Forecasting the Unpredictable: Water Planning Under Deep Uncertainty

“It’s tough to make predictions, especially about the future.” -Yogi Berra

Predicting the future is difficult for water planning. One side of the uncertainty coin is the weather. What will it be like in the future, and what is the timeline? Another side of the coin is human behavior. How many people will be here? How much water will they need? Will they do what they need to do to ensure there’s enough water? Climate change amplifies and expands these weather and people factors. How will a warming climate affect temperature and the amount and timing of precipitation? How will people adjust their behaviors to minimize temperature increases? And how will a warming climate affect the countless interwoven natural processes on the planet?

To plan, you need to know what you are planning for. In Texas, we plan for a repeat of the drought of record. This approach has a lot going for it, namely that the drought of record is a concrete event that has occurred in the past. With stationarity (climate without a warming or cooling trend) and some statistical assumptions, you can use the record to quantify the probability that the drought of record will occur again. However, the “record” is focused on the period of time in which we have consistent measurements of rainfall and/or streamflow, which is about 100 years, give or take a decade or two (or three). Furthermore, one hundred years is just a blink of an eye in the 11,000 years of this period of history we call home (the Holocene).

Indicators for rainfall, such as tree-ring thickness, allow us to extend the rainfall record deeper into the past. And these tree rings suggest that we’ve had far worse droughts.  For example, a study of tree rings in Texas that reached more than 1,000 years ago showed that decade-long droughts occur, on average, once every 100 years and that at least four mega-droughts, droughts of 15 to 30 years, have occurred over the last 1,200 years. We don’t need global warming to have a drought worse than the drought of record.

With global warming, the climate is not stationary: it is getting, on average, warmer every year. In Texas, that probably means less water running into rivers and lakes, less water refilling aquifers, and longer and stronger droughts. Using the drought of record in a non-stationary, warming world means we are always behind in ensuring water supplies for the state. But how do we plan for that?

The International Panel on Climate Change employs different warming projections to include various possibilities for the future, with each projection having its own cloud of uncertainty around it. Historically, these have been called representative concentration pathways, commonly abbreviated as RCPs. More recently, these have been replaced with shared socioeconomic pathways, or SSPs (which are, in my honest opinion, an ill-advised, and sometimes unintentionally comical, attempt to merge political storylines into concentration pathways). Because we are in a transition period from one (RCPs) to the other (SSPs), you will see both frolicking amongst the literature and popular press.

RCPs and SSPs, while slightly different, are essentially the same. You will almost always see them written followed by a number: RCP-8.5, SSP-4.5, SSP-1.8, among others. That number is the change in radiative forcing between 1750 and 2100 in watts per square meter. Radiative forcing is the difference between the amount of energy that hits the Earth from the sun minus the amount of energy that reflects off the Earth and back into space. Therefore, the larger the number, the more heat is retained in the planet’s atmosphere.

Figure 1 shows the historical average global temperatures and where we might be going in the future under different projections of radiative forcing. I’ve included a conversion from metric to FreedomUnits® (Imperial Units, i.e. Fahrenheit) for temperature. The halos of color or grey around the lines (“the spread”) show the uncertainty of measurements and projections, the latter as reflected in running the various climate models for the different scenarios. You might notice another number tacked on to the SSP before the dash—this ties into the socioeconomic scenario: 1 is sustainability, 2 is middle of the road, 3 is regional rivalry, 4 is inequality, and 5 is fossil-fueled development. While RCP/SSP-8.5 is often characterized as a worst-case scenario, things can get worse.

Figure 1 Historic global temperature averages and projections shown in Celcius and Fahrenheit

Figure 1. Historic global temperature averages and projections shown in Celcius and Fahrenheit.

OK. So perhaps the best way to include climate change in water planning is to identify the path we are on and then plan accordingly. Climate Action Tracker provides an estimate of where we are going based on “policies and action” and “pledges and targets” (see below). First, note that we’re doomed to keep warming below 1.5° C (2.7° F). That fried chicken will soon fly the coop, but there’s still hope for staying below 2.0° C (3.6° F), the upper limit of the Kyoto Protocol. But for planning purposes, let’s be conservative and go with the more solid “policies and action,” recognizing that these rely on notoriously unreliable human decision-making. That currently places us between 2.5° and 2.9° C (4.5° and 5.2° F) of warming by the year 2100. Comparing to the plot above, we see that it puts us on the RCP/SSP-4.5 line.

Figure 2 Line graph outlining warming projections in relation to global emissions from 1990 to the year 2100.

Figure 2. Line graph outlining warming projections in relation to global emissions from 1990 to the year 2100.

So far, so good, right? Wrong. The climate projections in Figure 1 assume a well-behaved climate: we turn the temperature up on the stove, and the resulting warming follows a smooth, predictable, almost pleasant curve. However, that is a best-case scenario for whichever RCP/SSP scenario we may wander down. Climate change is more like my cat, Clara (named after the thereminist, Clara Rockmore). As part of our morning ritual, Clara will entice me to pet her; however, she is one of those unpredictable cats who will suddenly become overstimulated and angrily lash out with her murder mittens. Given that I’ve known Clara for upwards of two years now, sometimes I can pet her without facing her wrath, and other times, after one stroke, she violently conjures her inner Demon Cat. I still don’t understand what sets her off.

Figure 3 The adorably unpredictable Clara the Cat.

Figure 3. The adorably unpredictable Clara the Cat.

Similarly, the climate also has trigger points that we don’t fully understand and that the models do not include in a practical way. This is why so many scientists are worried about any increase in temperature: the next fractional jump could trigger an event that amplifies warming or causes a substantial change to the climate and water cycle. Even worse, hitting a trigger could cause cascading events, one event triggering another event, and so on. The figure below shows a meta-analysis (an analysis of analyses) of tipping points with ranges and central estimates.  Some of these tippers (such as permafrost thaws and rainforest dieback) will cause rises in greenhouse gas emissions or retention, resulting in additional, potentially instantaneous, warming.

Figure 4. Note that the current level of warming (2023) is 1.36° C.

Figure 4. Note that the current level of warming (2023) is 1.36° C.

This world of warming, tipping points, and cascading events is the world we live in as water planners (and citizens of the planet). Humanity is conducting a global-scale, real-time experiment on the Earth, and we are both the observers and the rats in the maze.

Frustratingly, it’s impossible for actuaries (analysts) to determine the probability of attaining any given climate future. In the water world, people want (and often delude themselves that they have) a 100 percent reliable water resource. Even without climate change, that has not been the case in most cases. With climate change, the certainty of your water resource—if you know it—is most likely decreasing as each year trickles by. And it could suddenly decrease catastrophically. In the academic world, we call this immense uncertainty “deep uncertainty,” which is really just a hoity-toity way of saying, “We have no freakin’ idea what’s going to happen.”

The National Oceanic and Atmospheric Administration, through the U.S. Climate Resilience Toolkit, defines deep uncertainty as “when decision-makers and stakeholders do not know or cannot agree on how likely different future scenarios are.” I don’t like the focus this definition has on decision-makers and stakeholders because the scientists don’t know either. No one knows. So, folks have invented something called “decision making under deep uncertainty (DMDU)” under the principles of (1) planning should consider multiple futures, (2) robust plans perform well over multiple futures, and (3) plans that are flexible and adaptive perform better (U.S. Climate Resilience Toolkit).

Austin followed a decision-making-under-deep-uncertainty approach with its own water planning that winces 100 years into the future and incorporates climate change. Austin included an array of projections from multiple pathways, including something of an uncertainty cloud around RCP/SSP-8.5. This could be considered over-planning for a well-behaved climate since, as discussed, we are currently on the path for RCP/SSP-4.5. However, with unknown-yet-possible triggers and cascades, Austin is ready for far more than our current understanding of RCP/SSP-4.5.

Although planners consider a range of scenarios with decision-making under deep uncertainty, it’s ultimately the most severe scenario that must be planned for (although an ensemble of scenarios that are worst in their own way could make up “the most severe scenario”). Regardless of which scenario we use, we don’t know its probability: too much is not known to quantify this. However, there is one place on the uncertain uncertainty distribution that we know the probability of: that of 100 percent certain water.

A firm yield for a reservoir provides 100 percent of the volume 100 percent of the time, based on the period of record, which we already know is not 100 percent certain. Certain groundwater resources could be considered 100 percent certain over a period of time (until the economics of producing that groundwater dry up) or 100 percent certain if produced sustainably and recharge is independent of climate (generally the case with the confined parts of sandy aquifers). This is why climate folks often see groundwater as a “cure-all” for uncertainty, but that cure-all only pans out when groundwater is actually managed to achieve long-term goals. Seawater desalination is another source of 100 percent reliable water, assuming infrastructure remains viable (ditto for groundwater). Water sources that depend on precipitation and are victimized by drought—such as rivers, lakes, and karst aquifers—are intrinsically not 100 percent reliable and may be far less reliable under climate change.

So, what should water planners and communities do? At a minimum, make sure you are ready for a repeat of the drought of record: according to the state water plan, many communities in Texas are not. Don’t be lulled by the drought of record. One danger of water planning and setting a target is the false sense of security that will be enough. As a baseline, make sure you understand what warming does to your water resources under SSP-4.5 and SSP-8.5 and plan accordingly. And for the gold star, think about what you would do when your fickle water resources—those affected by drought—are no longer available. Are there things you can do now to ensure resilient water for the future? What is Plan B, and when would you need to start it? Though expensive, building truly resilient water resources is priceless; not having resilient water is the most expensive water there is.

Figure 5. Effects of drought on the Blanco River, 2011. Photo by Earl McGehee.

Figure 5. Effects of drought on the Blanco River, 2011. Photo by Earl McGehee.

Leave a Reply

Your email address will not be published. Required fields are marked *