水环境参数定量遥感反演空间尺度误差分析

摘要

对地观测技术的快速发展为水环境定量遥感监测提供了多源遥感数据的有利支撑,然而多源数据空间尺度差异引起的遥感数据和定量产品的不一致性严重制约了水环境科学研究和业务化应用.针对多源遥感数据的空间尺度转换和尺度误差问题,本文提出一种模拟遥感成像过程点扩散函数 (PSF) 的多源遥感数据空间尺度转换方法.以高空间分辨率GF-1卫星16m遥感数据为基础,模拟了常用的内陆水环境监测卫星Landsat TM/ETM+/OLI(30m)、Terra/Aqua MODIS (250m、500m、1000m)数据,系统研究了高动态浑浊水体(以鄱阳湖悬浮颗粒物监测为例)多源定量遥感监测的空间尺度误差,并对比分析了常用的多源数据尺度转换方法(基于遥感反射率数据平均法和基于悬浮颗粒物产品平均法)的有效性.结果表明,基于点扩散函数的空间尺度转换方法与传统方法具有较高的相关性,基于遥感反射率数据平均法的水环境定量遥感产品的误差水平低于基于悬浮颗粒物产品平均法;相对于南海等相对平稳水体的低空间尺度误差水平(<0.5%),在高动态浑浊的内陆或近岸水环境遥感监测中,由空间尺度变化引起的产品误差可达±5%左右.因此,在高精度水环境定量遥感发展应用的需求和多源多尺度遥感数据协同的背景下,本文研究对于提高多源遥感监测产品的一致性和应用能力具有重要理论和现实意义.%The rapid development of the Earth Observation (EO) technology offers substantial data for remote sensing monitoring of water environment, however, the research and applications are often impeded by inconsistent products from multi-resolution remote sensing dataset.A point-spread-function (PSF) based spatial resolution transformation approach was proposed, to mitigate scale errors from multi-platform sensors of varied spatial resolution.Using the high spatial resolution data of GF-1 (16 m) as reference, medium to low spatial resolution data were simulated, including Landsat-like and Terra/Aqua MODIS-like images.Comparisons between the PSF based scale transformation method and classical method revealed significant correlation, which also prove the efficiency of the proposed method.The scale errors of the remote sensing reflectance (RRS) average method are lower than the suspended practical matter (SPM) average method.Higher than 5% scale errors were produced by spatial scale transformation in high dynamic turbid waters, while for calm ocean waters, the error was less than 0.5%.Therefore, it is crucial for selection of proper scale transformation method, to achieve consistent remote sensing products from multi-source data.