Remote Sensing Innovation Workshop – After-Action Report
Dates: August 28 – August 29, 2019
Location: FEMA HQ * FEMA Conference Center, Conf. Room C
400 C Street SW, Washington, DC 20472
Participation: In-Person, By Invitation Only
Public Sector R&D Participants
Convene key leaders in public sector research and development (R&D) with existing efforts focused on advancing the use of different types of imagery/remote sensing to support the disaster management mission.
Develop a baseline understanding of the leading public sector R&D efforts underway, and the unique and specialized expertise each offers.
Share the key mission and business requirements for how imagery/remote sensing applies to the disaster management mission and is applicable for addressing to answer the highest priority problem areas.
Identify potential approaches and innovative solutions for addressing the top where all types of imagery/remote sensing can be brought together to best support disaster management.
Develop a coordination strategy and action plan for working together; public sector R&D groups, technical users, and decision-makers to fully use remote sensing products prior to, during, and after an incident.
Remote Sensing-Oriented Problem Statements
Remote Sensing Applications for Real-World Challenges
The public sector R&D groups presented innovative remote sensing solutions for emergency response in the fields of damage assessments, transportation, and debris detection. After presentations, the R&D participants and FEMA Staff and Support Staff broke out to begin initial discussions on these three problem statements to assess the questions being asked, the solution requirements, the challenges for developing a solution, and the business process where the solution will be used.
What is needed? Remote Sensing / Artificial Intelligence (AI) derived product that indicates damage across all structure types within pre-post 48h to support the emergency declaration.
Criteria. It must be easy to access, understand, and repeat.
Gaps. R&D needs infrastructure datasets, awareness and access to the imagery available for damage assessments, Damage Assessments AI algorithms are complex and require expensive and complex imagery, understanding of insurance requirements, and collaborative environment to operationalize acquisition of damage assessments products from partners/public sector R&D groups/stakeholders.
Goal. Public Assistance and Individual Assistance to pay grants based on estimated damage assessment products.
Innovative Ideas. An approach based on timeline and availability of imagery, i.e., satellite initially then combined Synthetic Aperture Radar (SAR), optical sensor approach, from Civil Air Patrol (CAP) and Federal Partners to make decisions based solely on remotely
sensed products. Multi-nodal sources to further refine, e.g., flood model extents, field
surveys. Add social vulnerability factors to point counts to provide the, “so what”?
What is needed? Identification of road blockages, extracting possible isolated communities, road damage proxy products, and a system for continuous tracking of the operational status of the road network. In addition, routing analysis for given natural disaster events and routing analysis for evacuations.
Criteria. During a disaster, foundational data becomes transactional (live, operation status updates) and then back to Foundational. Data must be temporal and continuously updated. Theoretical “Road Outage Modeled Product” will leveraging multi-modal inputs to extract and map road blockages, e.g., landslide data, Imagery detected flood extends, predicated modeled
flood extents, traffic cameras, Urban Search and Rescue (US&R), Crowdsourcing, etc. Weights of confidence points can be applied to data given the source. Each product is a rest service and is automated. Determining Area of Interest (AOI) input via FEMA’s Prioritizing Operations Support Tool (POST). AOI then reinforced based on observations, models, network analysis of known routes.
Gaps. When to turn on and activate SAR Sensors. Temporal and spatial restrictions on imagery, return time gap too large to collect enough imagery our imagery does not cover AOI to produce a road damage proxy product. Lack of transactional transportation datasets. Definition of isolated communities and targeting isolated communities at risk due to road outages. Automated service to run “transportation impact” analysis. Lack of Routing Analysis designed for Natural Disaster Events for a given area.
Goal. Automated process for continually tracking road status that uses variable inputs from remotes sensing, models, machine learning, and crowdsourcing sources.
Innovative Ideas. Remote Sensing techniques beyond imagery sources. For example, Airborne lasers looking for vibrations on a stop-sign to detect traffic, interference of FM/AM modulation to get a proxy measurement for the level of traffic on a road. Automated feedback loop within modeling to further refine.
What is needed?
- Priority Needs: Debris detection workflow and high-resolution imagery
- Identify the location of debris, the quantity of debris, types of debris and its impacts to Critical Infrastructure.
- Use this information to assist with declaration thresholds.
Criteria. Debris detection workflow must offer a repeatable process and be technically accessible for use by State, Local, Tribal, and Territorial (SLTT) agencies.
- Temporal, spatial, low resolution, and cloud cover imagery restrictions for creating a debris detection product.
- Lack of standards and guidelines for developing a debris detection algorithm with imagery.
Goal. Development of standardized processes and workflows that can be easily transitioned for use by SLTT agencies for damage assessment efforts – for the purposes of expediting disaster declarations and public/individual assistance, while reducing overall recovery costs.
Innovative Ideas. Scalable imagery collection and sensor type based on geographic extent and question, i.e., where is it, what type is it, how much is there? UAS for measuring debris piles over time, e.g. debris collection sites.