本文将遥感信息与作物模型同化实现作物生长参数的时空域连续模拟,进而监测生长参数的时空域变化.首先将作物模型WOFOST(World food studies) 与冠层辐射传输模型PROSAIL 耦合构建WOPROSAIL 模型,利用微粒群算法(PSO) 通过最小化从CCD 数据获取的土壤调节植被指数观测值SAVI(soil adjusted vegetation index) 与耦合模型得到的模拟值SAVI’之间差值优化作物模型初始参数.通过MODIS 数据反演实现参数的区域化,并将区域参数作为优化后作物模型输入参数驱动模型逐像元计算生长参数,实现生长参数的时空域连续模拟与监测,最终建立区域尺度遥感-作物模拟同化框架模型RS-WOPROSAIL .结果表明：同化模型解决了作物模型模拟空间域和遥感信息时间域的不连续问题.模型模拟的叶面积指数(LAI) 、穗重(WSO) 、地上总生物量(TAGP) 等生长参数较好地体现了水稻生长状况时空域变化,研究区水稻模拟产量与实际产量的误差为27.4% .

Continuous simulation of crop growth parameters at spatial-time scale is a key technique for monitoring crop growth status and precision agriculture. This paper realized the spatial-time scale continuous simulation of growth parameters with the assimilation of remote sensing information into crop growth model, monitoring growth parameters changes on spatial-time scale. Construct a model named WOPROSAIL with the coupling of crop growth model WOFOST and canopy radiative transfer model Prospect+Sail (PROSAIL). Then particle swarm optimization (PSO) algorithm was used to minimize difference between observed values of soil adjusted vegetation index (SAVI) derived from CCD data and simulated values of soil adjusted vegetation index (SAVI’) calculated by coupling model for optimizing initial parameters of WOFOST. Regionalization of parameters was achieved with MODIS data retrieval, then by inputting these regional parameters, optimized WOFOST model, initial parameters of which were optimized, was driven for each pixel and then regional growth parameters were calculated, achieving continuous simulation of crop growth parameters on spatial-time scale. Finally, a region scale remote sensing-crop simulation assimilation framework model named RS-WOPROSAIL was constructed. The results indicated that assimilation model solved the discontinuity of spatial scale simulation by crop growth model and time scale retrieval by remote sensing information. Growth parameters simulated by optimized crop growth model, including leaf area index (LAI), weight of storage organs (WSO) and total above ground production (TAGP), preferably refl ected the changes of rice growth status on spatial-time scale, and the relative error between simulation yield and actual measurements was 27.4%.