Row Away from the Rocks

Events like this will happen again - we have to choose to prepare

As part of a special session of the Texas Legislature, a legislative committee was convened last Wednesday to hold a hearing examining the catastrophic Texas Hill Country flooding of July 4th that took at least 136 lives. After seeing some comments by those who testified at the hearing and in various articles about the tragedy, I find myself thinking about the saying, “call on God, but row away from the rocks.” It seems to me like in our response to this disaster that we are seeing a lot of calling on God - both figuratively and literally - but not much about how we could have and should in the future “row away from the rocks.”

This CNN article appears to provide a good overview of the testimony that occurred at the hearing. I was particularly disquieted by the comments made by Republican state representative Wes Virdell, who is the state representative for Kerr County, the county that was of course hardest hit with over 100 fatalities when the Guadalupe River experienced a massive flash flood early in the morning hours of July 4. A summary of his comments were:

Government’s natural inclination is to overreact, and so I want to make sure that we don’t have that happening…You talk to the locals, and they don’t feel like there was anything more that could have been done at that point…That’s not saying every local, but (it) seems to be the general consensus...TDEM and all the agencies that came in — Texas Parks and Wildlife, DPS, National Guard — all of those guys worked very well together. It was impressive to watch, and the local officials with them. Outside of Kerr you see several people trying to point fingers and what not. You go talk to the people that live there, and they’re not blaming people.

I have said repeatedly in prior posts about this tragedy that examining this event should not be done to generate “blame,” but rather to learn everything we possibly can that will help reduce the impacts when the next flash flood like this happens, either here or in other parts of the country. The comments by Rep. Virdell to me border on declarations by other officials, including White House press secretary, that this was an “act of God” with an implication that it was unpreventable.

To me, that is a defeatist perspective; I encourage you to read this interview with sociologist Eric Klinenberg about the danger of this sort of attitude. With the technology and communication that we have today, there is absolutely no doubt in my mind that there is much more that could have been done to reduce the loss of life from this catastrophe, both at the strategic level of planning and mitigation prior to the event, and at the tactical level while the rain was falling and the flooding was beginning. I am going to take you through my thoughts on some of the key conclusions that I have reached about potential improvements after more than three weeks to examine the event.

First, though, I want to stress again that my goal here is to try to demonstrate how as a professional with meteorological and emergency management experience I think that we can and should be thinking about events like this, not to assign blame. I also want to stress that these are my personal opinions and thoughts - I am sure there are a lot of various perspectives here, and I would be glad to hear other thoughts via comments on this post (it is open to comments from all).

I want to start off by looking again at the rainfall that caused the flash flooding on the Guadalupe River, and the outlooks and forecasts that preceded it. The MRMS multi-sensor rainfall product shows that 6-9” of rain fell in 3 hours ending at 4 am CT July 4th pretty much right atop the South Fork of the Guadalupe River. Again, it is important to see how relatively isolated this maximum in rainfall was (areas of even heavier rain fell well to the north in central Texas around San Angelo), and understand how sensitive a flash flood event like this is to the rainfall occurring in that exact location. As a retired NWS hydrologist pointed out to me, if you shift this rainfall maximum just 10 or so miles to the southeast, it would have occurred in the Medina River basin, and the Guadalupe River would not have even seen any significant rise in river levels.

As I discussed in this earlier post about the event, we simply do not have the skill right now to be able to anticipate the location of an intense rainfall event like this at the precision to know which flash flood level stream basin will be impacted until the storms are beginning to develop (modeling systems like Warn-on-Forecast could provide that sort of skill with some hours of lead time in the future). In retrospect, I do think that discussions being issued by the Weather Prediction Center the evening before the event should have resulted in more aggressive messaging from local NWS offices in the region about the flash flood event. While that did not happen within the public flash flood watch product from NWS Austin/San Antonio, it is quite possible that it was happening in direct briefings to emergency managers and other local entities through Slack and other methods the NWS uses to keep local officials informed but that are not public facing.

One thing that I want to mention here as a possible “strategic” improvement would be to look at how flash flood events like this are messaged by the NWS at the outlook and watch time period. Flooding is a particularly challenging hazard for the weather community to message because unlike tornadoes which by their nature are localized events, flooding can occur on very large scales like multi-day mainstem river flooding that are more predictable to true rapid onset flash floods driven by very small scale convective rain events that are very challenging to forecast with any sort of lead time. I have long wondered if the weather community should look at a paradigm where true flash floods are messaged more like Storm Prediction Center messaging for severe thunderstorms and tornadoes. In other words, we would use the Weather Prediction Center outlooks leading up to an event to communicate the potential to the public with slight/moderate/high risk areas, but only issue a flash flood watch to highlight the more imminent threat of thunderstorms capable of flash flood producing rainfall over the next 6-9 hours. This sort of an approach could result in the public having more awareness of when and where they truly need to be on guard for rapid onset flash floods driven by intense rainfall from thunderstorms. I stress this is just an idea based on my own experience and a few conversations with colleagues over the years - it would need a lot of research from social and physical scientists to determine if it is a worthwhile approach.

Because of the challenge of forecasting these sorts of events, it is not really unusual in our current flood messaging paradigm for significant flash floods to occur without any sort of a watch before them. That is why the warning is so important, as it provides the last “safety net” to hopefully spur protective actions. One of the points that was made at the hearing and by others since the event is that people may not have reacted to the flash flood warning issued by the NWS at 1:14 am because they have “become desensitized” to flash flood warnings or other alerts on their phones.

This is certainly quite possible, but with regard to flash flood warnings specifically, I used the outstanding Iowa Environmental Mesonet website to look at how frequent flash flood warnings have been for Kerr County. Over the last 5 years dating back to 2020 and prior to the event on July 4th, there were only 8 flash flood warnings issued for Kerr County: 1 in 2020, 1 in 2021, none in 2022, 1 in 2023, 2 in 2024, and 4 so far this year. Considering we are talking about an area known as “flash flood alley” it is hard to see how this could be considered overwarning that would cause alert fatigue. It is certainly possible that the use of cell phone alerts for multiple hazards and purposes could cause people to “miss the one that matters” as the headline for this article states, and I am hoping that social scientists will be funded to more deeply study and understand this potential issue. An option for people who want notifications of weather warnings without relying on cell phone is NOAA Weather Radio. Weather radio receivers can be purchased relatively inexpensively, and programmed so that you only receive weather warnings for the county or parish you live in.

For those people living near and operating businesses such as campgrounds and resorts along the Guadalupe River, the knowledge of the serious danger that flash flooding presents should result in extra awareness of flash flood forecasts and especially warnings. Serious flash floods have occurred along the Guadalupe River multiple times in the last 50 years, including the July 1987 flash flood that swept away a camp bus resulting in the deaths of 10 teenagers in the same area as the July 4th event. A fact sheet put together by the Guadalupe-Blanco River Authority and FEMA after a major flood in 1997 noted that the river experienced major floods in 1936, 1952, 1972, 1973, 1978, 1987, 1991 and 1997, and that “if you live in the Guadalupe River Basin, you also live in one of the three most dangerous regions in the U.S.A. for flash floods!”

Given this history, it is particularly crucial that people in positions of authority and responsibility as local officials and business owners take flash flood warnings seriously and have well established plans for what to do when a flash flood warning is issued and/or flooding becomes imminent. To be frank, what has disturbed me the most in the aftermath of this tragedy is the apparent lack of serious planning by many of these responsible people to be able to promptly receive flash flood warnings and then to have a well-established plan of action in the event of a flash flood.

Here, I must start with the Kerr County officials with the primary governmental responsibility for public safety in an event like this. As was reported in the CNN article, both the Kerr County sheriff and the Kerr County emergency manager were apparently asleep during the most critical time when the heavy rainfall was occurring and flood waters were rising. “The county’s emergency operations center wasn’t up and running at around 1 a.m. to 3 a.m. when the most intense flooding started, Leitha (the Kerr County sheriff) said. ‘We’re in a rural area; we don’t have a 24-hour service, or anything open like that.’”

While I certainly do not expect that a rural county is going to have a 24/7 emergency operations center, I do expect it to have a way to receive and act upon critical weather warnings - especially when at any given time you have literally hundreds of people (including many children away from home) in an area that is you know is highly vulnerable to deadly flash floods. The fact that both the sheriff and the emergency manager slept through the key period of this disaster - as well as other communication issues in the county as I discussed here - certainly suggests that Kerr County did not even have this minimum amount of emergency preparedness.

In this timeline of the flood event constructed by USA Today, you can see that the NWS was issuing increasingly urgent statements between the time the initial flash flood warning was issued at 1:14 am and when the warning was upgraded to the catastrophic flash flood emergency at 4:04 am. In the NWS Directive (NWS policy document) that provides NWS meteorologists with their instructions on how and when to issue a flash flood emergency, the first bullet of criteria states:

(When) emergency manager(s) of the affected county(ies) or the state emergency management association declare a state of emergency and have confirmed that rapidly rising floodwaters are placing or will place people in life-threatening situations. The state of emergency for the affected areas may have been previously relayed by the emergency manager(s) or the state emergency management association through the WFO in a Non-Weather Emergency Message. These might include a Civil Emergency Message (CEM), an Evacuate Immediate (EVI), or a Local Area Emergency (LAE).

While I do not know if this is the case at the Austin/San Antonio office, my understanding is that at many NWS offices there is at least an informal policy to not issue a flash flood emergency without coordination with the county and/or city emergency managers to get ground truth reports and ensure that the emergency wording is collaborated. Obviously, that was not possible - nor were any of the other potential actions discussed in the directives bullet above - in this event if the emergency manager and sheriff were asleep and nobody was clearly designated to act in their place. Of course, the flash flood emergency was issued eventually, apparently without that collaboration between the NWS and emergency management.

The need for improved local emergency management procedures was acknowledged in Wednesday’s hearing, as shown in this summary of comments from Texas Department of Emergency Management head Nimm Kidd in Inside Climate News:

Committee members sought to understand why, as TDEM deployed resources, so many people were caught unprepared for the wall of water coming down the Guadalupe River. Kidd explained that under Texas law, disasters are locally managed and that the state’s role is to support local efforts. The mayor or county judge typically serves as the emergency management director. That individual can appoint an emergency management coordinator. Kidd was careful not to specifically implicate any officials, but spoke of a general problem.

Kidd said that TDEM has the contact information of emergency managers at the city and county level across the state. But he said that doesn’t ensure communication is always smooth. “That doesn’t mean that we have any way of guaranteeing that the responsible adult and local officials are awake and seeing the same information that our people who are paid to be awake at 2 o’clock in the morning are seeing,” he said. “There is no system in place to ensure that County Judge X or Mayor Y is getting the same information that we are getting from the National Weather Service,” Kidd said in response to questions from Sen. José Menéndez (D-San Antonio). “You do see the problem with that?” Menéndez asked. “I do,” Kidd responded.

As I have in prior posts about FEMA and emergency management, Kidd is making the point that disaster management is already delegated to the local level of government, despite the current insistence of the Trump administration to supposedly move that responsibility from the federal level to the state and local level. Making FEMA smaller and less responsive will only ensure that when the local and state levels are overwhelmed and federal assistance is needed, it will not be there or will be too slow in responding - criticism that has been levied of FEMA in their response to this disaster.

Kidd suggested to lawmakers to consider setting minimum qualifications for emergency managers at the city and county level, something that does not exist in Texas except for municipalities with populations over 500,000. This could be a helpful step that should be looked at more broadly and at all levels, as without any baseline standards for emergency management programs at the local and state levels we continue to create a situation where people who live in or are simply visiting a county with less resources and a sub-standard emergency management program are in greater danger if disaster strikes than those lucky enough to be in a location with good emergency and disaster management.

There has obviously also been a lot of discussion about warning systems and how effective they might have been. To look at this aspect, one must start by looking at the flood wave and how quickly it progressed. The above map was compiled by the Houston Chronicle and shows the locations where flash flood fatalities occurred on the morning of July 4th based on 911 service logs. The floodwave was moving down the South Fork of the Guadalupe River from southwest to northeast, so it first hit the area around Camp Mystic - where a number of young girls and their counselors perished - and then moved downstream to Hunt, at the confluence of the south and north forks of the river.

We know exactly when and how the floodwave hit Hunt, because there is a US Geological Survey (USGS) streamflow gage there, operated in collaboration with the Texas Water Board, the Upper Guadalupe River Authority, and US Army Corps of Engineers. The river level started rising about 1:50 am CT, but the initial rate of rise was relatively slow, going from 7.8’ at 1:50 am to 9.1’ at 2:50 am. After this point, the main floodwave started to hit Hunt, and the water level rose insanely fast. In the next hour, it would rise 10 feet, reaching 19.4’ by 3:50 am. Over the next 80 minutes, it would rise another 16 feet, reaching a record level of 37.52’ at 5:10 am, at which point the gage was overwhelmed.

Unfortunately, we do not have any similar specific information upstream of Hunt, because there are no gages in this area. The USA Today timeline notes 3 am as the time that water began to overwhelm the River Inn Resort, which is just upstream from Camp Mystic and likely where the first fatality dot is located. Hence, it appears that there was just under 2 hours from when the initial NWS flash flood warning for life threatening flash flooding was issued until the flood wave began overwhelming locations where people were staying or camping.

While some have stated that there was no reason for people along the river to have been prepared for a flood of this magnitude, that perspective seems dubious to me. Again, there is a long history of major flash floods causing significant damage along this stretch of river; the San Antonio Express in this article talks about news stories from a flash flood in 1932 that destroyed cabins at Camp Mystic. This article from NPR about FEMA removing some of the structures at Camp Mystic from the 100-year FEMA floodplain makes clear the threat at Camp Mystic and other nearby areas:

Regardless of FEMA's determinations, the risk was obvious. At least 12 structures at Camp Mystic Guadalupe were fully within FEMA's 100-year flood plain, and a few more were partially in that zone, according to an AP analysis of data provided by First Street, a data science company that specializes in modeling climate risk. Jeremy Porter, the head of climate implications at First Street, said FEMA's flood insurance map underestimates flood risks. That's because it fails to take into account the effects of heavy precipitation on smaller waterways such as streams and creeks. First Street's model puts nearly all of Camp Mystic Guadalupe at risk during a 100-year flood

Given the clear risk these properties were under, emergency management procedures would suggest to me that both public and private sector entities responsible for human safety in this area should have had redundant methods to receive warning information and clear action plans to execute in case of flash flooding. It is important to recognize that people did not need to be completely evacuated out of the area to survive. They simply needed to get to high enough ground to avoid being swept away by the rising river; dozens of people were rescued from trees and other high areas after the flood wave passed.

Additional river gages on the river above Hunt could have provided both the NWS and local entities with clearer information on the magnitude of the flood wave and how quickly it was building and moving down the river. There are siren or alert systems that can be connected to such gages to automatically activate when the gage is rising at a particular rate or when it reaches a certain threshold. Even without an automated system like this, just having basic emergency procedures clearly established as described above could make a huge positive difference in getting people to higher ground even with just minutes of warning lead time. However, given the statements by multiple people in the San Antonio Express article, and statements about lack of clear plans at other facilities along the river and communication snafus in the county emergency services, it seems likely these were not in place at a county government level and for at least some of the private sector campgrounds and resorts.

Obviously, you can see in the Houston Chronicle map of flood fatalities that the area around Camp Mystic was the first area to see fatalities, and clearly that was the area that had the least amount of warning and was most vulnerable to having loss of life even with more warning and better preparedness. However, many of the fatalities occurred well downstream of Hunt, including in the area around Kerrville. The USGS gage at Kerrville did not start rising rapidly from floodwave until 4:45 am, a full 40 minutes after the NWS had issued a flash flood emergency, and did not peak until 6:30 am. In contrast to the upstream areas, it seems more clear that with better preparedness and warning communication that lives could have been saved downstream of Hunt where there was much more lead time and information about the magnitude of the flash flood.

I want to finish this post by talking about a quote in the San Antonio Express article from a parent whose daughter attended Camp Mystic in the past:

I would be very surprised if something like this ever happened again because it was such a freak event. “I don’t think anyone was negligent in this situation at all,” she added. “I don’t know what could have even been done to prevent something like this.”

I am not going to comment about “negligence” as that is a term with specific legal meaning, but the idea that an event like this cannot happen again or that nothing could have been done to prevent the impacts of this event is the kind of thinking that, in my opinion, likely contributed to the loss of life in this catastrophe. An event like this will happen again on this river. It may not be in the next decade, it may not be for 50 years, but it will happen. In fact, scientific evidence strongly suggests that climate change is increasing the likelihood and frequency of major flash flood events such as this one, and even without climate change, increasing population naturally puts more people into danger from these types of events.

I do not want to see more than 100 families going through the heartbreak and despair that they have suffered over the last 3 weeks. Given the growing risks, we have to recognize that hope and telling ourselves these are freak events or acts of God are not the answer to reducing the devastating impacts of these events. We have the scientific knowledge to understand our vulnerabilities, and most importantly, we have agency. We can make decisions at political levels and at personal levels to mitigate these events, prepare for the impacts we cannot mitigate, and help those who sadly end up still being impacted. The first step, though, is accepting that with knowledge and agency comes responsibility; when events like this tragedy happen, we have to examine them with clear eyes, and then take the lessons learned and make things better going forward. We should all be willing to commit to work together to “row away from the rocks” and help ensure that tragedies such as this become less frequent and impact fewer communities and families.