基于GF-1数据的夏玉米FPAR遥感动态估算

为探索高分一号卫星（GF-1）估算农作物光合有效辐射吸收比率（Fraction of absorbed photosynthetically active radiation, FPAR）的潜力，以田间小区与大田夏玉米为对象，基于GF-1卫星的16 m空间分辨率宽视场（Wide field view, WFV）传感器光谱响应函数对地面实测冠层高光谱反射率进行重采样，获取GF-1 WFV的模拟反射率，构建宽波段植被指数，利用与FPAR极显著相关且具有较高相关系数的植被指数，建立不同生育期夏玉米FPAR的一元与多元逐步回归模型，筛选FPAR估算的最适模型，并在此基础上实现县域尺度不同生育期的FPAR动态估算。结果表明：模拟宽波段光谱反射率与GF-1 WFV光谱反射率间的相关系数|R|为0.967～0.985,决定系数R²为0.935～0.969;基于模拟反射率构建3波段植被指数与FPAR的相关性优于2波段植被指数，增强型植被指数（EVI）、土壤调节植被指数（MTVI2）、可见光大气阻抗植被指数（VARI）、综合植被指数（TCARI/OSAVI）等3波段植被指数与FPA...

Dynamic Estimation FPAR of Summer Maize Based on GF-1 Satellite Data

The fraction of absorbed photosynthetically active radiation (FPAR) is a key parameter in various of photosynthetic capacity and productivity potential of crops. The great progress of quantitative remote sensing and various data products make FPAR products widely used in carbon cycle and vegetation research in different regional scales. In order to explore the FPAR estimation capability based on the GF-1 wide field view (GF-1 WFV), the canopy spectral reflectance and FPAR of summer maize were measured from regional field and field plot experiments, including five nitrogen fertilization levels and two summer maize varieties from jointing to maturity stage. Firstly, the multi-spectral broadband reflectance was simulated by using the measured canopy hyperspectral reflectance based on spectral response functions of GF-1 satellite with a spatial resolution, and then the vegetation index was established by this simulated reflectance. Secondly, totally 17 vegetation indices were selected on the basis of previous studies, and the quantitative relationship between FPAR and vegetation indices at different growth stages was analyzed. Thirdly, the vegetation indices with high correlation coefficient and extremely significant correlation with FPAR were selected, and estimation models of summer maize FPAR by a linear regression model or multiple stepwise regression model respectively, by analyzing determination coefficient (R2), standard error (SE), root mean squard error (RMSE) and relative error (RE) of the estimation model and validation model, the optimal model for FPAR estimation was screened. Finally, the optimal estimation models were used to estimate the FPAR dynamic variation and spatial distribution for summer maize from jointing to maturity stage by GF-1 satellite data. The results showed that the correlation coefficient (|R|) between simulated broad band spectral reflectance and GF-1 spectral reflectance was 0.967~0.985, and the determination coefficient (R2) was 0.935~0.969, it showed that these was highly consistent between simulated spectral reflectance and GF-1 spectral reflectance. There was a good correlation between FPAR and the vegetation index constructed based on simulated reflectance, and the correlation coefficient of 3-band vegetation indexes was better than 2-band vegetation indexes, in particular, it was extremely significant (P<0.01) between FPAR and enhanced vegetation index (EVI), modified soil adjusted vegetation index 2 (MTVI2), visible atmospherically resistant index (VARI), TCARI/OSAVI, and |R| was 0.813~0.925. The simple linear regression model and multiple stepwise regression model of FPAR were established by EVI, MTVI2, VARI, TCARI/OSAVI, and the coefficient of determination (R2) of estimation model was 0.762~0.843, the coefficient of determination (R2) of validation model was 0.839~0.880, and the relative error (RE) was 7.037%~9.571%, it showed that the multiple stepwise regression model was better than simple linear regression model, and the multiple stepwise regression model could better estimate FPAR at different growth stages. The optimal model was used to estimate the spatial distribution and dynamics of FPAR at regional scale, and the measured values were validated. The R2 between the estimated and measured values was 0.819~0.856, and the relative error (RE) ranged was 8.41%~13.37%. These results indicated that the spatial distribution and dynamic variation of FPAR estimated based on simulated GF-1 WFV of hyperspectral reflectance were consistent with the actual spatial distribution, which provided a scientific basis for estimating regional FPAR and production potential of maize based on high resolution remote sensing data of GF-1 WFV.