- About Us
- SW Climate
New divisions for monitoring and predicting climate
Published July 25, 2007
Climatologists have long questioned the accuracy of the current climate divisions (CDs) in the United States in representing regional climate. To address their concerns, we embarked on a long-term effort in 2003 to create a more rational, statistically-based set of national CDs that would help improve drought monitoring and climate forecasting. The result, thus far, is an experimental map of new CDs that more accurately represents U.S. climate.
Near-real time climate monitoring, longterm climate change assessments, and statistical climate predictions are often based on data aggregated into CDs. CDs come from century-long efforts to organize climate observations across the country to match up with crop reporting districts, county lines, and/or drainage basins; the CD boundaries were finalized in the 1950s. Perhaps surprisingly, given their use, climate, based on objective groupings of long-term observations, was not the primary consideration in determining the CD boundaries (Guttman and Quayle, 1996).
The vast majority of data used in climate analyses comes from stations that are part of the voluntary Cooperative Observer Program (COOP) at NOAA. This network of stations has been collecting daily high and low temperatures, precipitation, and snowfall since 1890. CD data are computed by simply averaging all available, representative COOP station data within each division since 1931 into single monthly values. Data prior to 1931 were derived from statistical relationships between current division data and state-wide averages. CDsare used in many climate-related monitoring products, like the U.S. Drought Monitor (see page 8), because they allow for an easy calculation of regional averages and a comparison of recent climate anomalies against a century-long record. The NOAA Climate Prediction Center (CPC) has used so-called “megadivisions,” which are based on merging smaller CDs, as targets for climate predictions and for verifying forecasts.
The 344 U.S. CDs allow for up to ten divisions per state; however, they cover the conterminous United States rather unevenly (Figure 1a). Many states, such as Wyoming and Idaho, have ten divisions, but some rather large states do not. Arizona, a large state with complex topography, is represented by only seven CDs, some of which may not accurately represent regional climates. For example, the northeast third of the state, from the Mogollon Highlands and San Francisco Peaks, across the Painted Desert to the Four Corners is represented by a single division. Similarly, the southeastern Arizona CD stretches from the parched deserts of Organ Pipe National Monument, across the lofty Sky Island mountain ranges to the Gila River headwaters. Decisions about how to organize CDs were made on a state-by-state basis rather than from a national perspective (Guttman and Quayle, 1996).
Although the CD data provide a long, consistent, and gap-free record, climatologists have long questioned the assumption that the simple averaging of COOP stations into CDs is optimal for depicting regional climate, especially precipitation. To examine this issue, we correlated individual COOP station data with divisional averages (Figure 1b). Results show that much of the high elevation Interior West, especially along the Rocky Mountains down into New Mexico, is not well represented by divisional averages. During the wintersnow accumulation season in parts of the Interior West, there are poor correlations between individual stations and the associated CD (Figure 1b), and the situation is even worse in the summer.
Low correlations between individual COOP stations and divisional averages translate into poor reliability when largescale drought assessments or ENSOrelated forecasts based on these divisions are scaled down to the station level. This is one reason why drought monitoring and seasonal climate forecasting are difficult in the Interior West. In addition, some of the higher elevation Snowpack telemetry (SNOTEL) sites, operated by the USDA Natural Resource Conservation Service, may correlate negatively with their CD time series. This is due to orographic effects in high mountain areas: during the winter season, strong westerly winds yield large snowfall amounts on the windward side of mountain ranges, while the valleys to the east may experience windstorms and dryness. Because most COOP stations are located in valleys, CD averages may end up with precipitation deficits when compared with long-term averages, whereas SNOTEL-based precipitation assessments may show precipitation surpluses. This type of precipitation pattern is not well captured by the current CDs.
Analogous maps for seasonal temperature correlations do not show the same disparity between station and CD data, most likely because temperature variations are similar over larger regions than precipitation variations. Nevertheless, wintertime regional temperature anomalies are also not well represented by climate divisions in the orographic regions of the Interior West.
In 2003, we launched a project to create a different set of national CDs that would help improve drought monitoring and climate forecasting in the U.S. (see sidebar for methodology). The result is a map of new CDs, based on temperature and precipitation station data, which are no longer bounded by state lines (Figure 1c). Note the divisions along the borders of California, Arizona, Utah, Nevada, New Mexico, and Texas. For example, the map shows divisions that encompass the climatic similarity of the southeast corner of Arizona and the southwest corner of New Mexico. Both have similar ecosystems and year-to-year precipitation variations.
In addition, there is no upper limit of ten divisions per state. One of the goals of this project was to integrate SNOTEL sites into the analysis. We found that SNOTEL data correlates well with the new CDs, and most of the SNOTEL sites match up nicely with the nearest COOP-based CD.
With the creation of the joint temperature and precipitation maps (Figure 1c), this project is almost complete. The remaining stage is to fine-tune the new division boundaries with precipitation data from SNOTEL and precipitationonly COOP stations. For more information on the new climate divisions, visit the NOAA Earth System Research Laboratory web site (Figure 1c). We are working on the additional products, including additional time series of temperature and precipitation averages in each new climate division, from 1978–2006, and from 1948–1978, based on new climate divisions for that period, and final new climate division maps, including boundaries, spatial coverages (in percent of area), and new state-wide averages.
Guttman, N.B., and R.G. Quayle, 1996. “A historical perspective of U.S. climate divisions.” Bulletin of the American Meteorological Society, 77, 293–304.
Wilks, D.S., 1995. Statistical Methods in the Atmospheric Sciences. Academic Press, San Diego, 467pp.