Toggle Navigation
Digital Imaging: Imagine the Possibilities
Teledyne Logo

Flood Imaging: A Hard Rain’s Gonna Fall

Satellite-based hyperspectral imaging can help us understand and minimize Houston’s next flood.

Houston had a special problem. It was inevitable, the result of a long-term gradual development. It was also a single, sudden, once-in-a-millennium event.

August 26–30, 2017: In only two days, the equivalent of a cube of water 2 miles high, wide, and deep (almost exactly the size of Star Trek’s Borg Cube) was dumped on Houston and southeast Texas. Pictures of Hurricane Harvey’s effects are mind-boggling, with interstate freeways turning into seemingly permanent river systems. Early estimates put the damage at US$150 billion to $180 billion. Thousands of people in the United States’ fourth largest city are homeless, and government services are still overwhelmed.

The reasons this happened are not intuitive. Many catastrophic weather events like Katrina, Sandy, Rita, and Tōhoku cause storm surges that breach shores, seawalls, and levees. But when they pass, the threat has passed, too. Protecting against the storm is a matter of keeping the water out for a limited period of time.

But in this case, the water kept rising, even after the storm had passed. Floods are far and away the most common natural disaster worldwide, and impact of flooding in densely developed areas is far more pernicious than the drama of a storm. Which is why floods account for more deaths than any other type of natural disaster. Cities flood – and stay flooded for a long time – not because the water comes in, but because of how the water gets back out again.

How bad will the next one be? When will it arrive?

Just like dams, bridges, and skyscrapers are designed to bear certain loads, water management systems are designed with specific limits based on probable storm events. Scientists ask themselves: what’s the worst storm that could happen in the next ten years? The next 25? They look at the historical data and make reasoned, evidence-based predictions.

But as the climate changes, the historical record is becoming less useful to predict the future. In the case of Hurricane Harvey, the resulting flooding was considered a once-in-500-years flood. Maybe once-in-a-thousand-years. And it’s the third one in the last decade. In fact, there were six “1,000-year” floods in the US over the five years from 2010 to 2014; in 2015 and 2016, there were at least three each year.

To better predict storms in the future, we need better data about what’s happening now. That requires collecting and using up-to-date data. Satellite-based hyperspectral imaging is proving to be one of the best ways to do that.

Understanding a changing city

Keeping data up-to-date is difficult, especially in a city like Houston. First, it’s enormous, stretching across more than 600 square miles. It’s a classic case of American urban sprawl: available land made development easy at the ever-expanding edges. In the process, a lot of land gets covered, both by developments and the roads that take people there.

The United States is about 75% open land, and less than 1% is urban development. Normally, it’s the open land that soaks up rainfall. It gets taken in by lawns, parks, empty lots, farmlands, and anywhere else there is permeable, exposed soil. Prairies have been shown to absorb up to a foot of water an hour. But in cities, that 1% of pavement goes up to 40%. Urban development covers the soil with concrete, asphalt, houses and office buildings, so that the water cannot soak down—it has to go somewhere else. The diversity and complexity of Houston’s famously un-zoned and un-designed cityscape make this even harder to understand.

Second, Houston is a city that has undergone a lot of change. From 1990 to 2016, Houston’s population grew from 1.6 million to 2.3 million people. The greater metro area went from 3.6 million to 6.5 million.

The effects of urban development on the natural ecosystem and its link to the increased flooding in Houston, Texas have been shown. Using neural network techniques, a 2005 study used four Landsat Thematic Mapper images to reveal five land classes in Houston from 1984 to 2003: vegetation, bare ground, water, concrete and asphalt. Results show that asphalt and concrete increased:

  • 21% between 1984–1994
  • 39% between 1994–2000
  • 114% between 2000-2003

Vegetation across Houston has suffered an overall decrease. More than 40% of its wetlands disappeared. Even the soil can become inundated if too much water falls too fast. It the water can’t soak through, it will follow the topography and flow to low-laying areas and rivers, lakes, and seas. For Houston, natural drainage is provided by a series of slow-moving rivers, called bayous. But since the city’s landscape is mostly flat, the water does not travel anywhere quickly.

Buffalo Bayou, looking towards downtown Houston

That’s where a city’s manmade stormwater system is supposed to take some of the load and channel the water away from impervious surfaces and the structures built atop.

In anticipation of this near certainty of flooding, Houston has built drainage channels, sewers, outfalls, ditches, and detention ponds to hold or move water away from local areas. But, according to some experts, Houston’s storm water systems are in many places woefully inadequate, able to handle only 1.5 inches of water an hour.

When these storm water systems become overwhelmed, as they did during Hurricane Harvey, the streets and highways of Houston are unable to help direct and control the water, resulting in the images of interstate freeways turned into rivers. (While this is an emergency scenario, it also greatly limits the ability of the city’s residents to evacuate in the case of a flood.) The outcome is that water finds its way into places no one intended: towards downtown; into chemical plants that explode, and into homes.

Cairo, Illinois, Mississippi River from the Worldview-2 satellite. False color (left) and natural color (right) imaging of a flooded area shows much more clearly flooded areas can be identified with multispectral imaging.

The result was increased runoff and chances of flooding. And the city has continued to change. The US Federal Emergency Management Association had updated its models for Houston in 2016, but the city itself was working off older data from 2012. Without the right data, it was difficult to tell which areas where most in danger, and who was living there.

Even in 2011, there were concerns that the amount of water being held by Houston’s two dams were rendering them unstable. During Houston’s flood, U.S. Army Corps of Engineers made the decision to release water to save the dams, flooding thousands of homes that had not been affected by the flooding until then.

Further reading on Hyperspectral Imaging: