Estimation of regional surface heat fluxes (sensible heat flux and latent heat flux) using remote sensing data is of important application value in the field of global climate change, water resource and ecological environment etc. The Moderate Resolution Imaging Spectroradiometer (MODIS) data has lower spatial resolution; however the geometric and physical properties of the Earth's surface also have a high degree of heterogeneity, which in practice are facing serious scaling problems. This paper discusses two methods of estimating the pixel fluxes using multi-source satellite data (the high-resolution Landsat TM data and low-resolution MODIS data), combining the land cover information or remotely sensed vegetation index provided by Landsat data with MODIS data to correct the spatial-scale errors. The remote sensing data and ancillary data used in this paper to evaluate the methods was obtained in comprehensive experiment held at Hei′he Watershed in 2008, validation data was surface flux data from the different surfaces during the experiment, including Eddy-Correlation (EC) data and Large Aperture Scintillometer (LAS) data. The results showed that, the combination method that use high resolution land class or vegetation index data can provide better estimation of surface heat fluxes, especially at the boundary of different land cover and heterogeneous land surfaces. By contrast, the method of using remotely sensed vegetation index provided by Landsat data to decompose temperature is more applicable and has a better validation result. The application of both Eddy-Correlation (EC) data and Large Aperture Scintillometer (LAS) data were analyzed by contrasting in the scaling process of fluxes validation by TM and MODIS. According to the pixel resolution of MODIS, LAS measurement can provided useful data at the scale of several kilometers, and can be used to validate MODIS fluxes directly. On the other hand, EC data need to be compared to the TM fluxes first, and then the downscaled TM fluxes can be used to validate MODIS fluxes. Finally, the uncertainties in the validation of fluxes using LAS data are mainly from the following aspect, ① the error from the positioning of the LAS site in the image; ② uncertainty in the contributing pixels of LAS observation in the image.