Study On Rape Discrimination And Multiple Cropping Index Extraction Based On Modis-ndvi Time Series Data

Rape cultivation has a long history in China, one of the original countries. Currently, rape is the most important oil-bearing crops with more than half of the total planting area and more than 40% of the total yield of oil-bearing crops in China. Hubei Province is a major rape planting province, where the rape planting area and total yield have ranked the first in China for 19 consecutive years. Jianghan Plain is one of the major rape planting areas because of its superior natural conditions. Therefore, research on the rape planting conditions in Jianghan Plain is important for rape production management, rape layout regulation, agricultural guidance policies making of relevant authorities and the sustainable development. Remote sensing technology has the advantages of fast reaction, wide coverage and non-destructive monitoring, which has been widely applied to agricultural production. Moreover, the objective and scientific remote sensing technology can accurately extract large-range land coverage information. And it also can reduce human disturbance, which has been proved. Therefore, it can monitor the dynamic changes of rape planting and control the growth status of rape.In this paper, the study area was located in Jianghan Plain, the main rape cultivation area in Hubei Province. The MODIS data of the study area from 2002 to 2014 was collected. The research was carried out by the following ways: firstly, MODIS-NDVI time series data was calculated and generated by image stitching, projection transformation and image cutting, etc. Secondly, the noise was removed by empirical mode decomposition. Based on land survey data and other auxiliary information, the coverage types of other main surface features were distinguished during the same period according to the phenological characters of rape. In addition, the mixed pixels were decomposed by using linear spectral mixture model to identify the rape in the study area. Finally, according to the identification results, the rape planting area and its space distributions over the years were extracted. Then, the multiple cropping index of rape planting farmland was calculated, and the characteristics of its spatial and temporal distribution were analyzed. The main research results are as follows:(1) The noise can be effectively filtered from MODIS-NDVI time series data by Empirical Mode Decomposition(EMD).The results accurately showed the main phenological characters and growth circle of rape, rice, wheat, etc.(2) The linear spectral mixture model(LSMM) can be used to separate the endmembers of rape, wheat and forestland, etc, as well as their abundances satisfactorily. The results described the annual changes of rape planting. Furthermore, the identification precision of crops with remote-sensing image of medium spatial resolution was improved.(3) According to comparison between interpreted area and statistical data of rape in the study area, the estimation accuracy for the total rape planting area was over 92.5%. Meanwhile, linear regression analysis was carried out on the interpreted rape planting area, which presented remarkable correlation, with 2> 0.9. Furthermore, the rape planting change was accurately described by the planting space distribution over the years: rape planting area increased on the whole, but the planting area reduced significantly as affected by policies and climates from 2006 to 2008. In addition, rape planting tends to gather in several minor cities.(4) The multiple cropping index of the planting rape fields violated with verification precision about 78%, which illustrated the trend of planting patterns in the study area. As affected by market guidance and policies, the index constantly increased around 2006. However, in recent years, with the rural labor force migrating to cities, the multiple cropping index gradually reduces and tends to stable. For the rape fields in the study area, the main planting pattern is one-year-two-harvest production. Generally, rice is planted in the second season.The methods in this paper effectively identified the rape planting area. Moreover, the results can be used for analyzing the time and spatial dynamic changes of rape planting area and multiple cropping index, which provides effective means of remote sensing technology for rape monitoring.