基于星地联合观测的高动态变化湖泊水环境时空采样方法研究

Optimal sampling strategy is of crucial importance for high-precision monitoring and predicting of earth surface parameters, for instance, of water quality monitoring of inland lakes. And as the key component of earth observation network, scientific and efficient field sampling is the key for remote sensing modeling and products validation. Lack of prior knowledge for earth surface processes, subjective methods based on experts’experience are excessively adopted for field sampling. Taking Poyang Lake as a typical research region for its high spatio-temporal dynamic properties, multi-source satellite–ground joint data are adopted for spatio-temporal pattern analysis as prior knowledge, this project aims to: 1) cross-calibration of multi-source data and products, and develop a method for spatio-temporal pattern mining of suspended particulate matter in high dynamic waters using time series data, which takes both the spatial and temporal dimensions into account simultaneously, and to build the spatio-temporal prior knowledge of suspended particulate matter patterns; 2) analysis the scale pattern of the of the high dynamic suspended particulate matter，as well as the scale uncertainties for multi-scale sampling, and to establish the quantitative relation between sampling scales and spatio-temporal variations; 3) build a quantitative assessment index for spatio-temporal representativeness of field sampling, and to explore a method for optimal sampling strategy, including number of samples, spatial distribution of samples, sampling times and intervals, using spatio-temporal information from multi-source observation. This research will provide method for objective sampling of environment monitoring and remote sensing modeling/validation, and basis for ocean color remote sensing.

科学、高效的地面采样是优化资源配置，提高地表环境观测、预测和决策水平的前提，同时是定量遥感建模和产品验证等的关键。本研究针对高动态变化湖泊水环境监测的采样方法问题，选择时空变异特性显著的鄱阳湖为典型研究区域，以多源多尺度的星地联合观测数据为基础，拟1）开展多源观测数据和产品的归一化校正，研究鄱阳湖悬浮颗粒物时空变化规律提取和分区方法，构建悬浮颗粒物时空变异特征先验知识；2）解析悬浮颗粒物观测的时空尺度效应，分析时空变异影响下的地面采样不确定性，量化采样时空尺度变化对采样观测误差的影响关系；3) 建立采样有效性定量评价指标, 评估传统采样方法的观测精度，研究时空先验知识辅助的悬浮颗粒物采样方法(采样布局、采样数量、采样时间及频次) 。本研究可解决传统采样方法依赖主观经验的问题，为客观的环境监测采样、遥感地面建模验证等提供科学依据，为提高水环境监测能力，完善水色定量遥感方法体系提供理论基础。

针对高动态变化的湖泊水环境采样问题，以时空变化特性显著的我国第一大淡水湖—鄱阳湖为典型研究区域，利用多源多尺度遥感数据和地面观测数据：(1) 提出了一种考虑BRDF效应的多源遥感数据辐射交叉定标及归一化处理技术，可显著降低辐射定标的不确定性影响，多源卫星遥感产品的不一致性降低到5%以下；(2) 提出了一种基于优化估计的辐射定标-大气校正-水质参数耦合反演模型，悬浮物定量遥感产品的平均相对误差降低为15%左右；(3)开展了基于时序多源遥感数据的悬浮物时空规律及影响因素分析研究，解析了降雨、风、人类活动等因素对近岸水体悬浮物的影响机理；(4)定量结算了观测时空尺度对悬浮物观测的不确定性影响，结果表明针对悬浮泥沙日内变化的应用和研究，建议观测频次为每天4次或以上，当月均观测覆盖率提高到30%，观测误差可降低到20%左右；(5)实现了一种基于随机过程模拟的考虑时空变异特性先验知识的湖泊悬浮颗粒物采样方法，通过利用SSA优化采样位置，全湖悬浮物估算误差比系统采样和分层采样平均降低了18.11%和29.34%。遥感作为一种理想的手段，显示出巨大的潜力，可以提供空间上连续和全面的数据，作为有效的采样设计的先验知识。随着更多的卫星数据，遥感的能力得到了扩展。利用这些数据，通过评价现有监测站对水质的反映，或在今后的研究和日常监测中优化新的取样网络，水质监测的取样设计可以更加精确和有效。