Day 3 HFIP Annual Sessions 5 and 6:
Session 5 Summary & Key Ideas:
SBES Discussion:
Need to address misperceptions, better understanding of uncertainty, connectedness to information, and how to
respond to various levels of impacts. Laura Myers & Chandra Clarkvery useful slides on the value of SBES and how
we should think about it.
Renato: very useful slides on the dollar value of HFIP, which shows huge cost benefit of the program.
Coastal Impacts and Vulnerability Discussion:Craig Setzer Royal Caribbeanmore probabilistic wind speed
probability products! Multi-model and ensemble approach. Model forecasts and data are generally good, need more
products to visualize and convey the information.
Lee County EM: some storm-surge vulnerable residents did not take the risk seriously in Ian because storm surge
was minor in Charley and Irma due to similar intensities but much smaller wind field. 2 ft of water is roughly 1 mi
further inland in SW FL. Top priority is improving RI prediction at longer lead times. Probabilistic storm surge
more than 48h prior, 72h is helpful.
Session 5A: Social and Economic Impacts
- With regard to providing information to insurance companies, how much can actually be mitigated? Can't move a house/car out of the way. What is percentage of damage that's actually avoidable vs immovable structures that will inevitably be damaged
- $5M in losses avoided per storm, accounting for property damage from counties, etc. All damages are probably unavoidable or cannot be attributed to this cost forecast. Cannot assume what people in houses are doing. All we can do is track what govt is doing rather than indiv person. Sources of information can influence what people are doing. Cannot measure it or forecast.
- Please explain ex-ante value? And saw in basis for some of damages said 90% of damages are from wind and precip but thinking storm surge is more than 10%
- One of limitations is observing this is how wrong where, and how to copy that error that's exposed. Track error, track damages, then correlate. What is value of improving or having better info before damage course. If you have forecast of wind intensity over space. Can make forecast so that spread of prediction is more center over a certainty. How valuable is it. Rather than how big your error was after it's past.
- Storm surge damage from Ian was significant
- Storms in sample account for 90% of damages for hurricanes. People said why do you have all the storms? Mis-accounting for damages.
- Not assigning specific forecasts just the fact that there was a hurricane
- Does seem like product not advanced yet. Trouble in models that didn't allow us to track performance with storm surges.
- More information needed on duration of impacts and cessation of threats. One we do internally is time of departure from trop storm winds. Brief for emergency managers like coast guard but reluctant to provide info publicly in case people might think it's an “all clear” and go back home.
- Are we being too paternalistic?
- Research findings not enough information to find that the public would find that info useful. Partners like EMTs are interested in having some cessation or most likely in time so that they can plan on when they will resume services. From information and data and findings would say info can most benefit partners rather than public. One thing to counter conversation is that we didn't' believe for a long time that the public can handle probabilistic information but now we see that we can. If we've had information in the past we didn't think they could handle but they actually can then cessation info might also be beneficial but we don't have direct evidence
- When people were equally not likely to evacuate. Referring to those who do vs not. Some are more vulnerable than others trying to shelter in place. Want to recommend more education and outreach w regard to particular vulnerabilities
- HAFS system online now. Talking about track and intensity. What more products are we need for probabilistic prediction? What other things do you want to link with what we do for forecasting?
- Social scientists and users say they need probabilistic information. Economists say ROI is massive for that information. That's what congress wants to hear. Need to revise modeling systems to provide the inputs to produce that. Drive those products and meet user requirements. Congress wants a societal connection. Have to say “I've been a model that will solve a societal problem and here's the ROI” rather than just “I built a model”. Next few years will be brutal if we don't make that argument
- We're not running a business. Talk about value of info that we are providing. Move NWS to talking about value of things we're doing. ROI means something specific in economics world.