Study Of Leaf Area Index Inversion Based On Multi-temporal And Multi-angular Remote Sensing Data

Research on the application of remote sensing has evolved from the static stage to the dynamic stage,from single perpendicular observation to multi-angular observation.On the one hand this advanced technology provide plentiful data sources for research on monitoring dynamic change of the land surface by remote sensing%quantitatively describing scaling effect with topographic data and exploring the characteristics of directional reflectance of land surface target; on the other hand the comprehensive utility of multi-angular and multi-temporal remote sensing data could provide more reliable and affluent target information of temporal-spatial variation、 spatial distribution and three-dimension structural traits.As an essential indicator in reflecting the individual characteristics and group characteristics of vegetation LAI has been one of key variables in land surface process models. Aimed to resolve the low prescion of LAI (leaf area index) inversion of crops, this paper employed multi-temporal、 multi-angular data from material spectral as well as satellite observation try to develop the methods, accordiong to the changing traits of crop (winter wheat) growth stage and canopy structural traits, to improve LAI inversion prescion of crop. The final conclusions are presented as follow:(1)Given full consideration of the dynamic changes of winter wheat LAI during the whole growth stage of winter wheat,this study established segmented LAI inversion models based on correlation analysis between vegetation index and LAI in different wheat growth stages.The results suggested that the determinant coefficient (R2) and RMSE between LAI inversion values and the true values were0.5585and0.3209respectively during the whole growth duraton. The mSR index was suitable enough to inverse LAI during the earlier growth stages (before jointing stage) for winter wheat.The correlation coefficient and RMSE between LAI inversion values and true values were0.7287and0.2971respectively. The SR index was chosen to inverse wheat LAI in the medium growth stages (from joingting stagess to heading stagess).The correlation coefficient and RMSE between LAI inversion values and true values were0.6546and0.3061respectively. While the NDVI index was proven to be fine to inverse wheat LAI during the later growth stage(from heading stage to ripening stage).The correlation coefficient and RMSE between LAI inversion values and true values were0.6794and0.3164respectively.Therefore, it was suggested that the LAI inversion results from segmented inversion models, using propriate vegetation index in the light of the variations in the wheat cover indifferent growth stages, were much better than the inversion result from single NDVI index during the whole growth stage for winter wheat.(2)This paper further analyzed the bidirectional reflectance characteristics of two types of geometric wheat, erective variety J411and loose variety ZY9507, based on the semi-empirical BRDF(bidirectional reflectance distributio function) kernel driven model and six multi-angular indices in red band (680nm)and in NIR band (800nm) employing field-measured experimental data. The results revealed that the NDVI of both erective and loose varieties showed the distinct anisotropy and the bidirectional reflectance of the two varieties performed differently in the two bands which were mainly ascribed to their own property of LAD(leaf angle distribution) spatial distribution wavelengths> illumination-viewing geometries and noise. Finally, it was concluded that the wheat canopy had significant geometric effect in red band and volumetric scattering effect in NIR band. Moreover the geometric effect of erective wheat was much stronger in red band whereas the volumetric effect of loose variety was more significant in NIR band.(3)In order to resolve the limitation of canopy geometry on wheat LAI inversion accuracy, this paper developed two hot indices, MNDHD (Modified Normalized Difference between Hotspot and Dark-spot) and HDRI (Hotspot and Dark-spot Ratio Index) and established the inversion models for the erective variety J411and horizontal variety ZY9507respectively. The inversion model was built based on combining the general vegetation index, such as NDVI, and hotspot index, such as NDHD. The results indicated that the combined indices were feasible enough for LAI inversion referring to crop geometry,(4)This paper further inversed the true LAI, which was inversed by considering the canopy clumping index, of land surface vegatation of China and extract the temporal and spatial changing information validated the LAI inversion as an extension of the inversion method, taking the dynamic changes of wheat growth stages and canopy structureal characteristics into consideration, employing SPOT-VEGETATION S10data from2000-2010year. Furthermore the inversed SPOT LAI of China was compared with the MOD15A2LAI data.The results showed the SPOT LAI value were obviously higher than the MOD15A2LAI value, especialy for the forest species. For the deciduous broadleaved forest the SPOT-LAI value was about27%higher than the MODIS-LAI and for the evergreen needleleaved forest30%higher than the MODIS-LAI, which means vegetation structural parameter, the clumping index, exerted a great effect on vegetation LAI and the clumping index greatly improved the reliability of vegetation LAI inversion result.