Comparison of LARS-WG and AAFC-WG stochastic weather generators for diverse Canadian climates

Abstract

Two weather generators-LARS-WG, developed at Long Ashton Research Station (UK), and AAFC-WG, developed at Agriculture and Agri-Food Canada-were compared in order to gauge their capabilities of reproducing probability distributions, means and variances of observed daily precipitation, maximum temperature and minimum temperature for diverse Canadian climates. Climatic conditions, such as wet and dry spells, interannual variability and agroclimatic indices, were also used to assess the performance of the 2 weather generators. AAFC-WG performed better in Simulating temperature-related statistics, while it did almost as well as LARS-WG for statistics associated with daily precipitation. Using empirical distributions in AAFC-WG for daily maximum and minimum temperatures helped to improve the temperature statistics, especially in cases where local temperatures did not follow normal distributions. However, both weather generators had overdispersion problems, i.e, they underestimated interannual variability, especially for temperatures. Overall, AAFC-WG performed better. © Inter-Research 2004.