Mapping of land-atmosphere heat fluxes and surface parameters with remote sensing data

A variational data assimilation scheme is used to infer two key parameters of the surface energy balance that control the partitioning of available energy into latent, sensible, and ground heat fluxes (LE, H, and G). Remotely sensed land surface temperature (LST) is the principal data source. Maps of diurnal energy balance components are presented for a basin with varied land cover (Arno Basin, Italy) for a 18-day period in July 1996. Given available energy, the major unknown (dimensionless) parameters required for partitioning among fluxes are: (1) Landscape effects on near-surface turbulence as captured by the bulk heat transfer coefficient C-BN under neutral conditions and (2) surface control of the relative magnitudes of LE and H as represented by the evaporative fraction E-F. The data assimilation scheme merges 1.1-km resolution remotely sensed LST images (based on optical, thermal and microwave measurements from two different satellite platforms) into a parsimonious model of heat diffusion. Both the measurements and the model predictions are considered uncertain. Posterior error statistics that represent uncertainty of the estimated parameters are also derived. Maps of C-BN show spatial patterns consistent with the dominant land use and basin physiography. Daily maps of E-F exhibit spatial variations corresponding to land cover and land use-the day-to-day variations in E-F show fluctuations consistent with rain events and drydowns experienced during the period. Based on these parameters and available environmental variables, maps of diurnal LE and H may be produced (in this paper daytime LE maps are reported). The application demonstrates that remotely sensed land surface temperature sequences contain significant amount of information of the partitioning of available energy among the fluxes. The variational data assimilation framework is shown to be an efficient and parsimonious approach without reliance on empirical relationships such as those based on vegetation indices.