As the Caltrans signs say, California is in a serious drought. And last Wednesday Governor Brown mandated California's first ever statewide mandatory water reductions.
There's a challenge though. How do we implement that 25% aggregate statewide goal and do justice to the widely different local circumstances around California and diverse array of actions already taken by local utilities?
Water is complex, and simple monthly gallons per capita used per month do not tell the whole story. Many utilities don't even read meters on a monthly basis and many others will read meters anywhere from every 20 to 40 days.
And as the State Water Resources Control Board, the body charged by the Governor with implementing those reductions, say in a bold on their drought data page, there's a wide array of other factors that need to be taken into account before making comparisons based on the data.
Yet the State Board website sys admins seem to have forgotten to forward this page to the staff that wrote the proposed regulations. Their conservation targets do exactly what their website says not to do: make direct comparisons between utilities based on residential gallons per capita per day (R-GPCD) in order to implement the Governor's mandate.
The shortcomings of monthly water usage data
Beyond ignoring all the complex factors the State Board mentions on their website, this framework is problematic for some pretty basic reasons. As mentioned before most utilities don't read the same number of meters every month so there's inherent noise in this monthly residential GPCD data.
Second, months fundamentally are a human construct in a solar calendar system. There's no physical logic like the time it takes the earth to go around the sun for months -- rather months have more to do with the number of Gods Julius Caeser felt like honoring after conquering Egypt. So as you'd expect, there's huge statistical variation inherent in the State Board's monthly data.
In fact, there's three times as much variation in the monthly State Board data as historical annual usage data. That's why you get all sorts of weird outliers like the 28 instances where utilities had year over year percentage changes over 50% in the State Board monthly data.
Not to pick on East Palo Alto but there situation is illustrative of the larger issues with using just monthly data. They actually used 11% more water June 2014 through Feb 2015 than June 2013 through Feb 2014 though is in the lowest SWRCB conservation category since they had a low Sept 2013 R-GPCD.
Yet have no fear! There's an annual data source!
As you may have picked up from reading the fine print on that histogram, the California Department of Water Resources has fifteen years of historical water usage data from utilities.
What's more is they actually spent over three years developing 20% statewide water reduction targets with countless expert hearings rather than the roughly 3 weeks allotted to debate the State Board's proposed regulations. In addition, DWR's utility by utility level targets address the wide array of complex factors mentioned above.
But wait there's more!
That data already comes with a robust conservation framework!
These targets were initially designed to be achieved by 2020 yet interestingly enough California's water utilities already achieved those targets on a statewide basis achieving a statewide GPCD of 153.6 annualized on available 2014/15 data.
That's a smidge lower than the 154 GPCD statewide GPCD targeted by DWR in February 2010 and note that's the statewide number that everyone from enviros to ag to the state legislature to water managers agreed on (quite the rarity in the water world). In the aggregate, the state's water utilities have met those targets.
So why not just say all utilities need to go beyond those community specific 2020 targets to a degree weighted by their current R-GPCD?
That deals with the complexity inherent in California water and asks those who haven't taking effective conservation actions to do more to catch up.
Since ARGO is nothing if not a collection of civic minded data nerds, we took the liberty of joining the DWR and SWRCB data and calculating the utility by utility water reductions to achieve an aggregate statewide reduction of 20%.
You can find an interactive dashboard mapping the results and showing residential usage trends for the analyzed utilities. In addition, here's a full excel workbook detailing all of our results in case you want to play with the numbers yourself. [UPDATED]
In running those numbers, we made a couple assumptions:
- Every community conserves a percentage beyond their 2020 targets to a degree weighted by their 2014/15 R-GPCD. [UPDATED]
- [2020 target ]*[conservation factor ]*(1-([annualized 2014/15 R-GPCD] - [40 GPCD floor])/ [Max annual R-GPCD statewide])
- No community must conserve more than 40% water since above that you're talking about restricting water for just health and human safety uses.
- (60%)*[ 2013 annualized GPCD]
- Every community must conserve a minimum of 10% since we're all in this together as Californians.
- (90%)*[ 2013 annualized GPCD]
- There's a flat floor of 40 GPCD built into both how far communities have to go beyond their 2020 targets and as a general floor for targets (no one hits the latter).
For the mathematically inclined, please see here for the formula used to calculate the 2015/16 annualized GPCD necessary to get to the governor's statewide 25% urban water use reductions:
And that's it! We recommend using a rolling average (preferably at least 3 months) in tracking how well utilities are doing in meeting those targets to smooth out the inherent noise in monthly data. Beyond that we'd just ask the water community to remember history.
Philosophically, we create plans so that we have something in place and don't need to generate something in the middle of crisis a crisis. So why not use the plans we have to implement the Governor’s mandate?
The last time California made natural resources policy from scratch in a rushed crisis environment we got the Enron contracts. So let's approach this challenge of implementing the Governor's mandate, both through these regulations and the actual you know actions to improve conservation, with clear eyes and level heads.
We need your help
The State Board will decide on final regulations by May 5th or 6th so please share this post if you believe California needs effective data science to tackle the drought.
We also recognize that we're just a few data scientists looking at the problem from 3,000 miles away in NYC at CUSP.nyu.edu. So we'd would greatly appreciate feedback from the water community on how we could incorporate the water community's expertise in improving this analysis.
If past is prologue and the climate models predicting less Sierra snow pack in the future are correct, we can expect California's water challenges only to increase in the future.
At least 20% of team A.R.G.O. plans to return to California soon and the rest of us are just as much concerned with humanity's ability to cope with climate change as any scientifically sensible human, so we'd like to do our part to help California cope with this crisis.
Please let us know what we can do to help with the inherent measurement challenges in fulfilling the Governor's mandate.
 For instance less than 10% of the utilities tracked by the State Board don't have data from DWR since they didn't participate in the Urban Water Management planning process. Any good ideas on what do with those folks? A simple across the board 25% reduction there would be our naive baseline suggestion.
 There was and still is, however, a good deal of debate about some of the methods for calculating the utility level 2020 targets.
EDITED: An earlier version of this post titled "Why not use California's existing water conservation plan as a framework to fulfill the Governor's mandatory reductions?" suggested that "Every community conserves the same percentage beyond the 2020 targets (24.3% since we're slightly below the 2020 targets already)."
We listened to several very persuasive members of the water community who argued that even if you're under your 2020 target, if you have a high (say 200+ R-GPCD) there's still a lot you can and frankly should do to conserve.
And we agree so we've updated our analysis to reflect that input. Now ARGO's proposal weights how far each community goes beyond their 2020 targets by their current annualized 2014/15 R-GPCD.