Class Project

GIS in Water Resources

Craig W. Miller

November 27, 2002

mailto:CraigMiller@Utah.gov

Identifying Indoor and Outdoor Water Use

Introduction

In the arid west, understanding the principle components of water use is the first step in water use is the first step in addressing conservation. As a result of studies of water use patterns, low-flow fixtures have significantly reduced indoor water use over the last decade. A similar emphasis on understanding outdoor water use has the potential to eliminate a much larger amount of needless water use over the next decades. Figure 1 shows the water use patterns in Salt Lake City for water years 2000 and 2001.

Figure 1. Average water use per household in water years 2000 and 2001 in Salt Lake City.

As can be seen from figure 1, most of the domestic water use in Utah occurs in the summer and spring for irrigating landscaped outdoor areas. A breakdown of water use for the entire state is shown below in figure 2. This data suggests that the key to water conservation is understanding and solving the problem of water waste in residential outdoor water use.

Figure 2. Public supply water use broken down into categories.

The Utah Division of Water Resources has been studying indoor and outdoor residential water use to determine where potential water savings can be realized. This remainder of this paper will focus on one method developed for this project that assists with analysis of outdoor water use.

Identifying Outdoor Water Use

Outdoor water use, for the purposes of this paper, is that water use which exceeds indoor water demands. Indoor water demands, as can be seen in figure 1, can be assumed to be the average indoor demand during the months of December, January, and February.

Some Utah water purveyors record water use for each month of the year. The procedure for identifying outdoor water use is less complicated where monthly data exists. To obtain outdoor water use find the water use per day for the months of December, January, and February and subtract from water use in the remaining months of the year. Some water purveyors read meters in November and again in April. In these cases the following equation was used to separate outdoor daily water use from indoor daily water use.

where:

i – month number

j – month of interest

n – beginning month of period considered, for example if meters are not read for five months in the winter, use 1 for n and 5 for m.

m – ending month of period

p_j – ratio of u_av to u_j for month j

u_i – daily use in month i

u_av – daily average use above winter use (u_wn) for months n to m

d_i – days of use in month i

et_i - et_turf in month i per day

Although many factors affect outdoor water use including homeowner perceptions, water price, weather, and irrigation methods, the procedure above assumes that et_turf is a major factor driving outdoor water use.

After the total outdoor water use is calculated, the area of outdoor landscape requiring irrigation is needed to determine the depth of water application. Studies throughout the Wasatch Front in Utah have established relationships between tax-assessor lot size and landscaped area. The challenge is to match water use records with other readily available databases to determine lot size. How to identify a parcel by its postal address is a topic covered later in this paper.

Use Per Capita

One of the statistics of interest to many planning agencies is water use per capita. In Salt Lake County water purveyors use billing records to determine total residential demand and divide that amount by system population determined by clipping out population values from a census coverage using system boundary maps. Unfortunately system boundaries overlap one another and cause populations determined in this fashion to overestimate the total county population by over 10 percent.

Figure 3 shows overlapping system boundaries as reported by various water purveyors throughout the Salt Lake Valley to illustrate the problem water purveyors have of accurately determining system population.

Figure 3. Overlapping Salt Lake County water purveyors (note: the turquoise color is green, New SLC Water, overlain on lavender, Old SLC Water).

In order to correctly identify system boundaries, water service addresses obtained from billing records were used in conjunction with county property GIS coverages.

Addresses Matching to Solve Water Resources Related Problems

The two problems presented above (determining the depth of irrigation application or the service boundary of a water provider) can be solved with address matching, in other words finding a specific lot that matches an address. Three methods were compared to determine which provided the most reliable results.

Method 1, Address Standardization Using In-House Methods.

This method standardized the spelling of street coordinates (East became E, West became W) and the postfixes used for street definition (Road became RD, Avenue became Ave, etc.) The following summarizes some of our findings with respect to this method.

Table 1. Summary of results of using in-house address matching methods.

Address Standardization Using In-House Methods
Advantages	Disadvantages
· Does not require purchased products · Works well for a majority of addresses.	· Does not find about 50% of all addresses · Requires programming and debugging of main program and tools. · Does not adequately resolve the problem of alternate spellings of roads and streets.

In trial runs, in-house methods found about half of the addresses in a county property database. We never did adequately determine exactly why this percentage was so low.

Method 2, Geocoding Using County-Provided Data.

Counties in Utah sometimes provide street data which have the required fields for geocoding. Address matching was attempted using this method. Table 2 lists the results of this trial.

Table 2. Summary of results of using geocoding to find city lots.

Address Matching Using Geocoding
Advantages	Disadvantages
· Free or purchased products are available. · Works well for a majority of addresses. · Does not require any programming	· Does not find about 15% of all addresses · Some address locations were miles off the mark. · The method is sensitive to right and left offsets, etc.

The geocoding products provided by Salt Lake County were adequate for geocoding, but did the offsets necessary to place points on specific lots were missing. The points identified with geocoding revealed approximate locations but did not allow us to identify specific city lots.

In order to geocode using a street database, in ArcCatalog select “Create New Geocoding Service, as shown below.

Figure 4. Creating a new geocoding service.

Since the Salt Lake County street file also included zip code information, in order to create the geocoding, the option “US Street with Zone (File)” was chosen as shown below.

Figure 5. Creation option chosen for geocoding service.

Figure 6. Options for geocoding with zip code.

In figure 6 above, the Left Zone and Right Zone were filled in with the appropriate field name for the zip codes from the even and odd numbered portion of the street. The results of geocoding are shown in the following two graphics.

Figure 7. Geocoding results

Figure 8. Geocoding results of single family residence water service accounts showing service boundaries.

As can be seen from the graphic above, the service boundary provided does not precisely match service location addresses. This is as we expected since each of the service boundaries provided by water purveyors appears to overlap with another.

Figure 9. Results of geocoding on a lot level

As can be seen above, geocoding is useful for identifying the general location of a lot, but often misidentifies or does not identify a specific parcel. Using geocoding to tie water billing records to a specific parcel should be done with a good deal of caution.

Method 3, Address Matching Using USPS Zip+4 Address Database.

In this method we coded a simple Visual Basic^© program which standardized addresses using the Post Office’s Zip+4 web site. The program successively queried the database and eventually obtained standardized addresses from the Post Office database. From this a unique key was produced which included the zip code followed by the address. Table 3 summarizes our experiences with this method

Table 3. Summary of Results Using USPS Zip+4 Address Standardization.

Address Matching Using USPS Standardized Addresses
Advantages	Disadvantages
· Post Office provides free software to perform this task. · Works for about 95% of all residential addresses. · Requires little or no pre or post processing. · Standardizes variant spellings of street names. · With such a high percentage of first pass matches, database cleanup is much simpler.	· Post Office provided software requires access to a web server. · The method is time consuming. Hundreds of thousands of addresses will require days of USPS database querying. · Alternate street names are not resolved. · Alternate city names are often not resolved.

If data is in short supply, this method will provide the greatest quantity of address matches of all the methods. It identified system boundaries with great precision and does not misidentify plots of property.

The user interface of the program is shown below. It uses the VB Browser control to successively interrogate the Post Office Zip+4 interface.

Figure 10. Getzip user interface.

For this study we used the same addresses geocoded from Salt Lake City’s billing records to identify lots within the service boundary. The addresses from Salt Lake County’s tax assessor database and the Salt Lake City Water addresses were standardized using the Zip+4 database. The resulting databases were used to identify the lots within the county served by Salt Lake City. This is shown in the figure below.

Figure 11. Geocoded points superimposed over lots identified by Zip+4 address standardization.

As can be seen from the graphic above. Both methods have their strengths. If there is any ambiguity in an address, Zip+4 will not work well. Geocoding works well with ambiguous addresses but is not ideal for identifying a specific lot.

Discussion and Conclusions

Databases exist which can help water planning agencies identify indoor and outdoor water use. Additional data is often needed to clarify and refine our understanding of available data. One important piece of information is which lot is associated with a specific address. Three methods of address matching were presented. Each of these three have advantages which would make them more desirable in once circumstance or other. For the purposes of identifying a specific lot, Zip+4 address standardization appeared to work best. For identifying system boundaries a combination of Zip+4 address standardization and geocoding appears to give good results.