Abstract:
Abstract: Township is the main direct part of duty for straw burning. It is very important for an effective method to estimate the total amount of straws in small-scale region, such as Chinese townships. Research of township regional estimation methods for crop straw production would provides an effective approach for the township straw resources' planning and storage facilities location. This study aims to estimate straw production and their spatial distribution by using MODIS time series data and free world map image data in Chemen experimental site of Sihong County. The estimation process included three steps. Firstly, an uniformly distributed and relatively large field patch were chosen as the sample plot according to the electronic map. A field survey was conducted to investigate the current planting types. Then the "Crop type-EVI parameter" discriminant function and rules were established by setting crop type as classification variable and annual time series EVI as discriminatory variable, its MODIS-EVI time series data of the sample plot was collected online, and the accuracy rate was 97.37% in the case of setting confidence level of 95%.Secondly, sample plots' images were recognized based on the field surveys data. The discriminant analysis was used to implement the "Crop type-EVI parameter" discriminant function and rules. There blocks were discriminated as rice stubble wheat planting type, and 8 for dryland wheat type, free lands exactly judge right, The correct total of 109 blocks, the judge accuracy rate of 90.8%; after the exclusion of free lands, the judge accuracy rate of 88.2%.In order to improve the classification accuracy, this paper adopted assistant judgment by the Chemen township autumn map world remote sensing image data in 2012. Meanwhile, field investigation and verification were combined with, drawn the image of Chemen township straw resources type distribution map. The residents, woodland, rivers and other objects classify through on-the-spot investigation. Wheat, rice stubble wheat, dryland wheat, free lands were classified by discriminant classification rules, and then combined with time series data to revise the field investigation results. Thirdly, we extracted the free world map image data corresponding to each field patch and judged the crop types in the region according the rule approached. Subsequently, straw resource area and straw yield and of the whole county were estimated, which included 1.77 million tons of rice straw resources, collecting rice straw resources volume of 1.27 million tons, 2.78 million tons of wheat straw resources, collection of wheat straw resources amount to 1.51 million tons, the dryland of Gangzhu, Tuanjie village planted corn, soybean, peanut mixed cropping, but unable to statistics identify and estimate. The results showed that this method could get more accurate straw production area, resources yield and the spatial distribution data in small-scale regions, compared with the traditional investigation method, this method is faster and easier. All operations of the process have been developed into plug-in, which is integrated into the software Croplab. Users only need simple operation knowledge can be completed, and does not require too much knowledge of statistics and technical, especially for investigation and planning of rural area of straw resources.