Submitted to: American Meteorological Society
Publication Type: Proceedings
Publication Acceptance Date: January 1, 2000
Publication Date: January 1, 2000
Interpretive Summary: Dry land farming in the Southern Great Plains is sensitive to seasonal and inter-annual climate variations. Decision-making in agriculture, as well as related planning and management activities, could greatly benefit from new seasonal climate forecasts that are provided by NOAA. However, climate forecasts are provided for an entire region and are expressed as departures from long-term mean conditions. For farm scale applications, an interpretation of the regional forecast for farm-scale use is needed. Such an interpretation requires the quantification of the climate variations within the region, which is commonly referred to as spatial variability. This study quantifies the spatial variability of monthly precipitation for the central Oklahoma region and illustrates the implications of the spatial variability for establishing confidence in the one-month climate forecast. Spatial variability of monthly precipitation between stations and the region was found to be large, reaching up to 30% of the year to year variability. It was further found that when the climate of a region was categorized as wet, normal or dry, not all stations in the region were in that same category. For example, when the regional climate was categorized as wet, about 75% of the stations in the region were indeed in the wet category, whereas 20% were in the normal category and 5% in the dry category. These results demonstrate that the additional uncertainty associated with spatial variability within a region is large and must be considered when using regional climate forecasts for local applications.
Technical Abstract: Monthly precipitation time series of twelve climate stations are used to quantify the differences between station observations and divisional precipitation values for the Central Climate Division of Oklahoma. The systematic spatial differences between stations and divisional values due to the precipitation gradient are eliminated by standardizing the monthly precipitation. Differences between station and divisional standardized values are relativiely high with 25% of the differences above half standard deviation, and 6% above one standard deviation. The mean difference is equivalent to 36% of the mean temporal variation of the precipitation. Differences are higher during the summer when precipitation is mostly in the form of localized storms, and lower during the winter. These values of spatial variability explain the large differences observed between local and divisional precipitation. The statistical characteristics of the difference identify the range and uncertainties in local precipitation values when these are derived from divisional precipitation. The probability of observing local precipitation near the divisional value is higher from October to March. Thus, the risk of using regional precipitation for local decision making is lower during the winter than during the summer. Current climate forecasts provide the probabilities that forecasted regional monthly precipitation falls in one of three categories (dry, normal, and wet). When the divisional value is in the wet or dry category, near 75% of station observations are in that category, whereas about 20% are in the normal category and up to 5% are in the opposite category during the summer months. Correspondence is lower when the division value is in the normal category.