Most readers of this blog know that I’m a somewhat atypical engineer. I often look at problems as sketches on a map or layers in a geographic information system (GIS). Other times I consider the long-term economics of any particular project — think investments in water conservation. Maybe I was a planner in a prior life?

With my planning hat firmly in place, this post will outline one approach the Harris County region might take to decide where to invest in future flood risk reduction projects. This is about projects beyond the current 2018 bond program. This is about what we might try to accomplish with another large bond initiative.

While I am a member of the Harris County Community Flood Resilience Task Force, this post is my personal view and does not represent the views of the Task Force.

BASELINE CONDITIONS

Before we decide where to invest, we need to create a heat map that will clearly illustrate areas that need more flood risk reduction investment. Today’s mapping software can create heat maps from any type of location data. Google Map displays traffic speeds using a green to red color scale based on cell phone velocities. Pedestrian injuries can be mapped using a heat map format, as shown in the example below from the Los Angeles Times.

So what data should we use to create a Harris County area heat map showing areas that need flood risk reduction investment?

If it were up to me, I would use the following three datasets, combined in some weighted fashion, to generate a composite heat map. The following sections describe the three datasets I’d like to use.

Annual Chance of Inundation

First I would suggest using the annual chance of inundation. Almost all ground surface in Harris County has an annual chance of inundation ranging from 0.1% (higher ground elevations away from bayous) to perhaps as high as 100% (low areas next to bayous, newer suburban streets that are designed to carry water away from houses).

I believe that no ground surface in Harris County has a 0% annual chance of flooding, except perhaps the very top of a few constructed hills. The top of the Memorial Park Land Bridge (when it’s completed) might come close.

Maps used to govern the National Flood Insurance Program (NFIP) provide a very coarse understanding of inundation likelihood – only along bayous and channels above a certain size. Those maps typically only show three “zones.” They show areas with an annual chance of flooding of less than 0.2%, areas with between 0.2% and 1%, and areas with more than 1%. About 37% of the county’s land area has greater than a 0.2% annual chance of inundation. The image below is from Harris County Flood Control District’s Flood Education Mapping Tool and shows the current NFIP flood zones associated with Carpenter’s Bayou. Notice how we have no information about the areas far from the bayou? We can’t tell if those areas have a 0.19% annual chance of inundation (just barely less than 0.2%) or if they have a 0.019% annual chance of inundation (10 times less risky). We also can’t identify areas with zero flood risk, because there aren’t any of those areas in our county, no matter who promises to “make you safe” or to “protect you from flooding.”

The Harris County Flood Control District is currently working on a detailed remapping of the flood risks in our county. The effort is known as the Modeling, Assessment and Awareness Project (MAAPnext). The output from MAAPnext will be an updated map of flood risk for every portion of the county’s ground surface. This will include all areas, not just areas adjacent to bayous and channels and lakes. I’m not sure how fine-grained the inundation risk levels presented will be, but I suspect we may be able to see color-coded maps showing areas with 0.2%, 1.0%, 2%, and 10% annual chance of inundation. My sources tell me this will be shown using square grids 9 feet on a side.

So based on MAAPNext results we will be able to create a heat map of flood risk across the county. I would suggest red for areas with greater than a 10% annual chance of inundation, orange for areas between 10 and 2%, yellow for areas between 2 and 1%, light green for areas between 1 and 0.2%, and dark green for areas with less than 0.2% annual chance of inundation. Using mapping software we can “overlay” this information with the other data layers to create a composite heat map.

MAAPNext information will also be critical in calculating a new proposed comparative index that I will describe next.

Flood Mitigation Benefit Index

The second dataset I would suggest using was originally proposed by Dr. Earthea Nance, when she was with Texas Southern University and Iris Gonzales of the Coalition for Environment, Equity and Resilience (CEER). It was originally called the flood benefits index, but the name was changed to Flood Mitigation Benefit Index by the Harris County Community Flood Resilience Task Force for clarity and to avoid any association with the Federal Bureau of Investigation – which is really not that involved in flood risk reduction efforts.

This is a new index value that will be calculated for all U.S. Census Tracts in the county to allow “apples to apples” comparisons between areas of the county with similar populations. The Flood Mitigation Benefit Index — or FMBI — because who doesn’t love acronyms — is calculated as follows:

The total prior investment in flood mitigation should be expressed in inflation-adjusted dollars and should be the sum of all investments in the area of interest intended to reduce flood risks. This should be the sum of all city, county, state, and federal investments in the area of interest. This is the most difficult data to obtain because the sum needs to include investments going back to 1937 when HCFCD was created to serve as the local sponsor for projects with large amounts of federal funding, which arrived starting in the 1940s and continues to this day.

The population should be the current population living in the area of interest, based upon US Census Bureau information.

The annual chance of inundation should be the current percentage value converted to a number from 0 to 100. This means that the FMBI would be calculated using 0.2 in the denominator if the area of interest had a 0.2% annual chance of inundation. An area with a 4% annual chance of inundation would use the value 4 in the calculation.

Everything else being equal, areas with higher FMBI’s have received higher prior investments in flood risk reduction and currently have a lower likelihood of inundation compared to areas with lower FMBIs. Here’s a table of hypothetical situations to illustrate how the changes to the input variables (prior investment, population, and inundation likelihood) impact the calculated FMBI result.

To evaluate these examples, consider how you would feel if you were one of the 6,000 people living in one of the example U.S. Census Tracts. In which tract would you want to live? Which tract should get the next flood risk reduction project?

The green shaded examples illustrate what happens when we increase prior investment, hold population constant, and decrease inundation likelihood. The FMBI goes up dramatically with increasing prior investment and decreasing inundation risk.

The blue shaded examples illustrate what happens when we only increase population in the area of interest. This shows that the FMBI is a per capita value.

The yellow shaded examples illustrate what happens when we only increase the prior investment in the area of interest. This shows how more prior investment drives the FMBI higher.

The grey shaded examples illustrate what happens when we lower inundation likelihood. Lower risk areas have higher FMBIs.

The current “goal-post” for floodplain management and development — the standard of care for engineers — requires us to do no harm to any properties and structures both upstream and downstream of our new project — this means no increase in inundation depths. The standard of care also requires us to design new stormwater management facilities around structures so they have less than a 1% annual chance of inundation. Can you tell which tracts in the list above meet the current standard of care?

A review of the annual chance column should help us determine the answer. Reading down that column we can see certain tracts have annual chances of less than 1%. They are Tracts 4, 5, 11, 12, 13, 14, 15, 18, 19, and 20. These tracts are currently meeting the standard of care and have “good” flood protection. Their FMBI scores range from 3,333 to 166,667.

All of the other tracts, with higher likelihoods of inundation, need help. They have between two to ten times the risk of flooding than the standard of care. Notice how low the FMBI values are for those tracts? It would not be good to live in those areas, right? The areas with inundation chances of more than 1% don’t look so good. Their FMBI scores range from 8 to 9,000, with most below 2,500.

Like MAAPNext results, we will be able to use FMBI values to create a heat map for the county. If we examine the entire range of calculated FMBI values, we can divide them up into five buckets. The lowest 20%, the next highest 20%, and onwards to the top 20%. I would suggest red for the lowest quintile, orange for the next, yellow for the next, light green for the next, and dark green for the top 20%. Here’s a hypothetical “mock-up” of how the FMBI heat map might appear. Both the cost and risk input data are still being collected, so this hypothetical map is purely for illustration purposes.

Social Vulnerability Index

The third, and last, dataset I would suggest we use is the Social Vulnerability Index (SVI) published by the Agency for Toxic Substances and Disease Registry of the Centers for Disease Control. SVI values for US Census Tracts are calculated from 15 different input variables including poverty rates, employment rates, income levels, educational achievement, age, disabilities, family structure, racial composition, spoken languages, type of dwelling units, crowding, access to vehicle, and group living conditions (for example nursing homes).

SVI values are normalized to range from 0 to 1. A value of 0 means the area is very resilient to disasters, with no poverty, high employment rates, high income levels, high educational attainment, few disabilities, etc. A value of 1 means the area is very vulnerable to disasters, with high poverty rates, low employement rates, low income levels, little education, more disabilities, etc.

Like MAAPNext results and the FMBI results, we will be able to use SVI values to create a heat map for the county. I would suggest green for areas with the first quartile of SVI values (least vulnerable), light green for the second quartile values, orange for third quartile values, and red for fourth quartile values. Here’s what this looks like for the entire county using 2018 SVI values.

COMING UP NEXT

In Part 2 of this post, I will describe how we might build the composite heat map and how it could be used to identify areas for future flood risk reduction investments.

Part 2 has been posted here.