Two models are developed based on Lumb’s formula. Model 1 was developed by establishing cloud depth factors which are dependent on temperature alone, while model 2 was developed by establishing cloud depth factors which are dependent on temperature and humidity. Based on meteorological data measured with Davis weather station and TES solar meter, the original Lumb’s model predicted surface solar radiation with a mean square error (MSE) of 54,845,911.4. The modified Lumb’s model also predicted solar radiation with a mean square error of 61,519.4482. Developed models 1 and 2 (DM1 & DM2) recorded mean square errors of 44,500.436 and 43,620.40822, respectively. Data from random days were used for verification and results showed the least mean square error of 11,057.64914 for DM2. The importance of considering depth as a factor when estimating cloud cover is proven by analyzing existing models which were observed to predict poorly because cloud cover was estimated as a surface entity. DM2 proved the best model for hourly surface solar radiation predictions as it has the least mean square error. This empirical model should be adopted by environmental engineers in predicting hourly surface solar radiations because it considers cloud depth.