Model: AWE-GEN

An advanced stochastic weather generator for simulating 1-D high-resolution climate variables

Records of meteorological variables around the world are often very short, with substantial gaps and low spatial coverage. This creates a problem of data inadequacy in numerous  applications. To overcome such problems, weather generators, as the tools capable of generating consistent time-series of climatic variables, have been proposed and used in the past.  Weather generators infer from meteorological variables of an observed climate set of parameters that allow to numerically reproduce continuous and longer time series  of such  climate. Several studies on  agricultural and crop productivity,  water resource evaluation, flood risk analysis, etc., requires the use of weather generators. Other possible applications are related to the generation of inputs to hydrological models, ecosystem models, or in long-term land management and erosion studies.

AWE-GEN (Advanced WEather GENerator) is an hourly weather generator  capable of reproducing low and high-frequency characteristics of hydro-climatic variables and the essential statistical properties of these variables. The weather generator  employs both the physically-based and stochastic approaches and is a substantial evolution of the model presented by Ivanov et al., 2007.  It includes a formulation of the precipitation module based on the Poisson-Cluster process; stochastic-physical based modules of cloud-cover and air temperature. It further includes modules simulating vapor pressure, wind speed, atmospheric pressure, and shortwave radiation partition into various type and spectral bands including the simulation of the photosynthetically active radiation. AWE-GEN is also capable of simulating the inter-annual variability of the precipitation process and a wide set of statistics including extremes. Furthermore, a procedure to take into account non stationary change of climate has been incorporated in the AWE-GEN framework. The procedure is based on a stochastic downscaling of GCM predictions.