Assessing drought vegetation dynamics at the landscape scale in semiarid grass- and shrubland using MESMA

Keywords: remote sensing, dryland ecology, New Mexico, grassland, shrubland, MESMA
Session Type: Virtual Paper
Day: Wednesday
Session Start / End Time: 4/7/2021 08:00 AM (Pacific Time (US & Canada)) - 4/7/2021 10:20 AM (Pacific Time (US & Canada))
Room: Virtual 29

Abstract

New Mexico experienced significant impacts of regional-scale drought from 2011-2014. Global climate change may make such events a new normal for the southwest: drought events are expected to increase in both frequency and severity over the coming century. While semiarid grasslands recover quickly from short-term drought, the cumulative impacts of climate change may reduce the resiliency of these systems over time. Remote sensing methods can allow efficient and cost-effective comparison of ecosystem recovery from drought events over time using long-running imaging systems like Landsat. We investigate the efficacy of using multi-endmember spectral mixture analysis to quantify the impacts of drought events on the fractional cover of green vegetation (GV), non-photosynthetic vegetation (NPV), and soil (S). Field spectra of dominant vegetation species were collected at the Sevilleta National Wildlife Refuge over six field sessions from May – September 2019. Four endmember selection methods were tested to optimize the spectral library, as well as three thresholding adjustments to unmix Landsat imagery from 2009 (two years pre-drought), 2014 (final year of drought), and 2019 (five years post-drought). The best fit model had high levels of agreement for all three classes with R2 values of 0.89 (NPV), 0.71 (GV), and 0.81 (S), respectively. Image differencing showed increases in S and decreases in NPV fractions throughout the study area that were unaccompanied by a return to baseline cover in the post-drought period.