Comparison of Soil Mineralogy From EMIT Imaging Spectroscopy and Ground‐Based Measurements in the Deserts of Southern California

Abstract The highly variable spatial composition of desert soils remains a major source of uncertainty in estimates of dust radiative forcing. Existing soil mineralogy data sets used in climate models are limited by sparse ground observations and coarse spatial resolution. NASA's Earth surface Mineral dust source InvesTigation (EMIT) addresses this challenge by providing near‐global maps of surface mineralogy across dust source regions using imaging spectroscopy at ∼60 m spatial resolution. Here, we evaluate EMIT 60‐m mineral mass fraction estimates against ground‐based measurements derived from X‐ray diffraction and iron wet chemistry analyses of soil samples collected from active dust sources in the Sonoran and Mojave Deserts of Southern California. We also compare particle size distribution measurements with EMIT 60‐m soil texture estimates used for estimating mineral mass fractions, including spectrally inactive and thus unretrieved minerals, mainly quartz and feldspar. Both EMIT 60‐m mineralogy and soil texture estimates exhibit limited variability compared with ground measurements. Ground measurements reveal a strong Spearman correlation ( r s = 0.84) between the mass fraction of minerals unretrieved by EMIT and the sand‐size mass fraction. We apply empirical adjustments that constrain the unretrieved mineral fraction using sand content and rescale iron oxide mass fractions, significantly improving agreement with ground mineralogy, including an increase in the iron oxide correlation from 0.15 to 0.68. We further develop an empirical method to partition unretrieved minerals into quartz and feldspar. Together, these empirical adjustments provide an efficient approach to enhance EMIT 60‐m mineralogy estimates and better represent fine‐scale spatial variability in soil mineralogical composition.