The contiguous United States has experienced both warming temperatures and a general increase in precipitation during the period 1950–2006. During that time drought has been a recurring phenomenon with a number of large droughts occurring, starting with the major drought in the 1950s in the Central United States and culminating with the persistent drought in the western portion of the country that started in the late 1990s. Here we examine the influence of the multi-decadal warming trend on drought coverage and the possibility that the general increase in regional and contiguous U.S. precipitation since about 1980 has masked the tendency for increasing drought driven largely by increasing temperature. Results indicate that without the increase in precipitation, severe to extreme drought would have affected as much as 50% more of the U.S. during some months in the most recent drought period.
 Drought in the continental United States (U.S.) is a recurring phenomenon that has shown large natural variability on both the instrumental and paleoclimatic time scales [Andreadis et al., 2006; Woodhouse and Overpeck, 1998]. Estimated costs of drought in the U.S. run to $6–8 Billion per year [Witt, 1995] with some, such as the 1988 Midwest drought, exceeding $40 Billion. Recent research examining global changes in drought suggests that drought conditions have increased globally in the last 30–40 years [Dai et al., 2004b]. In the U.S. as a whole there is no indication that drought has become more common. However, in some regions, such as the Southwestern U.S., there is evidence that drought has become more prevalent, while in other areas drought has become less common [Andreadis et al., 2006]. Furthermore, Webb et al.  estimated that the ongoing drought in the western U.S. is likely the most severe in the last 500 years.
 Over the 20th century the contiguous U.S. has experienced statistically significant increases in both the annually averaged mean temperature and the total annual precipitation (Figure 1). Recent work attributes the observed warming over the contiguous U.S. and particularly the record warmth of 2006, to increased greenhouse gases [Hoerling et al., 2007]. However, even though the increase in precipitation is consistent with expected precipitation changes due to greenhouse gases [Trenberth et al., 2003], no studies have yet confirmed direct attribution of these changes to increased greenhouse gases.
 Research into future patterns of drought under increased greenhouse gas forcing suggests that, even with a more vigorous hydrologic cycle leading to higher annual rainfall totals in many mid-latitude continental areas, the rise in temperature will increase evapotranspiration resulting in increased drought conditions, particularly during the warm season [Meehl et al., 2007]. Although the root cause of drought is an extended period of reduced precipitation, it is usually accompanied by increased temperatures and lowered atmospheric humidity that leads to enhanced evapotranspiration and drying soil.
 There are a number of drought indices designed to characterize drought for a particular region, some of which are purely rainfall deficit indices, such as the Standardized Precipitation Index [see Heim, 2002], but most often it is characterized for scientific and monitoring purposes by an index that incorporates measures of both precipitation and evaporation. The most commonly used index is the Palmer Drought Severity Index (PDSI) [Palmer, 1965], which incorporates both precipitation amounts, and an estimate for evapotranspiration, and requires only monthly temperature and precipitation as input.
 Here we examine the influence of increasing temperatures on drought coverage and the possibility that the observed increase in precipitation has masked a tendency for more drought in the U.S. since 1950. The starting point of 1950 is chosen to include the major droughts of the early 1950s and also is a period in the U.S. temperature record that is roughly the start of a period of slight temperature decline until the 1970s when U.S. temperatures, like global temperatures, began a strong positive trend (Figure 1). Similarly, Figure 1b shows the U.S. annual total precipitation for the continental U.S. for the 1901–2006 period which shows a positive trend of 4.6mm/decade.
2. Data and Methods
 The data used here are monthly averaged temperature and total precipitation for 4000 stations from the U.S. Cooperative Observer Network (Coop) over the contiguous United States. The temperature data for each station have been checked and adjusted for inhomogeneities using the method described by Menne and Williams [2005, also Detection of undocumented changepoints using multiple reference series, submitted to Journal of Climate, 2007]. The station data were then aggregated by Climate Division to create one temperature and one precipitation time series of all months for the entire 1950–2006 period for each of the 344 divisions in the contiguous U.S. A second set of time series for each Climate Division was generated by detrending the temperature and the precipitation time series by subtracting the least-squares linear trend.
 PDSI values were calculated for each Climate Division using the following scenarios: (1) observed temperature and precipitation, (2) observed temperature and detrended precipitation, and (3) detrended temperature and observed precipitation. Graphs of the percent area in severe to extreme drought (PDSI < −3) through time for the entire contiguous U.S. and for each of the nine climate regions used by the National Climatic Data Center (NCDC) [see Easterling, 2002] were generated for each of the three scenarios. Graphs of the difference between the observed percent area in drought (scenario 1) and each of the other two scenarios were then generated, keeping in mind that scenarios 2 and 3 are hypothetical if there were no trends in temperature (scenario 3) or precipitation (scenario 2).
Figure 2 shows the percent area of the contiguous U.S. and each of the nine regions in severe or extreme drought for the 1950–2006 period. The contiguous U.S. shows a slight decrease in the area in severe or extreme drought driven mainly by the large drought of the 1950s where the area of the country in severe-extreme drought approached 50% at its peak. By contrast the most recent period which includes the ongoing drought in the western U.S., shows a peak of about 35% of the U.S. in drought. Regional results are mixed with some areas (such as the Central and South regions) showing a decline, while others (such as the Northwest and West North Central regions) showing a tendency to more drought. Interestingly, the West and Southwest show little change but have consistently been in drought, with only short intervals of wetter conditions throughout the entire period.
Figure 3 depicts the difference plots between the observed drought (scenario 1) shown in Figure 2 and the results using observed precipitation and detrended temperature (scenario 3 in blue), and observed temperature and detrended precipitation (scenario 2 in red). For the contiguous U.S. (Figure 3a) the percent of the U.S. in drought likely would have been less if there had been no increase in mean temperature. This analysis suggests that the increase in temperature has increased by 15% the area in severe or extreme drought in the more recent period. Similarly, (Figure 3a, red line) drought coverage likely would have expanded if there had been no increase in precipitation and temperatures had increased as observed. An increase in the area of the U.S. of close to 30% is seen for some months in the most recent drought period (1999–2006). In short, the increase in precipitation has masked the tendency for increasing drought driven largely by increasing since about 1980.
 Regional results are also shown in Figure 3. In some regions, including the West North Central (3f), and regions in the western U.S. (3h, 3i, 3j), detrending temperature (scenario 3) reduces drought coverage, indicating the observed warming is increasing evapotranspiration and general water demand. Similarly, detrending precipitation results in much greater drought coverage and shows that the increase in precipitation has been critical to reducing drought.
 The one region that has been experiencing an increase in temperature along with a small decrease in precipitation is the Northwest U.S. Both trends have contributed to increasing drought, shown by the results from detrending of both temperature and precipitation having the effect of decreasing drought. Finally, because there is little temperature change in the Southeast, Central, Northeast, and South regions, the main driver in these regions has been the increase in precipitation. Detrending the precipitation increases drought coverage by up to 50% in some months in some regions.
 In summary, it is clear that the observed increase in precipitation for the contiguous U.S. has masked a tendency for increasing drought due to increasing temperatures. Also, on a regional basis, areas that have experienced persistent drought since the late 1990s, especially the West and Southwest, severe to extreme drought coverage would likely be even more widespread without the observed increase in precipitation. Lastly, given the fact we have more confidence that temperatures will continue to increase due to increasing greenhouse gases than we do in continued increases in precipitation, it is likely we will see more persistent and stronger droughts in the future.
 The authors would like to thank Henry F. Diaz and an anonymous reviewer for their comments. Support for this work was provided by the U.S. Department of Energy, Office of Biological and Environmental Research under Interagency Agreement DE-AI02-96ER62276 and the NOAA Climate Program Office, Climate Change Data and Detection Element.