Rainfall predictive models for building simulation i – rainy time identification

Rainfall simulation is a key problem for the evaluation of the energy saving potential of passive cooling by natural rainfall evaporation in humid and hot areas. Using a set of data in Guangzhou Area from weather bureau as the training sample, and the data generated by Medpha (Meteorological Data Producer for HVAC Analysis) as the testing sample, distance discriminant analysis was applied to identify the rainy time. Dry bulb temperature, relative humidity, total solar radiation and direct solar radiation were chosen as the key factors determining rainy days identification and the simulated results were found to be in good agreement with the measured ones in the months before May, however adjustment was needed after May. Rainy hours identification was performed in daytime and nighttime separately using distance discriminant analysis and the predictive model was effective in the month that rainfall plays dominant influence.