Illustrating Hurricane Paths
Mar. 7th, 2019 08:11 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
Two different ways of illustrating hurricane paths can be interpreted differently by viewers. Each has its pros and cons.
Neither way gives precise probabilities of storm landing. What you see is either a blob or a bunch of lines showing where it's likely to go. What we need is a tonal map that says something like "90% likely to strike here," "75% likely to strike here," "less than 10% likely to strike here," etc. That's for the center of the storm. It would also be helpful to indicate similar gradients for damage, because the size of those zones will differ for storms based on their size and intensity. (These things are slightly different: you may have a lower probability of a direct hit but a high probability of a sideswipe.) Then use data collected from other storms to set guidelines for what level of response is advisable for towns and individuals. High-risk areas should evacuate at least the vulnerable people. Medium-risk areas should prepare for storm activity and be ready to scramble a higher response if it looks to get worse. Low-risk areas should watch in case it veers suddenly, and do things like make sure their shelters are prepped. If you want to get fancy, ram all that data into a processor that will take your zip code, spit out your hit probability and damage prediction, and give recommendations for different types of people. Remember you need to design the data processing for maximum surge demands, not everyday use.
Neither says anything about the intensity of the storm or its damage quotient. Those are separately measured with metrics that are also in need of upgrade. A direct hit will of course do more harm than a glancing blow from the same storm, but a glancing blow from a bad storm can be much worse than a direct hit from a weak one, and there are multiple factors which can make a storm very bad. These are all things we need to put on hazard forecast maps, separately, and then combine into an estimate of total danger, graduated as described above. Some of our current problems come from people thinking the wind speed is actually a damage quotient, that the cone map means the storm is getting worse, or that being some distance from the center means they can't get hit.
Some things we should do:
* Acknowledge Category 6 because now we are seeing some hurricanes with higher winds. Category 5 "everything bigger than 4" is now a wider range than the other categories, so it needs to be subdivided for accuracy. Be prepared to expand the categories whenever needed to account for worse storms, because they are getting worse. You have seen nothing yet.
* Measure and announce the speed at which the storm is traveling. A slow storm has more time to do more damage while it squats over a target. A fast storm may be bad, but then it's over and recovery can commence.
* Measure and announce the storm surge. Some hurricanes only raise the water a modest amount, while others tow massive amounts of water inland.
* Measure the precipitation coming from the storm and estimate what it is likely to dump on land. Announce this so people can plan ahead for heavy rains. Hurricanes tend to rain plenty, but some also throw hail (rare but possible), and some can drop several feet of water on one spot. Bear in mind that this is the most widespread threat and can overwhelm areas not "technically" in the hazard zone according to current measurements.
* Mark each of these factors on the hazard map: spiral wind speed, forward speed of travel, storm surge, precipitation. Score each of them separately. Then compile an overall danger rating. A hurricane of SWS 4, FST 5, SS 2, P 3 will quite probably do less harm than one of SWS 1, FST 1, SS 4, P5. The first one blasts through and flattens a lot of buildings but is gone in a day or two. It might be D2. The second peels off a few roofs but vomits several feet of water over hundreds of miles for a week. It might be D4. Further note that you should evacuate high locations for SWS 3+ while low ones mayb e sheltered from wind, and low locations for P3+ while high ones may be raised above floods.
* For the love of all good sense, everywhere in hurricane territory should have shelters sufficient for all souls on board, sturdy enough to withstand a direct hit by all the above forces, and stocked with enough food, water, medical supplies, and other necessities for at least two weeks. If you're not willing to do that, you don't get to cry when dead bodies float down the roads. Preferably, all the states bordering hurricane territory should have similar shelter capacity so the coastal states have somewhere to evacuate out of the way of the D4-D5 storms that will wipe everything off the face of the earth. Realistically, every town and state should have this kind of shelter preparation to cope with the barrage of storms, earthquakes, wildfires, and other mayhem we face. Individuals should stock for locally probable emergencies to the best of their ability. Failing to plan is planning to fail.
Neither way gives precise probabilities of storm landing. What you see is either a blob or a bunch of lines showing where it's likely to go. What we need is a tonal map that says something like "90% likely to strike here," "75% likely to strike here," "less than 10% likely to strike here," etc. That's for the center of the storm. It would also be helpful to indicate similar gradients for damage, because the size of those zones will differ for storms based on their size and intensity. (These things are slightly different: you may have a lower probability of a direct hit but a high probability of a sideswipe.) Then use data collected from other storms to set guidelines for what level of response is advisable for towns and individuals. High-risk areas should evacuate at least the vulnerable people. Medium-risk areas should prepare for storm activity and be ready to scramble a higher response if it looks to get worse. Low-risk areas should watch in case it veers suddenly, and do things like make sure their shelters are prepped. If you want to get fancy, ram all that data into a processor that will take your zip code, spit out your hit probability and damage prediction, and give recommendations for different types of people. Remember you need to design the data processing for maximum surge demands, not everyday use.
Neither says anything about the intensity of the storm or its damage quotient. Those are separately measured with metrics that are also in need of upgrade. A direct hit will of course do more harm than a glancing blow from the same storm, but a glancing blow from a bad storm can be much worse than a direct hit from a weak one, and there are multiple factors which can make a storm very bad. These are all things we need to put on hazard forecast maps, separately, and then combine into an estimate of total danger, graduated as described above. Some of our current problems come from people thinking the wind speed is actually a damage quotient, that the cone map means the storm is getting worse, or that being some distance from the center means they can't get hit.
Some things we should do:
* Acknowledge Category 6 because now we are seeing some hurricanes with higher winds. Category 5 "everything bigger than 4" is now a wider range than the other categories, so it needs to be subdivided for accuracy. Be prepared to expand the categories whenever needed to account for worse storms, because they are getting worse. You have seen nothing yet.
* Measure and announce the speed at which the storm is traveling. A slow storm has more time to do more damage while it squats over a target. A fast storm may be bad, but then it's over and recovery can commence.
* Measure and announce the storm surge. Some hurricanes only raise the water a modest amount, while others tow massive amounts of water inland.
* Measure the precipitation coming from the storm and estimate what it is likely to dump on land. Announce this so people can plan ahead for heavy rains. Hurricanes tend to rain plenty, but some also throw hail (rare but possible), and some can drop several feet of water on one spot. Bear in mind that this is the most widespread threat and can overwhelm areas not "technically" in the hazard zone according to current measurements.
* Mark each of these factors on the hazard map: spiral wind speed, forward speed of travel, storm surge, precipitation. Score each of them separately. Then compile an overall danger rating. A hurricane of SWS 4, FST 5, SS 2, P 3 will quite probably do less harm than one of SWS 1, FST 1, SS 4, P5. The first one blasts through and flattens a lot of buildings but is gone in a day or two. It might be D2. The second peels off a few roofs but vomits several feet of water over hundreds of miles for a week. It might be D4. Further note that you should evacuate high locations for SWS 3+ while low ones mayb e sheltered from wind, and low locations for P3+ while high ones may be raised above floods.
* For the love of all good sense, everywhere in hurricane territory should have shelters sufficient for all souls on board, sturdy enough to withstand a direct hit by all the above forces, and stocked with enough food, water, medical supplies, and other necessities for at least two weeks. If you're not willing to do that, you don't get to cry when dead bodies float down the roads. Preferably, all the states bordering hurricane territory should have similar shelter capacity so the coastal states have somewhere to evacuate out of the way of the D4-D5 storms that will wipe everything off the face of the earth. Realistically, every town and state should have this kind of shelter preparation to cope with the barrage of storms, earthquakes, wildfires, and other mayhem we face. Individuals should stock for locally probable emergencies to the best of their ability. Failing to plan is planning to fail.
(no subject)
Date: 2019-03-08 03:22 am (UTC)And either time lapse it or color code to show *when* the storm is likely to hit places.
You could probably do the same thing with precipitation and wind speeds each on their own 3d plot.
I expect that such plots updated in real time would be fascinating as the plotted surface would writhe and shift as different things affected the predictions.
Yes ...
Date: 2019-03-08 09:08 am (UTC)(no subject)
Date: 2019-03-08 10:07 am (UTC)The difference between fiction and reality
Date: 2019-03-08 03:23 am (UTC)Because this entire list seems basic, sensible and "why hasn't this been done before NOW?" obvious...
Yet, I can't imagine trying to write a story with this level of competent, PREPARED response in anything but FICTION.
(no subject)
Date: 2019-03-08 05:24 am (UTC)Thoughts
Date: 2019-03-08 05:43 am (UTC)Rotate batteries too.
(no subject)
Date: 2019-03-08 10:06 am (UTC)Shelters....
Date: 2019-03-08 04:16 pm (UTC)A better system for explaining the strength of storms and probability for different types of damage.
But none of that helps if people don't have a way to keep themselves safe during storms. The size and geography of Houston mean that it is basically impossible to efficiently evacuate. What I think we really need is a robust system of community shelters. Ideally, these would be spread throughout the community so that nobody lived more than a mile from a shelter.
In addition to keeping people safer during the initial event, a system like this would make emergency response easier as each shelter could have a medic on site and have radio contact with first responders so that you could avoid having a situation where overwhelmed 911 operators can't effectively tell the difference between people who really do need emergency help ASAP because someone is having a medical crisis and people who are freaked out because of the storm but are otherwise not in immediate danger.
(no subject)
Date: 2019-03-09 12:31 am (UTC)It's actually quite easy to design that access system in a way that makes it fairly trivial to scale up enough for millions of users. In fact, one of the easiest ways to build the system is also the most easily scalable.
In short*, I'd take the forecast data that already exists (to create the maps the media use), split it up into individual sections per zip code (5 digits is not really a lot in these terms), create a static page for each with the information/recommendations/etc. for that zip code, and upload those pages to a web server. (This can be repeated whenever the forecasts are updated, which IIUC is only once every few hours, so this process doesn't have to be fast enough to run in real-time.)
The web server then doesn't really have much to do - it's just sending the files the user asks for, without having to do any real processing, so it can serve quite a large number of users on its own, using software that is freely available. I suspect bandwidth would be a bigger problem than processing power, but I don't have any numbers to back that up.
When that one web server is not enough anymore, you can either throw more web servers at the problem, which is quite easy since they don't have to talk to each other in any way (unlike dynamic sites such as dreamwidth) - or use an existing CDN, which is basically exactly that, a bunch of web servers with the same files on them, just provided as a service instead of having to do it yourself.
Of course, if you start adding other requirements, like real-time data or user tracking or something, then the solution will probably have to change somewhat, in ways that are probably harder to scale.
*) At first, I explained in rather more detail, but it felt way too long... So I tried again, with similar results... Third time's the charm I guess.