Study On Soil Effects Removal Methods Of NDVI In Farmland Based On The Spectral Mixture Theory

The Normalized Difference Vegetation Index(NDVI)is one of the most widely used vegetation index in the area of remote sensing,which is used for crop growth monitoring,agricultural yield forecasting,drought forecasting and precision agriculture.While NDVI can eliminate the influence,such as atmosphere,shadow,sensor calibration,observation angle,etc,the soil background can make NDVI produces greater error.Because the winter wheat farmland NDVI which is in the same growth condition but grows up under different background of soil types has large difference,it is not easy for us to use NDVI to monitor wheat growth conditions effectively and evaluate wheat growth conditions precisely.Most of the existing methods to remove soil influence of NDVI is to develop and improve the existing NDVI model.Those vegetation index need consider the complex soil lines of research area,thus it can not form a product like NDVI and be used widely to monitor crop growth condition in a large and wide scale.Hence,study soil background removal method of NDVI is still hot and difficult nowadays.On the basis of simulating and analysing the influence of farmland background of typical soil types to winter wheat N-DVI under different vegetation coverage and mixed spectrum theory,the paper comes up with two soil background effect removal models(NDVIT)based on NDVI The paper uses winter wheat farmland in ChuZhou City,AnHui Province as research area,uses winter wheat in early jointing stage under the background of paddy soil and Yellow-cinnamon soil as study object,uses measured winter wheat canopy spectrum and leaf area index data,uses tranditional photographic method to calculate vegetation coverage and then study soil effects removal abilities of two models.Then,the paper uses winter wheat farmland in JiNing City,ShanDong Province as research area,uses winter wheat under the soil background of cinnamon soil,moisture soil,paddy soil and Shajiang black soil as study object,conbines with multi-spectral remote sensing data of Landsat-8 OLI satellite and finally confirms the applicability and validity of soil background effect removal models.The research contents and conclusions are as follows:(1)Conbined with typical soil types and vegetation coverage,it studys the influence of different soil background to winter wheat farmland NDVI.And it shows that different types of soil background causes great differences to winter wheat farmland NDVI,and the sensitivity of winter wheat farmland NDVI to vegetation coverage has obvious difference.The studys provide the basis for NDVI estimation frequency of all wheat growing period under diverse types of soil background.(2)Based on linear mixed spectrum theory,it establishes the soil background effects removal model(NDVI1T).And NDVI2T which is simplified by NDVI1T,is based on the theory of the dimidiate pixel model.It comfirms the applicability and validity of models.It quantitatively studies the abilities of soil noise influence resistance based on signal to noise ratio analysis,and it shows that NDVI1T owns better abilities of vegetation extraction and soil noise resistance.In the condition of low-medium LAI(LAI is less than 3),two models are more applicable to the situation which vegetational leaves is evenly covered or vegetation type is single.Both of the models have a good fitting relationship with NDVI and the correlation coefficients come to 0.9,which shows the feasibility of correcting NDVI in a large scale by using NDVIT.(3)It shows the confirmation of soil background effects removal model by using Landsat-8 OLI remote sensing image.It shows the feasibility of correcting the effects of soil background for NDVI products in a large scale by means of the fitting formula of soil effect removal model(NDVIT)and NDVI although it has four tpyes of soil background in research area.