Improving Forecasting of Hurricanes, Floods, and Wildfires

Overview Report on Supplemental Program FY-19

Photograph of Golden Gate bridge which connects San Francisco to Martin, CA.  Much of the hillside in Martin is ablaze in a wildfire.  There is also significantly reduced visibility due to smoke causing a orange-colored sky.

2020 was the most active Atlantic hurricane season on record. A record eleven named storms, out of the 30 total in the Atlantic, made landfall in the United States. Five Category 4 storms (also a record) formed in the 2020 season. The winds in 3 of the named storms - Iota, Delta and Eta - intensified by at least 100 mph in 36 hours or less. While the eastern U.S. was pummeled by tropical cyclones (known as hurricanes in the North Atlantic and central and eastern North Pacific), the western U.S. had raging wildfires. In California, over 9,279 fires burned over 4,358,517 acres making 2020 the largest wildfire season in California’s recorded history. Disasters related to weather/climate are becoming increasingly costly. Aside from the injuries and loss of life, a total of  22 weather/climate disaster events  impacted the United States in 2020, with losses exceeding $1 billion each.

Billion-Dollar Disasters by the Numbers (1980-2020). Credit: NOAA

Though not limited to hurricane and wildfires research, the Fiscal Year 2019 (FY19) Improving Forecasting for Hurricanes, Floods, and Wildfires (IFHFW) portfolio, certainly conducts world-class research projects in these areas. In an on-going effort to curb the cost to life and property, the U.S. Congress has passed a number of laws. One such law,  PUBLIC LAW 116–20 , passed JUN. 6, 2019, states for an additional amount for ‘‘Operations, Research, and Facilities’’ for necessary expenses related to the consequences of Hurricanes Florence and Michael, Typhoon Yutu, and of wildfires, $120,570,000, to remain available until September 30, 2020, and in part, $25,000,000 to improve (a) hurricane intensity forecasting, including through the deployment of unmanned ocean observing platforms and enhanced data assimilation; (b) flood prediction, forecasting, and mitigation capabilities; and (c) wildfire prediction, detection and forecasting.

This document will provide an overview of the $25 million described above. The Weather Program Office (WPO) within Oceanic and Atmospheric Research (OAR) of NOAA, takes the lead in organizing and managing this effort. World-class scientists are researching 11 separate projects over a 3 year period. Through continuous reporting and collaboration across multiple line offices, such as National Ocean Service (NOS), National Weather Service (NWS), National Environmental Satellite, Data, and Information Service (NESDIS), and Office of Marine and Aviation Operations (OMAO) and research laboratories, such as Atlantic Oceanographic and Meteorological Laboratories (AOML) and Earth System Research Laboratories (ESRL), the research projects are closely monitored. Many of these projects leverage the Unified Forecasting System (UFS) framework for research to operations transitions, which is mission critical for NOAA.  

Simple horizontal timeline of the FY19 Disaster Supplemental Program. Starts in 2019 with the Congressional Act supplying funds and concludes in 2023 with the last research projects finishing their work.

Timeline for FY19 IFHFW Portfolio

Below are links to each section of the report, including a breakdown of spending for the IFHFW portfolio, a map of partner organizations, 3 research focus areas (consisting of 11 projects), 4 vignettes (3 are short summaries of selected research projects, and 1 is on accelerated model transitions), and the report conclusion. 

Hurricane Laura (2020) impacting the Gulf Coast ( left ), 2020 California wildfire ( middle ), and 2018 flash floods in Maryland ( right ).

Financial Breakdown

Partner Organizations

2019 Improving Forecasting for Hurricanes, Floods, and Wildfires (IFHW) Portfolio

Adding to the work that began in the Improving Forecasting And Assimilation (IFAA) portfolio from the previous year, the IFHFW portfolio had two common focus areas with the IFAA portfolio - hurricanes and flooding, while undertaking a new focus area - wildfires.

Hurricane wind and surf battering a coastline.

Image Credit:  National Geographic 

I. Hurricanes. Research shows that properly representing ocean parameters (i.e. salinity, wave heights, and ocean heat content, etc.) in forecasts can reduce error in tropical cyclone (TC) intensity forecasts by as much as 50%. By expanding the use of underwater gliders and air-launched buoys and drifters, data gaps can be addressed. Through this increased and improved ocean data collection, improvements in hurricane intensity forecasts can be made along with an overall better understanding of hurricane rapid intensification (RI). Also, work is being done to provide accelerated development of the Hurricane Forecast and Analysis System (HAFS). Specifically, by improving data assimilation within HAFS, improved guidance will be provided in rainfall, storm surge, and tornadoes with land-falling hurricanes. Other benefits include improved skill in hurricane track, size, and intensity out to 7 days and to better quantify track, intensity, and size uncertainty.

Additional work within this portfolio will improve forecasting skill of hurricane track and intensity by using the vast amount of data from GOES satellites while leveraging artificial intelligence (AI) and machine learning (ML). This will lead to an increase in volume and quality of satellite observations in numerical weather predictions. Techniques will be applicable to multiple data assimilation frameworks such as Gridpoint Statistical Interpolation (GSI) and Joint Effort for Data Assimilation Integration (JEDI). This will allow NOAA’s global and hurricane prediction to extract more information content from satellite observations, resulting in less uncertainty in analyses and forecasts.

Images from research project's Quarterly Project Review showing model runs that have been completed on the "cloud".

RRFS Ensemble Test Runs on AWS Cloud

Most of the projects within this portfolio primarily use High Performance Computing (HPC) through the existing NOAA resources on the Jet, Hera, and Orion supercomputers. A few utilized HPC options through universities and cooperative institutes. Under the hurricane research focus area, one project utilizes a new HPC option, computer computation on the Cloud. This is specifically exercised for a prototype UFS (Unified Forecast System) based Rapid Refresh Forecast System (RRFS) model output. By establishing a modeling system on the Cloud, this project will help accelerate progress by providing computational resources for the research and further development of RRFS. This will allow the RRFS to replace legacy modeling systems and accelerate the transition to an operational UFS. Further, 3 kilometer resolution ensemble real-time test runs are enabled with the Amazon Web Services (AWS) Cloud. 

Image Credit: NOAA

II. Flooding. Greater density of data from the atmosphere is almost always seen as beneficial. The four flooding projects of this portfolio improve the flow of data about the atmosphere and the hydrosphere. Increasing our knowledge of oceanic heat content is just one aspect for the UFS to attain improved hurricane forecasting. By developing 4D data assimilation with Marine JEDI, high volume and high resolution ocean data will be ingested in air-sea coupled models within the UFS. Also, this air-sea model coupling will be focused in the US Atlantic and Gulf of Mexico regions, by conducting impact analysis for select historical hurricanes. Additionally, accelerating development of high resolution Ocean Data Assimilation techniques for oceanographic data from in situ and remote platforms will support improved hurricane intensity forecast models and surge predictions in the FV3-based next generation HAFS and Marine JEDI frameworks.

Flooding impacts the environment (erosion, habitat loss, pollutants, etc.) and is costly to the economy. It greatly impacts lives and property. Beyond the inherent challenges of QPF (Quantitative Precipitation Forecast) in a potential flood situation, there’s a need to know how much of the precipitation will soak into the ground vs. how much will contribute into the streamflow that can lead to flooding. The very best performing QPF is still limited without this additional knowledge. Projects within this portfolio will deliver accelerated improvement to the National Water Model (NWM) by developing prototype physics modules at Research Level (RL) -7 or higher representing heterogeneous infiltration processes. The key benefits will be increased forecast skill in the current and next generation NWM. Additional benefits include: accelerate flood water depth and flood mapping capabilities; improve NWM and inland flooding prediction through the development of Heterogeneous Infiltration Model(s); leverage/exploit satellite observations to improve flooding and inundation forecasts and monitoring; continue advancing ocean data assimilation and coupling of air-sea models in the UFS to support improved flood and inundation forecasting.   These improved predictions will help emergency management make better decisions during potential search and rescue missions, and they will have better knowledge of flood prone areas ahead of time.

NY Times Photograph of lone-fireman battling wildfire in western U.S. Fireman in foreground of massive fire and smoke.

Image Credit:  NY Times 

III. Wildfires. The U.S., on average, has over 48,000 Wildfires each year. These fires consume over 6 million acres and put about 4.5 million homes at risk. In addition, these fires release carbon dioxide (CO2), black and brown carbon, and ozone (O3) precursors. Wildfires can affect local weather conditions as well. Direct emissions of toxic pollutants can affect first responders and local residents. The formation of other pollutants as the air is transported can lead to harmful exposures for populations in regions far away from the wildfires. Smoke and air quality operational forecast models use a ‘persistence approach’ to produce wildfire emission forecasts. However, this simple approach leads to less than optimal forecast accuracies in regions suffering from wildfires. It is well documented that wildfire emissions are strongly modulated by the state of the surrounding vegetation for fuel and the present and future weather conditions. Projects within this portfolio will advance improvements in UFS modeling of wildfire smoke impacts; advance FV3-CAM to improve wildfire detection and prediction; and development and readiness of satellite observations for fire and smoke forecasts. Smoke aerosols greatly impact the environment and weather. Improved smoke models will allow scientists to better quantify smoke impacts on the atmosphere and improve predictions of PM 2.5 and ozone concentration.

Projects within this portfolio aim to incorporate Machine Learning (ML) to help assimilate the diurnal cycle of Fire Radiative Power, as well as improvements in microphysics, radiative properties, and Aerosol Optical Depth (AOD). Projects will improve the detection and forecasts of the conditions leading to the onset of fires, fire emissions after the onset of fires, and local weather forecasts in areas surrounding wildfires. This research looks to improve wildfire detection and prediction by increasing computing power and by assessing the land surface model state (i.e., dryness of vegetation, wind speeds, relative humidity, planetary boundary layer structure and potential lightning). One goal of this research will also yield improvements in the Fire Weather Index (FWI) to be used by Storm Prediction Center (SPC) in their Fire Weather Outlook. Additionally, it will improve air quality forecasts and improve biomass burning smoke forecasts. Geostationary satellites (GOES 16 & 17) provide fire products every 5 minutes. Fire emissions and characterization can be made available hourly. This research is benefiting NWS operational air quality and fire prediction models. Maintaining assimilation/model ready fire and smoke satellite products and disseminating these satellite observations with lower latency will allow better preventive measures to be taken by local, state, and federal agencies (time to make more informed decisions). Improved forecasts increase preparedness (i.e. more prepared fire fighters).    

The following 4 short summaries/vignettes are just a sampling of the research projects that comprise the FY19 Disaster Program portfolio. For a complete list of all 11 projects please use this link:  Complete list of FY19 Projects 

Vignette #1 - Hurricane Glider/Drifter Autonomous Observations for Hurricane Intensity Forecasts Project

Visible satellite image of hurricane Michael (2018) approaching the Florida panhandle as a strong category 4 storm.

Category 4 Hurricane Michael on Wednesday, October 10th 2018. Ocean conditions were favorable for intensification and have been shown as contributing to Michael's intensification ( Le Hénaff et al. 2021 ). Image Credit: NOAA

Many studies have shown the importance of including ocean information in hurricane intensity forecasting. The ocean is a key regulator of hurricane intensity. Essential Ocean Features (EOF) such as the Atlantic Warm Pool, Gulf Stream, Loop Current, and Mid-Atlantic Cold Pool cause rapid intensification and weakening of tropical cyclones (TC). Ocean data is especially critical for getting the forecast models to represent the ocean correctly, particularly when favorable EOFs are in the path of an approaching tropical cyclone. This project extends and enhances ocean observing data collection to support hurricane intensity forecast improvementsThis project will also build on existing efforts with underwater and surface systems and will deploy autonomous underwater profiling gliders and drifters that address critical data gaps and target features known to cause rapid intensity changes in tropical cyclones.

Storm surge from a tropical system flooding the coastal area, as well as strong winds generating large storm waves.
Two CIMAS employees deploying an underwater glider ahead of the 2020 hurricane season.
The track of a small wave drifter buoy is shown traversing up the U.S. east coast via the Gulf Stream current. Two additional photos show the small wave drifter boy being deployed in the Straits of Florida.

Deployment and track of wave drifter buoy.

This multi-partner project deploys gliders and drifters throughout the hurricane season to collect surface and subsurface water temperature and salinity, sea surface air pressure, waves, and winds. These data improve how the ocean is represented in operational models that are used to forecast RI in TCs. This project will continue and expand this ongoing, collaborative effort.

To carry out the goals of this project:

A schematic depicting how the underwater gliders navigate down to approximately 1000 meters in the ocean, collect data, and return to the surface to transmit the data back to its base station.

Representation of how the underwater gliders navigate their dives to collect valuable ocean data ahead of tropical systems. Image credit:  NOAA AOML 

  1. A continued collaboration among NOAA (NOS, OAR, NWS), U.S. IOOS Regional Associations, and Scripps Institute of Oceanography partners is needed to coordinate ongoing field plans for the Atlantic basin. For underwater glider deployments, the Mid-Atlantic Regional Ocean Observing System (MARACOOS) partners will coordinate ongoing field plans for the Mid-Atlantic, and the Caribbean Coastal Ocean Observing System (CARICOOS) partners along with AOML, will coordinate field plans for the Caribbean and tropical Atlantic. AOML is leading the coordination of a glider deployment in the Bahamas, as part of a new partnership with the Cape Eleuthera Institute. The data collected were highly valuable during the record-breaking 2020 hurricane season and have improved our understanding of EOF impacts on TC intensification.
  2. The network of gliders and drifters is again actively collecting data for the 2021 Atlantic Hurricane Season, which is expected to be an above-normal season. The 2021 season will be the first season in which the observation data are assimilated into the operational ocean model used to represent the ocean in regional hurricane models.

Underwater glider at the ocean surface (left, Credit:  NOAA AOML ). Drifting buoy deployed in front of Tropical Storm Isaias 2020 (right, Credit:  NOAA AOML ).

Vignette #2 - Advance FV3-CAM to Improve Wildfire Detection and Prediction

Photograph of helicopter dropping fire retardant over wildfire in western U.S. Numerous tall pine trees ablaze in foreground.
Wildfire drawing ominously close to neighborhood homes in western U.S. Reduced visibility due to significant smoke from wildfires.
Table showing recent decades of Wildfire billion-dollar events, damage and fatalities in the U.S., with the 2011 - 2020 decade with the most severe totals. This most deadly decade had 255 fatalities and $71.9 M in damages.

Billion Dollar Wildfire Events to Affect the United States from 1991-2020. *Unadjusted Total Cost Source: NOAA National Centers for Environmental Information (NCEI) U.S. Billion-Dollar Weather and Climate Disasters (2021).  https://www.ncdc.noaa.gov/billions/ , DOI:  10.25921/stkw-7w73 

Photograph of hillside ablaze, along with dramatic smoke, in western U.S. wildfire.
Map of CONUS showing areas of Elevated and Critical Fire Weather Conditions, with the worst Fire Weather Conditions in the southwestern U.S. Map also shows a smaller area of Elevated Fire Weather Conditions across SC and GA.
Animation of the experimental Fire Weather Index (FWI) for the same time period as previous image/map. Strong correlation with the Elevated and Critical risk areas shown in previous image (both in the desert Southwest and in the states of SC and GA).

As the frequency, severity, and impacts of wildfires increase, the importance of wildfire prediction and detection becomes even more critical. It then becomes essential to improve fire weather and smoke & AQ forecast accuracies in current operational models. This involves taking advantage of recent advancements in wind and lightning forecasts as well as improvements in land-surface (i.e. soil moisture, vegetation greenness/dryness, etc.) modeling. Additional improvements include enhancing the representation of other atmospheric states in models including temperature, moisture, etc. 

This project is accelerating wildfire prediction, detection, and forecasting by:

Bootleg fire and smoke as seen by the Geostationary Operational Environmental Satellite (GOES) in Pacific Northwest during the summer of 2021. Credit: NOAA

  1. Improving detection and forecasts of conditions leading to the onset of fires by integrating land surface model states (i.e. vegetation and dryness), wind speeds, relative humidity & Planetary Boundary Layer (PBL) structure, and potential lighting from explicit convection into the Convection Allowing Model (CAM). This will improve Fire Weather Index (FWI) forecasts. 
  2. Improving fire emissions after the onset of fires by integrating information from satellite-derived products into the CAM, improving plume-rise algorithm to determine injection heights, and developing Machine Learning (ML) algorithm to improve the diurnal (day-night) cycle of the Fire Radiative (FRP) power.
  3. Improving weather forecasts from wildfire smoke by (a) implementing additional wildfire-weather feedback mechanisms by adding the interaction with precipitation physics, (b) including all other relevant aerosol sources (dust, sea salt, and anthropogenic emissions) needed to further improve radiative properties and microphysics, and (c) introducing the assimilation of aerosol optical depth (AOD) in the CAM.

Vignette #3 - Accelerate the Exploitation of Satellite Observations to Improve Flooding and Inundation Monitoring and Forecasts

Photograph of severe flooding. Shows the inherent dangers of trying to drive through flooded areas with a car in a precarious position.
Arial photograph of extreme flooding with failed containment.
Table showing new record river crest and date for Midwest Flood Event of March 2019.

Preliminary record breaking river crests reported to National Weather Service during the flood event of March 2019 in the Midwest region (Table credit:  NWS ).

Side-by-side aerial views of the Elkhorn River at Platte, MO showing before and after of the March 2019 Midwest flood event.
Map of Texas Coastline in September 2019 showing inundated areas from Tropical Storm Imelda. Also shows SAR data from the National Water Model.

NOAA NESDIS STAR utilized the capabilities of the SAR instrument to detect areas submerged due to Tropical Storm Imelda (2019). The blue areas represent flooded areas identified by SAR. (credit:  NOAA NESDIS )

Reliable observations of flooding and inundation spanning larger extents are key in advancing the NWM capability. This project aims to address this need by utilizing satellite data to fill spatial gaps crucial for improvements in the NWM, which will aid forecasters, emergency managers, and local officials in decision making. Remotely sensed data from the Visible Infrared Imaging Radiometer Suite (VIIRS) on the JPSS satellites, Advanced Baseline Imager (ABI) on the GOES satellites, and Synthetic Aperture Radar (SAR) on the Radarsat and Sentinel satellites are being used to meet project goals. SAR data is possibly the best up and coming data for these purposes due to its ability to “see” through clouds and night conditions, as well as its high spatial resolution that can get down on the order of 10 meters.

To accomplish the goal of advancing flood and inundation forecasting skill of the NWM, this project plans to:

  1. "...Develop new and improved satellite products, with an emphasis on inundation, and deliver them in a reliable, and efficient manner to the [National Water Center] NWC and other interested offices of the NWS."
  2. "Innovative methods to compare the generally coarser satellite data to the NWM will be developed, as well as methods to integrate the various sources of satellite inundation maps from a variety of sensors."
  3. "Other satellite parameters and data sets related to hydrologic parameters (e.g., soil moisture, precipitation, vegetative state, etc.) will also be developed and tailored to NWC needs."

Vignette #4 - Accelerating Model Transitions

An objective of the FY19 IFHFW Portfolio is to support the acceleration of forecast model implementations that will help to ensure the timely implementation of Supplemental model developments into operations. To this end, $3 million of FY19 funds was allocated for staffing to support the transition to operations of model components and forecast updates being delivered by both FY18 and FY19 Supplemental funded research and development projects. While model updates are being developed, staff is also addressing any implementation backlogs that will prepare for the implementations needed for the Supplemental program. Through this funding, NOAA’s National Centers for Environmental Prediction Central Operations (NCO) is providing support for supplemental and non-supplemental product implementations into operations. 

To date, Supplemental supported NCO staff have transitioned 15 numerical weather model updates into operations. Of the 15 transitioned model updates into operations, 4 have been Supplemental funded projects while 11 model upgrades were non-supplemental projects. Supplemental funding has also allowed 2 other supplemental model upgrades to be operationalized by non-supplemental funded NCO staff. These model upgrades include the Real-Time Ocean Forecasting System and the Rapid Refresh (version 5)/High Resolution Rapid Refresh (version 4).

An example of a Supplemental funded model operationalized by Supplemental funded NCO staff is the Nearshore Wave Prediction System (version 1.3). The Nearshore Wave Prediction System provides high resolution nearshore wave model guidance to the U.S. coastal Weather Forecast Offices. The version 1.3 upgrade introduces the first hourly probabilistic hazardous rip current guidance up to 6 days out for the East and Gulf Coasts of the United States along with Puerto Rico, Hawaii, and Guam. This enhancement will help save lives as more than 100 deaths each year can be attributed to rip currents. The Nearshore Wave Prediction System version 1.3 upgrade also introduces the first hourly probabilistic erosion and overwash guidance for up to 6 days out on the U.S. East and Gulf Coasts. This is important because coastal property loss due to erosion costs approximately $500 million each year. 

Nearshore Wave Prediction System model viewer display of map of CONUS with nearshore wave observations (yellow diamonds) along with NWS forecast office area of responsibilities (red boxes).

Nearshore Wave Prediction System model viewer. The yellow diamonds are observation locations and the red boxes represent various Weather Forecast Offices county warning areas. (Credit:  NOAA )

For more information visit the following links:  Nearshore Wave Prediction System ,  Version 1.3 Upgrade 

CONCLUSION

National Oceanic and Atmospheric Administration (NOAA) logo

Much of the research performed by the Supplemental Programs, conducted with NOAA and its partners, is gaining an understanding of ever-increasing impactful atmospheric phenomena. Drought, severe storms, tropical cyclones, and wildfires resulted in 262 deaths and significant economic impacts. For the past 110 years, the mean U.S. Climate Extremes Index (CEI) was 20.58%. The CEI summarizes U.S. trends in temperature, precipitation, drought data, and includes tropical system activity based on the wind velocity of landfalling hurricanes and tropical storms. In 2020, the CEI was 44.63%. Between 1910 and 2020, the CEI has eclipsed 40% only 3 other times (1998, 2012, and 2017). Using this metric, an alarming trend would portend that the work done by this portfolio will see increased importance in the coming years and will keep within NOAA's mission of science, service, and stewardship that is directed to a vision of the future where societies and their ecosystems are healthy and resilient in the face of sudden or prolonged change.

Bar graph showing Climate Extremes Index (CEI). Apparent from the graph is the increasing trend in recent years of the CEI, including the plot of the 7-year average. The CEI takes into account all types of extreme weather/climate events, including tropical cyclones.

U.S. Climate Extremes Index 1910 - 2020

Credits:

The Supplemental Program Team is managed by the NOAA Line Offices of: Oceanic and Atmospheric Research / Weather Program Office (OAR/WPO), National Weather Service / Office of Science and Technology Integration (NWS/STI), and National Environmental Satellite, Data, and Information Services / Center for Satellite Applications and Research (NESDIS/STAR). For questions and/or feedback, please email the Supplemental Program Team at hurricanesupp.wpo@noaa.gov.

Billion-Dollar Disasters by the Numbers (1980-2020). Credit: NOAA

Timeline for FY19 IFHFW Portfolio

RRFS Ensemble Test Runs on AWS Cloud

Image Credit: NOAA

Image Credit:  NY Times 

Category 4 Hurricane Michael on Wednesday, October 10th 2018. Ocean conditions were favorable for intensification and have been shown as contributing to Michael's intensification ( Le Hénaff et al. 2021 ). Image Credit: NOAA

Deployment and track of wave drifter buoy.

Representation of how the underwater gliders navigate their dives to collect valuable ocean data ahead of tropical systems. Image credit:  NOAA AOML 

Billion Dollar Wildfire Events to Affect the United States from 1991-2020. *Unadjusted Total Cost Source: NOAA National Centers for Environmental Information (NCEI) U.S. Billion-Dollar Weather and Climate Disasters (2021).  https://www.ncdc.noaa.gov/billions/ , DOI:  10.25921/stkw-7w73 

Preliminary record breaking river crests reported to National Weather Service during the flood event of March 2019 in the Midwest region (Table credit:  NWS ).

NOAA NESDIS STAR utilized the capabilities of the SAR instrument to detect areas submerged due to Tropical Storm Imelda (2019). The blue areas represent flooded areas identified by SAR. (credit:  NOAA NESDIS )

Nearshore Wave Prediction System model viewer. The yellow diamonds are observation locations and the red boxes represent various Weather Forecast Offices county warning areas. (Credit:  NOAA )

U.S. Climate Extremes Index 1910 - 2020