Differences in station and regional precipitation assessment

Precipitation varies from one point to another within a region. This variation is commonly referred to as spatial variability. Spatial variability is defined as the mean difference in standardized precipitation between station values and the average for the region. Standardizing the precipitation data with long-term mean and standard deviation has eliminated the systematic spatial variability due to the gradient of the long-term mean precipitation. With the assumption of spatial homogeneity within a region, spatial variability is statistically equal between equally spaced stations. However, the average spatial variability between all stations within a region increases with the size of the region because the mean distance between stations also increases with the size of the region. On the other hand, spatial variability decreases with increasing time scale. Generally, monthly precipitation values are more variable than seasonal values, which in turn are more variable than annual values.

Assessing and quantifying these properties of spatial variability is necessary to interpret the reliability of regional climate forecasts for applications at the farm-level within the region. Regional climate forecasts are provided at monthly and seasonal time scales. In this study the spatial variability at the “climate division” and “super climate division” scales are analyzed and compared to assess the magnitude of variability within individual regions, as well as the change in magnitude between regions of different size. The analysis is conducted at hand of four climate divisions and one super climate division in Oklahoma. The time steps for the precipitation values are monthly, seasonal (3-months) and annual.

For a single climate division the spatial variability at the monthly time scale was found to range from 20 to 29 % of the mean monthly precipitation, depending on the month of the year; at the seasonal scale it ranges from 13 to 17% of the mean seasonal precipitation, depending on the season; and at the annual scale it was 8% of the mean annual precipitation. This demonstrates the decrease in the spatial variability of precipitation with increasing time scale. On the other hand, if the spatial variability is related to the mean temporal variation of precipitation, the spatial variability is of the same order of magnitude (about 35%) for the monthly, seasonal and annual time scales.

The increase in spatial variability with size of the region is illustrated for climate and supper climate divisions. At the climate division scale the spatial variability is around 35% of the mean temporal variation of precipitation of the climate division, whereas at the super climate division scale the spatial variability is 47% of the mean temporal variation of precipitation of the super climate division.

This significant spatial variability within a region leads to large localized departures from the average regional values. This study clearly demonstrates the critical influence of the spatial variability and the additional uncertainty it introduces in the downscaling of regional climate forecasts to local agricultural applications.