INFERRED DUST SOURCES AND FLUX DURING THE 1930S DUST BOWL DROUGHT

The 1930s Dust Bowl was an epic drought reflecting both extreme climate variability and human-induced landscape degradation. The dust generated during this event was a regional health hazard, exacerbated warming and increased hemispheric dust loads. Building upon prior work, this study utilizes land-surface classifications from black and white aerial imagery taken during the 1930s over southwest Kansas to infer PM_2.5 and PM₁₀ sources from reactivated eolian landforms and/or adjacent agricultural fields.

The minimum and maximum estimates of effective dust emissivity was calculated for these classified surfaces, based on dust flux measurements from analogous surfaces in the western USA. The defined parameters reflect the availability of the surface to emit PM_2.5-10 that can be suspended in the atmosphere for hours to months. The relative proportion of the surfaces classified for southwest Kansas was extrapolated to the wider Dust Bowl region to calculate probable areas with varying dust emissivity. A spatially constrained record of surface cover for 1935 and 1939 Dust Bowl counties was developed by integrating this analysis with agricultural census data. The geometrical mean of the emissivity rate was calculated to the 95% upper and lower confidence interval to estimate the minimum and the maximum range of dust emission by land surface type and by year.

The estimated emissions of total suspended particles (TSP) was between 198 million and 1.7 billion metric tons for the Dust Bowl area in 1935 and between 82 and 706 million metric tons in 1939 for a shrinking drought area, consistent with estimates from the Soil Conservation Service during the DBD. The majority, 65 and 52% of TSP in 1935 and 1939 were from uncultivated, antecedent dune landforms with the remainder from cultivated, and other human modified surfaces. The estimated PM₁₀ and PM_2.5 emissions from all surfaces in the Dust Bowl area was 3,000 to 16,500 μg m^-3 d^-1 and 257 to 2,700 μg m^-3 d^-1, respectively. These minimum emissivity estimates are consistent in magnitude, but higher than values assumed in modelling the climate effects of dust loading during this drought. The land surface attributes are noticeably heterogeneous, with dust sources preferentially from uncultivated areas, rather than solely from cultivated surfaces, as assumed in climate models.