Some observations on some recent quick turns around the Disaster Tango dance floor:
In discussing how we should think about the most extreme severe events, I suggested that we focus on a collective look at recurrence times. This suggestion, though it has a certain logical sense to it, has some pretty serious practical weaknesses, and thinking about them will prove to be instructive in itself.
The case at hand (extraordinary rains in South Carolina) has already developed an interesting and revealing twist which can shed some light on the difficulties.
We’ve been talking about how NOAA data indicate that rainfalls during the Joaquin event were at 1000-year repeat times at numerous places in the state. You can play around with this widget .
Please note that the distribution from 10 year to 1000 year events is fairly steep.
For instance, at the station I am looking at, (CHARLESTON WSO AP, just north of Charleston) a two-fold increase in precipitation at most time scales is enough to change an event from a 50-year event to a 1000-year event.
In what is for me an unprecedented turn, I found an interesting link via Marc Morano (a.k.a. “Climate Depot”) via a retweet.
It seems that though according to the National Weather Service many places in South Carolina attained to a “1000-year rainfall”, according to the US Geological Survey,
“USGS provisional data and preliminary analysis show NO indication that a 1000-year flood discharge occurred at any USGS streamgages”.
While that was the lead-off, the article went on to be a bit more equivocal.
“However, based on that analysis, it does appear that the USGS streamgage on the Black River at Kingstree, SC and the one on the Smith Branch at Columbia, SC both measured peak floods in the neighborhood of a 500-year flood. Currently, there appear to be a few more streamgages experiencing a 25-year to 50-year flood, but the majority of USGS streamgages had flood peaks that were less than 10-year floods.”
“How can we have a 1000-year rain that does not result in a 1000-year flood?
It comes down to a number of factors, including the pattern of movement of the rain storm in each particular watershed, the conditions of the soil and plant matter on the ground in the watershed, and the timing of rain storm in one watershed versus other watersheds, among other things. An example would be that ground that is saturated before 1 inch of rain fell would result in more water going into the stream that if the ground was dry and could soak up more of the rain. Also, less water will runoff into streams from 1 inch of rain falling in the summer with the trees full of leaves versus the winter when there are no leaves to intercept the rain. This is all the science of hydrology, which is the study of the movement and distribution of water on the earth. Of course, in South Carolina, many of the watersheds have streams that are regulated by dams.”
Marc Morano, a dedicated ClimateBall player, knows his moves, and skips the bits about 500 year vs 1000 year floods, and the bits about the differences between flood repeat intervals and rainfall repeat intervals, in writing about this, cherry picking the most denial-friendly points, as is entirely to be expected.
Feds declare no climate link to floods – SC’s ‘1000 year flood’ only a 10 year flood!
U.S. Geological Survey: ‘No linkage between flooding & increase in GHGs’
Dr. Robert Holmes, USGS National Flood Hazard Coordinator: ‘The data shows no systematic increases in flooding through time’ – ‘USGS research has shown no linkage between flooding (either increases or decreases) and the increase in greenhouse gases. Essentially, from USGS long-term streamgage data for sites across the country with no regulation or other changes to the watershed that could influence the streamflow, the data shows no systematic increases in flooding through time.’
I can imagine some reasons for this null result other than that there is no signal to be found, but let’s be fair and acknowledge this is the official position of USGS at this time.
On the other hand, there is a recent trend to more severe rainfall events in the US, as elsewhere. Whether that is attributable to greenhouse gases is complicated. It certainly doesn’t regress directly, and I could easily contrive to get a null result, but the recent uptick could very well be a nonlinear response to climate forcing.
And then there’s the very peculiar meteorological history of this event (and its coincidence with various other large climate anomalies elsewhere, notably flooding in France, record wildfires in the western US (*), and an astonishing early-spring heat wave in Australia.) Each event is disconcerting in itself, but at what point are we on solid ground claiming there’s a pattern?
Anyway, a number of issues are raised by the USGS vs NWS data, among them
- Should we treat meteorologically bizarre events separately in some way? (this has come up in discussions already)
- Should we focus on flooding recurrence or on rainfall occurrence, or both? Flooding has the advantage that it’s specific; river flows are river flows, while rainfall can be record-valued at anything from 5 minutes to 60 days on the NWS website, and of course, other time scales in principle as well? If we focus on flooding, though, we lose sight of the high intensity events in favor of high-duration ones. And one of the questions we might want to ask is whether high-intensity events are increasing.
- How much do we really know about the tails of the distributions anyway? We only have a few decades of records so the very long repeat intervals in which we are most interested are not very well constrained. Are these estimates even consistent between flood data and rainfall data?
Now I don’t expect Morano to take such questions up. He is an advocate, not a scientist. His role is to cherry pick evidence and reassure deniers that they ought to be denying. So he collects the stuff he thinks his readers want to see and feeds it to them.
However, it’s more than a little discouraging to see Prof Roger Pielke Jr., who is supposed to be an academic and indeed famously advocates for science to be “an honest broker”, to follow Morano’s lead and dance the naysayers’ part. He tweets:
It wasn’t a “1000-year flood” in South Carolina.
More climate porn & shoddy scientific claims on extremes.
It was certainly a 1000-year rainfall (on various time scales up to 3 days) in a great area of South Carolina, even if it wasn’t a 1000-year flood. That’s not “climate porn”, it’s data.
I often disagree with Roger, but I sometimes find that he has good observations as well. This isn’t one of those times, to say the least.
Despite how Roger writes about this material, there is no law of Nature that guarantees that there will never be trends in severe events associated with climate change. To the contrary, there are many reasons to expect it.
If we want to improve the realism of discourse, to put good information ahead of policy conversation rather than as an weapon in a debate, we don’t get to pick evidence from one side of the dance to the exclusion of the other side’s legitimate evidence. I don’t expect better from Morano, but from Pielke this is extremely disappointing. It certainly supports those who have been claiming that Pielke has been an advocate for no-policy all along.
Interestingly, Roger retweeted Mike Shellenberger’s tweet of this infographic, with whose message I heartily agree (**)
Roger could do with a bit more practicing what he preaches.
* This is disputed; there’s a peculiar graphic on yet another federal website (the Forest Service) that indicates we are far from record territory. I have to say that it looks wrong and the brief gloss in the text sounds wrong. As far as I know the provenance of this graph is unknown.
** (even if this is the sort of low-information-density infographic I could do without)
(not sure how long I can keep up this vaguely-related pop song lyric blog post title thing…)