Calculation of county-level cultivated land productivity based on NPP index corrected by topography

This study aims to estimate the productivity of arable land in a quick, accurate and cost-saving way, for the quality/productivity evaluation of arable land, and the implementation of the “trinity” policy, i.e., requisition-compensation balance of cultivated land. An attempt was made to calculate the productivity of cultivated land by using the net primary productivity (NPP) index, in order to increase the accuracy of the evaluation system, while saving time and cost. Taking the county level as the research scale, and Binyang County as the research area, the obvious terrain difference can indicate the variation of solar radiation subjected to topographic factors, and fill the research lack of NPP database at the county level. A CASA (Carnegie-Ames-Stanford Approach) model was used to calculate the remote sensing and meteorological data when extracting NPP index. The influence of topographic factors (terrain) on solar radiation was also considered in a modified CASA model. A geographic weighted regression approach was selected to compare the obtained NPP data with the utilization index of cultivated land, in order to verify the application of NPP data for the production of cultivated land. A comparison analysis of local correlation coefficient was made to determine the region with a large difference between the terrain and solar radiation, further to find the main advantages of the modified CASA model. The results showed that in the productivity distribution of cultivated land, the high NPP index was generally in the direction of southeast-northwest axis in the central region, and on the southern plain, whereas the low NPP index was on both sides of northeast and southwest in the research area. There was little change in the topographically modified NPP index, but the concentration of distribution increased, while the productivity has been extended to dry land and paddy fields. The geographic weighted regression between the indexes of cultivated land use and NPP showed that there was a strong correlation in the same geographical location, and the correlation coefficient of dry land can reach 0.87, while paddy field was 0.80, indicating that NPP index can well connect with the original cultivated land use index. It infers that the spatial autocorrelation of NPP index can be strong and sensitive to the factors affecting the productivity of cultivated land, such as soil conditions, terrain characteristics, crop differences, and transportation. The calculation of NPP index can be directly applied to the evaluation for the productivity of cultivated land, with the high efficiency and accuracy in the system. The proposed method can be more quickly and accurately applied to the dynamic estimation of cultivated land, the delimitation of basic farmland, the evaluation of land improvement benefits, and the transformation of medium and low yield farmland.