Spatial Clustering and Outlier Analysis for the Regionalization of Maize Cultivation in China

摘要

Regionalization has been the foundation of large-scale plantation and local optimization for crop cultivation. Current regionalization approaches practiced mainly rely on qualitative analysis and heuristic methods, which cannot meet the increasingly challenging demands. In this paper, we demonstrate the use of spatial clustering method on the regionalization of crop cultivation, with the maize growing in China as an example. In the proposed method, we adopt four indicators [that is, elevation, effective accumulated temperature (EAT), precipitation and yield] which are the major factors reflecting the maize cultivation differences, hi addition, by taking into account the spatial information of counties in the clustering process, we achieve a more spatially coherent clustering result. As a post-processing step, adjustment with the help of a Geographic Information System (GIS) eliminates regions that appear inconsistent with the vicinity. With the proposed approach, we classify the 2,831 counties of China into 7 regions, and show that the result is highly consistent with the conventional regionalization of maize cultivation in China. This result proves the feasibility of our approach, and suggests its possible application on other crops. Furthermore, we carry out outlier analysis for each of the regions to identify the counties that show abnormal behaviors in maize cultivation, and further analyze the possible causes. This study provides valuable information for cultivation region selection in large-scale crop plantation.