Grassland ecosystems are essential components of the global ecosystem. They may efficiently reduce CO₂ concentrations in the atmosphere and play a vital role in mitigating climate change. The objectives of this study were to reveal the spatial distribution features of net primary production (NPP) and net ecosystem productivity (NEP) under climate change in the Inner Mongolia grassland ecosystem, China, and to devise effective management strategies for grassland ecosystems. Based on the multiscale geographically weighted regression (MGWR) model, this study investigated the spatial variation features of NPP and NEP along with their driving factors. The results showed the following: (1) The annual average NPP in the Inner Mongolia grassland ecosystem was 234.22 $\mathrm{gC}\cdot {\mathrm{m}}^{-2}\cdot {\mathrm{a}}^{-1}$, and the annual average NEP was 60.31 $\mathrm{gC}\cdot {\mathrm{m}}^{-2}\cdot {\mathrm{a}}^{-1}$ from 2011 to 2022. Both measures showed a spatial pattern of high values in the northeast and low values in the southwest, as well as a temporal pattern of high values in summer and low values in winter. (2) The normalized difference vegetation index (NDVI) and solar radiation had promoting effects on NPP, where NDVI had the largest significant positive correlation area. In addition, precipitation and temperature on the influence of NPP were significantly negative with a larger area. (3) The area with a significant positive correlation of NDVI, solar radiation, and precipitation on NEP was larger than that with a significant negative correlation, while the area with significant negative correlation of temperature was larger. This study used the MGWR model to explore the relationship between NPP, NEP, and multiple factors. The results showed regional variation in NPP and NEP under the combined effect of various drivers. This contributes to a better understanding of carbon sinks under climate change in the Inner Mongolia grassland ecosystem.

A geographically and temporally weighted regression (GTWR) model is an effective tool for dealing with spatial heterogeneity and temporal non-stationarity simultaneously. As an important characteristic of spatiotemporal data, spatiotemporal autocorrelation should be considered when constructing spatiotemporally varying coefficient models. The proposed local autoregressive geographically and temporally weighted regression (GTWRLAR) model can simultaneously handle spatiotemporal autocorrelations among response variables and the spatiotemporal heterogeneity of regression relationships. The two-stage weighted least squares (2SLS) estimation can effectively reduce computational complexity. However, the weighted least squares estimation is essentially a Nadaraya–Watson kernel-smoothing approach for nonparametric regression models, and it suffers from a boundary effect. For spatiotemporally varying coefficient models, the three-dimensional spatiotemporal coefficients (longitude, latitude, and time) inherently exhibit larger boundaries than one-dimensional intervals. Therefore, the boundary effect of the 2SLS estimation of GTWRLAR will be more serious. A local–linear geographically and temporally weighted 2SLS (GTWRLAR-L) estimation is proposed to correct the boundary effect in both the spatial and temporal dimensions of GTWRLAR and simultaneously improve parameter estimation accuracy. The simulation experiment shows that the GTWRLAR-L method reduces the root mean square error (RMSE) of parameter estimates compared to the standard GTWRLAR approach. Empirical analyses of carbon emissions in China’s Yellow River Basin (2017–2021) show that GTWRLAR-L enhances the adjusted $R^2$ from 0.888 to 0.893.