National Institute for Water Resources, Water Resources Research Institute Program
Sensible and latent heat fluxes are the key variables in energy and water vapor exchange between the land surface and the atmosphere. Latent heat flux is the coupling link between the surface water, energy, and carbon exchanges with the atmosphere. Several techniques (e.g., lysimeters, eddy covariance systems, Bowen ratio methods, and large-aperture scintillometers) have been used to measure surface heat fluxes (Liu et al., 2011, 2013). However, in situ measurements of heat fluxes are costly and are therefore distributed sparsely, and cover only limited time periods. Consequently, a number of models have been developed to estimate surface heat fluxes from remotely sensed land surface temperature (LST) observations.
LST lies at the heart of the surface energy balance (SEB) equation. All components of the SEB equation (i.e., sensible, latent, and ground heat fluxes as well as net radiation) are related to LST. Recently, Bateni and Entekhabi (2012a) showed that LST observations contain implicit information on the partitioning of available energy among the SEB components. LST observations have been utilized in three main groups of studies to estimate surface heat fluxes. The first group of studies is diagnostic. These studies use LST to solve the SEB equation and retrieve surface energy fluxes (Norman et al., 1995; Anderson et al., 1997; Bastiaanssen et al., 1998a, 1998b; Su, 2002; Liu et al., 2007; Jia et al., 2009; Ma et al., 2012). The ground heat flux is usually taken as an empirical fraction of the net radiation. Additionally, surface heat fluxes can be retrieved only for instances in which remotely sensed LSTs are available. The second group is known as triangle approaches. These studies attempt to estimate the surface evaporation using empirical relationships between LST and vegetation indices such as the normalized difference vegetation index and leaf area index (LAI) (Jiang and Islam, 2001, 2003; Nishida et al., 2003; Wang et al., 2006; Tang et al., 2010; Sun et al., 2013). These methods need to define the dry and wet edges of the triangle space, which is site specific.
The third group of studies estimates the surface heat fluxes by assimilating sequences of LST measurements within a variational data assimilation (VDA) framework using the parsimonious force-restore equation as a constraint (Castelli et al., 1999; Boni et al., 2001; Caparrini et al., 2003, 2004a, 2004b; Crow and Kustas, 2005; Qin et al., 2007; Sini et al., 2008). In contrast to the diagnostic and triangle approaches, this group of methods does not require any empirical or site-specific relationships and can provide temporally continuous surface heat flux estimates from discrete spaceborne LST observations.
The VDA utilizes combined-source (CS) and dual-source (DS) schemes to simulate interaction between the land surface and the overlying air and to retrieve surface heat fluxes. The CS scheme does not distinguish the difference between soil and canopy temperatures and treats LST as the effective temperature of a mixed soil-vegetation medium. In contrast, the DS scheme accounts for the difference between soil and canopy temperatures and considers the interactions of the soil and canopy with the overlying atmosphere separately.
Bateni and Liang (2012) and Bateni et al. (2013a, 2013b) significantly advanced the CS and DS VDA approaches by using the full heat diffusion equation as a physical constraint instead of the simple force-restore equation. However, the CS and DS VDA approaches by Bateni and Liang (2012) and Bateni et al. (2013a, 2013b) have been tested so far at only two humid sites with grassland vegetation cover (i.e., the First International Experiment and the Southern Great Plains sites). In this study, the performance of the recently augmented CS and DS VDA frameworks is assessed in detail using surface heat fluxes collected at six FluxNet sites with different vegetation covers (grassland, cropland, and forest) and climate conditions. These sites are chosen because they sample different climatic and vegetative conditions in an effort to evaluate the robustness of the VDA schemes in various hydrological environments.
Sequences of daytime LST observations have various diurnal amplitudes depending on the available energy and the relative efficiency of SEB components (Bateni and Entekhabi, 2012a). Hence, an accurate characterization of the LST diurnal cycle is of vital importance for the reliable performance of the VDA methods. In this study, LST data from Geostationary Operational Environmental Satellites (GOES) are assimilated in the CS and DS VDA schemes to estimate surface heat fluxes. GOES can accurately characterize the LST diurnal cycle by providing LST data every 30 min and thus can significantly advance the robustness of the VDA framework. GOES LST can be accurately retrieved (Sun et al., 2004) and proved to be a significant data set for improving turbulent flux estimates of the land surface model (Xu et al., 2011).