Traditional pixel-based algorithms, considering only spectral information and ignore spatial information, have limitations to provide better accuracy of surface heat fluxes from high-resolution images. Based on the high-resolution satellite images, this paper systematically analyzes the feasibility of combining the object-based approach with traditional physical model to estimate surface heat fluxes.Sentinel-3 surface temperature and Sentinel-2 multi-spectral data were input to the energy balance Two-Source Energy Balance (TSEB) model to estimate the surface heat fluxes. Pixel-based TSEB model was firstly employed at 10m. An multi-layer experiments framework was constructed to explore the applicability of the object-based method. The object-based approach is introduced into TSEB model and the multi-scale segmentation algorithm of eCognition is used to segment the images and extract the surface objects. Two object -based strategies, estimating heat fluxes before or after aggregating objects properties, were used to analyze the influence of the different strategies on the results. Object-based method and different inversion strategies are compared with pixel -based results. The results show that, comparing with the pixel-based method, the object-based method is beneficial to map surface heat fluxes and can improve the estimation accuracy of the TSEB model, which is mainly influences by the vegetation-related parameters.
|