农作物种植格局对遥感分类精度的影响

Effects of Crop Planting Structure on Remote Sensing Classification Accuracy

摘要:

研究不同作物种植成数、田块形状和田块破碎度对作物遥感分类精度的影响，是科学评价作物遥感分类精度的基础。采用GF—1遥感数据，以时序植被指数的主要农作物分类结果为基础，对研究区冬小麦—夏玉米作物种植区的分类精度与种植成分、田块形状和破碎度的关系进行了研究。结果表明，种植成数与分类精度呈正相关，田块破碎度、田块形状指数与分类精度呈负相关。

关键词:作物种植面积;遥感分类;种植成数;田块形状指数;田块破碎度;GF—1

Abstract:

The study of effects of different crop acreage proportions, crop field shape index and crop field fragmentation on accuracy of crop classification by remote sensing provides a basis for scientific evaluation of the latter. Using GF—1 remote sensing data and based on the major crop classification results of the timeseries vegetation index, the relationship between classification accuracy of crops (including winter wheat and summer maize) and crop acreage proportion, crop field shape index as well as crop field fragmentation was studied. The research was based on 14 GF—1/WFV NDVI time series data. The timing vegetation indexbased crop classification knowledge rules were utilized on the basis of the best NDVI threshold interval of crops to be classified to complete the crops classification and make spatial distribution map. Then, totally 14 classical villages of Quzhou county were selected as sample plots, which included winter wheat—summer corn plots. The landuse ownership boundary map for the 14 classic villages was obtained according to 1∶50000 Quzhou county present landuse map, which was prepared by Quzhou County Land Resources Bureau and China Agricultural University jointly. The spatial distribution map of winter wheat—summer corn and landuse ownership boundary map among landuse survey maps were used to take image masking, and the lots and sample points of winter wheat—summer corn of each classical village region were obtained. Hence, the crop acreage proportion, crop field shape index and crop field fragmentation of winter wheat—summer corn in 14 villages were obtained, and classification accuracy, Kappa index were calculated. In addition, totally 14 groups of sample plot related data were acquired and graphs of relation between all influencing factors and classification accuracy were prepared. The results showed that the crop acreage proportion was positively correlated to classification accuracy, while the crop field fragmentation and crop field shape index were negatively correlated to classification accuracy.

Key words:crop planting acreage;remote sensing classification;crop acreage proportion;crop field shape index;crop field fragmentation;GF—1