Temporal dynamics of turbidity in China' s lakes and reservoirs: Distinct trajectories and corresponding driving mechanism (Necessary data and codes)

In this database, we uploaded the dataset and turbidity inversion code used for a research article entitled: Temporal dynamics of turbidity in China's lakes and reservoirs: Distinct trajectories and corresponding driving mechanisms, which includes: Modeling datasets (Train_ data_for_modeling.xlsx, Test_data_for_modeling.xlsx, Validation_data_for_modeling.xlsx), the 2003-2024 year-by-year turbidity inversion results of the XGBoost model (based on the median of the year), named XGB_MODIS_Water_Yearly_turbidity_full_TIFF_ from2003to2024.zip, Shapfile_Lakes_reservoirs_boundary_in_China_area_above_10_km2.zip) and point shpfile dataset (Shpfile_Lakes_and_Reservoirs_Points.zip), The results of the evolution trend of lakes and reservoirs and the collection dataset of eight types of drivers (Turbidity_yearly_timeseries_from_2003to2024_by_XGBoost_after_data_cleaning.xlsx), in addition to the Python code (based on XGBoost, random forest and KNN turbidity inversion models, named XGBoost_Code.py, Random_Forest_Code.py andKNN_Code.py). Finally, we upload the driving analysis R language code using Bayesian additive regression tree (BART) model combined with conditional permutation importance (CPI) and partial dependence plots (PDP), named BART_Model_with_CPI_and_PDP. R, we hope that the above data and code can promote academic exchange and progress.

本数据库上传了一篇题为《中国湖泊与水库浊度的时间动态：独特演化轨迹及相应驱动机制》的研究论文所使用的数据集与浊度反演代码，具体包含以下内容：建模数据集（Train_data_for_modeling.xlsx、Test_data_for_modeling.xlsx、Validation_data_for_modeling.xlsx）；基于中分辨率成像光谱仪（Moderate Resolution Imaging Spectroradiometer, MODIS）与极限梯度提升树（XGBoost）模型得到的2003-2024年逐年浊度反演结果（以年度中位数为统计基准），对应压缩包为XGB_MODIS_Water_Yearly_turbidity_full_TIFF_from2003to2024.zip；中国面积≥10 km²的湖泊与水库边界形状文件压缩包Shapfile_Lakes_reservoirs_boundary_in_China_area_above_10_km2.zip，以及点位矢量数据集压缩包Shpfile_Lakes_and_Reservoirs_Points.zip；湖泊与水库演化趋势分析结果，以及8类驱动因子的收集数据集Turbidity_yearly_timeseries_from_2003to2024_by_XGBoost_after_data_cleaning.xlsx；此外还包含基于极限梯度提升树（XGBoost）、随机森林（Random Forest）与K近邻（K-Nearest Neighbor, KNN）的浊度反演模型Python代码，分别命名为XGBoost_Code.py、Random_Forest_Code.py及KNN_Code.py；最后上传了结合贝叶斯加性回归树（Bayesian Additive Regression Tree, BART）模型、条件置换重要性（Conditional Permutation Importance, CPI）与偏依赖图（Partial Dependence Plots, PDP）的驱动分析R语言代码，文件名为BART_Model_with_CPI_and_PDP.R。我们期望上述数据与代码能够促进学术交流与研究进步。