基于荧光-光学特征的内陆湖泊溶解性有机碳遥感估算方法研究

Recently, the carbon cycle in freshwater ecosystem and inland lakes has raised worldwide interest and attention. Dissolved Organic Carbon (DOC) is the largest organic carbon pool in the water, and can be comparable to the carbon storage in the atmosphere, which plays an important role in the carbon cycle in water. So the study on the tempo-spatial distribution of DOC in inland lake would have significant positive impact on the research of radiation attenuation of UV-B caused by dissolved materials and carbon cycle in inland water. As a result, In this proposal, the optical difference in the inland water would be intensively studied based on several in situ observations and laboratorial analysis in different seasons. Furthermore, the proposal will conduct research on the co-variation relationship between absorption feature of CDOM and it's fluorescent properties, which often used to indicate the origins of CDOM, and develop the optical method to classify the water type based on the different DOC origins using CDOM absorption characteristics. Then, specific estimation algorithm for different water types would be finally built. On the basis of above research, the expected results in this proposal is to build an universal model to retrieve DOC concentration for Chinese inland lakes using remote sensing, i.e., the water type would be firstly determined based on the remote sensing reflectance, then the DOC concentration can be estimated using specific algorithm for definite water type. And the developed model can be widely applied using satellite images. This proposal will address the key DOC remote sensing method that has the ability to map the DOC tempo-spatial distribution in inland lakes with a large scale, which will contribute to the global and regional carbon cycle research regarding the role of lakes.

近年来淡水生态系统碳循环研究引起了国际上广泛关注。溶解性有机碳(Dissolved Organic Carbon，DOC)是水体中最大的有机碳储库，与大气中的碳储量相当，在水体中碳循环中起着重要的作用，利用遥感监测湖泊中DOC浓度的时空分布对于研究溶解性物质对UV—B辐射的衰减、内陆水体碳循环等方面具有重要的意义。因此本研究拟基于多次不同时相野外原位观测实验和室内分析，研究表征DOC来源差异的荧光特性与遥感可捕捉的CDOM吸收特性的共变规律，分析不同DOC来源水体的光学特征差异，建立基于CDOM吸收特征的DOC不同来源光学判别方法，针对不同DOC来源的水体构建其DOC遥感估算模型，发展内陆水体CDOM吸收特征反演算法，从而建立考虑不同来源的具有较强鲁棒性的内陆湖泊DOC遥感估算方法，即先进行不同DOC来源水体判别再进行遥感反演的方法。实现利用遥感手段对湖泊中DOC浓度的宏观、连续监测。

溶解性有机碳(Dissolved Organic Carbon，DOC)是水体中最大的有机碳储库，在水体碳循环中起着重要的作用，淡水生态系统DOC 是全球碳循环的重要组成部分，其变化将对全球碳循环产生重要的影响。构建普适性的内陆水体DOC遥感监测算法，研究湖泊中DOC浓度的时空分布对于研究溶解性物质对内陆水体碳的地球化学循环等方面具有重要的意义。在2019-2022年期间，共进行了21次野外采样，从太湖、巢湖、洪泽湖、千岛湖等不同富营养化程度的湖泊水体中获得了526个水样的光学和水质参数数据。通过分析内陆湖泊有色可溶性有机物(Colored dissolved organic matter, CDOM)的荧光指纹图谱信息，发现内陆湖泊中有机物类腐殖酸组分占据主导地位，在CDOM组成结构中占据了27-53%，均值为39%；分析内陆湖泊CDOM荧光和吸收的光学响应规律，发现CDOM外源荧光组分富里酸物质与254nm处的CDOM吸收系数(acdom254) 之间呈现高度正相关，表明湖泊CDOM吸收特征主要由外源组分主导，内源组分为辅；在此基础上，本研究利用M值作为判别水体中CDOM来源信息的光学指标，并将内陆湖泊水体分类两大类：当M值大于8时为Type I类型水体，CDOM以内源输入为主；当M值小于8时为Type II类型水体，CDOM以外源输入为主，且建立了基于遥感反射率的判别不同CDOM来源的水体分类策略。基于野外采样数据，本研究构建了基于荧光分类和CDOM吸收系数的DOC遥感估算半分析模型、经验模型以及机器学习(RF模型)，通过模型精度评估和对比，发现本研究构建的反演模型精度最高可达UMRD =17.35%，URMSPD=20.27%。最终把构建的模型应用至2016-2020年的长三角地区的内陆湖泊OLCI遥感影像上，探究了长三角地区湖泊DOC浓度时空演变规律，并揭示了人类活动和气侯变化对长三角地区湖泊DOC浓度变化的影响机制。