In preparation for completing its Conservation Master Plans and Urban Water Supply Master Plans, California Water Service (Cal Water) is assessing the impacts of climate change on water demand and water supply. The following summarizes the approach for analyzing the impact of climate change on water demand.
This modeling effort combined the framework of existing Cal Water district demand forecasts — district service area and customer-class specific forecasts of water consumption per connection, with modern statistical methods to explain the variation in water demand attributable to changes in water rates, personal income, weather, demographics, passive and active conservation programs, and economic conditions.
Response to weather is influenced by the nuances of microclimates, which include highly specific geographical factors as:
For this reason, each Cal Water district was matched to NOAA weather stations within their region. An additional analytical step was used to “fill” missing daily data in recorded NOAA weather stations. Regression models were used to predict any missing daily precipitation or maximum temperature values based on data from nearby weather stations with non-missing values.
The demand models were used to estimate the effect of weather on Cal Water customer water demand. Mechanically this is accomplished by comparing predicted demand conditional on observed weather with predicted demand conditional on normal weather.
The logarithmic predictions are retransformed to the raw scale on a monthly level and the difference between the two predictions constitutes the predicted effect of weather on a monthly basis for each service area and each customer class. The predicted weather effect can then be summed on an annual basis and expressed as percentage of annual weather-normalized water demand. An estimate of the variance of annual departures from weather-normal water demand were developed for each district and customer class. These estimated variances of class-specific water demand to weather can be weighted and summed across classes for an aggregate district level estimate of the standard deviation of water demand induced by weather variation.
These district level estimates of the extent to which demand responds weather were used to define the wet-year (minus one standard deviation), dry year (plus one standard deviation), and multi-dry year (plus 1.6 standard deviations). The estimated length of 1.6 standard deviations is derived from the ratio of the worst dry year in the sample and the magnitude of the estimated standard deviation averaged across districts.
The basic approach can be described as a conditional water demand forecast, in which the models of Cal Water demand are statistically estimated based on actual weather measurements. This model is then used to make a forecast into the future where changes to long term temperature and precipitation are possible; the forecasts can be interpreted as expected water demand conditional upon the assumed weather. This approach provides an empirical basis for defining how future climate change would affect future unconstrained water demand.
Based on the implications of potential climate change, the Cal Water Long Term Water Demand Forecasting model defines climate change scenarios that imply between a 2 to 3 percent increase in unconstrained water demand by 2040. Cal Water is continuing to utilize this model to assess various alternative demand scenarios.
Fourteen Cal Water districts were identified for detailed study of potential climate change impacts. The fourteen chosen service areas represent 85% of Cal Water’s 2014 production, and reflect a wide variety of geography and hydrologic systems. Water supplies for the fourteen service areas vary, with different combinations of local groundwater and surface water, as well as water purchased from different wholesalers.
Local climate change impacts will be estimated using a combination of historical data, both provided by Cal Water and publicly available, and projected temperature and precipitation data. Projection data is derived from SimCLIM, a climate change analysis application that can spatially represent and build databases of climate projections for a variety of pertinent parameters. SimCLIM uses 40 of the latest global circulation models (GCMs) run as part of the 2012 climate change assessment completed by the IPCC (CMIP5 generation of modules), published in 2013. SimCLIM allows for selection of one of four Representative Concentration Pathway (RCP): 2.6, 4.5, 6.0, and 8.5. An RCP is a projection for greenhouse gas concentration trajectories, which were updated in 2014. The 8.5 RCP will act as the worst-case scenario in this analysis. While the 2.6 RCP is the best case scenario for emissions, achievement of this pathway requires aggressive removal of atmospheric carbon1. Because the achievability of this RCP is debated, use of the 4.5 RCP as the best-case scenario is likely. SimCLIM allows a user-specified model sensitivity (Low, Mid, High), which is a measure of how responsive surface temperature is to changes in atmospheric CO2 concentrations.
Projecting impacts on groundwater supplies due to climate change is a complicated process. Groundwater basins are dynamic systems which respond to changes in climate at many different timescales: seasonally, annually, and over many years. Perhaps more importantly, natural responses to climate are complicated further by agricultural and municipal water demands on groundwater supplies. Demands on groundwater supplies can be amplified by surface water shortfall, by the duration of droughts, and increasing populations. For this reason, groundwater projections will be made in terms of long term trends. Similarly, the climate projection data from SimCLIM are only available in terms of statistical averages and long term trends.
The supply and delivery systems of the wholesale suppliers are generally very complex. We will rely on available data, including the results of any climate change modeling that these suppliers themselves have done or on other indicators of what the impacts of climate change might be.