近红外比值法在对流层航空遥感数据大气水汽估算中的改进

Improvement of near infrared ratio method in troposphere water vapor estimation with airborne remote sensing data

利用近红外波段大气窗口通道和水汽吸收通道辐亮度比值反演大气柱水汽含量,是卫星遥感大气水汽估算的通用方法之一.但对于对流层内的航空遥感水汽估算,直接套用卫星遥感水汽估算近红外比值法会引入飞行平台到大气顶层水汽的影响.根据航空遥感成像特征,利用Modtran和热力学初始分析资料(thermodynamic initial guess retrieval,TIGR)大气廓线库数据,分别构建入射路径上,航飞高度到地表的水汽透过率与太阳到地表水汽透过率的对数之比G与航飞高度内大气水汽与整层大气水汽之比R,以及入射路径上的航飞高度到地表的水汽透过率,与出射路径上地表到入瞳处水汽透过率的对数之比H与太阳入射角qs的函数关系,结合下垫面特征,建立对流层航空遥感水汽估算模型.以1614组TIGR廓线为输入模拟航飞入瞳处辐亮度,利用本文模型估算对流层内大气水汽,并与廓线数据直接计算值对比,结果表明,当航飞高度在1.0~7.0 km时,模型估算值的总体精度为0.22 g/cm^2,且精度优于0.5 g/cm^2的样本占总样本数95.30%.利用2014年5月28日郑州上街航空遥感试验获取的影像进行水汽分布估算,并与同步大气探空数据计算到的水汽进行对比,结果表明,各样区估算值与探空值的RMS误差为0.16 g/cm^2(12.8%),且对下垫面覆盖条件的先验了解能够提高模型估算精度.本文模型消除航空遥感飞行高度以上大气的影响,增大了模型的精准度与适应性,为热红外航空遥感数据实时大气校正提供了可靠的输入.

Using near-infrared radiance ratio between absorption channel（940 nm） and window channel（860 nm） to estimate water vapor content is a common method to retrieve water vapor with remote sensing data. For troposphere water vapor estimation with airborne remote sensing data, ready-made near-infrared ratio method introduces the influence of moisture between flight platform and the top of atmosphere. According to the aerial infrared camera image characteristics, using Modtran and TIGR data, define the following parameters： 1） R, the ratio between the water vapor content from surface to flight platform and the total atmospheric water vapor content; 2） G, the logarithm ratio between the water vapor transmittance from flight platform to land surface and the one from sun to surface on the incident path; 3） H, the logarithm ratio between the water vapor transmittance from flight platform to land surface on the incident path and the one from surface to flight platform on the emergent path, and build functional relation between G and R, H and solar incident angle respectively, and considering surface characteristics, build troposphere water vapor content retrieving model. Modtran simulation shows that the estimation precision is 0.22 g/cm^2 in a camera height of 1–7 km, and precision less than 0.5 g/cm^2 account for 95.30%. The measured data from airborne remote sensing experiment in Shangjie, Zhengzhou on May 28, 2014 shows that the estimation error is 0.16 g/cm^2（12.8%） compared with simultaneous balloon sounding data, and priori knowledge about underlying surface improves model precision. This paper eliminates the influence of atmosphere above flight platform, increases model accuracy and adaptability, and provides reliable input for real-time thermal infrared remote sensing atmospheric correction.