LIDA: A Land Integrated Data Assimilation Framework for Mapping Land Surface Heat and Evaporative Fluxes by Assimilating Space‐Borne Soil Moisture and Land Surface Temperature

Abstract

Land surface heat and evaporative fluxes exchanged between the land and atmosphere play a crucial role in the terrestrial water and energy balance. Regional mapping of these fluxes is hampered by the lack of in situ measurements (with the required coverage and duration) and the high spatial heterogeneity. In this paper, we propose a Land Integrated Data Assimilation framework (LIDA) based on the variational data assimilation technique to estimate the key parameters of surface heat and evaporative fluxes by jointly assimilating Soil Moisture Active Passive (SMAP) data and Geostationary Operational Environmental Satellite (GOES) surface temperature data into a coupled parsimonious land water and energy balance model. The method is implemented over an area of 31,500 km2 in the U.S. Southern Great Plains, and its performance is evaluated through consistency tests, comparison tests, and uncertainty analyses. The maps of retrieved heat and evaporative fluxes are used to analyze a range of feedback mechanisms in land‐atmosphere interaction, such as the dependence of evapotranspiration on vegetation and water availability.

Estimation of Turbulent Heat Fluxes and Gross Primary Productivity by Assimilating Land Surface Temperature and Leaf Area Index

“…The third‐sixth terms represent the difference between R , EF S , EF C , and c g estimations and their prior values, respectively. In the six study sites, values of 0.01, 0.7, 0.7, and 0.01 are used as the initial guess for C HN , EF S , EF C , and c g , respectively (Abdolghafoorian & Farhadi, 2020; Bateni et al., 2014; Xu et al., 2014). The VDA obtains the optimal magnitudes of C HN , EF S , EF C , and c g by minimizing the difference between the LST and LAI estimates and their corresponding observations.…”

Estimation of Root Zone Soil Moisture Profile by Reduced-Order Variational Data Assimilation Using Near Surface Soil Moisture Observations

“…The calculation of the gradient of the cost function in VDA/adjoint method is very efficient as it is independent of the number of unknown parameters and requires only a single simulation of the model forward in time and a single simulation of adjoint model backward in time (e.g. [47], [69]- [73]. However, the integration of the adjoint of a large scale model backward in time may require several forward model simulations that increases the computational burden and expense of this technique [74]- [76].…”

A Variational Framework for Coupled Estimation of Evapotranspiration and Recharge Fluxes by Assimilating Land Surface Soil Moisture and Temperature

“…However, ground-based observations of this flux are sparsely available both in spatial and temporal scales due to cost and difficulty of in-situ measurement [13]. It is not possible to extend these data to large areas due to land surface heterogeneity and dynamic nature of transfer processes [14]- [16]. Direct measurement of recharge can be made by using lysimeters, seepage meters, and chemical tracers.…”

“…Abdolghafoorian and Farhadi [1], [14] developed a framework known as land integration data assimilation (LIDA) framework that resolved the two shortcomings of VDA-based heat and evaporative fluxes retrieval by i) coupling water and energy balance model and ii) estimating the uncertainty of parameters and fluxes to guide it toward well-posed optimization problem. They assimilate the SM data into moisture diffusion equation in addition to assimilating LST into heat diffusion equation.…”

“…They assimilate the SM data into moisture diffusion equation in addition to assimilating LST into heat diffusion equation. Abdolghafoorian and Farhadi [14] tested algorithm for a large area in US southern great plain by assimilating the remotely sensed LST and SM data.…”