Simulating daily precipitation and temperature: a weather generation framework for assessing hydrometeorological hazards

Stochastic weather generators simulate synthetic weather data while maintaining statistical properties of the observations. A new semi-parametric algorithm for multi-site precipitation has been published recently by Breinl et al. (2013), who used a univariate Markov process to simulate precipitation occurrence at multiple sites for two small rain gauge networks. Precipitation amounts were simulated in a two-step process by first resampling observations and then sampling and reshuffling of parametric precipitation amounts. In the present study, the precipitation model by Breinl et al. (2013, J. Hydrol. 498: 23–35) is implemented in a weather generation framework for daily precipitation and temperature. It is extended to a considerably larger gauge station network of 19 stations and further improved to reduce the duplication of historical records in the simulation. Autoregressive-moving-average models (ARMA) are used to simulate mean daily temperature at three sites. Power transformations reduce the bias of simulated temperature extremes. Precipitation amounts are simulated by means of hybrid distributions consisting of a Weibull distribution for low precipitation amounts and a generalized Pareto distribution (GPD) for moderate and extreme precipitation amounts. The proposed weather generator is particularly suitable for assessing hydrometeorological hazards such as flooding as it reproduces the spatial variability of precipitation very well and can generate unobserved extremes.