Lai Scale Effect Study In Heihe Oasis Based On CASI Data

Leaf area index is one of the key parameters to characterize crop canopy structure.Accurate acquisition of LAI is of great importance for crop growth monitoring and yield estimation.Remote sensing provides technical support for LAI estimation in continuous regional scale or even global scale.However,the nonlinearity of the LAI inversion model and the spatial heterogeneity of remote sensing data,leads to a certain scale error in the LAI inversion results,which restricts the unified application of LAI products among different remote sensing data sources.Therefore,a challenging problem which arises in this domain is to study the scale effect of LAI.Taylor series expansion model,which is widely applied to scale effect research has an advantage of quantificationally describing the degree of nonlinearity of LAI inversion model and the comprehensive influence of spatial heterogeneity on LAI scale effect.However,it can only use one-dimensional variables as correction factors,which leads to a certain influence on correction accuracy.In this paper we apply this method in Heihe region,carry out LAI multiscale inversion based on LAI statistical model,discuss the difference characteristics and mechanism of LAI scale effect.Combined with Taylor series expansion model quantitatively analyze the scale error,which is caused by the LAI inversion model's nonlinearity and Spatial heterogeneity.Setting up scale error correction models based on one dimension variable and two dimensional variable,achieve the scale difference correction of CCD-LAI remote sensing products.The main contents and conclusions are as follows:First,Using Airborne Hyperspectral CASI image calculate four kinds of vegetation index combined with measured LAI establish six statistical models to invert LAI.The correlation between the LAI estimated from each inversion model and the measured LAI is analyzed.Select the combination of red band and near infrared band,which has the highest sensitivity to the measured LAI value,by using simultaneous matrix.Obtain LAI Inversion model of the best three order polynomial regression model.Second,discuss the mechanism of LAI scale effect and multilevel analysis Multispace LAI scale inversion's difference based on the scale difference characteristics of the two level of Repolymerization after inversion and Back inversion after polymerization and three methods of ascending scale.The results show that the overall distribution of LAI values is roughly the same under the three methods.However there is still a certain scale difference caused by the nonlinearity of LAI inversion model and NDVI variable nonlinearity and the scale effect of LAI remote sensing products(Nonlinearity Inversion model and Spatial heterogeneity).Third,refine the mechanism of scale effect and expand the Taylor series of the optimal LAI inversion model,quantitative analysis the scale effect of multiscale LAI inversion.The results show that the two order derivative term of the Taylor series expansion model can better explain the scale effect caused by nonlinearity in LAI inversion model,the variance terms can better explain the scale effect caused by spatial heterogeneity.Fourth,based on Taylor series expansion model,then use two order derivative and variance term as correction factor of One dimensional variable(NDVI)and Two dimensional variable(Near-infrared and Red-band reflectivity),the scale effect in CCD-LAI remote sensing products has been effectively eliminated and the inversion accuracy of the CCD scale LAI has been improved.The result of two methods of scale correction accuracy evaluation shows that the Taylor series expansion model based on two dimensional variable has better correction effect.In summary,this research is devoted to mining and effectively utilizing effective information in hyperspectral data.Take it as a sub-pixel,applied Taylor series expansion model,corrected the scale effect of large scale and low resolution LAI remote sensing products.Finally,remote sensing monitoring of large scale and highprecision LAI can be realized.Aslo can provide technical support for crop dynamic monitoring.