MODIS数据辅助的GF-1影像晴空光合有效辐射反演

面向农作物产量监测对中高分辨率遥感数据光合有效辐射(photosynthetically available radiation,PAR)反演的实际需求,该文选择山东省禹城市2014年1月至2014年12月共13景GF-1/WFV卫星影像作为数据源,基于中分辨率成像光谱仪(moderate-resolution imaging spectroradiometer,MODIS)地表反射率产品作为辅助数据源,开发了适于业务运行的WFV数据气溶胶光学厚度(aerosol optical depth,AOD)及PAR的反演算法。算法核心是采用6S(second simulation of satellite signal in the solar spectrum)大气辐射传输模型,建立包括AOD在内的大气参数与查找表(look-up table,LUT),结合大气顶层太阳入射辐照度及卫星入瞳处辐射亮度值反演地表反射率数据,通过与WFV蓝光波段地表反射率数据对比获取大气参数。通过反演的大气参数计算400~700 nm连续光谱区间的PAR值,并建立WFV数据离散红、绿、蓝光波段与连续光谱区间PAR的转换系数,实现WFV数据PAR的反演。其中,WFV蓝光波段反射率数据与MODIS地表反射率数据关系、离散到连续谱段PAR的关系可以从美国地质勘探局(United States Geological Survey,USGS)提供的典型地物波谱库数据理论计算获取。利用中国生态系统研究网络(chinese ecosystem research network,CERN)禹城站地面观测值进行验证结果表明,该文提出的算法总体精度达到92.63%,平均绝对误差为14.56 W/m~2,平均相对误差7.37%,具有业务应用的潜力。

GF-1 image clear sky PAR inversion assisted by MODIS data

Abstract: GF-1 satellite was launched in April 2013. Since then various researches and operations using the GF-1 imagery have been conducted. This study tries to establish the inversion model for the biophysical variable photosynthetically active radiation (PAR) derived from GF-1 imagery. The inversion of the variable aerosol optical depth (AOD) is first carried out using the surface reflectance product of MODIS (moderate-resolution imaging spectroradiometer) (it is MOD09A1). Based on the typical ground-object spectral library data provided by the United States Geological Survey (USGS), and spectral response function of the GF-1/WFV sensors, the study has found that the difference in terms of the blue band surface reflectance generated by 2 sensors of GF-1/WFV and MODIS is very small. MODIS data can thus be assimilated to create GF-1 surface reflectance library. The second simulation of satellite signal in the solar spectrum (6S) atmospheric radiation transfer model is applied to produce an atmospheric parameter look-up table (LUT). By comparing solar incident irradiance at top of atmosphere (TOA) and radiation brightness value at the entrance pupil of the sensor, and using surface reflectance product from MODIS, the authors have achieved the ground-air decoupling and the inversion of AOD. After establishing AOD library, the atmospheric parameters such as atmospheric transmittance, atmospheric hemisphere albedo, and atmospheric diffuse transmittance are computed and interpolated to LUT. Combined with the calculation of sun incident radiation intensity, the ground solar radiation intensities at red, green and blue bands of GF-1 are produced. By analyzing the relationship between the ground solar radiation intensities and overall 400-700 nm PAR values provided by MOD09A1, 3 conversion coefficients are computed. The coefficients have the values of 0.09156, 0.09951, and 0.1007 respectively for the blue, green and red bands, realizing the inversion from the ground solar radiation intensity of 3 discrete wavebands to PAR. By selecting Yucheng region, Shandong Province as a study area, and using 13 GF-1 images registered from January to December in 2014, the authors generated the values of PAR for the study region and conducted the accuracy assessment for the inversion using the ground measured values of PAR. The results show that the PAR acquired by the inversion approach described in this paper has an overall accuracy of 92.63%, with an averaged absolute value of error of 14.56 W/m2 and an averaged relative error of 7.37%. Study shows that GF-1 data can be effectively applied for the inversion of PAR; the establishment of reflectance library is feasible by including MOD09A product, and generating LUT through the 6S model. Finally this PAR inversion approach can provide reliable data support for crop growth monitoring, and parameter estimation based on GF-1 imagery.