Abstract:
Surface energy fluxes (also known as surface heat fluxes) consist of net radiation flux (Rn), soil heat flux (G), sensible heat flux (H), and latent heat flux (LE). Estimation of these fluxes is essential for climate modeling, disaster monitoring, plant water demand assessment, plant growth modeling, irrigation management and also in determining the large-scale atmosphere and ocean circulation patterns which eventually drive weather and climate. In situ measurements of these fluxes are expensive and available only over a limited number of field experiment sites. The Surface Energy Balance Algorithms for Land (SEBAL) is a well-known remote sensing-based model of estimating fluxes. Few studies have employed SEBAL in assessing heat fluxes in Bangladesh for few selected dates but none of them compared their results with any field measured data. This study aims at evaluating the performance of new automated version of SEBAL model to estimate energy fluxes in the context of Bangladesh.
The SEBAL model has been set up for 2006 to 2012 using Landsat images of seven different wavelengths and climate data. Albedo, radiometric surface temperature, Soil-Adjusted Vegetation Index (SAVI) have been then computed from the pre-processed Landsat images. Net radiation (Rn) has been computed from downwelling solar radiation and land surface temperature obtained from Landsat thermal images. Then ground heat flux has beencomputed from Rn and Normalized Difference Vegetation Index. Sensible heat flux (H) is calculated as a function of observations such as wind speed, vegetation type and roughness and surface to air temperature differences. The hot and cold pixel necessary for H computation has been automatically calibrated. Then latent heat flux has been estimated as the residual in the surface energy balance. Necessary data have been collected from Asiaflux, United Stated Geological Survey (USGS), Bangladesh Meteorological Department (BMD) and Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2). Surface energy fluxes estimated from SEBAL model were then converted to daily fluxes and compared with the field measurements from an Eddy Covariance Tower located at an agricultural field of Bangladesh Agricultural University, Mymensingh. Estimated surface energy fluxes from SEBAL shows very good correlation with observed data of Eddy Covariance Tower at agricultural field. For net radiation, coefficient of determination (R2) and root mean squared error (RMSE) are 0.95 and 20.01 W/m2, respectively. Daily latent heat flux also shows good correlation with observed data. R2 and RMSE for latent heat flux are 0.75 and 18.45 W/m2, respectively.
Temporal variation and mapping of surface energy fluxes for four types of land uses i.e., agricultural field, waterbody, urban area and forest area have been also been carried out. Net radiation reduces by 39.58%, 37.41%, 34.36% and 32.68% respectively for agricultural field, waterbody, urban area and forest area from summertime average to wintertime average from years 2006 to 2022. Latent heat flux also reduces by 48.44%, 50% and 0.01% respectively for agricultural field, waterbody and forest area from summertime average to wintertime average whereas LE increases by 19.18% for urban area over the years. Ground heat flux reduces by 61.22%, 76.57% and 104.14% for agricultural field, urban area and forest area respectively from summertime average to wintertime average over the years. Sensible heat flux reduces by 22.92% and 1.10% from wintertime average to summertime average for agricultural field and waterbody respectively whereas reduces by 47.85% and 82.72% from summertime average to wintertime average for urban area and forest area respectively. This study indicates that remote sensing-based surface energy balance algorithm, SEBAL can be a promising alternative to ground measurements of heat fluxes in the data-scarce regions.