On the Cause of the 1930s Dust Bowl

See allHide authors and affiliations

Science  19 Mar 2004:
Vol. 303, Issue 5665, pp. 1855-1859
DOI: 10.1126/science.1095048


During the 1930s, the United States experienced one of the most devastating droughts of the past century. The drought affected almost two-thirds of the country and parts of Mexico and Canada and was infamous for the numerous dust storms that occurred in the southern Great Plains. In this study, we present model results that indicate that the drought was caused by anomalous tropical sea surface temperatures during that decade and that interactions between the atmosphere and the land surface increased its severity. We also contrast the 1930s drought with other North American droughts of the 20th century.

In the United States, the 1930s were characterized by a decade of rainfall deficits and high temperatures that desiccated much of the land surface of the Great Plains. The drought and its associated dust storms created one of the most severe environmental catastrophes in U.S. history and led to the popular characterization of much of the southern Great Plains as the “Dust Bowl” (1, 2).

While progress has been made in understanding some of the important processes contributing to drought conditions (37), the mechanisms by which a drought can be maintained over many years are not well established. A number of studies have used the historical record of meteorological and oceanographic observations to identify statistical relations between slowly varying Pacific Ocean sea surface temperatures (SSTs) and precipitation over the Great Plains (8, 9). The record of observations, however, is too short to provide definitive results for long-term drought. Understanding the causes of the 1930s drought is particularly challenging in view of the scarcity of upper-air meteorological observations prior to about 1950.

Several recent studies using state-of-the-art atmospheric general circulation models (AGCMs) have shown how SST anomalies can produce prolonged drought conditions. Tropical SST anomalies, in particular, were found to contribute to recent prolonged drought conditions over much of the northern middle latitudes (10), to drought in the Great Plains (11), and to drought conditions in the African Sahel region during the 1970s and 1980s (12).

The importance of the Pacific SSTs (the pan-Pacific pattern) in forcing long-term precipitation variations in the Great Plains led us to expect that this pattern would be an important factor during the Dust Bowl when drought was most severe. SST anomalies, however, were surprisingly weak throughout the tropical Pacific during the 1930s. This prompted a much closer look at the relationship between SST anomalies and the generation of the Dust Bowl.

Our study is based on a number of century-long simulations carried out with the NASA Seasonal-to-Interannual Prediction Project (NSIPP) atmospheric general circulation model (13), the same model used in (11) and (12), although run here at a somewhat coarser horizontal resolution (14). The basic model simulations are an ensemble of fourteen 100-year (1902 to 2001) runs forced by observed monthly SSTs (15). These simulations will be referred to as the C20C runs, because they were carried out as part of the Climate of the 20th Century project (16). The runs differ only in their initial atmospheric conditions. As such, the degree of similarity in the runs (the “signal”) provides us with an assessment of how much the SSTs control Great Plains climate variations, while the disagreement among the runs (the “noise”) provides us with an estimate of the unpredictable component of the climate variability.

Figure 1 shows time series of the precipitation averaged over the Great Plains and filtered to retain time scales longer than about 6 years, using data from observations (17) and from the 14 simulations. The time series of the observed and simulated anomalies show considerable variability, with extended periods of both above- and below-normal conditions throughout the century. The correlation between the observed and ensemble mean anomalies is 0.57. The correlation between the ensemble mean and the individual simulations, a measure of the highest correlation we can expect with the observations, ranges from 0.53 to 0.79. While there is considerable scatter among the ensemble members, there are periods during which all the curves tend to follow one another. In particular, during the 1930s almost all of the runs show a tendency for dry conditions, consistent with the observations. The dry conditions of the 1930s are followed, in the early 1940s, by a rapid transition by all ensemble members to wetter conditions, again consistent with the observations. In general, the simulations agree with the observations to the extent that the observed anomalies fall within the scatter of the ensemble members.

Fig. 1.

Time series of precipitation anomalies averaged over the U.S. Great Plains region (30°N to 50°N, 95°W to 105°W; see box in insets). A filter (28) is applied to remove time scales shorter than about 6 years. The thin black curves are the results from the 14 ensemble members from the C20C runs. The green solid curve is the ensemble mean. The red curve shows the observations. The maps show the simulated (left) and observed (right) precipitation anomalies averaged over the Dust Bowl period (1932 to 1938). Units, mm/day.

Figure 1 includes maps of the ensemble mean and observed precipitation anomalies averaged over the Dust Bowl period (1932 to 1938). The observations show deficits exceeding 0.1 mm/day covering much of the central United States, with peak deficits exceeding 0.3 mm/day centered on Kansas. The simulated anomalies are similar to the observed, with peak deficits of similar magnitude and again centered over Kansas. The main discrepancy is the large deficit simulated over Mexico that does not occur in the observations (18). The simulation also fails to capture the full spatial extent of the drought, particularly the dry conditions that were observed over the northern Great Plains and parts of Canada. An inspection of the individual ensemble members shows that this discrepancy occurs as a result of the ensemble averaging, which acts to filter out the unpredictable (noise) component of the simulated anomalies. In fact, there is wide variability in the spatial pattern of the dry conditions among the ensemble members, with some showing negative precipitation anomalies (<–0.1) extending well into Canada and covering an area that exceeds that of the observed anomalies.

Figure 1 also reveals some rather peculiar model behavior during the mid-1970s. While the observations and 2 of the ensemble members show a tendency for slightly wet or neutral conditions, 12 of the ensemble members predict a drought even more severe than that during the 1930s. This highlights the probabilistic nature of the drought prediction problem and suggests, if we believe the model results, that the central United States was lucky to have had near normal conditions during the 1970s, because the probability of having a major drought was rather high (12 chances out of 14). In fact, the historical tendency for droughts to occur in the Great Plains roughly every 20 years (1910s, 1930s, and 1950s), together with the very dry conditions that existed over parts of the central United States by the mid-1970s, collectively led to speculation at the time that we were about to enter an extended dry period (19, 20).

Figure 2 shows our best estimate of the global SST anomalies (15, 21) that occurred during the Dust Bowl period. It must be emphasized that this time-averaged field is based on extrapolations of a rather limited number of ship observations using empirical orthogonal functions. Figure 2 shows that the anomalies are negative in most places, including the tropical Pacific and North Pacific, as well as much of the Southern Ocean. Positive anomalies occur in the tropical and North Atlantic Oceans and in some regions of the South Pacific. Surprisingly, the tropical anomalies tend to be small, generally less than 0.3°C. The largest anomalies occur in the North Atlantic and just off the coast of Asia, where they exceed 0.5°C.

Fig. 2.

The global SST anomalies averaged for the Dust Bowl period (1932 to 1938). The boxes delineate the various subregions (tropical oceans, Indian Ocean, Pacific Ocean, Atlantic Ocean) used in designing the idealized SST forcing experiments. The anomalies are the differences from the 1902 to 1999 SST climatology (15). Units, °C.

To understand the importance of each of these features to the 1930s drought, we carried out a number of idealized experiments in which SST anomalies, averaged over the 1932 to 1938 Dust Bowl period, were applied only to particular subregions (see outlines in Fig. 2). The remainder of the ocean was assigned climatological SSTs. Table 1 summarizes the experiments. Our aim was to separate the contributions to the drought from each of the three tropical basins (Indian, Pacific and Atlantic) and the extratropics (22).

Table 1.

The idealized SST experiments. The anomalies are the time-averaged (1932 to 1938) deviations from the 1902 to 1999 mean (22). All runs are 100 years in length. The regions are defined in Fig. 2.

Control Climatological (average of 1902 to 1999)
Global Global anomalies
Tropical Anomalies confined to tropics, climatological elsewhere
Pacific Anomalies confined to tropical Pacific
PacAtl Anomalies confined to tropical Pacific and tropical Atlantic
Paclnd Anomalies confined to tropical Pacific and tropical Indian Ocean
Fixed Beta Global anomalies and atmosphere—land surface interaction disabled

We first examined whether forcing the model with the 1932 to 1938 time–mean SST anomalies produces the same time mean response in the Great Plains precipitation as that obtained in the original C20C ensemble. The top two panels of Fig. 3 compare the ensemble mean precipitation anomalies from the C20C runs averaged from 1932 to 1938 with those from the Global run. The results are quite similar. In addition to the dry anomalies, the idealized forcing also reproduces some of the wet anomalies in the Pacific Northwest and along the southeast coast. There are some discrepancies between the runs. Most notably, the rainfall deficits that extend into Alabama in the Global run are absent in the C20C runs. Despite such minor differences, which must be artifacts of the Global run, it appears that the basic drought conditions simulated by the model in the Great Plains during the 1930s can be explained as a response to the time–mean global SST anomalies and that the year-to-year variations of the SSTs in that decade played at most a secondary role in shaping the drought.

Fig. 3.

(Top left) Ensemble mean precipitation anomalies averaged for the period 1932 to 1938 from the C20C runs. (Top right) The 100-year mean precipitation anomalies from the Global SST run (Fig. 2 and Table 1). (Bottom left) Same as the top right panel except for the Tropical SST run. (Bottom right) The difference between the precipitation anomalies from the Global SST and the Tropical SST runs. Units, mm/day. In all but the C20C panel, values are shaded only if they are significant at the 10% level based on a t test. The contour intervals are the same as the shading intervals (see color bar), with dashed contours indicating negative values.

We next tried to distinguish between tropical and extratropical effects. The results of the tropical run (lower left panel of Fig. 3) show that the main features of the drought are reproduced with the tropical SST forcing alone. The Global-minus-Tropical difference map (lower right panel) shows that the impact of the extratropical SSTs is much smaller. Extratropical anomalies tend to broaden the region of drought conditions, especially to the east and south of the central Great Plains. They also increase the region of wet anomalies in the Pacific Northwest.

The precipitation anomalies from the different model runs, including the contributions from the different tropical basins and the extratropics, averaged over the core Dust Bowl region (the rectangles in Fig. 4) are shown in fig. S1. The results show that the contributions from the tropical Pacific and tropical Atlantic are significant, while those from the tropical Indian Ocean and the extratropics are not (22). Figure S1 also shows the results of repeating the Global run, but in this case disabling the interactions between the atmosphere and land surface [the Fixed Beta run(22)]. Preventing this feedback reduces the precipitation deficit by 50%. Thus, land-atmosphere interaction appears to be responsible for much of the drought severity. The much tighter confidence intervals associated with the Fixed Beta results (fig. S1) show that without soil moisture feedbacks, precipitation variability is greatly reduced, consistent with previous studies employing the same model (7, 11). The spatial distribution of the precipitation anomalies from the Fixed Beta run shows that the reduced deficits span the Great Plains (Fig. 4). These results are consistent with previous work (7) showing that the Great Plains region is particularly sensitive to soil moisture changes.

Fig. 4.

(Top) Mean precipitation anomalies from the Global SST run. (Bottom) Mean precipitation anomalies from the Fixed Beta run. Shading indicates significance at the 10% level based on a t test. The contour intervals are the same as the shading intervals (see color bar), with dashed contours indicating negative values. The boxes indicate the core Dust Bowl region (36°N to 39°N, 99.5°W to 105°W). Units, mm/day.

The results presented so far have been for annual-mean conditions. It is well known, however, that most of the rain in the Great Plains tends to fall during the spring and summer seasons. We show in Fig. 5 the seasonal distribution of the area-averaged precipitation anomalies from observations, the Global run, and the Fixed Beta run. The results from the Global run are quite similar to the observed (23), though there is a general tendency to underestimate the deficits. The largest deficits occur during the warm season, with about half the deficit occurring during the summer months. Somewhat surprisingly, the fall season has larger deficits than the spring season. The winter season precipitation anomalies are by comparison rather small; in fact, the observed winter anomaly is slightly positive. The main impact of disabling the interactions with the land surface is to reduce dramatically the deficit during the summer. Figure 5 includes the ensemble mean and 90% confidence intervals of the precipitation anomalies from the C20C runs. Comparing those results with those from the Global run provides further evidence that the Global run reproduces the basic features of the C20C run— in this case, the annual cycle of the anomalies. The fact that the confidence intervals for the C20C anomalies encompass zero for winter and spring confirms that the significant precipitation anomalies for the core region are confined to the summer and fall season.

Fig. 5.

The seasonal variation of the mean precipitation anomalies averaged over the U.S. Great Plains (see box in insets in Fig. 1). The black bars show the observed values averaged over the Dust Bowl period (1932 to 1938). The light gray bars are the 100-year averaged results from the Global SST run. The dark gray bars are the 100-year averaged results from the Fixed Beta run. The small boxes and associated thin bars are the ensemble mean and 90% confidence intervals, respectively, of the 1932 to 1938 precipitation anomalies from the C20C runs. The labels on the abscissa indicate the seasons, where DJF is December, January, February; MAM is March, April, May; JJA is June, July, August; SON is September, October, November. Note that for DJF, the Fixed Beta run produces anomalies that are too small to show up in the figure.

These results suggest that we must look to the summer and fall seasons to understand the mechanisms linking tropical SST anomalies to the Dust Bowl region (24). An analysis of the summer circulation changes (not shown) suggests that the role of the cold Pacific SST anomalies was to generate a global-scale response in the upper troposphere (negative height anomalies in the tropics and a tendency for positive height anomalies in the middle latitudes) that suppressed rainfall over the Great Plains. The warm Atlantic SST anomalies produced two upper-level anticyclonic circulation anomalies on either side of the equator, with the northern anomaly extending across the Gulf of Mexico and the southern United States. In the lower troposphere, the response to the warm Atlantic SST anomalies was a cyclonic circulation anomaly positioned to suppress the supply of moisture entering the continent from the Gulf of Mexico. It is noteworthy that the Atlantic response is confined almost exclusively to the summer and fall season.

While the severity, extent, and duration of the 1930s drought was unusual for the 20th century, proxy climate records indicate that major droughts have occurred in the Great Plains approximately once or twice a century over the past 400 years (25). There is evidence for multidecadal droughts during the late 13th and 16th centuries that were of much greater severity and duration than those of the 20th century (25). For example, tree-ring analyses in Nebraska suggest that the drought that began in the late 13th century lasted 38 years (26). An analysis of the other major central U.S. droughts of the 20th century (11) suggests that a cool tropical Pacific is common to all. Only the Dust Bowl drought, however, combined cool Pacific SSTs with a warm Atlantic Ocean. Figure 1 shows that since the early 1980s (with the exception of 1987 to 1989), the Great Plains generally experienced above-normal precipitation. On the other hand, much of the western (especially the southwestern) United States, including some parts of the Great Plains, experienced below-normal precipitation during the past 5 years, leading to moderate or extreme drought conditions (27). The cause of this most recent drought is unclear, although a preliminary look at the relation between SSTs and long-term precipitation variations over the southwestern United States from our C20C runs suggests a strong link to the pan-Pacific pattern discussed earlier. One difference compared with the Great Plains is that the southwestern United States appears to have a stronger link to the Indian Ocean SSTs.

Supporting Online Material

Materials and Methods

Fig. S1

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article