Since launching in January of 2016, the California Data Collaborative has made substantial progress towards its long term vision of integrating the entire lifecyle of water use data.  Focused on the historic water efficiency actions taken by water managers in the drought, our phase one pilot has focused on urban customer water use to accelerate learning what works at the local level to inform smart statewide efficiency.

First, it will accelerate the measurements that water managers need to effectively manage water demand. Currently, research into the effectiveness of price, messaging, workshops and other water demand management questions are too often tackled by looking at old research surveys and doing quick back of the envelop calculations rather than looking at actual analytics customized to a utility's unique service area.

Second, California currently collects a variety of reported utility-level water usage metrics through the State Water Resources Control Board conservation program (monthly averages since 2013), Department of Water Resources Urban Water Management Plan program (annual averages last updated in 2010), and the Clean Drinking Water program.  Our data infrastructure will streamline the time intensive reporting requirements mandated on local water utilities by the state through automated queries and table creation.  

Third, this data infrastructure will enable the deployment of targeted marketing and customer engagement strategies to influence conservation.  This sort of behavioral science is standard practice in web companies and is becoming increasingly prevalent in government as evidenced by Obama’s executive order on September 15, 2015.

Together this new data infrastructure enables a new and more modern way to manage water as we outlined in our feature June 2015 paper in the journal of the American Water Works Association on how data can help utilities adapt to future water uncertainty.

A simple, scientific approach to efficient water use

 
Graphical illustration of the MWELO targets.

Graphical illustration of the MWELO targets.

 

The Model Water Efficient Landscape Ordinance (MWELO) targets provide a scientifically determined allocation of reasonable indoor and outdoor usage. These customized targets provide and effective and equitable framework for conserving our state's precious water resources.

To demonstrate how the state currently performs with regards to the MWELO targets, Data Collaborative staff prepared an estimate of efficiency targets for 340 large water agencies across the state. The chart below compares our aggregate, estimated efficiency target against actual aggregate water production for these utilities.

  • Blue line -- actual residential production
  • Black line -- estimated efficiency target with error bounds

As you can see, in aggregate, California's urban areas met MWELO efficiency targets in 2015 according to our estimates. However, approximately 30% of individual urban retailers did not meet those benchmarks.   

Methodology

The MWELO performance standard is calculated using the formula for existing development:

55 * [Population] * (365/12)  +  0.8 [Monthly ET] [Irrigable Area] 0.62

The variables in brackets are utility specific data aligned with the assumptions outlined above.  55 represents indoor usage of 55 gallons per capita per day and multiplying by 365 and dividing by 12 is used to convert that to monthly usage.  0.8 represents the conservation factor used for current development in MWELO and 0.62 is used to convert into gallons.

1.         This analysis uses the MWELO formula for existing residential and does not factor in the different new development calculation. 

2.         Irrigable area is estimated for each tax parcel using a statistical model incorporating parcel size, population density and neighborhood income. Specifically, a linear regression model was fit to actual measured irrigable area obtained from participating agencies. Variable selection was performed by comparing error scores obtained through five-fold cross validation. This approach yields approximately 70% predictive accuracy after aggregation at the agency level.

3.         Evapotranspiration (ET) for each month is calculated as the inverse distance weighted average of the ten nearest CIMIS stations.  Monthly ET is calculated as the sum of daily ET reported by the CIMIS during that month.

4.         Residential production is simply the total production for each agency multiplied by the reported percent residential usage. These data are both publicly available in the State Water Resources Control Board (SWRCB) conservation reporting data.

5.         All water production data is in gallons and is pulled from the SWRCB statewide conservation reporting excel spreadsheet.  340 agencies out of the 411 agencies in the SWRCB database matched with our statewide parcel shapefiles.

California has spent over half a billion dollars in cash rebates for home owners to tear our their lawns, the largest such rebate program in American history, yet these program's true impacts and the path to turf market transformation remain unclear.  One of the Data Collaborative's first priorities is to determine the efficacy of these turf rebate programs, as well as to explore alternate cost effective strategies for making California friendly landscaping mainstream.  Buying every square foot of lawn would run in the tens of billions of dollars, and a pragmatic goal involves a combination of encouraging efficient irrigation practices and focusing landscape conversions on ornamental rather than all lawns.

Usage and Irrigable Area by Usage Percentile

The height of the bars shows the percent of total single-family residential (SFR) usage consumed by a given percentile of customers. The bar area shows the percent of total SFR irrigable area within that same percentile. For example, in MNWD the top 5% of water users collectively use over 16% of all SFR water, and have 23% of the irrigable area.

 Difference from Non-Participating Median Usage by District

To determine the true water savings from a turf rebate, it is not enough to only look at usage before and after a lawn is removed. The difference-in-difference approach shown here is a simple way to control for some of the variation in water usage over time resulting from non-rebate factors such as weather, public education efforts, rate shifts, increased media attention on the drought in 2015, and other conservation actions.  

To create this chart, the water usage of the median non-participating SFR customer within each utility was subtracted from the usage of the median customer participating in a turf rebate. This acts to normalize for external, non-rebate factors because non-participating accounts were equally subject to those external effects. This difference is then shown over time relative to the point when the rebate action is finalized (shown on the chart as 0 Months Since Rebate).

One can see that turf rebate participants tend to be above-median consumers before their rebate and below-median consumers afterwards, although these effects can vary greatly between utilities.


With lower water sales in the drought, many water retailers are struggling financially.  Fixed revenue has big implications for water utilities in times of declining water sales.  This quick quiz offers a way to polish your knowledge on fixed and variable terminology.  

Below we offer a simple tool to help utilities think through how to make their rates resilient to times of drought by illustrating the importance of fixed revenue sources like meter service charges. You can set:

  • The percentage fixed costs like O&M and capital improvement make up of total costs,
  • And change the percentage fixed revenue to see how Anytown's finances shake out in high, low and normal water sales scenarios.

Go ahead and try changing the fixed revenue percentages and see how it affects Anytown California's finances!

 

What if Anytown just raises volumetric rates when it’s losing money from low water sales?

The challenge with raising volumetric rates is economics 101.  Raising the price of a good like water will decrease the demand for it and can ultimately reduce revenue like in the graphic below.  How much water sales are affected is a function of many complex factors such as the economy, local culture, climate and other important considerations.  Measuring and predicting the impact of price on water demand is a core goal of the data collaborative.  

With trends towards increased water scarcity, pricing water will need to rely increasingly on a "subscription" model to ensure revenue regardless of high or low water sales.  Retail water volumetric costs range across California though are uniformly on the order of a small fraction of a penny per gallon.  What Californians really are paying for is not the water so much as the reliability of the underlying infrastructure.  

Note setting water rates also involves a host of other considerations.  For instance, Proposition 218 requires a nexus between the cost of water supply and the structure of water rates.    Our integrated analytical plan envisions building on this initial educational tool to provide an integrated suite of analytics and services to support water managers in achieving revenue stability regardless of water scarcity.