. Firstly, the spectral data were preprocessed using fractional order derivation (FOD), standard normal variate (SNV), and normalization (Norm) to enhance the spectral response characteristics of the water quality parameters. Second, a method combining the Pearson’s correlation coefficient and the variance inflation factor (PCC–VIF) was utilized to decrease the dimensionality of features and improve the quality of the input data. Again, based on the screened features, a back-propagation neural network (BPNN) model optimized using a mixture of the genetic algorithm (GA) and the particle swarm optimization (PSO) algorithm was established as a means of estimating water quality parameter concentrations. To intuitively evaluate the performance of the hybrid optimization algorithm, its prediction accuracy is compared with that of conventional machine learning algorithms (Random Forest, CatBoost, XGBoost, BPNN, GA–BPNN and PSO–BPNN). The results show that the GA–PSO–BPNN model for turbidity (TUB), ammonia nitrogen (NH₃-N), total nitrogen (TN), and total phosphorus (TP) prediction exhibited optimal accuracy with coefficients of determination (R²) of 0.770, 0.804, 0.754, and 0.808, respectively. Meanwhile, the model also demonstrated good robustness and generalization ability for data from different periods. In addition, we used this method to visualize the water quality parameters in the study area. This work provides a new approach to the refined monitoring of water quality in small rural rivers.

Ecosystems in arid and semi-arid areas are delicate and prone to different erosive effects. Monitoring and evaluating the environmental ecological condition in such areas contribute to the governance and restoration of the ecosystem. Remote sensing ecological indices (RSEIs) are widely used as a method for environmental monitoring and have been extensively applied in various regions. This study selects the arid and semi-arid Loess Plateau as the research area, in response to existing research on ecological monitoring that predominantly uses vegetation indices as monitoring indicators for greenness factors. A fluorescence remote sensing ecological index (SRSEI) is constructed by using monthly synthesized sun-induced chlorophyll fluorescence data during the vegetation growth period as a new component for greenness and combining it with MODIS product data. The study generates the RSEI and SRSEI for the research area spanning from 2001 to 2021. The study compares and analyzes the differences between the two indices and explores the evolution patterns of the ecosystem quality in the Loess Plateau over a 21-year period. The results indicate consistent and positively correlated linear fitting trend changes in the RSEI and SRSEI for the research area between 2001 and 2021. The newly constructed ecological index exhibits a higher correlation with rainfall data, and it shows a more significant decrease in magnitude during drought occurrences, indicating a faster and stronger response of the new index to drought in the research area. The largest proportions are found in the research area’s regions with both substantial and minor improvements, pointing to an upward tendency in the Loess Plateau’s ecosystem development. The newly constructed environmental index can effectively evaluate the quality of the ecosystem in the research area.