A Border Wall Could Cause Deadly Flooding. Officials Plan to Build It Anyway

Engineers have long concluded that a wall in the flood plain of Starr County, Texas, could be catastrophic.

On July 12, 2008, in the border town of Nogales, Arizona, the afternoon turned from sun-drenched to ominously dark in a matter of minutes. A summer monsoon is typically a welcome respite in the desert heat, but the approaching storm quickly became violent. Rising waters began creeping into downtown businesses. The public library was inundated. At the international port of entry, the road leading into Mexico began to buckle, and a commercial bus waiting to cross started sinking. Passengers frantically scrambled out, and the driver managed to back up onto firmer ground, escaping the newly formed ten-foot-wide sinkhole.

Just across the border, in neighboring Nogales, Sonora, people watched in horror as city streets morphed into raging rivers strong enough to sweep away cars. A churning mass of caramel-colored water and debris began to fill the shops in the busy commercial district near the river. Mexican soldiers and firefighters rushed in to pull people from the floodwaters with ropes. One eyewitness later said it looked like a “bomb fell.”

The next morning, as residents of both towns awoke to the scenes of devastation, many were struck by the stark differences in damage between the two sides of the border. In Arizona, although mud and debris filled the streets, most of the water had dissipated, while in Mexico, entire streets remained submerged. A second, much larger sinkhole had formed near the border fence on the Mexican side. Mexico declared Nogales a disaster zone, and roads remained closed for weeks as workers hauled away mangled cars and rubble. It would take months to repair the damage, which was estimated at roughly $8 million.

The sister cities of Nogales, which were built along a floodplain, had been battered by heavy rains before. Longtime residents still remember other major floods, like one in 1983 that raged across southern Arizona and killed around a dozen people. The 2008 storm, on the other hand, was a typical monsoon, about two inches of rainfall in two hours, and locals wondered why it had been so catastrophic. They also wondered why it had been so much more devastating on the Mexican side.

They would soon have their answer. Underground, near the massive sinkhole, Mexican engineers discovered that a storm-runoff channel had been blocked by a five-foot-tall concrete barrier that had been constructed a few months earlier by U.S. Customs and Border Protection. As a result, the storm runoff in Mexico was unable to pass through the channel to a shared water treatment plant in Arizona. As pressure mounted, water burst up into the street like a geyser. The nearby border fence had compounded the damage. After getting clogged with storm debris, it had acted as a dam, capturing the floodwaters on the Mexican side rather than allowing the water to run its course on both sides of the border.

One hundred fifty miles to the west, the same storm system had caused another disaster to unfold that July afternoon. A recently built fifteen-foot steel mesh border fence became choked with several feet of debris during the downpour. Like the fence in Nogales, it effectively turned into a dam. Floodwaters rose as high as seven feet along the fence before pouring into Organ Pipe Cactus National Monument and the nearby towns of Lukeville, Arizona, and Sonoyta, Sonora. Local businesses suffered severe damage.

  • “Mitigating the impacts of flooding from the U.S. side of the border is unattainable,” wrote a CBP contractor in a September 2008.

It was a disaster foretold. A year earlier, officials at the Organ Pipe monument, a desert tract managed by the National Park Service, had warned CBP and the Department of Homeland Security that during a storm, a fence built there would become plugged with debris, causing the water to back up and flood the surrounding areas. But CBP had dismissed their concerns, issuing a report soon after claiming that its own environmental assessment had found “no significant impact.” According to CBP, the fence “would not impede the flow of water” or cause flooding. So the agency proceeded as planned.

After the flood, CBP commissioned Michael Baker Corporation (now called Michael Baker International), an engineering and consulting firm, to examine what had gone wrong. According to the report, engineers found that in some cases, as much as six feet of debris had collected on the fences, causing a “water-fall effect” on the other side. Afterward, CBP and the U.S. Army Corps of Engineers spent several million dollars retrofitting fences with gates that could be opened to relieve the flooding. But in the summer of 2011, heavy monsoon rains tore through a forty-foot stretch of fence near Organ Pipe, washing the fence away. In 2014, floods in downtown Nogales, Sonora, caused several million dollars in damage. This summer, heavy flooding once again struck Nogales, killing two people. What was once a rare occurrence has become a devastatingly regular phenomenon.

CBP has said little publicly about the floods, but at the time the barrier in Nogales was built, CBP contractors were under pressure to meet a deadline mandated by the Secure Fence Act of 2006: 670 miles of fencing had to be erected across the Mexican border by December 31, 2008, before George W. Bush left the White House.

The agency had nearly met its planned quota in Arizona, but in Texas, it had encountered stiff resistance from landowners. After the catastrophes along the Arizona border, meeting the deadline for the planned sections in Texas appeared even less plausible. CBP was primarily concerned about Starr County, a largely rural stretch west of McAllen, where most communities sit in the Rio Grande floodplain. Just across from them are more densely populated cities in Mexico.

Two months after the devastating 2008 Arizona floods, engineers within the agency quietly concluded that building a wall in Starr County was too dangerous. “Mitigating the impacts of flooding from the U.S. side of the border is unattainable,” wrote a CBP contractor in a September 2008 document obtained by the Sierra Club through a Freedom of Information Act lawsuit. “The risks associated with the potential flooding on the Mexican side of the fence could range from minor property damage to the loss of life.”

But in the decade since that 2008 report, CBP has persisted in trying to build a border wall through Starr County anyway. It has been repeatedly stymied, first because the wall violated a binational treaty, and then because the agency ran out of money. Then, this April, Congress approved $1.6 billion to increase security measures along the border and build 100 miles of wall, including 33 miles in Texas.

It came as no surprise to Roberto Salinas when CBP came calling this spring. Salinas is the mayor of Roma, Starr County’s second-largest city, which was built on a sandstone bluff high above the Rio Grande with a commanding view of its Mexican neighbor, Ciudad Miguel Alemán. Spanish colonists settled the area in the 1760s; locals will proudly tell you that the town existed long before the Rio Grande became an international boundary. In the nineteenth century, it was an important steamboat destination for river merchants. In more recent times, its historic district, an ensemble of nearly two-hundred-year-old white and terra-cotta adobe buildings, has lured filmmakers. (A generously mustachioed Marlon Brando, playing Emiliano Zapata, once fought the Mexican Revolution there in the movie Viva Zapata!)

CBP asked Salinas to attend a meeting with other local mayors. He had already met with officials from the agency three or four times in the previous two years about the potential construction of a border wall. “But this time, they told me their plans were more definite, because they had the funding and they had decided on how many miles would be built,” Salinas said.

When Salinas arrived at the meeting a few weeks later, he learned that CBP planned to build up to twelve of those miles in Starr County, much of it through his town. Salinas and the other mayors told CBP officials that they were worried about how the wall would impact their communities. “They told us that everything would depend on the engineering and design of the wall,” Salinas said. “They said there really wasn’t a lot they could tell us.”

View of downtown Nogales, AZ, with the border wall in the back on July 2012.Image: Sandy Huffaker/Getty Images

Somewhat perplexingly, however, they were able to describe what kind of wall would be built. The officials said that it would be a bollard barrier, made up of vertical steel posts, similar to an eighteen-foot fence built a decade ago in Brownsville, 115 miles east of Roma. “A structure of that type could very easily accumulate debris and cause flooding,” Salinas told me.

Salinas had never seen the 2008 CBP document predicting disaster if they built a wall through his town, but he didn’t need to. He’d lived along the river his entire life. And he knew what had happened in Arizona.

When I met Salinas at his office this August, he told me that although most of his constituents are in favor of increased security along the border, many are opposed to the wall, not only because of potential flooding but also because their claims to the land go back more than eight generations, to a time when the Spanish controlled the territory. Throughout the centuries, their ancestors had shed blood and fought bureaucratic battles with the governments of Spain, Mexico, the Republic of Texas, and the United States to retain rights to their land. “People don’t want their property divided or taken from them,” Salinas said.

He hoped the engineers hired by CBP would come to their senses once they arrived and saw what they had to work with—the eroded sandy bank along the river, the creeks that drain into the Rio Grande, and the expansive floodplain. But I told him what CBP had recently confirmed to me in a statement: that after years of reports concluding that a wall here would exacerbate major floods, the agency was still intent on building the wall through the floodplain in Roma. It would also be built through nearby Rio Grande City, which has about 14,500 residents, and the smaller community of La Grulla. A CBP analysis of the wall in Starr County had demonstrated “negligible impacts,” the statement said.

“Well,” Salinas told me, pausing, “we believe it would have a pretty bad impact.”

Later that afternoon, I walked across the international bridge and looked out toward Miguel Alemán. At more than 20,000 residents, its population is twice as large as Roma’s. Several neighborhoods sprawled near the banks of the Rio Grande, as well as churches, a doctor’s office, a grocery store, and a large customs building. Once I was across the river, the secretary of tourism picked me up and gave me a ride to the mayor’s chambers at the Palacio Municipal. There, I was ushered into the office of Mayor Rosy Corro, who in 2016 had become the first female mayor in the city’s history. (She lost her reelection campaign and left office October 1.)

  • “We have a part of the city that is very low near the river, and a lot of people live there. They will lose everything.”

Behind Corro was a large framed portrait of her husband, Raúl Antonio Rodríguez Barrera, the former mayor of the city. Corro followed my gaze. “I am a widow,” she said. Her husband, I knew, had been gunned down in 2012, in front of their home. Miguel Alemán had been on the front line in a battle over territory between the Gulf Cartel and Los Zetas. During my recent visits to the area, there had been soldiers in military trucks patrolling the streets. But in mid-August I saw no sign of military patrols, and Corro told me things had cooled.

I asked Corro if she had spoken with anyone from the U.S. or Mexican government about the wall being built near her city, and she said no. She was surprised to hear that funds had already been allocated for a wall in Starr County. “The truth is, they don’t tell us anything,” Corro said, shaking her head. “The only thing we’ve heard is from reporters from the United States who come here, like you.”

She told me that in 2010, many of Miguel Alemán’s neighborhoods near the river had been inundated during Hurricane Alex. She worried that a wall in Roma would make things worse.

“We have a part of the city that is very low near the river, and a lot of people live there,” she told me. “They will lose everything.”

For the past 129 years, through a series of evolving treaties, the International Boundary and Water Commission, a little-known but vitally important binational agency in charge of boundary demarcation, water rights, and flood control, has helped keep the peace between Mexico and the U.S. For many years, it was the only thing preventing CBP from building in the floodplain in Starr County.

The IBWC maintains a U.S. section in El Paso and a Mexico section in Ciudad Juárez, each with its own commissioner. The agency has more than two hundred employees, including engineers and hydrologists. Today, one of the IBWC’s most important jobs is to uphold the Boundary Treaty of 1970, which, in addition to setting guidelines for territorial disputes, requires both countries to agree on any border structure built in a floodplain. (CBP had failed to notify the IBWC or Mexico when it built the concrete barrier in the storm tunnel under Nogales, which resulted in the catastrophic flooding and triggered a diplomatic complaint.)

For decades, the IBWC has been admired internationally for its ability to solve seemingly intractable problems, so much so that it has served as a model for Middle Eastern countries fraught with international boundary and water conflicts. But recently the border wall has sown division within the IBWC, especially when it comes to Starr County.


Documents obtained through the Sierra Club lawsuit show that from 2008 to 2010, the U.S. and Mexico sections of the IBWC were allied in their opposition to building a wall in the county, even as CBP pressed for its construction. During that time, CBP commissioned a series of reports from Michael Baker assessing the impact of a border wall built in the floodplain. Each of the reports looked at different potential alignments and designs of the wall. And all came up with the same findings: the fence would exacerbate flooding.

In 2009 C. W. Ruth, then the U.S. commissioner of the IBWC, sent a letter to CBP saying that it would not sign off on the agency’s building a fence in the floodplain. The following year, Edward Drusina, the new commissioner of the U.S. section, reiterated the IBWC’s opposition. “Given the model results obtained by CBP and the potential impacts that may be caused on the international boundary, the USIBWC is not in a position to approve construction of the [Starr County] fence projects,” Drusina wrote to CBP.

Two years later, however, the U.S. section of the IBWC made a sudden reversal, surprising its Mexican counterpart. CBP had commissioned yet another report on the floodplain—this one also from Michael Baker—in August 2011, but unlike the others, this report relied on a proprietary modeling software called FLO-2D instead of the open-source software HEC-RAS, which is preferred by the Army Corps of Engineers and has been used for decades by other parties doing flood analysis along the Rio Grande. The new report came up with a dramatically different conclusion: building a fence in the floodplain “should result in minimum hydraulic impacts to the Rio Grande on the Mexico side.”

In February 2012 the U.S. section of the commission sent a letter to Scott Recinos, a civil engineer with LMI, a private consulting firm based in Washington, D.C., that manages many of CBP’s border-wall projects. The commission told Recinos it no longer had any objection to building a fence in the floodplain. “After an in-depth and thorough review, the USIBWC has concluded that the proposed fence projects will not cause significant deflection or obstruction of the normal or flood flows of the Rio Grande,” wrote John Merino, then the principal engineer of the U.S. section.

Border Patrol agents patrol the Rio Grande under the international bridge spanning the U.S.-Mexico border on April 2016 in Roma, Texas.Image: John Moore/Getty Images

In response, Loren Flossman, the director of CBP’s wall program, sent out a celebratory email to his bosses: “Sirs, the attached letter to Scott Recinos from IBWC is the successful conclusion of over 3 years of singular dedication and tenacity by Scott. His commitment has resulted in IBWC’s approval of the . . . fence segments, which are part of the 670 miles of proposed fence.”

I sought further explanation on the reversal from CBP and the IBWC, but interview requests with both agencies were declined. In a written response, a CBP spokesperson, Carlos Diaz, explained, “All pertinent findings were taken into consideration for the alignment and design of the planned wall.”

In a follow-up email, I asked what Recinos had done specifically to get the IBWC to sign off on the approval. In another written statement, the agency simply said that the IBWC had “concurred with CBP’s hydraulic analysis.”

When I called Stephen Mumme, a political science professor at Colorado State University who has spent much of his career studying U.S.-Mexico relations in regard to the Rio Grande and who knows officials from both sections of the IBWC, he said the U.S. side was continually being pushed to approve CBP’s plans to build the wall. “The U.S. section of the IBWC was under a lot of pressure from the Department of Homeland Security to allay Mexico’s fears,” he told me.

I did manage to finally reach Drusina, the commissioner of the U.S. section of the IBWC who had reversed his position and signed off on CBP’s plans in 2012. In May, he’d been asked to resign by the Trump administration, which he said hadn’t come as a great surprise, since he’d been appointed under Barack Obama, in 2010. “I served at the pleasure of the president,” he said. (His successor, Jayne Harkins, was sworn in on November 1.)

I told Drusina I found it odd that in every study prior to the August 2011 Baker report, the IBWC had consistently concluded that a wall would violate the 1970 treaty. Why had he suddenly withdrawn his objections? Had he been pressured by CBP?

Drusina said he’d never been pressured. He felt his agency had developed a good hydraulic model in conjunction with CBP. “We used good science and engineering to demonstrate how much of an impact the fence would have on floodwaters,” he said. He also told me that Mexico had been advised every step of the way. “The Mexican section understood that we had a job to do,” he said.

  • “Anyone who has lived alongside the river has no problem understanding what happens when you stick something in the way of the water.”

Documents from that same period dispute some of his claims, however. After the 2011 Baker report was released, Luis Antonio Rascón Mendoza, the principal engineer with the Mexico section of the IBWC, sent a series of diplomatic letters protesting the results. “The decision to use the FLO-2D model when submitting projects located in the floodplain must first be agreed upon through the commission,” he wrote in a December 2011 letter to Drusina. “We reiterate our opposition to the construction of the proposed fence in the Rio Grande floodplain.”

The U.S. section responded in a letter that Mexico had seven days to prove that the data was wrong—an impossible task, which seemed like a further insult. Mexico did eventually run its own data through the same FLO-2D software, according to a February 22, 2012, diplomatic letter signed by Rascón Mendoza. His engineers found that if a wall were built in Roma, the floodwaters could increase by 40 to 100 percent on the Mexican side of the fence. “Again we want to express our disagreement with the U.S. section in the unilateral way it has agreed to this project without the agreement of the full commission,” wrote Rascón Mendoza to his U.S. counterpart, John Merino.

Rascón Mendoza did not respond to interview requests, but in April 2017 he told NPR he was still against the wall. “If they plan that type of project, we will oppose it,” he said. So far, Mexican officials have been reluctant to elevate their grievances beyond diplomatic letters of protest filed with the U.S. State Department. But Mumme believes they’re weighing further action. “What I am dead sure of, and based on the conversations I’ve had, is that Mexico is paying close attention,” he said. If the treaty is violated, Mexico could lodge a formal complaint that could ultimately be tried before the International Court of Justice in The Hague, Netherlands. A more likely recourse, however, would be for Mexico to take the fight public. “This could become very politicized, especially if it appears that the U.S. is gaming the treaty at Mexico’s expense,” Mumme said.

When I spoke to Drusina, I also asked about the author of the 2011 Baker report, Padinare Unnikrishna, who accepted a position at the IBWC shortly after publishing the contested findings. Drusina said his hiring had nothing to do with CBP’s influence or the report’s favorable conclusions.

The IBWC declined to make Unnikrishna available for an interview and sent a written statement instead. “Dr. Padinare Unnikrishna severed all ties with his previous employer on Sept. 30, 2011,” wrote Lori Kuczmanski, a public affairs officer for the IBWC’s U.S. section. “The U.S. Section’s Legal and Ethics Office has determined there is no conflict of interest.”

Not long after the IBWC approved the plan in 2012, it held a “citizens’ forum” in Starr County to inform landowners that the new flood model showed that a fence would have little impact on their communities. Unnikrishna was the presenter.

Scott Nicol, a Sierra Club borderlands volunteer in McAllen who worked on the border wall Freedom of Information Act litigation, recalled that the meeting was tense. “There were a number of landowners who wanted to know how this was going to impact them, what it was going to mean for their livelihood and their safety.”

Nicol said that Unnikrishna never mentioned he’d been the author of the report and that he had seemed condescending. “I think this is why the meeting went so badly so quickly,” Nicol said. “He came off as saying, you know, ‘There’s a lot of jargon in this report. This is all over your heads, but just trust us. Everything is going to be fine. Trust the experts.’ But it wasn’t over anybody’s head. Anyone who has lived alongside the river has no problem understanding what happens when you stick something in the way of the water.”

Last year, Texas RioGrande Legal Aid, which represents landowners along the border, hired its own civil engineer to look at the contested 2011 Baker report and other documents obtained from CBP through the Sierra Club litigation. But after CBP announced it was altering its alignment of the wall segments in Starr County yet again, TRLA decided to have the engineer wait on his report until the agency’s plans were finalized.

Nonetheless, the consulting engineer hired by TRLA, Al Blair, said he’d already conducted an extensive literature review and read all of the documents. He had examined the 2011 Baker report and felt confident in his assessment, which he agreed to discuss with me. Blair has worked on drainage and other water projects along the Rio Grande for decades, and his critique of the Baker report was withering. “There is no value in this report,” he said. “I don’t understand why they did this, other than they wanted a new model so they could dispute the other one. In my opinion, this is a worse model of what actually occurs during a flood in the Rio Grande.”

What particularly stood out, he said, was that for CBP to reach its favorable findings, the data assumed only a 10 to 25 percent degree of blockage when debris hit the fence during a flood. Previous models commissioned by CBP had assumed 50 and even 100 percent blockage, which was more likely, he said. “Any reasonable engineer who has worked on the Rio Grande knows the river is extremely vegetated,” he said. “During a flood, everything’s in the water, and there’s trash. So if you put a filter in, like a fence—a strainer, so to speak—you can imagine that it’s going to clog up almost immediately.”

  • “In my opinion, this is a worse model of what actually occurs during a flood in the Rio Grande.”

CBP already knows this, he said, because officials have seen what has happened to the border walls in Arizona. “And that’s in a desert environment, not semitropical like the Rio Grande Valley.”

Blair said he had another concern. “I haven’t found any reports about what happens on the American side of the wall,” he said. “A lot of debris and trash will flow toward the Rio Grande, and it will clog up the fence, and if that gets plugged up, the water is going to back up right into the middle of town. That’s going to be a big problem.”

CBP’s estimates are even more problematic, Blair said, because climate change has rendered most flood models obsolete. Future rainfall events will be far more severe than anything engineers have seen in the past, something the agency’s predictions haven’t accounted for. “Nowadays we’re all so scared because the thirty inches of rain from [Hurricane] Harvey means all our previous estimates of flooding are too low,” Blair said. (The IBWC, Unnikrishna, and CBP declined to respond to Blair’s assessment of the 2011 Baker report.)

It’s unclear when CBP will finalize plans for Starr County, though construction in neighboring Hidalgo County will begin in February. It is certain, however, that a wall will have lasting repercussions. “You can run the studies and put the parameters in for any answer you want to justify, but when Mother Nature strikes, that’s the real test,” Mumme, the Colorado State professor, told me. “And then you’re talking about human lives.”

This article was reported in partnership with the Investigative Fund at the Nation Institute, now known as Type Investigations, where Melissa del Bosque is a Lannan reporting fellow.

About the reporter

Melissa del Bosque

Melissa del Bosque

Melissa del Bosque is a former Lannan reporting fellow with Type Investigations.