MSE-Stack Downscaled 1-km Monthly XCO₂ Dataset over the Mongolian Plateau, 2017–2020

The dataset was generated using a three-layer stacked ensemble learning framework (MSE-Stack), which integrates OCO-2 XCO₂ retrievals with multi-source meteorological, ecological, topographic, and anthropogenic variables.The dataset is designed to characterize the spatial, seasonal, and interannual variability of atmospheric XCO₂ over the Mongolian Plateau and to support regional carbon-cycle and environmental studies.

It provides high-resolution and spatially continuous XCO₂ estimates, helping to address the spatial and temporal discontinuities associated with sparse satellite observations.The dataset was generated as part of the study “A three-layer stacked ensemble learning method for high-resolution XCO₂ downscaling using OCO-2 and multi-source data.”