Drought Research

Was the Dust Bowl predictable?

Richard Seager
Lamont-Doherty Earth Observatory of Columbia University

"Dust Bowl" by Alexandre Hogue (Missouri, 1898 - 1994)

Recent modeling work (Schubert et al. 2004a,b, Seager et al. 2005) has claimed that the Dust Bowl drought of the 1930s was forced by small changes in tropical SSTs, essentially several persistent years of La Niña-like conditions in the Pacific and a warm subtropical North Atlantic Ocean, both of which are assumed to have had natural origins. If this is so then it means that, had the SSTs been known in advance, it would have been possible to predict that the drought was to occur and, perhaps, the environmental and social catastrophe of the Dust Bowl could have been ameliorated.

Fundamentally that would have depended on being able to predict the SSTs and we currently do not know if tropical SSTs are predictable to any useful degree beyond the interannual timescale with only a few studies having looked into this (Karspeck et al. 2004, Seager et al. 2004). Here we take a different tack and place ourselves in 1929 and ask the more limited question: If the SSTs in the 1930s were foreknown could the drought have been predicted? The paper (Seager et al. 2007) appeared in the Journal of Climate.

Using ensembles of atmosphere model simulations to assess predictability

To answer this we compare 16 member ensembles of model simulations of the 1930s forced by historical SSTs variously imposed globally, in the tropical Pacific alone and in the tropical Atlantic alone. These simulations are compared with another ensemble which starts with the atmosphere and land initial conditions in January 1929 (taken from the global SST forced simulations) and integrates forward to 1940 but with the SST climatology of 1856-1928 imposed. This latter ensemble therefore ignores the influence of SST variations in the 1930s and looks to see if some combination of the atmosphere and land initial conditions and internal atmosphere-land variability could have produced a near decade-long drought.

Figure 1 shows maps of the ensemble mean precipitation anomaly averaged over 1932 to 1939 relative to the precipitation climatology for that ensemble for 1856-1928. The observed precipitation anomaly 1932-1939 minus 1856-1928 is also shown. The observed precipitation reduction covered most of the United States and was centered in the northern and central Plains. Unlike most historical droughts it also impacted the northern Rocky Mountains states and the Pacific Northwest. The modeled drought with global SST forcing (GOGA) has the right magnitude but was centered and extended too far south and did not impact the Pacific Northwest. The modeled drought was forced by both the tropical Pacific (POGA-ML ensemble) and tropical Atlantic (TAGA ensemble). In contrast, the ensemble with climatological SSTs (COGA) produces largely incoherent and weak precipitation anomalies emphasizing that the real drought was in large part SST-forced. Hence, advance knowledge of SSTs would have led (with this model) to a prediction of a serious drought but one that extended too far south and did not impact the Northwest.


Figure 1. The observed (top left) and modeled precipitation anomalies during the Dust Bowl (1932 to 1939) relative to a 1856 to 1928 climatology. Observations are from GHCN. The modeled values are ensemble means from the ensembles with global SST forcing (GOGA), tropical Pacific forcing (POGA), tropical Pacific forcing and a mixed layer ocean elsewhere (POGA-ML), tropical Atlantic forcing (TAGA) and with land and atmosphere initialized in January 1929 from the GOGA run and integrated forward with the 1856-1928 climatological SST (COGA). Units are mm per month.

Figure 2 shows statistics for the ensemble spread, ensemble mean and observed precipitation averaged over the Great Plains, the Southwest and the entire West (see caption for details). The ensemble with climatological SSTs produced precipitation anomalies that are not separated from climatology. In contrast the ensembles with SST forcing produce droughts that are well separated from climatology though clearly too strong in the Southwest and not strong enough in the West as a whole. Both Pacific and Atlantic SSTs contributed.


Figure 2. The lower quartile, median and upper quartile (shown by the box) and the lower and upper ranges (shown by the whiskers extending below and above the boxes) of the model ensembles for GOGA, COGA, POGA and TAGA for the Great Plains region (top), the Southwest (middle) and entire West (bottom). The observed value is shown as an asterisk. Units are mm per month.

Figure 3 shows maps for the surface air temperature anomaly. All the model simulations miss the remarkable extent and strength of warm anomalies in the 1930s. The observed warm pattern closely follows the pattern of precipitation reduction and is most likely a response to it. Thus the temperature prediction error follows from the precipitation prediction error. (As support for this, during the 1950s drought the model simulated quite realistic patterns of both precipitation reduction and temperature increase and again their patterns coincide.)


Figure 3. Same as Figure 1 but for temperature. Units are Kelvin.

The curious spatial pattern of the Dust Bowl drought

No model ensemble member produced a spatial pattern of drought akin to that observed during the Dust Bowl. That is, no mix of SST-forcing and internal variability was able to produce a Dust Bowl-like drought. Instead the model produces droughts centered in the Southwest and the southern and central Plains. These are the regions most connected, via atmospheric circulation, to tropical Pacific and Atlantic SST anomalies. That pattern of teleconnections is quite similar to what can be deduced from observations and, hence, the northern and northwestern parts of the Dust Bowl drought are hard to explain in terms of SST forcing alone. Other models (NSIPP, GISS, GFDL), to the best of our knowledge (since only one is published), also produced Dust Bowl droughts centered too far south. The 1950s drought is one that was centered in the regions of strongest teleconnection to tropical SSTs.

The inability to correctly model the spatial pattern of the Dust Bowl drought, and its deviation from the typical pattern of tropical SST-forced drought, makes us wonder how unusual the Dust Bowl pattern was. Analyses of instrumental data since the mid nineteenth century - shows that none of the other five multiyear droughts in this interval (1856-65, 1870-77, 1890-96, 1948-57 and post 1998) had a similar pattern. For a longer record we examined the updated version of the gridded tree ring records within the North American Drought Atlas. Computation of the spatial correlation between the Dust Bowl pattern and six year low pass filtered data revealed only three prior droughts that had an anomaly correlation exceeding 0.4 (Figure 4). All were in the Medieval period. No other droughts had an AC exceeding 0.31 presenting a clear separation between Dust Bowl type droughts and the more commonly occurring pattern with a center in the Southwest and southern and central Plains.


Figure 4. The tree ring reconstructed summer Palmer Drought Severity Index for the Dust Bowl drought and three prior droughts in the last millennium that had similar northern-centered patterns and without strong drought in the Southwest or Mexico. The data is from the updated version of the North American Drought Atlas of Cook and Krusic (2004).

  1. SSTs during the 1930s forced a severe multiyear drought in North America.
  2. Both tropical Pacific and tropical North Atlantic SSTs were involved in forcing the model drought
  3. The model drought is statistically significantly distinct from anything that can happen in the presence of climatological SSTs.
  4. The modeled Dust Bowl drought was further south than that observed and was not accompanied by the near-continental scale warming of the observed drought
  5. The observed drought had a pattern that has not been seen since Medieval times when three droughts with this pattern occurred.
Clearly aspects of the Dust Bowl drought could have been predicted in 1929 had advance knowledge of SSTs been available. This motivates increased effort to extend the predictability of tropical SSTs beyond the interannual timescale. The particular pattern of the Dust Bowl drought deserves special attention - including the possible role of the widespread devegetation and land surface degradation and dust aerosol loading during that decade, see our Dust Bowl page and the reference Cook et al., 2008. Also, Ben Cook has created an interesting movie of the dust loading in the GISS simulations of April to August 1934. View Movie. It is notable that the last time droughts with that pattern occurred (during the Medieval period) there is also evidence of devegetation and dune mobilization (Forman et al. 2001).

Picture from Worster (1979) 'Dust Bowl; The Southern Plains in the 1930s', Oxford University Press. The 'they' refers to the Dalhart (TX) Chamber of Commerce which thought the Dust Bowl series of paintings by fellow southerner, Hogue, defamed the South and tried to buy this painting to burn on the streets of Dalhart.

  • Cook, B.I., R.L. Miller and R. Seager, 2008: Dust and sea surface temperature forcing of the 1930's 'Dust Bowl' drought. Geophysical Research Letters, Vol. 35, L08710, doi:10.1029/2008GL033486. PDF
  • R. Seager, Y. Kushnir, M.F. Ting, M. Cane, N. Naik and J. Velez, 2008. Would advance knowledge of 1930s SSTs have allowed prediction of the Dust Bowl drought? Journal of Climate, 21, p. 3261-3281. DOI: 10.1175/2007JCLI2134.1. PDF
  • Seager, R., Y. Kushnir, C. Herweijer, N. Naik, and J. Velez: 2005. Modeling of tropical forcing of persistent droughts and pluvials over western North America: 1856-2000. J. Climate, 18, 40684091. PDF
  • Cook, E. R. and P. J. Krusic: 2004. North American Summer PDSI Reconstructions. Technical Report 2004-045, IGBP PAGES/World Data Center for Paleoclimatology Data Contribution Series, Boulder, CO, USA.
  • Karspeck, A., R. Seager, and M. A. Cane: 2004. Predictability of tropical Pacific decadal variability in an intermediate model. J. Climate, 17, 28422850. PDF
  • Schubert, S. D., M. J. Suarez, P. J. Region, R. D. Koster, and J. T. Bacmeister: 2004a. Causes of long-term drought in the United States Great Plains. J. Climate, 17, 485503.
  • Schubert, S. D., M. J. Suarez, P. J. Region, R. D. Koster, and J. T. Bacmeister: 2004b. On the cause of the 1930s Dust Bowl. Science, 303, 18551859.
  • Seager, R., A. Karspeck, M. Cane, Y. Kushnir, A. Giannini, A. Kaplan, B. Kerman, and J. Velez: 2004. Predicting Pacific Decadal Variability. Earth Climate: The ocean-atmosphere interaction , C. Wang and S.-P. Xie and J. A. Carton, ed., American Geo-physical Union, Washington, DC, 115130. PDF
  • Forman, S., R. Oglesby, and R. S. Webb: 2001, Temporal and spatial patterns of Holocene dune activity on the Great Plains of North America: megadroughts and climate links. Glob. Planet. Change, 29, 129.

Maintained by: Virginia DiBlasi-Morris , Lamont-Doherty Earth Observatory of Columbia University

LDEO home | search | ocp webmaster
Copyright © 2011 by The Trustees of Columbia University in the City of New York, Lamont-Doherty Earth Observatory.
All rights reserved.