Fig. 6: Spatial matching and balance assessment.

a Delineates the distribution of project sites and their matched counties, connected by lines to emphasize the geographic pairing. b Compares the standardized mean difference of covariates between treated and control counties before and after matching. The open circles represent the standardized mean differences before matching, while filled circles (blue) indicate these differences after matching. The vertical dashed lines demarcate standardized mean differences of −0.1, 0, and 0.1, providing reference points for assessing the quality of covariate balance. The covariates include forest coverage (forest), industrial structure (1^st Ind., 2^nd Ind.), and economic indicators (revenue, per_GDP, GDP). c Provides an example of pinpointing individual project and matched pixels at 1 km × 1 km resolution (30 m × 30 m resolution is exhibited in Fig. 1c) where the blue pixels denote the project sites and the purple pixels denote the reference area. d Compares the standardized mean difference of covariates between treated and control pixels in buffer zones before and after matching. The covariates include land use of the pixel and the surrounding pixels (LUCC, LUCC Buffer), landform, where the proportion of each category is considered, as well as NDVI, precipitation, and solar radiation (SRAD).