Abstract Vegetation in mountainous areas contributes about 36% to the global gross primary productivity (GPP). However, the influences of topography on radiation and water redistributions in mountain ecosystems are so far ignored in existing global GPP data sets. Here, an eco‐hydrological model was adopted to simulate 30 m resolution mountain and flat GPP over 16 watersheds. Then, a topographical correction index (TCI) was developed based on simulated soil water redistribution (TCI water ), radiation redistribution (TCI rad ), and redistribution of climate factors (TCI clim ). Finally, the proposed TCI was applied to four GPP data sets. The mean‐bias‐error (MBE), determination coefficient ( R 2 ), and Root‐Mean‐Square‐Error (RMSE) between mountain GPP and flat GPP (or GPP data sets) were used for evaluation. Results showed that the MBE of flat GPP before correction (194 g C m −2 yr −1 ) was reduced to 126, 94, and 2 g C m −2 yr −1 after the corrections of TCI water , TCI rad , and TCI clim , highlighting the effectiveness of integrated redistribution information in correcting the topographical effect on GPP estimation. The relationship between mountain and flat GPP after the TCI correction was improved at the 30 m resolution (increasing R 2 by 0.09 and reducing RMSE by 90 g C m −2 yr −1 ) and 480 m resolution (increasing R 2 by 0.13 and reducing RMSE by 178 g C m −2 yr −1 ). Regarding the four GPP data sets after the TCI correction, the MBE of 183 g C m −2 yr −1 was averagely reduced to 17 g C m −2 yr −1 , and RMSE was reduced by 118 g C m −2 yr −1 at 480 m resolution. This study suggests that integrating topography‐induced interactions into current GPP data sets is a feasible way to understand the carbon budget in mountain ecosystems.
Plain Language Summary Vegetation in mountainous areas contributes about 36% to the global gross primary productivity (GPP). Currently, the topography‐induced interactions among pixels haven't been attempted in the generation of global GPP data sets. This study developed a topographical correction index (TCI) based on simulated soil water redistribution (TCI water ), radiation redistribution (TCI rad ), and redistribution of climate factors (TCI clim ). Results showed that the mean‐bias‐error (MBE) of flat GPP before correction was significantly reduced after the corrections of TCI water , TCI rad , and TCI clim , highlighting the effectiveness of integrated redistribution information in correcting the topographical effect on GPP estimation. Regarding the GPP data sets after the TCI correction, the MBE of 183 g C m −2 yr −1 was reduced to 17 g C m −2 yr −1 , and Root‐Mean‐Square‐Error was reduced by 118 g C m −2 yr −1 at 480 m resolution. This study suggests that integrating topography‐induced interactions into current GPP data sets is a feasible way to understand the carbon budget in mountain ecosystems.
Key Points A topographical correction index (TCI) based on soil water, radiation, and climatic redistributions was useful in correcting topographical effects on gross primary productivity (GPP) products The mean‐bias‐error (MBE) of flat GPP before correction (194 g C m −2 yr −1 ) was reduced to 126, 94, and 2 g C m −2 yr −1 after applying TCI water , TCI rad , and TCI clim The proposed TCI was effective for MOD17, Global LAnd Surface Satellite, Moderate‐resolution Imaging Spectroradiometer, and PML GPP, with the MBE and Root‐Mean‐Square‐Error reducing by 166 and 118 g C m −2 yr −1