To model the constituents of the energy balance for different surface types and to better understand how they change depending on surface properties, a regression analysis is used to develop models of sensible, latent and ground heat fluxes as a function of radiation components. Meteorological instrumentation was deployed at five sites with different surface types, including desert plain, grove, concrete surface with few low-rise buildings and streets in mid-rise downtown. Instrumentation included sonic anemometer, krypton hygrometer, net radiometer, soil heat flux plates, IR, relative humidity and temperature sensors. The linear regression analysis of short and long wave radiation including their gradients was applied only to half of the data collected at the Citrus Grove site and resulted in the new expressions for sensible, latent, and ground heat fluxes. These models/expressions allow for the estimation of surface fluxes by measuring only radiation components. The new expressions were tested on remaining data from the Citrus Grove site and the other four sites. Predictions are very good for daytime periods while nighttime performance varied from case to case. The comparison between the new model and other models such as Grimmond and Penman-Monteith Model is also presented using data from all five sites. We can recommend use of new models for estimate of sensible, latent and ground heat fluxes when only radiation data are available.