UrbanDemand


National Water Demand Predictions for the Municipal Sector

Developed by Maura Allaire

County-level predictions of municipal water use were made for year 2010. These predictions represent an improvement over county-level data available for domestic use from the USGS Water Census. The Water Census provides approximations of withdrawals within a given county. It does not provide information on end user water demand. This is particularly an issue in countries that are the source or receiver of large water transfers. In addition, in most water census years, county-level water use data for the domestic sector is not reported for several states. In these cases, approximations from previous Census years are used, with or without scaling for population change.

The Columbia Water Center developed improved predictions of county-level water use by creating a regression model of water sales on price, climate, and demographic characteristics. In year 2010, the demand regression included 265 water utilities and over 54% of the variance in water sales was explained by the specified model.

Data

Water use and price data were obtained from Water and Wastewater Rate Surveys produced by the American Water Works Association (AWWA). These biannual surveys provide utility-level information on water use and rates for public water systems in the U.S. The analysis in this study focuses on survey data from 2000 to 2010. Average water price was calculated based on the price faced by a user with average use for a given utility service area.

Climate variables, including average monthly precipitation and temperature, were obtained from daily 1/8-degree gridded meteorological estimates by Maurer et al. (2002). Gridded estimates were aggregated to a county-level by Tara Troy, using the VIC model

Demographic characteristics were obtained from the U.S. Census and include median household income, % college degree or more, % unemployment, # business establishments, average household size, housing density, median housing value, median year housing built, % owner-occupied dwellings.

Comparison with USGS Water Census

Columbia Water Center predictions tend to be higher than reported water use from the USGS Water Census (variable: dopspcp)

Comparison in the model

Construct the AWASH environment:


In [1]:
using DataFrames
include("../src/lib/readconfig.jl")
include("../src/lib/datastore.jl")
suffix = "";

In [2]:
mastercounties = readtable(datapath("global/counties$suffix.csv"), eltypes=[String, String, String]);

Attempting to load the data (showing what happens if you don't have everything setup):


In [4]:
config = emptyconfig()
config["urbandemand-path"] = "demand/urbandemand_May11.csv"
config["urbandemand-column"] = "per_capita"
config["urbandemand-transform"] = "repcap"
col2 = configdata("urbandemand", "demand/simulation2010demanddata.csv", :PS_WTotl, :regions);


LoadError: There are not 3109 entries, but no urbandemand-index configuration specified.
while loading In[4], in expression starting on line 4

 in error at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in configdata at /Users/jrising/research/water/model/awash/src/lib/readconfig.jl:100

In [5]:
config["urbandemand-index"] = "FIPS_county"
col2 = configdata("urbandemand", "demand/simulation2010demanddata.csv", :PS_WTotl, :regions);


LoadError: The PopulationDemand component needs to be loaded first.
while loading In[5], in expression starting on line 2

 in error at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in anonymous at /Users/jrising/research/water/model/awash/src/lib/readconfig.jl:35
 in configdata at /Users/jrising/research/water/model/awash/src/lib/readconfig.jl:94

Finally, load the data:


In [6]:
include("../src/PopulationDemand.jl")
col2 = configdata("urbandemand", "demand/simulation2010demanddata.csv", :PS_WTotl, :regions);

A quick comparison:


In [7]:
DataFrame(fips=mastercounties[:fips], surveyed=col1, estimated=col2)


Out[7]:
fipssurveyedestimated
1010017038.19646.785578166
20100331761.341999.195453175
3010055738.37440.535253222
4010076761.64971.888792205
5010095351.611952.297234796
6010113180.32133.49371306
7010133733.44791.047861436
80101532217.627618.908848248
9010175959.67539.569498999999
10010194825.75708.207924787
11010216761.68914.201153459
12010231880.53926.2748094089998
13010254355.67017.4179431699995
14010272295.32992.286900952
1501029774.32982.883627432
160103110508.811064.870254712
170103312154.212814.692311564
18010352336.83588.584012704
1901037414.82221.268588979
20010396844.59132.62074907
21010412834.62649.635208846
220104342933.818202.240487592
23010459471.79374.904197795
24010478158.111288.72540768
25010499831.216105.649138235
260105118459.515397.169525115
27010537853.910578.592175181
280105528166.225874.59300833
29010572806.94383.885569637
30010596319.16537.335061647999
&vellip&vellip&vellip&vellip