Rainfall identification and estimation in Guangzhou area used for building energy simulation

In uncomfortably warm areas and seasons of the year, passive cooling effects resulting from natural rainfall evaporation can greatly cut down on building energy consumption. To simulate the passive evaporation cooling effect and evaluate the relevant energy-saving potentials, hourly rainfall data are needed. However, in currently used building energy simulation software, such as DOE, EnergyPlus and DeST, no rainfall information is provided in the climatic database. This paper uses a limited set of monthly and daily rainfall distribution data in Guangzhou area to identify and model monthly, daily and hourly rainfall patterns. For a current weather database used by building energy simulation software, rainy days and rainy hours are distinguished using distance discriminant analysis, which uses measured data samples for rain identification. According to an autocorrelation analysis of rain sequences, a one-order AR model is suitable for monthly rainfall estimation by AIC criterion judgment. Distribution of daily rainfall month-by-month shows a Gamma distribution model agrees well with daily rainfall distribution. Using a Gamma distribution model and monthly total rainfall, daily rainfall is assigned stochastically. Analysis shows distribution patterns of hourly rainfall percentage, both in the rainy season and non-rainy season, coincide well with the Beta distribution. Using a Beta distribution model and daily total rainfall, hourly rainfall is assigned stochastically. A comparison of statistics features of simulated data to that of measured data validates the method.