5.3 DATA COLLECTION METHODS

Heretofore, this chapter has considered stormwater utility rate policy issues in order to define the types of data normally required to implement the utility fee or assessment. This section discusses how and where to collect such information. Emphasis is provided in the subjects of impervious and pervious area data, quality assurance, accuracy, and data sources.

5.3.1 General Principals

In essence, the goal of the data collection process for a stormwater utility fee is to provide sufficient data in order to collect the user fee or special assessment. Since the fees or assessments are collected monthly, quarterly or annually, the data collection process must include the ability to update the data; to allow review by the utility customer, the process must also include a record of the data collected. Finally, because customers may review the supporting data, a quality assurance process must be provided.

These issues are important to the general process of data collection since the data collection process is the most costly and time-consuming element of a stormwater utility start-up. Poor quality work, incomplete information, and over-attention to detail can undermine the public acceptance process and delay or even derail implementation. Therefore, stormwater managers, when considering the implementation process for the utility fee, should plan ahead for the data collection process with a focus on the goal (i.e., data sufficient to collect the fee). Strict use of data from other sources developed for other purposes can lead to difficulties (see discussion below).

Data Requirements

As presented in Section 5.2, the general list of data needs for a stormwater utility fee or special assessment includes the following items:

• Reference areas (impervious and total parcel area)
• Number of dwelling units
• Onsite stormwater facilities with maintenance record and type of facility
• Land uses (developed versus undeveloped, platted, vacant, etc.)
• Runoff patterns (for no discharge exemption)

Data Sources

Significantly more data are available now than were available when stormwater utilities started in Florida with the City of Tallahassee in 1986. Furthermore, technology has improved to allow for a higher-level electronic and digital data set to record and maintain utility information. The 1997 FASU Stormwater Utility Survey indicates some of the methods used by existing stormwater utilities. Of the 52 respondents, 33% used the Property Appraiser's Office data, 13% used aerial photographs, 10% used onsite measurements, 6% used other sources, and less than 2% used permit data.

The list of potential sources of stormwater utility data provided below is not exhaustive. The list does provide at least one source for each of the types of data required.

Database Development

Regardless of the types of data collected, the data should be accumulated into a permanent record or database. Over 10 years ago when stormwater utilities in Florida started, database management systems were complicated and spreadsheet programs were relatively simple. Today, the distinctions between database management systems and spreadsheets are blurring, so the ability to create a stormwater utility database has expanded whether spreadsheets (e.g., Excel, Quattro, Lotus 1-2-3, etc.) or database management programs (e.g., Access, dBASE, ARC/INFO, etc.) are used.

Databases generally have a primary field (key or index) to which all data is associated. For a stormwater utility database, the primary key depends on the final billing structure. For special assessments, the assessment is collected along with the ad valorem tax bill so that the primary key is related to the property owner. This means that the information collected must be related to the data which can be used by the Tax Collector for bill processing: usually this is the parcel identification number, not the owner name. On the other hand, user fees collected with other utility fees must be associated with the customer account number. The account number is usually the primary key for the user fee database.

To start the data collection effort, therefore, the foundation of the stormwater utility billing database needs to be developed. If funding is to be through a special assessment, then the foundation of the utility is the parcel number and, for each parcel number in the municipality, the database should be initially populated with data available through the Property Appraiser's Office. Minimum information includes parcel number, owner information, site information, legal description (6 lines), and land use codes. The data collection process will complete the database with reference area data, stormwater facility information, and other rate structure needs. Every parcel number must have associated utility data before the data collection process is complete.

For user fees collected through existing utility billing programs, most municipalities use a two-step process to assign reference areas and other information to each existing utility customer. The first step is to create a parcel specific data set, as noted above, using the parcel number as the primary key. The second step is to associate each existing utility customer with a parcel number. Many of the parcels have more than one customer assigned (e.g., a multi-unit strip mall is one parcel but may have five customers). The billing database is associated with the customer, not the parcel. The effort is to distribute the parcel-specific reference area equitably to the customers. For example, a single strip mall may have five water customers on the property. The stormwater utility fee is based upon the equitable distribution of reference areas to the five customers (methods for equitable distribution are discussed below). The billing database has a listing of each customer and associated reference areas, which may be a percentage of the area measured for the parcel on which the customer is located

.

Quality Assurance

In most cases, stormwater utility databases consist of billing information (generally collected and maintained from other sources) and customer specific information associated with the utility fee or assessment. Assurance of the quality of the information related to the customer is imperative for the credibility of the overall stormwater utility program. Therefore, a quality assurance program must be provided during the development of the initial billing database, as well as during maintenance of the database. A number of practices that can be followed to improve the quality of the stormwater utility billing database are discussed below.

Quality Review During Measurements
During the initial preparation of the utility billing database, all measurements should receive a quality assurance check. First, all measurements should be reviewed by someone other than the original measurer. For example, if impervious areas are digitized or planimetered, each section map should be reviewed by a quality assurance reviewer. If the magnitude of the parcels measured is significant, then at least 10% of the parcel measurements should be randomly reviewed.

Measurer Changes
More than one measurer should be used in the digitizing process. For example, digitizers and quality reviewers should periodically exchange responsibilities.

Periodic Database Reviews
Routinely, a quality reviewer should consider the numbers being entered into the utility billing database. If the data are entered manually, random checks on the data entry forms and the database should be performed. Also, simple database internal checks should be run. For example, the total impervious area for a parcel cannot exceed the total parcel area, and a simple test can be run to test for this occurrence in the database. Other database tests may include comparison of measurements to average values for other stormwater utility data sets from adopted utilities in Florida. For example, if the average reference area for single family in the subject municipality is 5,000 square feet, and the average reference area for other stormwater utilities in the region is 2,500 square feet, there may be reason to check the single-family unit reference area measurements.

Field Checks
If aerials are used to measure impervious and total areas, trees often block the measurer's ability to see impervious material. Field visits may therefore be required to confirm such materials. For quality assurance, if the measurer is unsure of the aerial information, a field check should be completed.

Top 100 to 150 Fee Payers
Discussed further below, the customers with the highest reference areas must be provided the greatest assurance that the measured data are accurate. Experience has shown that, even with the large customer base utilities, the quality assurance program should provide the most assurance to the 100 to 150 customers paying the largest fees. For this reason, once the initial database has been developed, the largest 100 to 150 reference areas should be revisited, both by re-measuring and by specific visits in the field.

Customer Review
Whether the stormwater utility rate structure is based on uniform or variable fees, many communities in Florida have provided the database information to the customer prior to the first billing. This is required for the use of special assessments where the basis for the assessment must be provided to the owner. Some communities also provide basic utility billing information to customers in either a mock bill or a mail-out of the reference area information to allow customers to check the measurements.

Accuracy and Precision

The accuracy of the stormwater utility billing database is improved by the quality assurance checks provided during the measurement and maintenance of the database. That is, the database can agree well with reality (accuracy) by allowing multiple methods of quality assurance. However, the next question is: "How precise should be measurements be?" With the new types of digital information available to stormwater managers, precision can be high. For the purposes of this discussion, precision means the number of decimal places the data carries, and reflects the degree of agreement of multiple measurements of a quantity (Webster's Third New International Dictionary, 1986). For example, the number 5.482 is more precise than is 5.5, but it may not be more accurate.

To answer the question of precision, there are two types of measurements that need to be considered. First, most stormwater utility rate structures are based on a residential reference area average as the equivalence. All or a sample of the residential parcel reference areas are averaged to derive a single equivalent value usually resulting in values between 1,500 to 3,000 square feet. Since the equivalent reference area is based upon an average, and the reference areas are measured from aerials with a 3-foot pixel size (at best), the precision of residential equivalent values should be on the order of 10's or 100's of square feet. For example, if the single-family average reference area is determined to be 2,524.7 square feet, the actual value recorded should be 2,520 or 2,500 square feet simply because of the measurement methods used and the use of the value itself do not warrant higher precision.

Secondly, reference areas are generally measured for each non-residential parcel. In almost every stormwater utility rate structure in Florida, the measured reference area is divided by a residential equivalent to determine an equivalent residential unit. For stormwater utilities in Florida, the residential equivalent is generally about 2,000 to 3,000 square feet with an accuracy of about 100 square feet. Using an illustration to address accuracy of the measured reference areas, consider a 1-acre parcel with 16,453 square feet of impervious area and an equivalent reference area of 2,500 square feet ? 100 square feet. Thus the ERU value could be 6.86, 6.58 or 6.33 depending on whether the "accurate" equivalent reference area is 2,400, 2,500 or 2,600 square feet. Similarly, fixing the 2,500 square feet equivalent area, then the three ERU values are the same as a measured impervious area of 15,825, 16,450 or 17,150 square feet for a range of over 1,300 square feet. This example illustrates that because the measurements are not perfectly precise and the measurements are to be divided by a 4-digit average reference area, requiring a 1- or even 10-square foot precision is "overkill." For the measurement of reference areas for a stormwater utility database, precision to the 100-square foot level is sufficient. Furthermore, most stormwater utilities in Florida round the ERU calculation to the nearest tenth (e.g., 6.86 is rounded to 6.9).

5.3.2 Impervious/Pervious Area Data

With the basic principles in mind, the next step is to consider the site-specific information for the foundation of the stormwater utility fee or assessment: impervious and pervious areas. For the purposes of this section, rather than considering pervious areas, the impervious and total areas of the parcel will be considered with the pervious component being the difference between the two. Also, both manual data collection and digital processing are considered below.

Manual Measurement

The minimum information needed to provide the manual measurement of impervious and total areas includes the parcel boundaries for all sections covering the municipality and recent aerials of the same scale covering the same area. Both of these data can be obtained from the County Property Appraiser's Office. Note exactly when the aerial photographs were taken. Usually flown in January, aerials are only as good as actual flying date. Reference area information that has changed from the date of the aerials has to be obtained from another source.

The second step is to identify the parcels for which reference areas are required. These may include a sample of residential parcels and all of the non-residential parcels. Once the parcels are identified on the parcel map, the appropriate aerial is placed over the parcel map on a light table and all of the required parcels are measured for the map. If there is sufficient staff, two teams should be working: the first team should identify parcels to be measured and provide quality assurance review; and the second team should be measuring reference areas. Periodically, the teams should switch. Often, it is not clear on an aerial whether an image is impervious area or not. When this occurs, the measurer should not slow the measurement down by attempting to resolve the image. Rather, the measurer should write the problem on a 3"x5" card and place it in a specified location. Once a week or so, the first team or a separate field team should pick up the cards and resolve the questions either in the field or through other research.

Using this method of reference area measurement, the average time taken for each parcel is about 20 to 30 minutes. Thus, for estimating time to develop the preliminary database, multiply the total number of parcels to be measured by 0.5 to yield the total staff hours needed. For example, if 5,000 parcels need measurements, about 2,500 staff hours are needed. If three staff members are assigned to the task, about 21 weeks (2,500 staff hours/3 staff/40 hours per week) will be needed to create the database. It should be noted that while this is very labor-intensive work, entry level or even temporary staff could complete the effort. If temporary labor is used, at least one municipal staff member should be assigned to oversee the work for quality assurance and for continuity.

Digital Measurement

With the advent of more easily available digital aerial and parcel boundary information, the measurement of reference areas for stormwater utilities has significantly improved. The process to measure digital images is essentially the same as manual measurement; however, with digital data, the computer can automate a number of the steps. First, from the digital Property Appraiser's data, the boundaries of parcels that need to be measured can be colored (highlighted) automatically, significantly decreasing the time needed to identify measured parcels. Secondly, the reference areas for a single photograph can be digitized in a short period of time, allowing the computer to calculate impervious and total parcel areas. It should be noted that since such programs calculate areas within a particular polygon, each polygon within a parcel should be characterized as pervious or impervious.

The digital process allows for reduction of the team structure. Since the computer identifies which parcels are to be measured, the parcel marking team is not needed. Thus, only one team is needed (delimiting impervious and pervious areas). The geographic information system will calculate impervious and total parcel areas and digitally dump the results into the utility billing database. In the same manner as the manual measurements, exception cards (parcels for which the aerials are not clear) should be written, and another team should resolve the questions identified.

Based upon recent experience with this digital process, parcel processing time is reduced to about 10 minutes per parcel. In the example given above, 5,000 parcels would require about 830 staff hours, or with three staff, about 7 weeks.

Using these basic measurement techniques, the stormwater utility database can be created. Further information is provided below on the elements of reference area measurements and database development.

Property Appraiser's Data

Property Appraiser's data are provided in two general databases. The Name-Address-Legal file (NAL) contains owner specific information including the parcel number, owner data, situs information (i.e., information on the parcel itself such as address), land use codes, and legal description. The Master Appraisal File (MAF) contains appraisal information on the parcel such as number of buildings, roofed areas, number of stories, land uses, etc. The two files together provide sufficient information to allow an appraisal for tax purposes.

Land uses are referenced to the 100 Department of Revenue (DOR) Land Use Codes. Codes 00 to 09 are residential, and 10 through 99 are non-residential land uses. A list of all codes is provided in Appendix A.

At first glance, it appears that general MAF data sets may have useful information for the stormwater utility. Experience has shown that the NAL and MAF are for appraisal purposes and that use of the data for any other purpose extends the accuracy of the data beyond the intent. This is not to say that the MAF is inaccurate; rather, the MAF is for appraisal purposes only. The stormwater utility database can be started from the NAL and MAF, but the reference data populating the utility database should be obtained through measurement.

To reinforce this conclusion based upon experience, each Property Appraiser has methods to store appraisal data that must be completely understood before the data are useful. For example, some Property Appraisers combine all floors in a townhouse into the base floor while others do not. This may also be true for some high-rise condominiums. Property Appraiser data does not usually include driveway data. Also, mobile home parks may be included in DOR Code 03 (mobile homes) or 28 (parking lots and mobile home parks). Sometimes, mobile homes have been "condo-ized" where each mobile home unit is listed separately on the tax records. These examples illustrate that the Property Appraiser's data are for appraisal purposes, and the data should not be directly used for the stormwater utility unless there has been a clear understanding between the Property Appraiser and the caretaker of the stormwater utility database.

Aerials and Digital Images

The measurement of reference areas is dependent on the availability of aerials. Of course, manual measurement of all parcels in the field is possible; nevertheless, most stormwater utilities are based upon aerial photography. Recognizing the value of such photographs, many water management districts are creating digital aerial images of their entire district starting with coastal or critical areas. The Property Appraiser's Office also has aerials, generally for sections at a 1 inch = 200 feet scale. These are flown every two or three years. The Florida Department of Transportation (FDOT) also has aerials, flown every other year and available as a 9" by 9" photograph (these are commonly referred to as DOT 9 by 9's). The DOT 9 by 9's can be scanned at a resolution of about 1,000 dots per inch, creating an 80-megabyte image that can be "rubber-sheeted" to match general parcel boundaries. Rubber sheeting is necessary since these photographs are not rectified. The accuracy of this process improves if the photographs are rectified to known survey points or an accurate roadway coverage.

As digital aerials become more and more available, the quality and resolution will also improve. Use of digital aerials should strongly be considered for the development of the database.

Field Observations

When an aerial is not clear or for quality assurance purposes and specific measurements are needed, municipal staff members can visit parcels in the field. Field observations can confirm accuracy as well as improve it, yielding a more reliable stormwater utility database. Two methods of field observations are employed depending on the question to be answered: drive-by window observations, and field measurements with a tape measure or measuring wheel. The first is normally used when there is a simple issue about whether an area is impervious or pervious (e.g., sand or shell) and is not overly time consuming since exiting the automobile is not required. The second is used to confirm an aerial measurement (for the top 100 fee payers, for example) and is very time consuming. Both methods may be used during the development of a utility database.

A distinct difference in the two methods is the contact with the customer. A drive-by window observation does not require contact with the customer. If the municipality is in the initial preparation of a stormwater utility public acceptance program, contact with the customer may be premature. When the actual measurement is done with a tape or wheel, there will be contact with the customer. If the municipality is not ready to announce the potential user fee or assessment, the field measurement will become the first public awareness effort. Sometimes a low-key effort will engender support especially when the customers see that the municipality is doing its homework. It is more likely, however, that public response will be negative because they will not have been educated about the stormwater needs. Refer to Chapter 3, "Gaining Community Acceptance", for further information on public interaction.

Finally regarding field measurements, staff needs to be careful about trespassing. If actual field measurement is to occur, the actions of the measurers need to be announced to the public ahead of time. Furthermore, the field staff should carry a letter of authorization to do the work, signed by a city or county administrative official.

Equitable Distribution of Reference Areas

As noted above, when an existing utility billing structure is to be used to distribute the stormwater utility fees, property-specific information needs to be equitably distributed to customers. Common examples of multiple customer properties include apartments, mobile home parks, strip malls, large malls, and industrial parks. Using a large mall as an example, the overall area is generally split up into parcels for each large anchor store, and one parcel for the multiple internal mall stores. The problem is how to equitably distribute the reference areas (especially the parking lot) to the multiple customers.

Historically, there have been a number of ways this question has been answered:

• Send a total bill to the mall management for payment in full. The mall management would be responsible for payment of the fee.
• Prior to billing, send the reference information to the mall management and allow the customers to divide the fees. Bills are then sent to individual customers. The distribution can be updated on an annual basis.
• Send a bill to each anchor store and one to the mall management for distribution. The parking lot is distributed to the anchor stores and mall management based upon roofed area.
• Periodically determine the rented space within the mall and distribute the parking lot reference area to each renter according to the percent rented space. Bills are then sent to individual tenants.

The methods are generally listed in order of complexity and administrative costs. The most expensive way to deal with this issue is to send individual bills to tenants since leased space frequently changes. Whichever way is selected, it is suggested that alternatives be discussed with major malls and other large parcels for which distribution is required.

5.3.3 Dwelling Units

For stormwater utilities that have a residential rate structure based on dwelling, each residential customer must be assigned an accurate number of dwelling units. Single-family units (DOR Code 01), condominiums (DOR Code 04) and sometimes, mobile homes (DOR Code 02) have one dwelling unit. Multi-family units (DOR Codes 03 and 08) and mobile home parks (DOR 28) must be assigned numbers greater than one depending on the number of dwelling units.

Most municipal codes define a dwelling unit to essentially be living quarters for a family including a food preparation area. Examples are a 200-unit apartment complex (200 dwelling units) and a duplex (2 dwelling units). More difficult examples include retirement homes (DOR Code 06) and boarding houses (DOR Code 07). Often municipalities assign one dwelling unit to these types of parcels due to the nature of the buildings and because these facilities generally only have one kitchen.

Sources of dwelling unit information are generally limited. Most Property Appraisers have some information on the number of dwelling units for multi-family residential properties; however, some research may be required to find the data since it is not normally contained in the NAL or MAF data sets. Other sources of such information include relocation guides, apartment finders' documents, the Fire Marshall and emergency management agencies (needed for evacuation emergencies), and commercial listings. Of course, each multi-family unit and mobile home park can be called.

It should be noted that Property Appraisers vary in how mobile homes and mobile homes in parks are treated. As noted previously, DOR code 02 is for mobile homes and code 28 is for mobile home parks. Sometimes, "condo-ized" mobile home parks, in which all units are individually assessed, are categorized as single-family residential units. The lesson to be learned is that mobile homes need special attention to confirm parcel-specific data and dwelling units.

5.3.4 Average Residential Reference Areas for Equivalence

Section 5.2.1 above notes that most stormwater utilities in Florida base the fees or assessments on either single-family unit equivalence (SFU) or equivalent residential units (ERU). Both are based upon an average reference area per residential dwelling unit (single-family only in the first case and all residential types in the second). One SFU or ERU represents the average reference area for the residential dwelling unit. For example if the average reference area for a single-family unit in a community is 2,500 square feet, then an SFU can be defined as 2,500 square feet.

To calculate the average residential reference area, either all of the residential properties need to be measured for the reference area, or a sample of the properties need to be measured in order to achieve a statistically representative result. In communities for which data on reference areas have already been defined, the average reference area is simple to calculate. Unfortunately, few communities have these data so most use statistical sampling methods to define the equivalent reference area. Provided below is a method to develop such a statistically significant result.

Starting with the single-family units first, measure (by whatever means available) the reference area for 30 single-family units. Calculate the average and standard deviation of these data. The statistical question is: "How many single-family measurements are needed to be 95% confident that the difference between the sample average and the average of the total number of single-family units is less than 5%?" The 95% confidence limits are:

where X_a is the average of the sample, s is the standard deviation, and N is the number of measurements (Statistics, 1961). An estimate of the total number of measurements needed for 95% confidence is therefore:

For example, suppose for the first 30 measurements the average is 2,500 square feet, and the standard deviation is 750 square feet. An estimate of the total number of measurements needed to be 95% confident that the sample average will be within 5% of the "real" average is 138.2 or 139. Since 30 have already been measured, an additional 110 should be sampled. For a 99% confidence, the "1.96" is replaced with "2.58." In the example, 99% confidence would require about 240 measurements. This estimate may need to be calculated more than once, as new data are collected.

Once the single-family average reference area has been determined, a similar process can be completed for other residential properties. The difference is that the average reference area is to be related to each dwelling unit. Therefore, choose about 5 to 10 properties in each of the other residential categories (multi-family, condominiums, mobile homes in parks, and mobile homes on single lots). Determine the number of dwelling units for each and measure the reference area. The average reference area per dwelling unit can be calculated, as well as confidence intervals and estimates of the number of samples needed. It is likely that for these types of residential parcels, the initial sample will be sufficient.

5.3.5 Stormwater Facilities for Credit Calculations

As noted in Section 5.2.1, some municipalities allow credits for maintained onsite stormwater facilities. The subject of this section is the source of such data for billing purposes.

During the initial feasibility stages of a stormwater utility fee, where alternative rate structures are being considered, an estimate of the impact an onsite facility credit might make can be derived from water management district information. Most districts have an electronic database of permitted stormwater facilities owned and operated privately. The databases have information on the type of facility and limited information on facility location. These data are not usually associated with individual parcels and considerable effort is needed to assign permitted facilities with parcel numbers. However, for the purposes of estimating impacts, the water management district data are useful. As the water management districts create better electronic GIS datasets, onsite facility data will be easier to obtain.

For the actual billing process, customers who may be able to receive a credit for onsite stormwater facilities should be required to provide the necessary information to the municipality. At the same time, the municipality can confirm proper maintenance of the facility. Confirmation of maintenance can be provided through photographs and certifications.

5.3.6 Data Adjustments

Once the stormwater utility billing database has been completed and either a mock bill or even the first bill has been delivered, it is likely that adjustments to the database will be necessary due to omissions and errors in measurement. Customers may come to the stormwater staff and review the reference area measurements for their property or number of dwelling units and subsequently request modification. A policy needs to be developed to make these adjustments to the database so that changes are not arbitrary and only one "original" utility database is used.

Most communities provide an adjustment procedure in the stormwater utility ordinance. Depending on the rate structure itself, the reasons for such an adjustment need to be specifically defined. For the most part, adjustments will be based on the number of dwelling units or the actual measurement of the reference area. Many municipalities require the fee payer to provide certified measurements by a registered surveyor to challenge the original municipal measurement. As noted above, missing information regarding credits should be clearly defined as the responsibility of the customer.