Mapping Major Crops Spatial And Temporal Distribution In China Using MODIS Time-Series Data

Over the past20years, great complex changes have been occurred in the quantity,quality and structure of Chinese land use, which require us to have a good grasp of thenational arable land resource status and food production trend to develop relatedpolicies. Crop is an important part of the geographical environment. Crop remotesensing is one part of the resources remote sensing, and it creates good conditions foragricultural development. Crop type identification lays the foundation for thedevelopment of agricultural remote sensing and ensures the healthy and orderlydevelopment of agriculture and it provides important information about agriculturaldevelopment and food production.In this paper, the research region is the whole china. Combiningagro-meteorological observation data and MODIS data with a resolution of500m at8-day intervals, we rebuilt the EVI (Enhanced Vegetation Index) time series curve andextracted Chinese cropping systems, and researched the identification and extraction ofmain food crop include winter wheat and maize. The main methods and results are:(1) MODIS data was pre-processed and then EVI time-series curves were built.The curves present a sawtooth waveform because of the noise affects. In this paper, weuse HANTS, which use the method of the Fourier transform and the generalized leastsquares as the core, to reconstruct the crop vegetation index curve. The results showthat time series curve can well reflect the characteristics of crop growth andmulti-cropping rotation, and it is feasible on the critical phenological charactersextraction.(2) Combining with agro-meteorological observation data, we used the method ofcurve turning point to recognize the key crop phenological stages and extractedcropping systems (multiple cropping and crop calendar) by establishing extractionmodel. Based on this, we can further large regional scale estimation of agriculturalproductivity. Analysis showed that the difference between the crop phenology based onremote sensing and phenological data of agricultural weather station were in±16days.It can be seen, MODIS/EVI curves are of great potential in recognizing regional agricultural cropping systems.(3) Based on the critical phenological phases and cropping system information formajor crops and combining with agro-meteorological observation data, we adoptedthreshold method and established the identification method model of the winter wheatand maize. Threshold was determined after repeated commissioning studies. Becausesimilar phenological characteristics of different crop types exist interference, weimprove the identification accuracy according to the Chinese crop comprehensiveregional distribution characteristics. Finally, the extraction results were validated andthe extraction accuracy is82.56%.