Louisiana has a long history of major floods. These are driven by Mississippi River floods, extreme rainfall events, and hurricane surge. There was a heavy rain event in March 2016 in the Shreveport area with major flooding. The Amite River Basin Flood Tracking Chart shows the history of flooding on Amite River, which is the major watershed draining this area. More than 30 years ago, a major rain event flooded much of the same area as the August 2016 flood. The 1983 flood was not as deep and did not flood as many houses, but was a harbinger of the August 2016 flood. The key question is whether the August 2016 flood was due to a freak rainstorm and could not have been anticipated, or whether it was largely foreseeable due to changes since the 1983 flood. If it was a freak storm, unlikely recur in our lifetime, then rebuilding the same way in the same places would make sense. If the August 2016 flood, while extreme, represents the natural increase in flooding to be expected because of changes in development and weather since 1983, then we should rethink rebuilding and future development in these flood prone areas.
Why 100/1000-Year Rain Events Underestimate Risk
In the August 2016 storm, some areas received more than 20 inches of rain, and one received more than 31 inches over 48 hours. These were record events in some places. This was reported as a 0.1% per year rainfall – a 1,000-year event. The August 2016 storm resembled Tropical Storm Allison, which flooded parts of Texas, including Houston, and Louisiana in 2001, and the South Carolina flood of 2015.
There are three major problems in assigning probabilities to extreme rain events. The first is that there is less than 100 years of good quality weather data for most parts of the United States. Major cities may have data as far back as 150 years, but recently developed swaths of Louisiana are unlikely to have good data going back more than 50 years. To really know what a 100-year rain event is, you need at least have twice as many years of data – 200 for a 100-year event, 2,000 for a 1,000-year event – as your projection. If you have 100 years of data, you can be pretty sure about the 50-year event, but for longer period (lower probability) events, you have to use models to predict the characteristics of the missing data. For most places in Louisiana, even 50-year projects require extrapolation from a too short data set. As you move to longer ranger projections, 100/500/1000-year events, the projection depends more and more on modeling rather than actual data. Traditionally, the modelers assumed that the weather in the period before reliable weather data was like the weather of today. This assumption is called stationarity – that the system is stable and that the past and future resemble the present.
Looking back 1000 years, using tree ring data and other paleoclimate indicators, you see much greater variability in extreme weather events and even significant climate shifts. There were megafloods on the Mississippi River that dwarf the 1927 and 2011 floods, and 80 year droughts in the west. This leads to the second problem – how do you incorporate these extremes into the long term forecasts? This is a policy question as much as a scientific question. The more extremes you build into the model, the bigger the 100, 500, or 1000-year events the model will predict. The bigger the 100-year event, the larger the area of the 100-year flood zones on FEMA flood maps. Larger flood zones limit the amount of land that can be developed and the layout of cities. In highly flood prone areas, such as southern Louisiana, this could drastically limit single family detached home development and result in a huge expansion of areas requiring flood insurance. Assuming stationarity and ignoring these events suppresses the prediction of outliers – extreme weather events – to favor development. This is not a criticism of weather modelers. They are under tremendous pressure by politicians to minimize predictions of flooding risk. No politician would accept dramatically expanding flood zones because scientists see bad weather 1000 years ago in tree rings.
The third problem with 100/500/1000-year rain projections is incorporating climate change into the projections of extreme weather risk. While we still maintain the stationarity assumption for historic weather, it is clear climate change will affect future rain projections. Droughts will be worse and extreme rain events will be more frequent or more damaging. The August 2016 storm had so much moisture to draw on because the Gulf of Mexico was at near record temperatures. The warmer the water, the faster moisture will move to the air and the more energy there is available to drive storms. The warmer the air, the more moisture it can hold, and thus the larger the potential rainfall. While it is impossible at this point in time to know how much the August 2016 rainfall was affected by climate change, it is clear that these events will be more common in the future. (Even if the overall climate in Louisiana is drier.)
Why is Flooding Increasing?
Rainfall alone tells us nothing about whether and how much an area will flood. The consequences of flooding are determined by the topography of the land and how it has been developed. For example, if are at your beach house, it could rain for days and you would never flood because the water would run into the ocean. If you are in a slot canyon – a deep narrow canyon that collects all the runoff for miles, even a small amount of rain upstream can be fatal. The 18th century plantation houses along the Mississippi were only elevated a few feet above the ground, yet this was enough to protect them from flooding. When the Mississippi flooded in 1850, it spread across its floodplain and became miles wide, but only a few feet higher. It was only when the river was confined by levees that it would rise 10-20 feet or more during a flood. Floods happen less often, but a levee breach will be catastrophic – the old floodplain is now filled with houses at ground level and the torrent from the river will be many feet deep as it spreads through the floodplain.
As the area that flooded in the August 2016 flood was developed over the past 100 years, houses, businesses, and roads gradually filled in the watersheds of the rivers that drain the area. As areas were paved and forests were replaced with cropland and housing developments, rain would run off into the watershed rather than being absorbed. The same amount of rain would produce more runoff and increased stream levels after development. Feeder streams were blocked or restricted to culverts, reducing the ability of local regions to drain into the rivers. The rivers were restricted by levees and flood control projects, which increases flooding upstream as the river height increases because the river cannot spread through the floodplain. Highways and other infrastructure act as dams, flooding areas upstream. Flooding gets progressively worse for any given amount of rain because there is more runoff and less ability for it to drain. Local areas, down to the level of individual homes, which had never flooded before start to flood because their drainage is blocked and there is more runoff with the same amount of rain.
This is a problem for every community in the United States that is subject to periodic flooding. There are many communities with millions of homes that have built themselves beyond the drainage capacity of their watersheds and face the same difficult choices about development and relocation as does the area currently flooded in Louisiana.
Floodplains versus FEMA Flood Zones
Prior to 1968, there were few federal restrictions on development in areas subject to flooding. The primary regulation was the private insurance market. Flood insurance was often part of the homeowner’s coverage and if the house was in a high risk area, insurance would either be very expensive or unavailable. This was a significant limitation on residential development in high flood risk areas.
Hurricane Betsy flooded New Orleans in 1965 and there were several serious river flooding events around the same time. In the eyes of local communities and their elected officials, high private insurance rates were interfering with rebuilding and redevelopment after the floods. Congress responded by passing the National Flood Insurance Program (NFIP) in 1968. The original intent of the program was to provide affordable (subsidized) flood insurance for homes already at risk of flooding in exchange for communities preventing any future development in high risk areas. (The subsidy has persisted, but the land use restrictions have proven ineffective.)
The NFIP could have used the natural floodplain as the basis for the insurance. Floodplains are the areas that a river or creek will overflow into during floods. The river might overflow into the floodplain near its course every spring but might only reach the outer edges of the floodplain every 1000 years. Had the natural floodplain been used, there would be much lower flood damage in the United States and we would be much better positioned for climate change. But the natural floodplain would have required many people to buy flood insurance who had a relatively low risk of flooding (at least as understood at the time) and would have dramatically limited development in many communities. As a compromise, the NFIP created the legal concept of a flood zone.
Congress decided that the flood zone for subsidized insurance would be where the risk of flooding was high enough that people worried about it. The threshold was a 1% chance of flooding in a year, which is commonly called the 100-year flood. (This is about a 26% chance of flooding during a 30 year mortgage.) This would protect the people at the greatest risk for flooding while allowing development in the areas of the floodplain with less than a 1% risk of flooding in a given year. FEMA also maps the 0.2% (500-year) risk zone, but does not restrict residential development in this zone and does not require flood insurance in the 500-year flood zone.
FEMA issues Flood Insurance Rate Maps (FIRM) – now in digital form – that show the areas covered by the 100-year and 500-year flood zones. (There are other flood zones, but they are not necessary for this discussion.) Flood maps are developed by combining the projections of the runoff risk for a region with a detailed topographic map showing the elevations for individual homes and the local drainage patterns. Preparing a new flood map is time-consuming, expensive, and politically controversial if it expands the 100-year flood zone. When a proposed flood map expands the 100-year flood zone, communities will often go to court to block the release of the map. FEMA is under great pressure to compromise and reduce the size of the 100-year flood zone below that predicted by their models.
FEMA does not have the resources to include the impact of every impediment to drainage caused by development in the flood maps. It will also be lucky if it can update the maps every couple of decades. Thus flood maps generally underestimate the risk of flooding in developed areas, and the gap between the flood map and the real risk of flooding increases each year that the flood map is not updated. In rapidly developing areas like some of those flooded in the August 2016 flood, the flood map will be nearly useless in determining real flood risk. Areas that have been in the 0.2% risk area (500-year flood) may now be in the 1%, or even 2% (50-year) flood zone if the map were recalculated and could incorporate all of the drainage issues.
The problems with the accuracy of flood maps was highlighted by a recent study looking at claims filed by flood insurance policy holders. It found that there was no difference in the probability of claims on policies in the 100-year flood zone and those in the 500-year flood zone: someone living in the 500-year flood zone was as likely to be flooded as someone living in the 100-year flood zone. While there are many factors that likely contribute to this surprising finding, one is that flood zone determinations are inaccurate and/or out of date.
The requirement for insurance for property inside the 1% risk zone but not in the much larger 0.2% risk zone leads many people to believe that they are not at flood risk if they are outside the 1% risk zone. Even if the maps are correct about the projected risk, a 0.2% chance of flooding is a 6% chance of flooding during a 30 year mortgage. This is higher than the 4% risk of a structural fire in the house, but no one questions the need for fire insurance.
Was the August 2016 Flood a Freak Storm or the Future?
While the August 2016 rain event set records in many places, it is impossible to say whether it is a 0.1% (1000-year) event for the purposes of deciding on rebuilding and relocation. Such a prediction will contain the uncertainties of incorporating paleoclimate data and the future effects of climate change. Given the bias in the modeling, it is certainly less than a 1000-year event. A risk-averse person might rationally decide that once you have flooded, you are going to flood again unless you relocate, elevate above the flood level, or there are massive changes in the drainage system. (Massive changes in the drainage system are not going to happen quickly, if at all.) It is clear that whatever the current risk, it will continue to increase through time.
If FEMA just pays out the emergency grants and the proceeds of the flood insurance policies, rebuilding will be under existing FEMA rules. These include some restrictions on rebuilding if a home is in the 100-year flood zone and is more than 50% damaged. The more important question is whether FEMA will reevaluate the flood maps for the region and expand the 100-year flood zone or put additional elevation restrictions on rebuilding. A FEMA study projected substantial saving over a ten year period if houses are built or rebuilt with two feet of free-board – two feet of elevation of the top of first floor above the base flood level. This will likely depend on whether Congress makes a special appropriation as it did with the Road Home Program for New Orleans after Hurricane Katrina.
Whatever FEMA decides to do, everyone who flooded should consider that there is some risk of future flooding. At a minimum, property owners who flooded should buy flood insurance. While relocation and elevation are options for severe flooding, there are many resilient construction options for homes with more limited flooding. These include using water resistant materials on lower floors, such as tile or concrete rather than carpet or wood floors, moving critical utilities higher in the house, and other strategies to minimize the damage from flooding.
Who is to Blame?
The mayor of Walker says that he will file a lawsuit against the state because the new I12 expansion dammed up the drainage for Walker. (Such lawsuits have been successful in the past, although the state refused to pay the judgement so the plaintiffs did not receive any compensation.) From a land use perspective, the primary problem was that local planning and zoning commissions have known that the areas are prone to severe flooding for at least the 30 plus years since the 1983 flood. Homeowners and small businesses were not warned about the risk and were not advised to buy flood insurance if they were outside the FEMA 100-year flood zone. They relied on planning and zoning commissions to determine if developers were building in safe locations.
While local officials are already blaming the Army Corps of Engineers for not building canals and other flood control structures, it is not the duty of the Corps to protect communities that build in high risk flood areas. The Corps only builds flood control projects when Congress directs it to by appropriating the money to build the projects. If a community believes it needs a flood control project, it should stop development in the area until it can raise the tax money to finance the project and get it built. Developers who disrupt drainage patterns also have some responsibility, as does the State when it builds highways that are dams.