Preliminary Analysis of Hurricane Harvey Flooding in Harris County, Texas

by Nicholas Pinter, Nicholas Santos, and Rui Hui

Located in Harris County, Texas, Houston is the 4th most populous city in the US.  The flooding now unfolding in the Houston area is a human and economic disaster likely to rank with Hurricanes Katrina and Sandy among the worst in US history.  At the present moment, little quantitative information is available about the extent of flooding or economic damages.  The inundation is so widespread that detailed assessment has not yet been possible.  Even measurement of the flooding by visible-light based satellite has been impossible because the remnants of Hurricane Harvey have stalled over the Gulf Coast for the past week, dumping record precipitation and obscuring the skies from satellite view.

Our team in the Natural Hazards Research and Mitigation Group at the University of California, Davis conducted two sets of analyses to provide preliminary information on the pattern of flooding and to provide some context on historical flood damages in the Houston area.  Specifically, we:

(1) Processed radar data from the Sentinel-1A satellite from Aug. 29, at the peak of flooding in many areas.

(2) Analyzed data from the US National Flood Insurance Program (NFIP) to look at the nature and pattern of flood exposure in Harris County, including Houston.

Satellite Analysis

The Sentinel-1 system, which consists of two satellites, was launched by the European Space Agency.  The Sentinel-1 satellites carry radar sensors, which can scan the surface at night and through cloud cover.  In addition, this radar imagery is sensitive to standing water, making it an ideal tool for mapping the extent (but not depth) of flood water covering an area.

A Sentinel-1 satellite passed over Houston on August 29, which we used to create a map that highlights water and flood inundation on the surface.  We overlaid the resulting map on flood-zone boundaries from FEMA’s National Flood Hazard Layer and on other geographical data. Finally, we further compared the surface water extent with known pre-storm extents by flood-zone in order to determine flooding extent in each area in the portion of the county we assessed.

The resulting inundation map shows the striking extent of flooding in Houston and provides context for the on-the-ground photos coming out of the area.  Reservoirs are overflowing, streams and rivers surging high over the surrounding floodplains, and highways and roads are now themselves rivers.

harvey_map

An image of the flood inundation map in Harris County, Texas, showing the flooding extent of Hurricane Harvey. Click to open and explore.

Table 1 shows the extent of flooding mapped by flood zones, as defined by FEMA.  The flooded area in square miles is not an accurate measure of total area underwater because image pixels with vegetation or buildings will not be classified as standing water.   For example, designated floodways (river channels, mostly) were certainly 100% full of water throughout the study area on 8/29, but the radar signal categorized only 30% of those pixels as water.  However, the distribution of water-designated pixels correctly illustrates the pattern of flood inundation.  In the future, we plan to create a raster map that shows the boundary between flooded and not-flooded land in Harris County, from which accurate total inundation areas can be calculated.

harvey_table

Table 1. Distribution of flood inundation in Harris County, Texas in different mapped flood zones.

In Table 1, the percentages of total flooded area illustrate the great intensity of flooding resulting from Harvey.  About two-thirds of the inundation is outside of FEMA’s Special Flood Hazard Area, which is the so-called “100-year” floodplain, or area with a 1% or greater chance of flooding in any given year.  The 13.44% of inundation in the 500-year floodplain is very similar to the total portion of the study area in the mapped 500-year zone, suggesting that on average, flooding on Aug. 29 completely filled Harris County’s 500-year floodplain.  More than that; over 50% of estimated inundation occurred outside of any mapped flood zone.

History of Flood Damage and Exposure in Harris County

Texas has a long history of damaging flooding, and Houston has seen some of the worst of it, well before Harvey.  Since 1964, Harris County has had 27 federally declared disasters related to flooding and/or coastal storms.  We obtained databases of NFIP policies nationwide back to 1994 and insurance claims back to 1972.  For the analysis here, we identified current policies and past claims in the City of Houston, Harris County, and Texas as a whole.  We also examined repetitive flood losses.

As of June 30, 2017, Harris County policyholders held 249,212 NFIP flood insurance policies, covering $70.34 billion of assets, and generating $138.4 million in annual premiums. Over 40% of Texas NFIP policies are from Harris County.  Texas-wide, there are 593,115 policies in force, covering $161.2 billion, with $ 364.0 million in premiums.  Texas represents about 12% of NFIP policies nationwide and about 10% of annual premiums to NFIP.  Notably, and in contrast to the trend nationwide, the number of NFIP policies both in Harris County and in Texas has declined during the past several years (Figure 2).  Numbers of policies peaked in 2008 and have declined every year since that time.  This decline occurred despite several large floods in Texas during that time, which tend to sharply increase flood-insurance penetration in other areas of the country.  In addition, policy totals in Texas began declining well before Biggert-Waters 2012, federal legislation that raised NFIP premiums for some policyholders.

NFIP policies in force, by year, in Texas and in Harris County.

From 1975 to early 2015 (span of our NFIP claims data), NFIP policyholders in Texas experienced 194,029 paid flood losses, totaling over $9.2 billion (2015 dollars) in claims. The largest number of losses and NFIP claims occurred in 1979, 2001, and 2008 (Figure 3).  From 1975 to early 2015, Harris County had 77,697 paid flood losses, totaling about $3.6 billion (2015 dollars).

NFIP flood losses, by year, in Texas and in Harris County (in 2015 dollars).

Repetitive Flood Losses in Harris County

In 1998, the National Wildlife Federation report, Higher Ground, identified a major challenge to the NFIP – policyholder who made flood claims again and again, with some of them receiving cumulative payments many times the structure’s value.  In 1998, the worst case was one home that was flooded 16 times.  Today, the largest number of claims on a single structure has risen to 40.  And Texas has a disproportionately large share of repetitive loss structures and losses.

FEMA maintains a list of properties flagged as Severe Repetitive Loss Properties (SRLPs); defined as those that have had at least four claims (each ≥$5000) or total claims exceeding the value of the structure.  Of about 30,000 designated SRLP properties across the US, 4889 are in Texas, the second highest of any state (after Louisiana).  SRL properties in Texas have received $962 million in NFIP payments, the 2nd highest of any state (after Louisiana).  It seems likely that Harvey will push Texas to #1.

A total of 2794 of Texas’s SRL properties are located in Harris County, where 1925 of those are in the City of Houston. This number, summed together, is more SRLPs than any in other jurisdiction in the US. The total number of NFIP paid losses for SRL properties in Harris County is 15,685 (10,321 in Houston), the largest of any area in the US. The total of NFIP payments for SRL properties in Harris County is $596,025,224 ($195,971,067 in Houston), also the largest total of any area in the US.

We have also examined individual SRL properties, tabulating the worst of these in a variety of ways.  By counting the maximum number of paid claims, one property in Houston has been rebuilt at taxpayer expense 29 times.  We have also calculated total claims per property as a multiple of that structure’s value.  We tabulated the 30 largest of these ratios for single-family residential structures nationwide (structures valued <$10,000 excluded).  Harris County and Houston have 9 of these 30.  For example, one Houston-area home, valued at $116,335, has received NFIP payments totaling $1,848,916, or 15.9 times the structure’s value.  Other properties show even larger ratios.

There is room for some optimism in Texas.  FEMA provides funding to reduce long-term flood losses by mitigating individual properties, including acquiring and demolishing the most flood-prone structures.  In Harris County since 2000, 996 such properties have been acquired, at a cost of $63.5 million.  This optimism must be tempered, however, by the vastly greater pace of new construction in the Houston region.

The catastrophic flooding in the wake of Hurricane Harvey will certainly generate extensive discussion.  By some calculations, the current flooding represents the third “500-year” flood in the Houston area in the past three years.  Harris County flood managers have suggested that these extreme events represent rare, but plausible expressions of natural and stationary hydrology.  In contrast, credible counterarguments focus on climate-change tipping points as well as the rapid and extensive suburban development.  The catastrophic extent of current flooding in Texas points not only to a truly extreme event this year, but to a pervasive pattern of repetitive flooding. This pattern, in Texas and the Houston area in particular, points almost inescapably to local factors such as runaway development and lack of balanced hydrologic planning.  Recognition of these root causes is a vital first step in reshaping policies and guiding recovery in the wake of Harvey.

Nicholas Pinter is the Roy Shlemon Professor of Applied Geosciences in the Department of Earth and Planetary Sciences and an associate director of the UC Davis Center for Watershed Sciences. Nick Santos is a GIS developer and researcher at the Center for Watershed Sciences. Rui Hui is a postdoctoral researcher with the Center for Watershed Sciences.

This entry was posted in Around the World, flood, Floodplains, Planning and Management, Tools and tagged , , . Bookmark the permalink.

31 Responses to Preliminary Analysis of Hurricane Harvey Flooding in Harris County, Texas

  1. Dennis Johnson says:

    Kudos on quick work and high level analysis. This work will be helpful to planners, insurers, builders, and legislators…if they read it and accept the scientific evidence.

    Like

  2. Jai Rho says:

    Preliminary estimates of Harvey-related losses exceed $160 billion (the total Texas-wide flood insurance policy coverage). Perhaps insurance companies will become more interested in lobbying for responsible climate change mitigation and flood control efforts. In the meantime, there will be a deluge of claims while the insurance companies recede from the scene.

    Like

    • Bernie Weber says:

      I live in Houston just west of Barker Reservoir and I can tell you that the flood maps you have are not correct. Not even close in some areas.

      Like

      • nickrsan says:

        Hi Bernie – thanks for chiming in – we expect that there will be plenty of discrepancies – this is a quick analysis with the available data. As better options come in for assessing the flooding, those can be used instead. In the first few days after a flood, there are few options though. For example, NOAA is now flying and releasing imagery from different time periods of the flood, but its extent is limited.

        Like

  3. Pingback: Early Data From Harvey Shows Epic Flooding | iTruck NEWS

  4. Pingback: Early Data From Harvey Shows Epic Flooding – AKANews – Featured Articles

  5. Pingback: Early Data From Harvey Shows Epic Flooding

  6. Pingback: Early Data From Harvey Shows Epic Flooding - Dekalb Chronicle

  7. Pingback: Early Data From Harvey Shows Epic Flooding-Times of News Italy

  8. Pingback: Early Data From Harvey Shows Epic Flooding – Nation & World News

  9. Pingback: Rick Perry says climate change debate is "secondary" - News 12 Now

  10. Pingback: Rick Perry says climate change debate is "secondary" amid Harvey destruction - World Big News

  11. Pingback: Rick Perry says climate change debate is “secondary” amid Harvey destruction

  12. bradynock says:

    Do you have any plans to extend the analysis to adjacent counties?

    Like

    • nickrsan says:

      Not at this time – if future time and funding permit, we’d be interested in similar analyses comparing flood maps to damages.

      Like

  13. Pingback: Rick Perry says climate change debate is "secondary" amid Harvey destruction - News 12 Now

  14. Pingback: US News Headlines Rick Perry says climate change debate is "secondary" | USA News

  15. Pingback: Rick Perry says climate change debate is “secondary” amid Harvey destruction – Castwb

  16. Syriac Mathews says:

    There may be a need for additional filtering of satellite data. I live on WIllowbend Blvd Houston TX 77035. The satellite overlay data from Sentinel 1 on August 29th data shows the intersection of Willowbend Blvd and West Blvd, which is 4 houses away from us, to contain street flooding. But the reality was that the street flooding on 28th of August was completely drained out and from August 28th there was no pooling or flooding on the intersecting roads. Could cloud cover affect satellite data?

    Like

    • nickrsan says:

      Hi Syriac, Thanks for providing that context – streets often show as confounded with water in this analysis – normally, one would subtract out the water from a previous satellite date to mostly remove that false positive, but we had some alignment issues with the data when doing that. Removing those locations using that method also creates its own problem of not showing *actually* flooded streets in a situation with widespread flooding like this.

      In this case, cloud cover is not a factor because we’re using radar data which passes through clouds.

      Like

  17. Pingback: NPR week 2 – Site Title

  18. regionswork says:

    There can be a great gap between what water volumes the system can handle within the floodway, 100 year, 500 year flood markers used for risk assessment, and the amount of water that is added by rain events. Here in the Shenandoah Valley, Any of the steep mountain watersheds whose small streams carry a fraction of the water that carved them, will flood with heavy rains – seven inches an hour for a short period of time. It is a sort of Russian Roulette as to which hollow a heavy storm will hit. A pure, natural environment, takes the rain or drought, and the terrain and living things there adapt as best they can. Very little of that around these days. The builders of the built environment will take precautions to the limit or ability of their knowledge and wisdom as they built. Coordination and planning with a long term horizon, that takes in the watersheds at a comprehensive scale, are likely to do better, since little is built with the intention that it not be permanent. When the region is receiving more water on a regular basis than the built environment and, if there has been development, a more fragmented natural environment can handle, the new data requires rethinking by the risk managers, public and private. Insurance is a bandaid, not a solution. Acts of God really don’t exist, that’s just religion for inadequate underwriting. All life is risk management. Thus, truly conservative approaches are required. Your optimism about building in a risky area should not dig into my wallet via government bailout of your losses when that optimism has proven unwarranted in a flood. .

    Like

  19. William Fraizer says:

    Professor Pinter et al:
    My compliments on your efforts to use satellite data to perform a rapid assessment of the extent of flooding in Harris County.
    My home is located in the center of the area used for the primary figure in your article. I inspected the flooded areas located on the north bank of Buffalo Bayou, on foot, during the peak of flooding on 27 August and again on 28 August. I also noted the extent of flooding from White Oak Bayou to the north. Water levels dropped noticeably during this period and fell further by 29 August, when the rains in central Houston ceased.
    Your mapping appears to show a number of areas as inundated which were not flooded, even during peak water levels, except perhaps some street ponding, during rain bands on 27-28 August.
    For example you show an area of inundation almost in my backyard and another along the main east-west rail corridor; neither of these areas had any flooding. The rail corridor is along the high ground so it is the last area to flood.
    I submit that without proper ground truth checks, your detailed statistical analysis may have a high degree of error, and your conclusions, although preliminary, cannot be supported.
    I would be happy to provide some specific observations to your team, if that would assist your efforts.
    May I also suggest you use the flood zone maps generated by the Harris County flood control authorities as your reference documents. I’m sure they would welcome the opportunity to work with you on this. These maps are available on their website; just search for the Harris County Flood Education Mapping Tool.

    Like

  20. Pingback: Early Data From Harvey Shows Epic Flooding - Campus Centro

  21. Pingback: Why governments miscalculate your risk of flooding - Technology & Science - Kasa News

  22. Pingback: Why governments miscalculate your risk of flooding - Times of News Canada

  23. Pingback: Why governments miscalculate your risk of flooding | MediaNavy

  24. Pingback: Why governments miscalculate your risk of flooding – Technology & Science – Castwb

  25. Pingback: Why governments miscalculate your risk of flooding – Technology & Science | PoshToo

  26. Pingback: Why governments miscalculate your risk of flooding – Technology & Science

  27. Pingback: Why governments miscalculate your risk of flooding – Technology & Science – CHB Blog

Leave a Comment (moderated)

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s