Robustness Of Disturbance Method For Main Dryland Crop Planting Pattern Automatic Extraction In China

The disturbance of difference in crop phenology,different planting patterns and changes in site conditions can lead to the problem of intra-class variability and inter-class similarity between the remote sensing index and the band temporal curves.Therefore,this paper establishes an anti-interference method for main dryland crop cultivation pattern automatic extraction in China,utilizing the 8-day maximum synthetic MOD09A1 data from 2005 to 2015 to carry out the research.The main contents are as follows:1.A method for extracting dryland planting area based on the ratio of vegetation index and SWIR increment was established.SWIR temporal curves of dryland crop compared with paddy crop in the growing period will have a noticeable rise and descent trend,based on this feature,the ratio index of SWIR band and vegetation index change was constructed.The disturbance of similarity between paddy crop and dryland crop temporal curves,and phenology variability within different type dryland crops was eliminated,and the dryland planting pattern recognition was realized.Using reference points,the overall precision of the single dryland crop recognition was 91.99%,the double dryland crop recognition was 90.86%.2.A new method of maize identification based on the temporal characteristics of remote sensing index in the peak period of crop growth stage was established.The water content of maize leaf is abundant,drought index temporal curve change obviously,compared with other dryland crop in the peak period of crop growth stage,based on this feature,constructs the index that can reflect the changing characteristics of the temporal curves of maize drought index rises and descent trend.Reducing disturbance of intra-class variability of temporal curve in maize,and the maize carries on the effective identification.Reference point data validate overall accuracy was 90.74%,Landsat8 OLI data classification results verify that the overall accuracy was 87.91%,and the fitting of statistic data was 0.8446.2005?2015 maize identification results showed that the total area of maize planting increased by 53.18%,mainly in Heilongjiang,Inner Mongolia and other regions.3.Based on the method of dryland planting area extraction and maize recognition,the study of the major dryland planting pattern in China was conducted in 2005-2015.The study showed that:Over 75%of the dryland planting area of single dryland crop in the past year.In the eleventh year,the area of dryland crop decreased by 6.76%,mainly in the single dryland crop,reducing the 137,667 km2,where spring maize decreased 16,151 km2,mainly distributed in Northeast China Plain;the double dryland crop increased the 92,006 km2,where "winter wheat plus summer maize" increased by 12.37%compared with 2005,mainly distributed in Hebei,Shandong and other provinces;a slight increase in the area of paddy-upland rotation and the change of region was mainly distributed in Jiangsu,Anhui and other provinces.