Upcoming Events
MS Thesis Defense - Sarah Ostertag
Nov 21, 2023, 11:00 AM - 12:30 PM
GMU Exploratory Hall, Room 2312
Zoom: https://gmu.zoom.us/j/96642402654?pwd=b2diWWVGVGlENnUwdUQrQlYxekxCdz09
Candidate: Sarah Ostertag
Master of Science in Earth Systems Science
Department of Geography and Geoinformation Science, Department of Atmospheric, Oceanic & Earth Sciences
TITLE: Geospatial Techniques for Assessing Select Wildfire Variables in California
Committee:
Thesis Director: Dr. Matthew Rice
Committee Members: Dr. John Qu, Dr. Daniel Tong
ABSTRACT:
This research highlights various geospatial techniques to gain insights from remote sensing data and vector, raster, multidimensional, and statistical processing tools to assess aspects of wildfire in California. The first chapter summarizes the results of 98 Mann-Kendall time series trend tests for wildfire occurrence and severity, including a comparison of 6 different severity metrics. A database capturing large wildfire perimeters (> 1,000 acres; n=2,857) in the state of California in the United States is used to document trends in fire occurrence (as frequency of large wildfires per year) and fire severity (using total annual area burned [AAB] as a proxy) per ecoregion for the study period 1950 – 2020. Fire severity was also assessed with five additional metrics based on Monitoring Trends in Burn Severity (MTBS) annual burn mosaics over a subset of the study period (1984 – 2020), based on data availability: annual area burned at high severity (AABHS), the annual area burned at low to moderate severity (AABLMS), the mean burn severity class per ecoregion (SEVmean), the proportion of total annual area burned at low to moderate severity (LMSprop), and the proportion of total annual area burned at high severity (HSprop). Approximately 20.67% of the total area of California has been burned by large wildfires during the study period; of this, approximately 8.01% of the total area of California has been burned repeatedly. However, large wildfires are not occurring more frequently across the entire state as originally postulated, but rather these increases in occurrence are centered in the Sierra Nevada and Klamath mountain regions. Mountainous ecoregions are the driving force behind public sentiments in California and the state’s wildfire trends, characterized by both increased large wildfire severity and, in some cases, large wildfire frequency, with a likely influence from an increasing incidence of lightning and dry lightning strikes in these regions induced by climate change. Using six metrics – Total AAB, AABHS, AABLMS, SEVmean, LMSprop, and HSprop – to assess wildfire severity rather than solely relying on total AAB as a proxy allows for a more nuanced analysis that reveals that even though more area is burning at a higher severity in California as a whole and in select ecoregions, more area is burning across the state in general, and a greater proportion of this increase in burned area is attributed to low to moderate severity fires, rather than high severity fires.
The second chapter explores another important wildfire phenomena, smoke, and more specifically smoke plumes that rise high into the atmosphere and travel large distances, which can have numerous effects. Deficient information on smoke plume rise can result in large uncertainties on air quality and climate model simulations of aerosol vertical profiles and smoke plume dispersion. Using MODIS and VIIRS fire pixels for context, top of smoke plume heights for northern California on September 9, 2020 were collected and measured from MISR, CALIOP, AOCH, MAIAC, and GFAS data products and were compared to results derived from the meteorological and air quality model presented in Li et al. 2021 and Li et al. 2023. Results from MAIAC and GFAS consistently plot smoke plume heights near the terrain surface or omit smoke particles altogether over the dense smoke plume visible in true-color imagery, as each of these schemes are inherently limited by MODIS cloud and smoke misclassification issues. Conversely, predictions from the WRF/CMAQ air quality model, retrievals of MISR, AOCH, and CALIOP compare favorably to one another and appear to identify stratified injected smoke layers more adeptly in the troposphere compared to MAIAC and GFAS. These observed differences highlight the difficulty in distinguishing between high altitude, optically thick smoke plumes and clouds during events with heavy atmospheric particle loads and the importance this limitation has on the sensitivity and effectiveness of different smoke plume height retrieval methods.