Generating Gridded Gross Domestic Product Data for China Using Geographically Weighted Ensemble Learning

Gridded gross domestic product (GDP) data are a crucial land surface parameter for many geoscience applications. Recently, machine learning approaches have become powerful tools in generating gridded GDP data. However, most estimation seldom consider the geographical properties of input variables. Therefore, this study, geographically weighted stacking ensemble approach was developed to generate Three algorithms—random forest, XGBoost, and LightGBM—were used as base models, linear regression replaced by locally fuse three predictions. A case study conducted China demonstrate effectiveness proposed approach. The results showed that downscaling outperformed models traditional learning. Meanwhile, it had good predictive power on county-level test with R2 0.894, 0.976, 0.976 primary, secondary, tertiary sectors, respectively. Moreover, predicted 1 km high accuracy (R2 = 0.787) when evaluated town-level Hence, provides valuable option generated from 2020 great significance other