Sandy assessment released

[CrazyC83]

Reading it now. Will have my opinions shortly.

http://www.nws.noaa.gov/os/assessments/pdfs/Sandy13.pdf

[CrazyC83]

1) Clearly, the warning process brought a lot of controversy. I personally would have, at least on paper, issued hurricane warnings. In fact, many believed they were really under one. But the technical proceedings would not have allowed it on short notice for a post-tropical cyclone. It was a catch-22 for sure by the NHC at the time. Perhaps the WFOs could have issued (inland) Hurricane/Tropical Storm Warnings while no coastal warnings issued? It was clear the NHC could not issue coastal warnings given the situation. Changing the software process takes weeks or months, not hours.

2) Confusing indeed. But a Storm Surge Warning would make the problem worse, not better, by adding yet another product. I wish they would read up what happened after Hurricane Juan in 2003 (no tropical products were issued by Canada at the time and it confused everyone and the awareness was minimal ahead of a historic storm).

3) Similar situation as in #2. A lot of media representatives were complaining and said it was confusing.

4) Agreed that social science is key. Look at the 2011 tornadoes as inspiration there. Learn to see what people did right and wrong.

5) Online presence is key! We knew NWS Mount Holly was on the ball all along, and other offices need to follow suit. They were simple and blunt yet polite and kept showing impact potential from many days out. Complicated formats simply tune people out.

6) Again, going back to points 1 and 2 yet has been discussed to death. People even went against the NHC and thought they were in a Hurricane Warning! Clearly, communication was good, but making it official is important too. The media was right in that they said this was a hurricane-level event.

Personal side recommendation related to 6:
* Consider eliminating the High Wind Warning product altogether (think back to Ike in the Midwest, which was not well-prepared at all). Use the terms "tropical storm force" and "hurricane force" to describe completely non-tropical events but make clear the levels of impact would be similar. Such would not be issued if in a Blizzard Warning or a tropical-level warning.

7) Good idea to have a storm.gov website, but perhaps have links to local media (accredited) in the affected areas, since those are often most reliable.

8) Same as with 7. Basically, what exists now is great from a meteorological perspective, but not from a public perspective. What Storm2k members see is not what the Average Joe sees.

9) Again, the media - mainly local TV - should always have a leadership role next to the NWS. The NWS should give them the information and they are the best ones (in most cases) at sending the information to the public.

10) That was a crucial failure that needs to be considered in the future. Perhaps Central or Southern Region could help out in a region-wide failure.

11) Agreed, the sites should be mobile-friendly as well. Ahead of Sandy, I was using my phone a lot to get information at many moments.

12) Indeed, social media was a game changer in Sandy. Look at how in the dark we were during Katrina and other storms for many hours before the devastation became clear. During Sandy, we knew everything right away. Expanding it could continue to change forecasting and reporting.

13) Agreed, storm surge risks need to be communicated at least 48 hours ahead of impact - ideally 72 to 108 hours ahead of impact, before evacuations begin. Perhaps an SPC-level risk at days 5, 4 and 3 (slight, moderate, high) before numbers.

14) Agreed, storm surge levels need to be communicated for different scenarios. For example, inundation in NYC at 6 (minor event), 9 (just below Irene), 12 (below Sandy), 15 (just above Sandy), 18 (1938-like event directly in NYC) and 21 foot (1635-like event directly in NYC) storm tide.

15) Extra staffing sure helps, but do they have funding for it?

16) Agreed, anywhere that a tidal influence exists, they need a storm surge model. Consider combining it with downstream rainfall for a worst-case scenario model.

17) Agreed, storm surge levels are confusing since they vary or people don't know what they mean. They should stick to simple uses (i.e. Zones A, B and C) based on storm tide (surge + high tide) worst case scenarios.

18) Coastal training is important for those who run coastal WFOs. They need to know what the threats are.

19) I'm against the flow here. Adding a Storm Surge Warning product I think would bring additional confusion. Plus what is the line? Other graphics and mapping tools need to be clearly identified though.
19a) I agree that HLS's should have been issued, notwithstanding the status. In addition, a similar product should be created for major storm events as a whole (i.e. the February blizzard, major tornado outbreaks) as I suggested in the Irene rebuttal. All offices should have issued HLS's as well.

20 and 21) Don't blame the NWS there. Blame Congress. They hold the purse strings.

22) Again, goes back to Irene on verification. They couldn't update the recommendations in time for Sandy, hopefully they will for 2013.

23) Simply put, the NWS admitted the assessment was incomplete. They need outside groups for assessments - a mixture of media, academia and business communities.

[Riptide]

Fascinating and revealing read, thanks for sharing.

[wxman57]

The big issue with issuing a detailed storm surge graphic 72hrs or more ahead of landfall is that surge height is highly dependent on the exact point of landfall. Consider that typical track error at 72 hours is over 100 miles left or right of the track. At best, any surge graphic issued so far out would have to assume that the landfall could occur over a 200 mile wide area, sort of like the SLOSH basin graphics. These graphics would depict the worst likely conditions IF the storm should make landfall near or to the west of any given point. Of course, most of the areas on such a graphic would experience a surge quite a bit lower than the graphic would indicate (or possibly none at all). The farther out before landfall such a graphic is issued, the greater the false alarm rate.

Knowing that many communities require 72 hrs or more to evacuate, they are just going to have to accept the fact that the false alarm rate will be high, and that in many cases there would be no significant inundation.

[CrazyC83]

The big issue with issuing a detailed storm surge graphic 72hrs or more ahead of landfall is that surge height is highly dependent on the exact point of landfall. Consider that typical track error at 72 hours is over 100 miles left or right of the track. At best, any surge graphic issued so far out would have to assume that the landfall could occur over a 200 mile wide area, sort of like the SLOSH basin graphics. These graphics would depict the worst likely conditions IF the storm should make landfall near or to the west of any given point. Of course, most of the areas on such a graphic would experience a surge quite a bit lower than the graphic would indicate (or possibly none at all). The farther out before landfall such a graphic is issued, the greater the false alarm rate.

Knowing that many communities require 72 hrs or more to evacuate, they are just going to have to accept the fact that the false alarm rate will be high, and that in many cases there would be no significant inundation.

They think they can get it reliably issued at 48 hours. Once beyond there, I think an SPC-like risk level (slight, moderate, high) accounting for both the probability and the degree of impact could be introduced at days 3, 4 and 5. Slight risk could be either a low probability event (i.e. less than 10%) or a high probability of a marginal event, moderate risk could increase both factors and a high risk could be a good chance of a high impact event. Once within 48 hours, they could get down to hard numbers.

The biggest disagreement I have is the Storm Surge Warning. Adding more products I think would lead to even greater confusion and lost messages. That is already covered well under Hurricane/Tropical Storm products. They could make it that a Hurricane Warning could be issued for coastal locations in major flood scenarios even if winds do not support such.

[Derek Ortt]

The big issue with issuing a detailed storm surge graphic 72hrs or more ahead of landfall is that surge height is highly dependent on the exact point of landfall. Consider that typical track error at 72 hours is over 100 miles left or right of the track. At best, any surge graphic issued so far out would have to assume that the landfall could occur over a 200 mile wide area, sort of like the SLOSH basin graphics. These graphics would depict the worst likely conditions IF the storm should make landfall near or to the west of any given point. Of course, most of the areas on such a graphic would experience a surge quite a bit lower than the graphic would indicate (or possibly none at all). The farther out before landfall such a graphic is issued, the greater the false alarm rate.

Knowing that many communities require 72 hrs or more to evacuate, they are just going to have to accept the fact that the false alarm rate will be high, and that in many cases there would be no significant inundation.

They think they can get it reliably issued at 48 hours. Once beyond there, I think an SPC-like risk level (slight, moderate, high) accounting for both the probability and the degree of impact could be introduced at days 3, 4 and 5. Slight risk could be either a low probability event (i.e. less than 10%) or a high probability of a marginal event, moderate risk could increase both factors and a high risk could be a good chance of a high impact event. Once within 48 hours, they could get down to hard numbers.

The biggest disagreement I have is the Storm Surge Warning. Adding more products I think would lead to even greater confusion and lost messages. That is already covered well under Hurricane/Tropical Storm products. They could make it that a Hurricane Warning could be issued for coastal locations in major flood scenarios even if winds do not support such.

That's how Canada does it

[Kingarabian]

Nice to see you again Derek!

[KBBOCA]

bump...

I assume I'm like many others here who kind of disappear in late November / early December until late May.

It would be good to keep this on the front page for awhile for further discussion now that more of us regulars are tuning in as June 1 approaches.

I'll be reading the assessment in just a bit. Hope others will chime in! This could be an interesting discussion...

[KBBOCA]

The assessment made for good reading. I'm impressed they got it out as quickly as they did, and think they hit the key points. Given the lack of funding, they were very fortunate that two outside firms conducted surveys of the general public about their perceptions of risk, etc. and that they generously shared the data. Without that data about how residents in NJ/NY viewed the storm and the warning system, the report would have been much weaker.

I was VERY VERY glad to see they've changed the plan re: how to handle warnings in the event of a storm transitioning to post tropical. The new decision flow chart allowing them to maintain tropical warnings for a post tropical storm if there remains a significant threat to life and property is very wise. KUDOS.

Strongly agree with the recommendations about the need to present storm surge data clearly. It's good to see that more resources are likely to be devoted to forecasting and communicating storm surge risk clearly. Obviously it is a major challenge given all the complex factors. But certainly this work is needed as it is surge / and battering tidal waves (along with inland flooding) that presents the greatest danger to life from tropical systems. The wind gets most of the focus... but it's the water that kills!

Was fascinated to read some of the details re: social media use and hits, and so glad to see the recognition that was given to EXCELLENT work by the Mt Holly office for their amazingly clear and well-publicized briefings.

Did anyone else catch that line about the NHC calling Emergency managers in NYC to correct them on storm surge information - which "opened their eyes" to the danger?!?! I don't think that was publicized before? Certainly all of us here were screaming when Mayor Bloomberg and the NYC EmOps folks failed to call for mandatory evacuations on Sat night... really glad to see NHC called and made sure they got the message. Here's that section from pp. 39-40

NYCOEM was confused about the storm surge forecast, what and where the impacts would be, and how high the water would rise . They specifically recollected a call from NHC on Saturday night with an updated surge forecast that was "an eye opener," noting it was an odd situation to be called directly by NHC, but was effective. This apparent communication breakdown
may have delayed critical decision making about evacuations
and led to potentially confusing messaging to the public.

All in all, a well done and useful assessment that seems to have covered many of the major issues and important topics in a very fair way. Nice work by the assessment team!

[KBBOCA]

Doing some browsing online to see if any of the weather blogs have commented on the Sandy Assessment.

Excellent discussion by Bryan Norcross at his blog on Wunderground, here:

http://www.wunderground.com/blog/bnorcr ... ntrynum=31

He makes some really good points! Go read it...

[KBBOCA]

In an earlier blog entry about Sandy and the assessment process, Bryan Norcross makes a really interesting suggestion about ways of communicating potential IMPACT and identifying key risk areas...

http://www.wunderground.com/blog/bnorcr ... ntrynum=30

Translated into Sandy terms, these recommendations mean changes in the way the threat from high-impact events is characterized and communicated. We need a new communications strategy that includes special terms, language, formats, and/or bells and whistles for unusual and extreme events.

The special language would include statements that explicitly convey what will happen if the forecast is correct. This technique was famously used to good effect before Hurricanes Katrina and Ike when the local NWS offices issued provocative statements promising “certain death” to motivate action from residents and public officials. But the tornado people have done more, and done it better. The Sandy assessment team should take note.

If the tornado system for extreme events were directly imported into the hurricane/tropical-storm communications process we would get:

• 72 to 96 hours before peak impact a Risk Area is defined
• 48 hours out a Watch area is defined
• 36 hours out a Warning Area is defined
• 12-24 hours before peak impact a Hurricane Emergency area is defined

In parallel, a Storm Surge Risk Area, Watch, Warning, and Emergency would be issued, as appropriate to the individual storm.

What do you all think?

[CrazyC83]

In an earlier blog entry about Sandy and the assessment process, Bryan Norcross makes a really interesting suggestion about ways of communicating potential IMPACT and identifying key risk areas...

http://www.wunderground.com/blog/bnorcr ... ntrynum=30

Translated into Sandy terms, these recommendations mean changes in the way the threat from high-impact events is characterized and communicated. We need a new communications strategy that includes special terms, language, formats, and/or bells and whistles for unusual and extreme events.

The special language would include statements that explicitly convey what will happen if the forecast is correct. This technique was famously used to good effect before Hurricanes Katrina and Ike when the local NWS offices issued provocative statements promising “certain death” to motivate action from residents and public officials. But the tornado people have done more, and done it better. The Sandy assessment team should take note.

If the tornado system for extreme events were directly imported into the hurricane/tropical-storm communications process we would get:

• 72 to 96 hours before peak impact a Risk Area is defined
• 48 hours out a Watch area is defined
• 36 hours out a Warning Area is defined
• 12-24 hours before peak impact a Hurricane Emergency area is defined

In parallel, a Storm Surge Risk Area, Watch, Warning, and Emergency would be issued, as appropriate to the individual storm.

What do you all think?

I like the thought of risk areas defined (SLGT, MDT, HIGH) as far as 120 hours (5 days) from impact. Perhaps those should be used throughout as well?

For example with Sandy:

October 24 - Introduce a SLGT risk of significant storm surge (warning criteria).

October 25 - Upgrade to MDT risk for the coast from, say, Delmarva to Rhode Island, and hatch it (potential for 10 feet or higher above mean sea level)

October 26 - Issue Hurricane Watch. Maintain MDT risk but for a larger area.

October 27 - Issue Hurricane Warning on surge (in addition to winds) with HIGH risk for the Jersey Shore and NYC area.

[KBBOCA]

Ah, just came across one other really interesting post-Sandy assessment - by none other that the "hero" from NWS Mount Holly, Director GarySzatkowski who issued the famous "personal plea" that was so effective.

He apparently put together a post-Sandy assessment and briefing. You can see his 52 slide powerpoint here:

http://focusonfloods.org/wp-content/upl ... kowski.pdf

[brunota2003]

I like the idea of the Slight/Mod/High risks in advance, but have it for overall threats (not specifically surge). Yes, surge needs more attention, but I also don't want to see attention taken away from other threats. I would hate to see surge deaths drop, only to watch inland flooding or wind deaths jump due to lack of focus.

I know military bases use threat codes for weather, why can't something like that be given for the public? For example, Fort Drum used 4 codes to describe road conditions and how to proceed with the day (they get really bad lake effect snows). Green meant roads are clear and Black meant emergency vehicles authorized for travel only (with other colors between).

Something similar could be in place leading up to the event:

The current "Hurricane Awareness Level" is _________.

Green:
All clear, no threats expected in the next 120 hours.

Blue:
Possible impacts from a tropical cyclone within the next 120 hours, residents should monitor the latest information and start checking their supplies. Cities should prepare for potential emergency operations. (For this, an area could be highlighted, and this does NOT match the cone! This is for all potential impacts)

Amber:
Impacts from a tropical cyclone are possible starting within the next 60 hours, residents should continue to monitor the latest information and prepare for adverse weather conditions. Cities should activate their emergency operations groups, if they have not done so already. Evacuation considerations should be made in accordance to local operating procedures.

This is where things diverge. Different storms have different impacts. A weak tropical storm may have little to no noticeable impact, whereas a strong hurricane may be utterly devastating. These next levels can be tailored to specific regions, where different threats pose more/less risk, the nature of the storm, etc.

Orange:
Minor to moderate impacts from a tropical cyclone are likely, starting within the next 24 to 36 hours. Residents should continue to monitor the latest information and rush to completion any preparations. Potential impacts include sporadic power outages (generally only a day or two for outages to be restored), minor to moderate wind or surge damage, and potential for minor to moderate inland flooding. Residents in these locations should be prepared for a possibility of an upgrade to Red if the storm changes course or if conditions appear they will be worse than currently forecast. Winds will generally be in the 35 to 75 mph range, with surge in the 1 to 7 ft range (altered for regional vulnerabilities).

Red:
Severe impacts from a tropical cyclone are likely, starting within the next 24 to 36 hours. Residents should continue to monitor the latest information and rush to completion any preparations. Impacts include, but are not limited to, long duration power outages (1 to 2 weeks or more), significant wind or surge damage, including complete destruction of some structures, and potential for moderate to major inland flooding. Evacuations should be underway and nearing completion, as severe impacts are expected in less than 24 hours. Residents within this area may be on their own for survival, and survival may be difficult to near impossible within some areas. Winds will generally be in the 75 to 115 mph range, with storm surge in the 7 to 15 ft range.

Black:
Extremely dangerous and life threatening weather conditions are imminent (within 12 to 24 hours) or ongoing, and potentially include high winds, large storm surge, etc. Impacts include, but are not limited to, long term power outages (greater than 2 weeks), complete destruction of houses or buildings are possible in many areas due to winds or surge, and potential for major to record inland flooding. Residents within this area will be on their own for survival until after the storm passes, and survival will be difficult to near impossible within most areas. Winds will generally be in the 115 to 170 mph range, with storm surge in the 15 to 25+ ft range.

I think this would serve the public better, in graphic form, in conjuncture with watches/warnings. Put the point out there bluntly that if you are within "x" area, you most likely will NOT be rescued, and survival may be extremely difficult to near impossible. Watches and warnings need to be modified as well, to better fit with the various threats tropical cyclones normally have, and to make them simpler to understand for the public.

As far as my levels go, not all storms will get to Red or Black, a quick moving, weak TS may only prompt an Orange. The exact details can be tailored to the needs, with Red/Black being more like very enhanced wording/PDS type scenarios. I would like to see these areas shaded away from the coastline, as well. This is not just specifically for coastal areas.

Example:
During Hurricane Isabel, coastal NC was not several impacted (to the left of the eye). There was some minor wind and surge damage, as well as power outages lasting a day or so. This would fit into Level Orange. However, in Central Virginia (which does not typically get tropical systems), there were many trees down with a lot of areas without power for up to 2 weeks. This would fit into Level Red.

This level of alerting could be dynamic during tropical storm and hurricane threats, giving the NHC and local EOCs a way to generally convey potential and ongoing impacts to the local public in a generally easy to understand format. By using something like this, it would also be a way to convey threats well inland on graphics that currently only show coastal watches/warnings, and neglect the inland portions.