Use of regional Precipitation forecasts for local applications

Over the last decade, our understanding of global atmospheric circulation patterns, long-distance teleconnections, sea surface temperature-atmosphere interactions and their impact on continental scale weather patterns has significantly increased. This new atmospheric/climatic information has enabled NOAA to produce regional monthly and seasonal precipitation forecasts. The climate-sensitive agricultural sector can benefit from these climate forecasts by incorporating new regional and local climatic information into agricultural planning and management strategies. To fully exploit emerging climate forecast information for local applications, the relation between the mean monthly precipitation over a region and spatial variability at local scales must be quantified. Such a relation is a prerequisite to downscale regional climate information for utilization at farm or field scales.

Monthly time series of twelve stations covering the 1914-96 period are used to quantify the differences between station observations and divisional precipitation values for the central climate division of Oklahoma. Systematic spatial differences between stations and divisional values due to precipitation gradient are eliminated by standardizing the monthly precipitation value. Differences between station and divisional standardized values are relatively high with 25 % of the differences superior to (+/-) half a standard deviation, 6 % superior to (+/-) one standard deviation, and maximum values near three standard deviation. The average of the absolute values of the differences is equal to 0.36, which implies that the mean difference between station and divisional series is equivalent to 36 % of the mean temporal variability of the divisional series. Differences are higher during the summer when precipitation is mostly in the form of localized storms. The above quantification of the spatial variability explains the observed large differences between local and divisional values. The characteristics of the statistical distributions of the differences identify the range and uncertainties in local precipitation values based on divisional precipitation. They are the expression of the relation between regional and local precipitation.

Current climate forecasts do not provide divisional precipitation values with corresponding range of uncertainties. They provide only the probabilities that the forecasted monthly precipitation falls in one of three categories (dry, normal, wet). However, we have demonstrated that station observations can significantly differ from divisional values. Thus, it is important to determine the probability that station precipitation values are in the same category as the divisional value. For the central climate division of Oklahoma, a statistical analysis shows that when the divisional values are in the wet category, there is an 80% probability that the station observation are also in the wet category, an 18% probability that they are in the normal category and a 2% probability in the dry category. Similar probabilities are obtained for the normal and dry precipitation categories. The probability of correspondence is higher during the winter month when precipitation events are frequently associated with cold fronts, and cover a large area. These probabilities quantify the increase in uncertainty of forecasted precipitation associated with the information transfer from regional to local scales.