Adaptive responses to climate and other risk factors require salient information to support agricultural decisions. Outcomes of alternative decisions in agriculture can be simulated through process (crop growth) models that often require daily weather as input. Historical daily series can be used, but often are short or unavailable. An alternative is to use algorithms known as “weather generators” that produce synthetic series of daily weather with statistical properties consistent with those of historical data.

We have developed an enhanced semi-parametric weather generator that combines parametric and non-parametric approaches. Precipitation occurrence is modeled parametrically as a first-order Markov chain with three possible states (dry, wet, and very wet). The boundary between “wet” and “very wet” days is defined dynamically as a given percentile (typically the 50- or 60-percentile) of rainfall amounts for a given period. On the other hand, values of relevant weather variables (minimum and maximum temperatures, precipitation amount and solar radiation) are generated non-parametrically, through resampling of historical series based on selection of “nearest neighbors” in multivariate space.

We have also adapted the generator so that it can produce output conditioned on a given seasonal forecast, for example those produced operationally by the International Research Institute for Climate and Society (IRI). These forecasts state the likelihood of regional precipitation or temperature in a period falling within certain categories (e.g., “above normal”, “below normal”). This allows us to generate multiple equally-likely realizations of interannual events (e.g., an El Niño event or a wetter than normal spring). These synthetic series subsequently can be combined with relevant process models to assess climate-related risk in socioeconomic sectors sensitive to climate variability.

An extended abstract (in Spanish) describing the hybrid generator and providing diagnostics of its performance was published in the Proceedings of the 9^th Argentine Meteorology Congress. A manuscript with more details on the generator and its performance was submitted to Water Resources Research.