comparing wildfire leo-geo and geo-geo stereo winds and

1
Model Domain Camp Fire Woolsey Fire Comparison Domain Mariel Friberg 1,2 , Dong Wu 1 , James Carr 3 , James Limbacher 1,4 , Yufei Zou 5,6 , and Susan O’Neill 7 1 NASA GSFC, 2 USRA, 3 Carr Astronautics, 4 SSAI, 5 University of Washington, 6 PNNL , 7 USFS RESULTS DATA AND METHODS Stereo Winds from 3D-Wind [1] Stereo Aerosols from MAGARA [2] WRF-CMAQ simulations [3] BACKGROUND New satellite observations provide mesoscale details of fire dynamics details, thanks to the latest breakthroughs in satellite stereo- imaging for tracking from the 3D-Winds algorithm [1] and the Multi-Angle Geostationary Aerosol Retrieval Algorithm (MAGARA) for aerosol properties [2]. Enabling this new the state-of-the-art algorithm is the accurate image navigation and registration (INR) from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Geostationary Operational Environmental Satellite (GOES-R) suite Advanced Baseline Imager (ABI) imager. The INRs facilitate the conversion of stereo information into plume height and winds with a high vertical resolution in the planetary boundary layer (PBL) where most wildfire plumes reside [1]. References 1. Carr, J.L et al., “GEO-GEO Stereo-Tracking of Atmospheric Motion Vectors (AMVs) from the Geostationary Ring,” Remote Sensing, 2020. 2. Limbacher, J.A. et al., “A Multi-Angle Geostationary Aerosol Retrieval Algorithm,” AGU, 2019. 3. O’Neill, S. et al. A Multi-Analysis Approach for Estimating Regional Health Impacts from the 2017 Northern California Wildfires. JAWMA, 2021. CONCLUSIONS Our preliminary results highlight the importance of our sub-daily (LEO-GEO) and sub-hourly (GEO-GEO) stereo retrieved observations for model simulations. Using stereo retrieved products to constrain model input parameters (data assimilation) such as (1) injection height and time, (2) plume wind speed and direction, and (3) aerosol loading (relating to emissions) and aerosol transport. Figure 4. Sep. 9 th snapshots of the explosive 2020 California Creek Fire pyroCb at 14 UTC (top left) and 16:31 UTC (top right). The GOES-16 &-17 3D-Wind retrievals show dynamic wildfire PBL variations and capture the pyroCb plume top height reached ~11 km above the terrain (bottom left) with a wind speed of ~15 m/s (bottom right) [1]. MOTIVATION Impact of plume injection height on air quality and risk assessment prediction of wildfires Facilitate furthering our understanding of complex and highly localized wildfires Increase spatiotemporal observations of wildfire convective plume dynamics Plume top height, wind velocity, wind direction, and aerosol composition Support improvements to regional and local dynamic meteorological models of intense wildfires Pyrocumulonimbus (pyroCb) clouds Comparing Wildfire LEO-GEO and GEO-GEO Stereo Winds and Aerosols to WRF-CMAQ simulations Figure 2. The images show the (a) GOES-17 RGB, (b) Fine-mode AOD and (c) AOD from MAGARA retrievals, and (d) modeled AOD (CMAQ) for the Camp and Woolsey fires. Injection time for the Camp Fire is a significant contributor to the difference between observed and modeled AOD plumes. While the Woolsey Fire observed and modeled injection time were similar, the AOD injection height and emission loadings most likely contributed to transport differences. Figure 1. MODIS-GOES 3D-Wind retrieved wind velocity and direction and plume top height during the 2018 Camp Fire on Nov 10 th . Left image shows the RGB image of the Camp and Woolsey fires. The top right image shows plume height is higher near the fire source. The bottom right image shows the shows cross-section at 37.8 deg N and the dense fire plume with a top of ~1.2 km above the surface covering the San Francisco Airport and Bay Area [1]. Figure 3. Preliminary scatter plots compare absolute values of MODIS-GOES Stereo and WRF wind speed, u- and v- components (m/s) for the Camp and Woolsey Fires Nov 8-16, 2018. Dashed black line show WRF domain. Red boxes show different comparison domains. Black stars indicate the ignition locations of the Camp and Woolsey Fires. The lower speed cluster appears to follow the 1:1 line, while higher speed cluster indicates WRF winds are slightly lower than those of Stereo. Height (m) Longitude Wind (m/s) Height U Wind V Wind Terrain 10 (m/s) MYD Band 1 to GOES-16 CONUS Band 2 Cloud Height (m) Sep.9, 2020, UTC=14:01Z Sep.9, 2020, UTC=16:31Z San Francisco Plumb Shadow Overshootin g Plume North Complex Fire Creek Fire North Complex Fire Creek Fire Plume moving south Twilight San Francisco pyroCb GOES-16 RGB Fine-mode AOD AOD Modeled AOD 16:00 UTC 17:45 UTC 19:00 UTC 21:45 UTC Camp Fire Woolsey Fire ________________ MAGARA ________________

Upload: others

Post on 01-Aug-2022

20 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Comparing Wildfire LEO-GEO and GEO-GEO Stereo Winds and

Model Domain

Camp Fire

Woolsey Fire

Comparison Domain

Mariel Friberg1,2, Dong Wu1, James Carr3, James Limbacher1,4, Yufei Zou5,6, and Susan O’Neill71NASA GSFC, 2USRA, 3Carr Astronautics, 4SSAI, 5University of Washington, 6PNNL , 7 USFS

RESULTS

DATA AND METHODS• Stereo Winds from 3D-Wind [1]• Stereo Aerosols from MAGARA [2]• WRF-CMAQ simulations [3]

BACKGROUNDNew satellite observations provide

mesoscale details of fire dynamics details, thanksto the latest breakthroughs in satellite stereo-imaging for tracking from the 3D-Windsalgorithm [1] and the Multi-Angle GeostationaryAerosol Retrieval Algorithm (MAGARA) foraerosol properties [2]. Enabling this new thestate-of-the-art algorithm is the accurate imagenavigation and registration (INR) from theModerate Resolution Imaging Spectroradiometer(MODIS) and the Geostationary OperationalEnvironmental Satellite (GOES-R) suite AdvancedBaseline Imager (ABI) imager. The INRs facilitatethe conversion of stereo information into plumeheight and winds with a high vertical resolutionin the planetary boundary layer (PBL) where mostwildfire plumes reside [1].

References1. Carr, J.L et al., “GEO-GEO Stereo-Tracking of Atmospheric Motion Vectors (AMVs) from theGeostationary Ring,” Remote Sensing, 2020.

2. Limbacher, J.A. et al., “A Multi-Angle Geostationary Aerosol Retrieval Algorithm,” AGU, 2019.3. O’Neill, S. et al. A Multi-Analysis Approach for Estimating Regional Health Impacts from the 2017Northern California Wildfires. JAWMA, 2021.

CONCLUSIONSOur preliminary results highlight the importance of our sub-daily

(LEO-GEO) and sub-hourly (GEO-GEO) stereo retrieved observations formodel simulations. Using stereo retrieved products to constrain modelinput parameters (data assimilation) such as (1) injection height and time,(2) plume wind speed and direction, and (3) aerosol loading (relating toemissions) and aerosol transport.

Figure 4. Sep. 9th snapshots of theexplosive 2020 California Creek FirepyroCb at 14 UTC (top left) and 16:31UTC (top right). The GOES-16 &-173D-Wind retrievals show dynamicwildfire PBL variations and capturethe pyroCb plume top height reached~11 km above the terrain (bottomleft) with a wind speed of ~15 m/s(bottom right) [1].

MOTIVATION• Impact of plume injection height on air quality

and risk assessment prediction of wildfires• Facilitate furthering our understanding of

complex and highly localized wildfires• Increase spatiotemporal observations of

wildfire convective plume dynamics • Plume top height, wind velocity, wind

direction, and aerosol composition• Support improvements to regional and local

dynamic meteorological models of intensewildfires• Pyrocumulonimbus (pyroCb) clouds

Comparing Wildfire LEO-GEO and GEO-GEO Stereo Winds and Aerosols to WRF-CMAQ simulations

Figure 2. The images show the (a) GOES-17 RGB, (b) Fine-mode AOD and (c) AOD from MAGARAretrievals, and (d) modeled AOD (CMAQ) for the Camp and Woolsey fires. Injection time for theCamp Fire is a significant contributor to the difference between observed and modeled AODplumes. While the Woolsey Fire observed and modeled injection time were similar, the AODinjection height and emission loadings most likely contributed to transport differences.

Figure 1. MODIS-GOES 3D-Wind retrieved wind velocity and direction and plume topheight during the 2018 Camp Fire on Nov 10th. Left image shows the RGB image of theCamp and Woolsey fires. The top right image shows plume height is higher near the firesource. The bottom right image shows the shows cross-section at 37.8 deg N and thedense fire plume with a top of ~1.2 km above the surface covering the San FranciscoAirport and Bay Area [1].

Figure 3. Preliminary scatter plotscompare absolute values ofMODIS-GOES Stereo and WRFwind speed, u- and v-components (m/s) for the Campand Woolsey Fires Nov 8-16,2018. Dashed black line showWRF domain. Red boxes showdifferent comparison domains.Black stars indicate the ignitionlocations of the Camp andWoolsey Fires. The lower speedcluster appears to follow the 1:1line, while higher speed clusterindicates WRF winds are slightlylower than those of Stereo.

Hei

ght (

m)

Longitude

Win

d (m

/s)

• Height• U Wind• V Wind• Terrain

10 (m/s)

MYD Band 1 to GOES-16 CONUS Band 2 Cloud Height (m)

Sep.9, 2020, UTC=14:01Z Sep.9, 2020, UTC=16:31Z

SanFrancisco

Plumb Shadow

Overshooting Plume

North Complex Fire

Creek Fire

North Complex Fire

Creek Fire

Plume moving south

Twilight

San Francisco

pyroCb

GOES-16 RGB Fine-mode AOD AOD Modeled AOD

16:00 UTC

17:45 UTC

19:00 UTC

21:45 UTC

Camp Fire

Woolsey Fire

________________ MAGARA________________