Research On The Inversion Of Vegetation Leaf Area Index Based On GOST2 Mode

Leaf area index(LAI)is an important vegetation parameter characterising leaf sparseness and canopy structure,and is one of the key indicators for studying the global and regional carbon cycle,hydrological cycle and regional response to climate change.Traditional LAI acquisition methods are mostly field surveys,which are time consuming and demanding in terms of weather and environmental conditions,whereas remote sensing technology can quickly and accurately acquire large scale feature information,so remote sensing is currently the main common method to acquire regional LAI.There are a variety of global LAI remote sensing products available,but there is still a great deal of uncertainty in the accuracy of the products,especially in areas with complex topography and subsurface.The GOST2 optical model takes into account the complex canopy structure and photon scattering mechanisms within the canopy,and uses“GO+ray tracing”to simulate the area ratio of the four canopy components,which not only has It is not only physically meaningful,but also more accurate in modelling the reflectance of canopies in complex terrain.Therefore,this paper presents a study of LAI inversion based on the GOST2 optical model and its application in the Jianghuai basin.The main conclusions are as follows:(1)The vegetation indices(NDVI,RVI,DVI,MSAVI)and LAI based on the simulation results of the GOST2 optical model all show some correlation,and the correlation coefficients between the vegetation indices and LAI are enhanced by the modelling strategy based on leaf type and the modelling strategy based on slope,therefore,this paper adopts the strategy based on slope and leaf type for the inversion of LAI.The correlation coefficients between the inversion models of low slope,medium slope and high slope under broad-leaved vegetation conditions and LAI were 0.81,0.79 and 0.81,respectively,while the correlation coefficients between the inversion models of low slope,medium slope and high slope under coniferous vegetation conditions and LAI were 0.85,0.84 and 0.85,respectively.(2)The spatial and temporal correlation coefficients between the MODIS-based LAI inversion products and the domestic LAI products GLOBMAP LAI,GLASS LAI and GLOBALBNU LAI are good,with the monthly average LAI correlation coefficients exceeding0.99.The inversion results are consistent with the domestic LAI inversion products in nine major basins,different DEM classifications,and the annual average values of different features.The results of the inversion are consistent with those of the domestic LAI inversion products,and can show the temporal variation of LAI in different regions.The correlation coefficients with GLOBMAP LAI,GLASS LAI and GLOBALBNU LAI are 0.91,0.85 and 0.86respectively,areas with large spatial differences are concentrated in areas with complex topography,such as South-eastern coastal river basins and woodlands.The AE and RMSE of the inversion results with GLOBMAP LAI and GLOBALBNU LAI are mainly concentrated below 0.5 m~2/m~2,and with GLASS LAI the AE and RMSE are larger,but all below 1 m~2/m~2.(3)The LAI inversion products based on Landsat8/9 data for the Jianghuai basin have obvious monthly variations:July-August>March-April>May-June>September-October>January-February and November-December,which is consistent with the monthly variations of domestic LAI products in the Jianghuai basin.The monthly variation of LAI in the northern and western and southern mountainous regions of the Jianghuai basin is similar to the national monthly average LAI variation,rising from January to August and falling from August to December.However,in the central cultivated areas,LAI has a decline in May-June due to the rice and wheat crop rotation in the Jianghuai basin,which leads to a downward trend in overall LAI in the Jianghuai basin from May to June as well.The absolute errors of the inversions using Landsat8/9 data and the measured LAI were 0.28 m~2/m~2,0.40 and 0.38 m~2/m~2,respectively,and the root mean square errors were 0.53 m~2/m~2,0.63 m~2/m~2 and 0.62 m~2/m~2,respectively,in the cultivated land of Shangqiu City,the Dabie Mountains of Lu’an City and the forested land of Pingdingshan City,which were all reduced compared with the inversions using MODIS data.The results of this paper indicate that LAI inversion based on the GOST2 optical model has some potential to be applied to both medium and high resolution remote sensing data,and can provide a reference for producing LAI inversion products with lower errors.