Why do hotter, drier areas struggle to meet their water use efficiency targets?
Irrigable-not-irrigated area drives visible trends in WUO exceedance.
TLDR: When using provisional data to model the impact of draft water use efficiency regulations, water suppliers in higher-ET areas are much more likely to exceed their objective— at least when focusing on the residential (largest) portions of the regulations. This trend almost disappears when 100% of measured “irrigable-not-irrigated” (INI) landscape area is included.
The California Data Collaborative has been estimating the impacts of the State’s Framework for “Making Conservation a California Way of Life” and helping water suppliers to prepare since 2016 when the framework was still an executive order and had not yet been signed into law. In the years since, having spoken to dozens of urban water suppliers and having helped more than 20 suppliers calculate their urban water use objective (WUO), one observation kept sticking in my mind: suppliers in cooler/coastal regions often had water use levels at or sometimes well below their WUO, while those in hotter, more inland regions seemed more likely to exceed their WUO — sometimes dramatically. The question in my mind was: why?
Background
Per the legislation in AB1668, the structure of the WUO formula is designed to explicitly control for several key drivers of water consumption (population, landscape area, climate, etc.) with the hope that whatever variation left over actually reflects “efficiency” in water uses. The legislation puts it this way:
“The Legislature finds and declares that this chapter establishes a method to estimate the aggregate amount of water that would have been delivered the previous year by an urban retail water supplier if all that water had been used efficiently. This estimated aggregate water use is the urban retail water supplier’s urban water use objective. The method is based on water use efficiency standards and local service area characteristics for that year.” [emphasis mine]
The legislation further calls to “[incorporate] precipitation data and climate data into estimates of a urban retail water supplier’s outdoor irrigation budget”. In recommendations and draft regulation so far, the State has chosen to use measures of effective precipitation (defined as the amount of precipitation that remains in the soil and is available for use) and reference evapotranspiration (or ET₀ which is the sum of evaporation and transpiration from plants) as the local service area characteristics that reflect climate.
As a data scientist (not a lawyer), my personal interpretation of this is that, if the framework equation truly controlled for the effects of climate on water consumption, no correlation between WUO exceedance and ET₀ would be expected. My own experience seemed to hint that such a correlation did exist, but until recently there was no easy way to test this using publicly available data.
Analysis
Fortunately, anyone now has the ability to dig into estimates of the WUOs for every urban water supplier thanks to provisional data disclosed by the State Water Resources Control Board (Water Board). I would caution anyone reading that this data is provisional. It is compiled from public data originally reported for a completely different purpose, so it may not accurately reflect the situation for every urban water supplier.
However, the data was deemed accurate enough to be used by the State to model the effects of the draft regulation and inform a large number of important decisions and recommendation for setting specific water efficiency standards, so it should be good enough for our purposes. And what we find when digging in is that contrary to our hypothesis, there appear to be clear trends towards increasing levels of projected WUO exceedance in higher ET₀ areas.
A January 4, 2024 report from the California Legislative Analyst’s Office (LAO) titled Assessing Early Implementation of Urban Water Use Efficiency Requirements shows this trend in the context of geography. Water suppliers in coastal hydrologic regions show much smaller average water use reductions required to meet their WUO under proposed 2035 standards.
These inland areas are also areas that tend to have higher annual ET₀. We can look at trends in WUO exceedance with respect to ET₀ in the charts below, prepared by CaDC. The charts show the “WUO ratio”, defined as qualifying water use divided by the WUO which is a way to show how close each water supplier is to meeting their WUO. A ratio greater than one indicates exceedance of the WUO while a ratio less than one indicates consumption within allowed levels. This ratio is plotted against annual ET₀. Each point represents one water supplier in the year 2021.
Only residential water consumption and the residential (indoor and outdoor) portions of the WUO are considered for this analysis because the focus of the analysis is on the outdoor portion of the WUO, and no high-quality landscape measurements are yet available for commercial, industrial, and institutional landscapes.
Figure 1 shows the WUO ratios with only the square footage of Irrigable Irrigated (II) areas included. II areas are areas that appeared to be actively irrigated in the aerial imagery used by the State to classify urban landscapes. The current draft framework proposal is to only include II areas after June 30, 2027. In Figure 1, a clear trend is visible where 10 inches increase in annual ET₀ is associated with, on average, an 11.7% increase in WUO ratio (more suppliers exceeding their WUO), with ET₀ alone explaining 18% of all variance in the data.
Figure 2 shows the WUO ratios with 20% of INI area included. INI is defined as landscaped area that could be irrigated, but did not currently appear to be irrigated in the aerial imagery used by the State do classify urban landscapes. The inclusion of 20% of INI area is the level allowed for in the current draft regulation for suppliers exceeding their WUO, but this inclusion of 20% INI is only allowed until June 30, 2027. In Figure 2, a clear trend is still visible where 10 inches increase in annual ET₀ is associated with, on average, an 8.6% increase in WUO ratio (more suppliers exceeding their WUO), with ET₀ explaining 10.6% of the variation.
Figure 3 shows the WUO ratios with 100% INI included. The trend is reduced so that 10 inches increase in annual ET₀ is associated with, on average, a 3% increase in WUO ratio (more suppliers exceeding their WUO) with ET₀ explaining only 1% of variation in the data. In other words, the trend has almost disappeared.
Discussion
As discussed previously, if the goal of the framework is truly to control for the effects of climate (and other factors like population, landscape area, and water losses) on water consumption, a clear correlation between WUO exceedance and ET₀ would most likely not be expected. An important exception would be if there are other, unobserved, drivers of residential water use that are correlated with ET₀, such as income or irrigation practices.
Unfortunately, it would take a much more detailed study to fully determine all of the drivers of WUO exceedance (anybody want to collaborate?). In the meantime, we can only use our prior knowledge to speculate on the cause of this correlation. There are many possible causes, but I am including two of my personal favorite hypotheses below. These are just guesses!
- Irrigation of “not irrigated” areas — the Department of Water Resources’ (DWR) September 2022 “Recommendations for Outdoor Residential Water Use Efficiency Standard” finds likely non-zero amounts of irrigation occurring on landscapes classified as INI. This is the source of the recommended 20% INI buffer. However, as shown above, including 20% of INI in the objective reduces the strength of the correlation but does not remove it. One possible explanation for this could be that the amount of irrigation being applied to INI areas is not a static percentage for all suppliers, but rather increases with annual ET₀. Regions with higher annual ET₀ may have landscapes that can absorb more irrigation without becoming visibly green and appearing as “irrigated” in imagery. This could be a result of different plant choices, or simply higher temperatures drying plant life more quickly.
- Not-irrigated areas classified as “Irrigated” — the converse of the first point is that areas with lower ET₀ may be more likely to have landscapes that are naturally green, even with minimal to no irrigation. This could inflate aerial measurements of irrigated area, causing them to be larger than the area that is truly being irrigated. This would in turn increase the WUO for areas with lower ET₀, making it easier to keep water use below the objective.
Interestingly, the previously mentioned LAO report includes the following among its recommendations that the legislature could consider:
Maintain Flexibility in Calculation for Irrigable Landscapes, Allowing Inclusion of Some Landscapes That Are Not Currently Irrigated. One of the components suppliers must use to calculate their WUOs is the square footage of landscapes that are irrigated. As described earlier, the proposed regulations would allow suppliers some wiggle room in calculating this factor — they can include up to 20 percent of the landscapes that ostensibly are not irrigated currently (based on DWR data), but are the type of landscape that could be irrigated in the future. However, the proposed regulations only would allow this data flexibility through June 30, 2027, at which point no land area that appears unirrigated could be included in the calculation. The Legislature could consider allowing this data buffer on an ongoing basis (as DWR had recommended in its report), given that there could be many reasons an irrigable landscape might not be — or might not appear to be — irrigated currently but could be in the future. This would help suppliers comply with requirements as it would… increase the WUO and it provides a reasonable buffer given uncertainties around the precision of these data.
While keeping the 20% INI buffer would indeed help suppliers comply with the WUO requirements, the analysis in this blog post shows that even a 20% buffer would not eliminate the trends that seem to favor suppliers in areas with lower ET₀.
Conclusion
The bottom line is that, without INI area included in the framework, there seems to be a clear trend working against higher ET₀ areas. Including a large enough fraction of INI in the framework reduces this trend dramatically, bringing it close to zero.
