基于静止气象卫星的河北平原实际蒸散量遥感估算

Estimation of regional evapotranspiration over the Hebei Plain using geostationary satellite data

本文提出利用中国第1颗可操作性静止气象卫星风云2号C星(FY-2C)数据结合中等分辨率航天成像光谱仪MODIS产品估算河北灌溉农田实际蒸散量(ET)的方法,其中FY-2C的第1、2波段用于反演区域地表温度,再结合16 d MODIS合成的植被指数产品(MOD13),得到地表温度与植被指数的三角空间分布图(Ts-NDVI)。通过Ts-NDVI空间分布的关系,利用改良三角算法得到区域的蒸发比(EF)。最后结合MODIS地表反射率产品MCD43估算得到的日净辐射量,根据能量平衡计算得到该地区的日实际蒸散量。模型结果与地表Lysimeter观测数据比较,显示该模型估算得到的蒸发比和日蒸散量结果较为合理,误差在可接受范围。此外,FY-2C用于估算地表ET,其时间分辨率具有较强的优势,从而为获得多幅无云蒸散图提供了有利条件。

Fengyun-2C（FY-2C） is the first Chinese operational geostationary meteorological satellite with one visible and four infrared channels of optical imaging radiometer.Image acquisition temporal frequencies of FY-2C are 1 h and 30 min for dry and rainy seasons,respectively.This allows for mapping diurnal variations in land surface temperature（Ts） at a relatively high temporal scale.Data from FY-2C satellite were applied in combination with Moderate Resolution Imaging Spectroradiometer（MODIS） satel-lite products to assess regional evapotranspiration（ET） over the Hebei Plain.The approach was based on the improved triangle me-thod,where the temperature-vegetation index space included thermal inertia.As the method requires information on temperature change,the emergence of geostationary meteorological satellite data capable of determining land Ts at high temporal scale had made it possible to estimate daily Ts change from optimal observations.Two thermal infrared channels（IR1： 10.3-11.3 μm and IR2： 11.5-12.5 μm） from FY-2C were used to estimate Ts via a split window algorithm originally proposed for the MSG-SEVIRI sensor.As it compared reasonably well with observations of Ts,the FY-2C high temporal resolution data were exploited to determine morn-ing rise in Ts.The chosen time windows were 8：00 and 12：00 local time.Combined with 16-day composite MODIS vegetation indi-ces product（MOD13） at a spatial resolution of 5 km and Ts difference（ΔTs） from FY-2C,evaporative fraction（EF） was interpolated in ΔTs-NDVI triangular-shaped scatter space based on the improved triangle method.Using energy balance,regional daily actual ET was eventually derived from the EF and available energy estimated from MODIS surface albedo products MCD43.Spatial variations in the estimated surface variables（Ts,EF and ET） corresponded well to land cover patterns and farmland management practices.Estimates based on this method were validated with field observations from Luancheng Agro-Ecosystem Experimental Station of Chinese Academy of Sciences in Hebei Province（37o53＇N,114o41＇E）.The estimated ET and EF also compared well with lysimeter data collected for the period from June 2005 to September 2007,with RMSE of 0.92 mm for ET and 0.14 for EF and corresponding R2 of 0.74 and 0.55.Of particular interest was the high temporal frequency of FY-2C acquired images,increasing the likelihood of acquiring cloud-free images with the potential for plotting estimated daily ET maps.The study demonstrated that FY-2C was highly valuable in water resources and agricultural management applications in China.