Monitoring policy-driven crop area adjustments in northeast China using Landsat-8 imagery

Abstract

To keep pace with increasing demands for food, China is advancing robust policies on food security, and policy-oriented adjustments in crop planting area are needed to strike a balance among environmental sustainability, socio-economic development, agriculture and food security. To this end, the crop area adjustment policy was introduced in northeast China since 2015 and the policy provides incentives for quick planting area changes between different crops. However, geospatial data that quantify the magnitude and direction of these crop area adjustments are grossly inadequate in content and accuracy, thereby limiting our understanding of the overall policy impacts. To fill these data needs, applying the random forest (RF) classification algorithm on temporal images of Landsat-8 Operational Land Imager (OLI) acquired over the Heilongjiang Province (1235 scenes) in 2015 and 2016, this study proposed an integrated approach to the identification and monitoring of adjustments in the planting areas of maize, soybean and paddy rice. Overall crop classification accuracies of 88.7% and 87.9% were obtained for 2015 and 2016, respectively, and our approach further accounted for localized error distribution patterns. The result indicated that the area of soybean and maize had a significant change in 2016 (maize area decreased by 22.0% with about 1,261,775 ha and soybean area increased by 39.8% with about 696,653 ha, compared to 2015). The expansion in soybean-planted area recorded in this study is mostly in regions that previously mapped suitable for soybean cultivation based on local agro-climatic factors. The decrease in maize-planted area in tandem with the increase in soybean-planted area is significant in the northern parts of the Songneng Plain where initial profit differentials between the two crops are minimal due to being more suitable for soybean, and where with the introduction of subsidies, soybean recorded higher profits per unit area than maize. This suggests that with more incentives to cover the profit margin between soybean and maize, or the direction of subsidies to areas where soybean naturally records higher yields as a function of land suitability, the crop area adjustment policy is expected to achieve its goal in Heilongjiang. The current study, having provided information on the relative shifts in agricultural land-uses as informed by agricultural policy and land suitability data at a biannual scale, more detailed studies that would apply the proposed approach to multiple years of satellite data are therefore encouraged to provide information on the dynamics of agricultural land-use as informed by climate-induced changes inland suitability, market demand and government policy.