[Update on September 7, 2017 – Added National Academy of Sciences Report on Attribution of Extreme Weather Events in the Context of Climate Change]

I wonder if Harvey was caused (or made worse) by climate change?

Now that I have your attention….

I previously have written a bit on whether or not certain studies of rainfall history suggest that extreme events are becoming larger or more frequent.  Here’s a quote from my op-ed in Trib-Talk from February 2017:

Our region has also been looking at the effect of climate change on rainfall patterns. In March 2016, Harris County joined a national study by the National Oceanic and Atmospheric Administration [NOAA] to recalculate the 1 percent storm size for Texas using rainfall records from a longer period of time. Similar work completed for the southeast and southwest regions of the country [by NOAA] showed that rainfall depths over time did not have statistically significant trends up or down — i.e., no climate-change effect was observed on those types of storms.

Since I reported on what NOAA did in New Mexico and Louisiana, many folks have apparently taken this to mean that I don’t believe in climate change. To set the record straight, here’s what I think about climate change and how we might deal with it.

Climate change will require us to do three things, in amounts that relate to each other:

1. Mitigate: Reduce the amount of carbon we emit. Hopefully we will do this using market forces and creativity. This might be done by establishing a cap on emissions and a price for the right to emit a certain mass of carbon. This would then create a business case for reducing carbon emissions, sequestering carbon, or otherwise controlling carbon and then selling the right to emit some.
2. Adapt:  Change our planning and design approach for infrastructure, development, and just about everything else we do, to account for anticipated future conditions. This could be wind speeds, rainfall intensities, rainfall depths, rainfall frequencies, sea levels, solar radiation, temperature, etc. Think where and how we grow food. How we get around.  How we plan development.  The size and materials used for infrastructure.  This also might also include retrofitting existing systems.
3. Suffer: Deal with the social, environmental, and economic consequences of the anticipated changes, many of which will be negative and will be felt mostly by lower income people. This might be water scarcity, flooding, new diseases, sea level rise, storms, more costly food, more costly transportation, etc.  Of course suffering will cost money also.

The amount of energy, thought, and time we put into each of these three things will depend on how much energy, thought, and time we put into the other two.  If we don’t mitigate or adapt, than suffering will increase.  If we mitigate a lot, then the required adaptation effort will go down a bit and the suffering will diminish.  This is not a new way to conceptualize our situation. I’m stealing this from some other much more well informed people.

Now, lets focus on development and drainage in the Houston area.

If you review historic rainfall amounts up through today, you can see statistically significant increasing trend lines for smaller rainfall amounts. For example, the number of days per year experiencing more than 3″ of rain is increasing over time. This suggests that the number of 12″ days or 13″ days (about equal to the current 1% annual chance event) is increasing, even though those events are so rare we can’t see the trend in only 70 years of data. That’s because, by definition, we would expect to only see one such event in 100 years of data. Hard to see a time series trend in one data point, right?

Now if we see increasing trends in 3″ days (which are very common), 4″ days, and 5″ days, then we can extrapolate the data for both smaller storms and larger storms out to any future year to see what kind of rainfall we can expect and what kind of stormwater management and floodplain management we should undertake. Thanks to Matt Berg and Katharine Hayhoe for this insight.

So in Houston we currently design stormwater systems to handle 12″ days, which is the depth of rain that has a 1% chance of occurring every year (based on the last time we calculated this value).  Let’s say we update our analysis of rainfall data and we find that the current 1% annual chance is actually a 14″ day.  Let’s further assume that we estimate that the 1% annual chance event in the year 2067 will be 16″ day (based on our extrapolated increasing trend line).  If these assumptions were true then we have the technical information to inform some important policy choices.

First, we would need to decide what level of risk we would like to use to design new infrastructure and developments. Perhaps we would want to change from a 1% annual chance level of service to a 0.5% annual chance level of service? This would mean that our designs would need to accommodate more rain in a single day. Building to a lower risk level would require a higher initial investment, but might reduce overall life-cycle costs, especially if repair and replacement probabilities were factored in.

Second, we would need to decide, based on the facility’s design life, what future conditions we should consider. A facility in operation in 2067 that needed a 1% annual chance risk level should be designed to handle the hypothetical 16″ day mentioned above.  A facility designed for a 10 year life might be designed to handle the current or short term predicted increase in the 1% annual chance rainfall.

Lastly, and perhaps most importantly, we would need to find the money to invest in retrofitting or modifying the older portions of our city that don’t currently meet our existing 1% annual chance rain event, much less the 1% annual chance event predicted to occur in 2067.

We can figure this stuff out. We just need to do it.

Oh, what about Harvey, which delivered four day rainfall totals depicted in the image below, from Harris County Flood Control District’s Flood Warning System.

Well there is an entire science used to determine how to attribute a particular outcome to certain input variables. Attributing any particular event to human-caused climate change depends upon the time scale and geographic size of the event being studied. I direct you to this paper about whether the August 2016 flooding in Louisiana was due to climate change and to the National Academy of Sciences report Attribution of Extreme Weather Events in the Context of Climate Change for additional insights and details.

I tend to think that human-caused climate change increased the probability of Harvey occurring, but I personally don’t have a way to estimate the magnitude of that increase. Perhaps some statistics experts and climate scientists can help us out?

[Update on September 7, 2017 – Added National Academy of Sciences Report on Attribution of Extreme Weather Events in the Context of Climate Change]