Investigating Land Surface Properties with Different Ecosystems Using Earth Observing Big Data
Date
2020
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Land surface is of vital importance in the energy exchange within ecosystems and plays a key role in regulating the long-term climate change. To conduct large-scale land surface monitoring, Earth observing platforms including satellite techniques have been widely applied to land surface properties, climate and ecosystem monitoring, since remote sensing can provide global spatial and temporal continuous measurements with good resolution compare with traditional ground-based measurements. The availability of high-quality satellite products in past fifty years has promoted the study of monitoring carbon stocks in the soil and carbon dioxide flux from soil due to climate change. In this dissertation, I target 1) estimation of land surface water content (vegetation water content, soil water content) with both high spatial and temporal resolutions through combining multi-sources satellite observations with a machine learning based downscaling model; 2) measurement of surface soil temperature in forestry regions; and 3) application of the satellite-derived high resolution land surface measurements to carbon dioxide flux monitoring through the building of semi-empirical models integrating ground-truth observations. There were three study areas in this dissertation: 1) one of the main U.S. agricultural areas in Iowa, 2) the Great Dismal Swamp wildlife refuge in the U.S., and 3) the Central Tibetan Plateau in China. Field experiments were conducted in three study areas collecting ground-truth measurements of soil moisture, soil temperature at various depth, carbon dioxide (CO2) flux from soil, and other related land surface properties. These in-situ measurements are used for model training, calibration and validation. This dissertation presents three scientific areas of work: 1) A novel algorithm is proposed to monitor bare soil’s water content and the water content of soil with vegetation cover at high resolutions by combining an improve downscaling model, vegetation water content retrieval model, the universal triangle model and water cloud model integrating microwave and optical remote sensing techniques; 2) Surface soil temperature within a forestry region is monitored through thermal/ optical satellite sensors, considering the annual variation of air temperature, land surface temperature, soil temperature and vegetation index. The success of the application of thermal/ optical observations within forestry regions can improve the resolution of monitoring results compared with microwave measurements used in previous ways; 3) A semi-empirical model monitoring soil carbon flux is built, with theory is based on the close relationship between soil carbon flux and several soil properties (e.g. soil temperature, soil moisture) revealed by ground observations. Compared with previous methods, by integrating microwave remote sensing techniques with optical/ thermal infrared satellite techniques, the algorithm proposed in this dissertation estimates not only land surface properties for bare soil, but also monitoring them with various land cover types in three study area with different regional climate. By combining multi-source Earth observations, the daily vegetation water content, surface soil moisture and surface temperature with high spatial resolution can be obtained. The monitored land surface properties play a significant role in retrieving soil carbon flux. Regional carbon flux monitoring is helpful not only in natural disaster monitoring and forecasting, but also in regional climate related studies.