Study On The Method Of Grass Yeild Model

This thesis chose Xilinguole grassland, the typical grassland in the semi-arid temperate zone, as a research area. The method of grass yield model was studied by some statistics methods, such as related analysis, single regression, multiple regression and ridge regression. The model's precision was validated. The yield of grass was estimated by the model which was chosen. The thesis chooses the hotpot problem as study content and has theory meaning and application value. The main development and innovation of the thesis are the followings:(1)The cubic remote sensing estimation model was found in MODIS-VI and square sample with regression method. y=-901.287+15548.395x-35026.104x2+34419.533x3y is the estimated grassland production in Xilinguole. x is the synchronous NDVI. The R2 is 0.394 and the precision is 69.31%.The remote sensing model was used by estimating the different types of grass yeild and validating the precision. The precisions of NDVI models were higher than EVI model. In addition, the precision of cubic and quadratic model was significantly higher than the linear model in the area of higher and lower grassland production.(2)The synthetic estimation model was found in meteorology data, MODIS-VI and square sample by using multiple regression method. The estimation model was chosen as the model of NDVI, valid precipitation and the valid dryness. y=1064.357+7003.184NDVI-1111.753 valid precipitation +353.123 valid drynessy is the estimated grassland production in Xilinguole. The R2 is 0.4350 and the precision is 70.66%.(3)The ridge regression model was built by meteorology data, MODIS-VI and square sample. The estimation model was chosen as the model of EVI, valid precipitation, valid temperature and the valid dryness. y=1897.02+10577.47EVI-521.02 valid precipitation -472.021 valid temperature +141.44 valid drynessy is the estimated grassland production in Xilinguole. The R2 is 0.4540. The precision is higher than multiple regression. The precision is 73.40%.(4)The ridge regression model i.e. the model of EVI, valid precipitation, valid temperature and the valid dryness was chosen as the grassland production estimated model in 2008 by contrasting the three models.(5) The meteorological factors were lead to the grass yield model. An ideal and methods was lead for discussing the effect of Phonology to grass yield.