Source Water Assessment
for the Cache la Poudre River

An EPA National Pilot

WORKING NOTES AND RECOMMENDATIONS

Introduction

The purpose of this section is to:

• Identify ‘lessons learned’ from this national pilot source water assessment (SWA); and
• Recommend actions necessary to continue finalizing the SWA for the Cache la Poudre (CLP).

This section concludes with a brief review of the:

• Parameters within which Project staff conducted the SWA; and
• Project accomplishments.

 Following this, the notes address the ‘lessons learned’ and generally are organized according to the following:

• General
• Expertise
• Methods
• Data
• Cartography
• Geographic Information Systems (GIS)
• Delineation
• Contaminant Inventory
• Susceptibility Analysis

Some redundancy might be apparent in the notes.  One reason for this is that many of the specific issues are related.

The section concludes with recommendations for finalizing SWA for the complete CLP SWAA, including all areas upstream from the City of Greeley intake, predominantly the North Fork CLP.

General Parameters Within Which Project Conducted

The purpose of this Project was to produce a SWA for the area associated with the Cache la Poudre (CLP) mainstem that supplies raw water to the City of Fort Collins public water system.  The overwhelming emphasis has been to produce a product suitable for placement on an Internet website to educate the public about from where their drinking water comes and potential threats to its quality.  See the more detailed description of the Project as first envisioned. 

One critical parameter is that the Project was started prior to EPA finalizing its guidance for conducting susceptibility analyses and just as the Colorado Department of Public Health and Environment (CDPHE) was beginning to develop the state plan under the federal guidance.  Given the primary public education purpose of the Project, CDPHE presented the ‘Clear Creek Watershed Project’ maps as a general model of the type of product it desired, and other Project participants agreed.  In addition, Project participants directed staff to try to develop the national pilot CLP SWA within parameters of the developing Colorado Source Water Assessment and Protection Program (CoSWAP) to the degree possible, recognizing that there would be differences given that Project work would be mostly or totally completed before the state program was finalized and approved by EPA.  Project participants and other factors strongly suggested that the Project rely on electronic data that was readily accessible from Internet websites, and particularly those provided by the U.S. Environmental Protection Agency (EPA) (e.g., BASINS).  This fit with the need of for this initial assessment to “… use only data that is necessary and available in a format that the Project can use easily within the time and funding parameters available for Phase 1 work” (5/22/98 Detailed Work Plan, p. 2).  Finally, the Project had a fixed budget.  A progress report submitted to the CDPHE on March 31, 1999 describes the early context and some of the challenges it presented in more detail. 

Project Accomplishments

The Project achieved the goals of this first attempt to develop a source water assessment for the City of Fort Collins Cache la Poudre source water assessment area (SWAA).

Among Project accomplishments, we:

• Followed what was then evolving as the national cutting edge approach to data supporting SWAs by obtaining, manipulating, and reviewing electronic data accessible from the Internet, primarily from EPA websites, and from commercial data providers.

• Invested significant effort with the Larimer County and Arapahoe-Roosevelt National Forest Geographic Information System (GIS) program staffs in identifying and characterizing the digital data holdings of these potential data providers in order to identify that potentially useful to developing the SWA.  We are particularly proud of the use we made of Larimer County parcel data to develop maps of land use, individual sewage disposal systems (ISDS) (e.g., septic), and agricultural land use.

• Contacted several state, federal, and private organizations and identified, characterized, manipulated, and where appropriate, used data from these sources.  Generally, we found that of the potential data providers we contacted, most did not have data that identified potential sources of contamination in the national pilot SWAA and/or did not have data that was readily useful within Project parameters.  Frequently, the task of identifying and researching this data still required a significant amount of time even to arrive at a judgment that a particular source did not have data that would support the national pilot SWA.  However, we did identify three databases (i.e., above, under ground, and leaking underground storage tanks), and developed a thematic map from this data, to illustrate what might be done in instances where appropriate data sets are identified. 

If we have disappointments, they have to do with our being unable to do much more and follow to resolution all the issues and potential additional lines of attack that we uncovered along the way.  In particular, this involves not being able to investigate the data holdings of all sources that may have data potentially supporting conduct of a SWA, and also not being able to process further some of the data we did obtain.  Another disappointment is being unable to invest time in verifying the data we did use in site-specific field investigations.  Many of these issues are not complex, and could be resolved with an investment of the time required to resolve them.  We hope that we will have the opportunity to address these issues further at another time, and particularly in the context of refining and expanding the CLP SWAA.

General

Identify and assess needs of all stakeholders fully and carefully prior to starting SWA.  Identify project goals and objectives, and specify products. Select methods for achieving these.  Develop an implementation plan, including identifying and characterizing required data, tasks, expertise, budget, time, and other resources (e.g., hardware, software, equipment, etc.) required.  Provide sufficient resources to accomplish the work.  This is straightforward, standard project management practice, applied in an ideal situation.  As explained above, we initiated the Project before EPA or the state finalized their programs.  Thus, the product was specified in general terms (i.e., a source water assessment, maps like the Clear Creek model), a primary tool taken as given (i.e., geographic information system (GIS)), and a fixed level of funding resources committed, prior to specification of stakeholders’ needs as represented by a final federal susceptibility analysis guidance or state program, or in-depth assessment of local needs.  With regard to the susceptibility analysis in particular, and to some degree the contaminant inventory, the Project in effect pursued a ‘moving target’.  The most striking example concerns the buffer zone requirement which in large part drives the accuracy required for mapping and the susceptibility analysis, and ultimately a host of factors like the scale and number of quadrangles of maps and data required, appropriate data sources, etc.  The Clear Creek maps were selected as a model for the Project to follow, and while appropriate for the initial public education purpose, were at a scale inappropriate to achieve the accuracy required by the CoSWAP susceptibility analysis method that was finalized after our work was essentially completed.

Particularly important, is the need to identify and characterize ALL data required to apply the SWA methods selected and produce the products specified, prior to attempting to develop a work plan and estimating budget and other resources required to implement it. 

Different stakeholders and different purposes have different cartographic implications.   This is an elaboration of the above themes.  For example, needs related to public presentation and education on a website compared to those related to site-specific resource management have different cartographic implications.  Methods appropriate and most efficient for meeting these needs vary and need to be clear at the outset.  The SWAP program as it ultimately evolved, seems to embody a mix of needs at both these extremes.  It seems that there is potential for producing material suitable for educating the public and motivating concern and action about source water protection using methods that are less complex and data, labor, and technology intensive than what the CoSWAP program requires ideally. 

Acquire and maintain appropriate expertise on your team.  Two interrelated points are made here: (1) have appropriate expertise on your SWA team; and (2) have the resources necessary to acquire and maintain the expertise on your team, or replace it from the open market, if necessary.

One expertise that is frequently overlooked is in cartography, in addition to expertise in GIS.  Geographers trained in both fields suggest that an increasing majority of individuals trained in GIS come to it from other fields and are not trained in cartography.  Expertise in cartography is particularly important given the important SWAP intent to communicate effectively with the public, while developing the analysis and material as efficiently as possible. 

Although having adequate resources in a timely manner is important for all aspects of conducting a SWA effectively and efficiently, the issue is particularly germane with regard to obtaining and retaining GIS expertise.  Individuals with technical expertise in GIS are in great demand; those with expertise in cartography and GIS are relatively scarce and in greater demand.  Anecdotal evidence suggests this is the case even in large organizations, including public agencies with a line GIS support staff.  The Project had retained and lost to other organizations and projects no less than three different sets of GIS technical staff during the approximately six month duration between notification of award and completion of the contract approval process.  The loss of these resources had a ripple effect throughout the conduct of this project, including being at least a partial factor in the Project losing its primary GIS resource person.  A lack of resources prevented the Project from obtaining still more GIS expertise from the open market.

Methods

Attempting to develop a SWA using readily accessible electronic data may not be the most effective and efficient way to develop a SWA for a water system in Colorado.  Frankly, if Project staff had to do the SWA over again, we would do it another way. 

Several issues related to characteristics of available electronic data (discussed below) made the effort inordinately complicated, labor intensive, and doomed to fall short of the CoSWAP ideal that eventually evolved.  Frankly, Project staff and the water system staff wondered if the whole process wouldn’t have developed a more complete and accurate SWA more efficiently, and in less time the ‘old fashioned way’ – i.e., relying on interviews with knowledgeable individuals, use of paper maps and other information, field work, and interviewing facility owners and managers.  Given the final character of CoSWAP, the ideal requires detailed data characterizing each individual PSOC and locational data of relatively high resolution (i.e., sufficient to identify the location of each PSOC relative to each factor for which the ‘sensitivity analysis’ accounts).  It seems that site-specific investigation will be required regardless of the reams of electronic data amassed, numerous manipulations made of them, and hours invested in trying to resolve and reconcile for differences and inconsistencies among them.  Perhaps it would be more efficient to forego this, and just develop the required data through site-specific investigation, then enter the data into a geographic information system.  Or perhaps a limited number of readily available and easily used electronic data sets meeting CoSWAP requirements exist, and perhaps the CoSWAP implementation process or some other effort will identify these -- if a decision is made to identify and characterize available data systematically and comprehensively. 

That being stated, a local jurisdiction still may find it worthwhile to quickly generate a ‘draft list’ of PSOCs or acquire one from a commercial source.  The Project was not able to identify an ideal data source, nor was it able to identify typical costs for available data for surface water sources.  Purchasing data for a surface source water area is potentially costly, regardless if the source is a public agency that charges for the data (e.g., Hazardous Materials and Waste Management Division, CDPHE) or a private business.  Regarding data from private businesses, the alternatives are very limited  (anecdotal evidence suggests maybe only 3-6 companies nationally).  Project staff experience suggests that costs to a local PWS will depend on factors like the number of records ultimately found, the desired format of the data, and/or the size and shape of the source water area.  In addition, the data providers, public or private, may not be able to develop a firm cost estimate at the outset.

Although the Project did not address a ground water source, our sources suggest the cost for obtaining a ‘data dump’ for a ground water source are low – ranging from $125 per search per well, and probably not more than four times this.  The cost will depend on the radius of the search area around the well, the specific types of data desired (e.g., currently permitted facilities, historical data.), and the commercial provider.  Data providers may provide this data only in report format, and not in a digital format. 

Data

Great quantities of data are available.  A large amount of data is available using the Internet.  Seek collections of potentially relevant data.  Most of the base (map) or reference data sets can be obtained from some form of the U.S. Census Bureau's Tiger/LINE data.  A significant amount of watershed data can be downloaded from EPA websites (e.g., BASINS) at no cost (in money!!).  Local agencies, or local units of federal agencies may have required data. 

Time is required to search for data that is appropriate for the specific needs of the SWA.  More time is required to determine if it really is useable, and still more to make it serve the particular purpose of the SWA. 

Sometimes acquiring the data has unforeseeable problems.  Of note, is one set of 1:100,000 data that we downloaded from the USGS Internet site.  The data was provided in three sets of compressed files for the Larimer 1 degree quadrangle.  When the Project GIS staff decompressed the downloaded files, some files were overwritten and data for the middle one third of the quadrangle disappeared.  We invested a significant number of hours on the telephone trying to locate USGS staff with the technical background necessary to discuss, understand, and explore the problem.  The problem resided with the protocol that USGS followed in that they assigned identical file names to specific files contained in different compressed files within the quadrangle. 

None of our efforts gave us the data we needed from this particular source.  Given our product-oriented mission, we had to move on to other alternatives.  We resolved the issue by acquiring other data.  For all we know, the problem lurks still, in the Larimer data, and in others, for the unwary data seeker.

Metadata is not readily useable or available for some data sets.  Many data sets lack straightforward metadata characterizing the data set.  This was found even regarding the basis for the classification schemes used to develop map features.  One example is the data set the Project used to develop the land cover map.  The U.S.G.S. website provided several technical references but did not describe the source data and how it was classified and manipulated in order to derive the parcel records from which the Project developed the map.  Other, locally produced data sets were available, but lacked written documentation regarding how they were developed.  The private business providers of business dataset apparently lack metadata and were able to only provide us with a verbal statement that they “thought the data was two years old”.

Data that a provider says will meet your needs may not be in a format that is useful.

For instance, data sets often include a set of address fields.  Frequently, these addresses are not useable for geocoding; they may be incomplete, inaccurate, or be a mailing address that is different than the location of the feature of interest.  It is useful to review tabular data before trying to plot it.

Attempt to document and remain conscious of specific characteristics of data collected.  Many organizations compile useful sets of environmental data from multiple sources and combine them into a product that they distribute.  However, without careful accuracy and methodology standards they often end up combining data that represents a broad spectrum of accuracy and even meaning.  For example, a list of addresses might contain feature locations, owner addresses, company addresses, and more.

Establish a management protocol at the outset for distributing, reviewing, and using GIS data in a team environment.  For instance, most GIS data tables don't have an explicit index field relating tabular data to the corresponding spatial data.  If tabular data is modified or sorted, it may not be possible to rejoin it with corresponding spatial data. 

Data sets for similar or identical features from different sources for are not likely to be equivalent.  The problems of stitching together data from different data sources is compounded because they may not be equivalent in the categories used, or data characteristics (e.g., scale, accuracy standards, currency).  Many data sets provide coordinate data using different coordinate systems.  This is not a big problem, but requires time for manipulating data to put it into a common format.

Many data sets provide locations using street addresses rather than coordinate data.  Time is required to manipulate this data to provide a coordinate for an address.  This process, called ‘geocoding’ is a potential source of error given that the source software may assign a coordinate for an identical street address at a different location in the SWAA or far outside its boundaries.  Likewise, geocoding may produce coordinates within the SWAA for facilities located outside the SWAA. 

Different data sets from different sources cover different geographic areas.  Electronic data from different sources has different geographic coverages.  None of these may encompass the SWAA entirely, and in all likelihood, a SWA will need to stitch together a quilt from several data sources.  The CLP has two excellent local data sources, the Arapahoe-Roosevelt National Forest (ARNF) and Larimer County GIS programs.  However, data available from each covered most, but not all, of the SWAA, and each covered somewhat different areas.  In addition, the SWAA contains relatively small areas, primarily transbasin diversions, for which neither organization’s data sets provide coverage.  Other data is available not on the basis of jurisdictional or natural physical boundaries, but on other bases, e.g., primarily by zip codes and census enumeration districts.  Using this data involves collecting and integrating the data for zones or districts  that encompass the entire SWAA, then ‘clipping’ it to capture and use that located only within the SWAA.

In the CLP SWAA, this difficulty will be compounded further, if a SWA truly responsive to the needs of local stakeholders is ever is developed.  The actual physical SWAA, not the political construct, extends into several different jurisdictions in the State of Wyoming.  (The federal and state requirements, and legal realities motivating them, to stop the SWA at the state line are hereby acknowledged.)  Integrating this patchwork from multiple data sources is another investment of time required. 

Initially, collect coverages extending a sizable margin beyond the SWAA boundary.  For considerations like efficiency, maximizing presentation of the SWAA on a computer screen, etc., the Project developed initial base map coverages with the minimum possible margin around the delineated SWAA.  After the completing the initial delineation and producing a base map, review by the local water system operator suggested that one transbasin diversion extended further west, and beyond the developed base map.  Project staff had to collect additional data sets and coverages, and develop a revised base map.  Coverages with a margin of safety around the SWAA are advisable too, given the uncertain quality of locational data, and uncertainties related to scale, e.g., concerning the precise location of a feature of interest in comparison to how it presents on the maps produced. 

Much of readily available electronic data is of dubious quality.  Quality assurance appears lacking.  In spite of providing coordinate data, one federal database suggested that identified facilities were verified only to be located in the correct county!

More specifically, the Fort Collins water system operator reviewed positive 'hits' from EPA's BASINS databases of regulated facilities.  Of the 14 records in, or very near the delineated Phase 1 SWAA, none were in reality anywhere near the coordinates that BASINS provided.  The coordinates possibly were as much as 10 - 50 miles away from the actual facility, and at least in one case, far outside the SWAA.  The data sets providing positive hits included: IDF - 11 records at one coordinate, RCLIS - 2 records at one coordinate, and 1996 CWNS - 1 record at 1 coordinate very close to the Fort Collins intake.  The latter is the most entertaining example -- a wastewater treatment plant that in reality is maybe 50 miles further east in the High Plains just south of the Wyoming border.

In addition to locational data for points identified as within the Phase 1 SWAA being 100% inaccurate, this raises the issue of what regulated facilities might be located within the SWAA that the BASINS databases do not contain.

The BASINS mining data, at a general level of accuracy (i.e., the distribution of facilities seems to be in the correct general vicinity), fits with the Fort Collins water system operator’s knowledge based on a lifetime traveling through the watershed.  Further verification of the accuracy of this data probably would require research of each record and field inspection to verify coordinates using a global positioning system (GPS) unit.  Most of this area consists of steep slopes and forest, and may require access by foot, horseback, or helicopter. 

These findings, unless a far cry from the norm, potentially have implications for the national and state programs proposing to rely on these readily accessible data bases of regulated facilities, even for a first cut SWAA.  

With regard to the national pilot, field verification is required to determine if specific PSOCs (at the site level, not at the land use level) are indeed located within the SWAA, and if they are, in which of each ‘sensitivity’ zones.  Site verification is necessary particularly near the boundaries of the SWAA and ‘near’ and ‘far’ zones where the highest concentrations of PSOCs occur in the national pilot SWAA.  The reason for this includes the number of inaccurate locations provided by data sources, the small scale of the Project maps, and the fact that some coordinate locations were developed by geocoding street addresses.  The latter could be an important source of error in the Glacier View Meadows area in particular where in theory geocoding could pinpoint an address on the highway that traverses both sides of the SWAA boundary, but where the actual PSOC is in all likelihood some distance either side of the highway, in or not in the SWAA.  Anecdotal evidence elsewhere suggests that in rural areas, even where an accurate coordinate is derived for a street address, the actual facility may be a mile away from the street.

The data that is available is changing constantly.  Most data providers are constantly adding to and improving the data sets they manage.  In most cases, by the time a SWA process is ‘completed’, additional, better quality data will be available.  Considerable time can be invested in ‘improving’ a SWA every time new or better data is discovered.  One source of data that Project staff found extremely useful is the ‘parcel’ data that the Larimer County GIS director developed from county Assessor’s data long after most Project resources were expended.   

Public agency data providers may not have the capabilities to provide required data in a timely and practical manner.  Project staff experienced examples of public agency staff that did not respond to repeated telephone calls for assistance.  Others expressed what Project staff interpreted as an inability to invest the time required for responding to requests from Project staff.  Others require cash payment in order to provide required data (e.g., Hazardous Materials and Waste Management Division/CDPHE).  Finally, providers may be able to provide data only in a paper format (e.g., Hazardous Materials and Waste Management Division/CDPHE).  These examples raise serious issues regarding the potential ability of contractors and/or local interests to develop SWAs achieving the state’s need to develop a SWA for every public water system in the state, and particularly to the CoSWAP ideal.  These findings raise serious issues concerning a potential contractor’s ability to develop a fixed cost estimate in responding to a CDPHE request for proposals (RFP).

Local agencies were very helpful in providing data and assistance.  Project staff wants to acknowledge and thank Arapahoe-Roosevelt National Forest (ARNF) and Larimer County GIS staff in particular for the time they invested in this Project.  Future developers of SWAs may find local data providers the more productive source for quality data and assistance.

Assessors’ data, and commercially available database of businesses, appear to be potentially useful.  The primary deficiency of assessor’s data is acknowledged -- given its purpose, i.e., of assessing and supporting the collection of property taxes, many think it is generally deficient regarding tax exempt parcels, including those owned by public, religious, and non-profit organizations.  One example are records for parcels owned by religious organizations; in most instances the parcel record does not provide sufficient data to identify if the site is developed, and if so, the use (e.g., church, campground). 

However, assessor’s data is a potential source for detailed data on every tax-producing parcel, potentially with some degree of uniformity throughout the state.  The Project used Larimer County assessor’s data to develop the ‘land use’, ‘ISDS’, and ‘agricultural lands’ maps by aggregating the aggregation of abstract codes that the Larimer County GIS expert developed.  Each parcel in the county is assigned an abstract code, under a system state system of oversight.  In addition, the data set contains other data elements, some of which might be useful, for example, parcel owner contact information, which could aid in surveying them. Given the ultimate state basis for this county level data, use of assessor’s data may hold some potential for use in CoSWAP implementation.  CDPHE might investigate this data system and its potential usefulness. 

The commercially available database of businesses also is potentially useful.  CDPHE and local governments would improve the benefit of using this data source if it revised Tables 4.1 and E.1 by providing Standard Industrial Classification (SIC) codes for the activities they list. 

Some public agencies charge for providing data.  One example is the Hazardous Materials and Waste Management Division/CDPHE.  Developers of future SWAs should evaluate the comparative costs and benefits of obtaining data from public vs. commercial sources.  The issue of data providers charging for data may pose potential problems for a potential contractor developing a fixed cost budget in response to a CDPHE RFP.

Data for many PSOCs are in varying stages of development.  Some data, at the time the Project researched these issues, particularly from public agencies, is not available, or in very rudimentary stages of development, and is not useable readily.  Specific examples noted include state data on agricultural chemicals application, and EPA data characterizing Class V Underground Injection Control Wells.  Some of the major quality assurance issues of the ‘better’ data sources are mentioned above.

Cartography

Obtain input from a cartographer in developing specifications for cartographic products, and obtain review from a cartographer of cartographic products.  Cartographic design principles are critical in producing a functional and user-friendly product.  Determine the needs of your end-user and design cartographic products to meet these.  Many of the aspects of good map-making are very subtle and have nothing to do with the mechanics of GIS.  Consider the color, shape, placement, etc. of each feature and how they relate to one another.  Avoid notations, abbreviations, and other labels with which the user is not familiar.

Use only a scale and level of precision that is appropriate for meeting the SWA needs.  Do not incur the extra costs of surpassing this.  The buffer zone specifications of the ‘sensitivity’ component of the CoSWAP susceptibility analysis exercise will be a primary driver of the resolution required in future SWAs developed under the state framework.  CoSWAP specified the buffer zone requirement long after Project map products were specified at a much smaller scale and the course for developing them set.  Project maps represent general locations only, and only to the degree of accuracy inherent in the data sources used.  The level of accuracy is sufficient for achieving the primary public education purpose of the Project as initially envisioned, i.e., of presenting maps on a computer screen.

Develop data and map only the minimum number of features necessary to accomplish the analysis meeting the needs of a specific SWA.  Develop a good base map before beginning to plot data required for analysis.  Ensure that it depicts all features required, in order to maintain a consistency in subsequent maps.  Strive to keep the SWAA on the base map as uncluttered and empty as possible, so that it can accommodate the colors, symbols, lines, and labels of the subsequent thematic overlays.  Include only the most significant reference features that are necessary for the end-user to locate the thematic features of interest.

For considerations of good cartographic presentation, collect, develop, and present base map features covering the entire rectangle (landscape or portrait) encompassing the SWAA. 

GIS

GIS files are prone to corruption.  GIS software has some weaknesses and it is sensitive to a number of factors.  The numerous data files being used can often exceed several hundred megabytes each.  The demands placed on the software can make them unstable and prone to corrupt files.  Adequate hardware is necessary and can speed data processing significantly.  Consider using separate project files for each map product to limit the implications of file corruption.

All team members should have access to maps and data using the GIS system.  It is probable that team members with the expertise required to make the decisions regarding tabularized data will not be the same as those having expertise in GIS.  The SWA will progress more efficiently if the former have direct access to GIS so they can review and investigate new data as it is collected and processed.

Technology is changing constantly.  Hardware and software may become available during the course of the SWA that was not available at its beginning or when tasks were completed.  The situation creates conflict between the needs of completing a SWA within fixed parameters and those to develop the best products possible given available technology.  Many shortcuts and resources exist that can save time and money.  EPA's BASINS package includes GIS software capable of handling simple SWAs.  Most GIS software has extensions that can be obtained to extend their functionality.  GIS software also allows for using a number of ‘visual tricks’ reducing data processing needed.  For instance, a second hill shade grid can be created with a hole in it to simulate the effect of clipping your coverages.  Also, layering two identical feature layers with different symbols can create the effect of a new, unique, symbol that cannot be otherwise produced.

Delineation

The most efficient method for delineating the SWAA will depend on the character of needs regarding a specific SWA.  For a single, simple SWAA, the delineation may be accomplished most efficiently the ‘old fashioned’ way by drawing it on a paper map of appropriate scale, then digitizing the boundary.  In other instances, recently developed computer models are available for delineating natural watershed boundaries. 

Contaminant Inventory

Include a classification for ‘forest land cover’ and/or ‘wildfire hazard’.  The Fort Collins water system operator has perceived this as the most significant potential threat to source water quality in the SWAA because of the potential for a catastrophic wildfire!

The Fort Collins water system operator’s recent experience reacting to and dealing with the consequences of the Bobcat Fire have increased his concern about this issue and the potential of the state and national SWAP programs to address it.  High intensity burn areas of the Bobcat fire are contaminating the Big Thompson SWA (another source of water for Fort Collins and most of the North Front Range region) with runoff that contains at contaminants at levels that downstream water systems never have confronted previously.  The contaminants include: total organic carbon (TOC), manganese, iron, zinc, alkalinity, pH, total suspended solids, ammonia and other nitrogen species, and phosphorus.  The contamination is so bad that water supply managers have had to set-up an automated early warning system to notify downstream users to close their intakes to withdrawing fire contaminated runoff into their reservoirs and treatment works.  The emergency reaction is the same that is taken when an eighteen-wheeler hauling fuel or hazardous chemicals spills into the river.  In this case, however, the emergency response will need to be repeated every time it rains for at least the next three years.

The contaminant inventory needs, first, to identify forest cover, then, where it is identified, develop additional detailed data for characteristics that relate to wildfire hazard potential, and ultimately to the potential implications of wildfire for water quality. 

The business database did not identify gas stations.  Local reviewers of the business database identified the possible presence of gas stations.  Project staff was unable to verify the presence of these by field inspection.  The business classification scheme of the database identifies the primary business type, e.g., lodge, store.  

Table 4.1 is problematic in its application to the contaminant inventory and susceptibility analyses.  CoSWAP Table 4.1 is not congruent with CoSWAP Table E.1, nor is either classification scheme congruent with those many data sources use. 

Experience using Table 4.1 suggests that the table needs additional refining.  This is one area where expertise in cartography and land use classification schemes might prove beneficial to CoSWAP.  Given that the classification schemes of most data sources are not congruent with those Table 4.1 uses, consistency in SWAs under the CoSWAP framework would be promoted if the Table provided a definition for each classification and its associated SIC codes. 

Some questions that occurred to the analyst using the table follow.  How is CoSWAP distinguishing between “government installation” and several potentially publicly owned and/or operated “residential / municipal” facilities?  How would Table 4.1 suggest classifying a rural, privately owned, commercial golf course?  What is the distinction between ‘lagoons and liquid waste’ and ‘wastewater’, and ‘wastewater’ and ‘septic systems’?   What is a ‘managed forest’?  (Is there a forest in the continental United States that is not managed?)  How is CDPHE distinguishing ‘managed forests’ from ‘crops, irrigated and non-irrigated’ agriculture?    Does this include all land covered by forest and managed by pubic agencies, including ARNF, RMNP, Colorado State Forest Service?

Table 4.1 does not contain a ‘public’ or ‘government’ category.  Many data sources classify public uses or activities separately from the major classifications of Table 4.1.  Government PSOCs are interspersed between “commercial / industrial”, “residential / municipal”, and “agricultural/rural”.  Publicly owned lands depicted by the Project’s land ownership and land use maps (including agricultural lands map) and the data sources, from which these were derived, certainly include facilities that Table 4.1 classifies in each of its three major classifications.

CoSWAP Table 4.1 lacks clarity in how to classify ‘park’, ‘recreational’, and ‘campground’ uses.  Depending on specific characteristics of a particular enterprise, it seems that enterprises having similar water quality implications could be classified in any of the three major classifications of Table 4.1.

CoSWAP Table 4.1 does not contain classifications for PSOCs that local stakeholders perceive as threats to source water quality.  One agricultural PSOC that local stakeholders perceive as a significant potential threat is ‘grazing’.  

Attempting to sort out what is the same facility and which characteristics represent one or more PSOCs in different data sets is extraordinarily time consuming.  A protocol needs to be established for identifying each specific PSOC and characteristics of concern, where these may be identified differently by different data sources.  Perhaps this is a semantic problem.  Is a PSOC one site with multiple potential contamination concerns?  Or is each potential contamination concern a separate PSOC?  The issue is exacerbated in that different data sources might identify one site differently.  The issue is exacerbated further because of uncertainty regarding data quality, e.g., where a data source identifies several PSOCs at the same coordinate. This issue too suggests that the SWA process may have potential to be accomplished more effectively and efficiently by doing a site specific field survey and not attempting to use, verify and correct, and reconcile data from a number of electronic data sources.

Some discrepancies seem to be present between the ARNF facilities depicted on the paper forest map in comparison to the electronic database of ARNF sites.  Some facilities on the ARNF paper map do not seem to be identified in the electronic data source, and some facilities present in the database were not found on the paper map.    Some facilities appeared in different locations on the paper map in comparison to that generated from the electronic database (e.g., waste-transfer station).

Susceptibility Analysis

The CoSWAP susceptibility analysis method seems intended to account for several potential concerns initially, but in its operationalization accounts only for health concerns.  Specifically, CoSWAP states “The susceptibility of a PWS will be determined by the possibility for a PWS to draw water that potentially could be contaminated at concentrations that may pose a concern to consumers of the water.”  Later, in developing its ‘threat’ and ‘hazard’ concepts, CoSWAP considers essentially only the “potential health implications of exposure to each” chemical.  The CoSWAP methodology excludes consideration of existing and potential contaminants that may have implications about which consumers are or may be concerned (e.g., aesthetic), but that do not pose a health concern.  However, the CDPHE hopes that these concerns can be accommodated when the time and resources are available to address these. 

Investing the effort required to complete the CoSWAP susceptibility analysis exercise based on incomplete electronic data of dubious quality may be an effort disproportionate to the value added.  The CoSWAP susceptibility analysis method in its ideal requires comprehensive, detailed, and accurate data regarding each PSOC and its location, particularly relative to surface waters.  Given that there are fundamental uncertainties about the basic data upon which the analysis is built (e.g., Are the PSOCs really located where data indicates they are located?  Are the PSOCs even located within the SWAA?), the results, and value, of this third step are uncertain at best. Completing every cell for every PSOC in every matrix required by the CoSWAP method is labor intensive.  Does the exercise, based on data of uncertain quality, produce value proportionate to the level of effort required to complete the matrices?  Could less labor-intensive methods produce proportionate value, more efficiently? 

Table E.1 is not congruent with Table 4.1 or all data sources.  In developing the ‘Threat Identification Matrix’, Project staff quickly noticed that a one to one matching of the classifications in Table 4.1 with those of Table E.1 is lacking.  The difficulties inherent in trying to use these required resources is exacerbated further given that these classifications do not necessarily match the classification schemes used by data sources.  For example, the analyst questioned if the business data base ‘tool and dye’ record should be matched with contaminant hazards of Table E.1 ‘machine shop’ or ‘metal plating/finishing/fabricating.  In addition, Table E.1 contains PSOCs that Table 4.1 does not contain (e.g., cemetery) that may be important to the analysis, and in some instances are considered important by local stakeholders (e.g., grazing), or EPA (e.g., wells).  In addition, Table E.1 does not include activities identified by the business data base (e.g., exterminator) that Project staff judged of potential concern, or PSOCs identified by Table 4.1 as high priority (e.g., AST, UST, LUST).  Attempting to match, and decide on the most appropriate matching among classifications of data sources, Table 4.1, and Table E.1 is time consuming, frustrating, and fraught with potential for inconsistent application in future SWAs.  To increase the probability that analysts conducting future Colorado SWAs apply the CoSWAP susceptibility method consistently throughout the state, CDPHE might consider providing a definition of each PSOC and associated SIC codes, as well as making Tables 4.1 and E.1 congruent with one another, and with data sources to the extent practicable.

The parcel data file does not always provide sufficient information to identify the type of PSOC present.  In addition to information deficient for the purposes of applying the CoSWAP susceptibility analysis method concerning tax-exempt parcel records, the parcel file does not contain sufficient information characterizing many commercial and industrial parcels to determine the type of PSOC present.

The criteria for assigning ratings in different portions of the CoSWAP susceptibility method could be defined more explicitly.  The Project generally assigned default values in the evaluative matrices because of a lack of verified site-specific information.  However, if CDPHE wants the method applied consistently by several analysts to all water systems throughout the state, decision criteria seem to require additional specification, and, in particular, this specification needs reflect the reality of the data potentially available to support the analysis.  For example, what are the specific decision rules to distinguish between a ‘likelihood of release’ rating of ‘likely’ vs. ‘unlikely’?  What construction materials, of what age, maintenance, and other factors support one rating as opposed to another?

Table 5.9 ‘Vulnerability as a Combination of Threat and Risk – Surface Water Supply Systems’ decision table seems logically and pragmatically problematic – i.e., not all ‘Highs’ and other ratings are equal – ultimately it all depends.  The Project analyst explained his concerns to the Colorado SWAP Coordinator verbally and in a series of emails.  The crux of the problem seems to do with the fact that the multi-attribute evaluative matrices that precede this one in the CoSWAP susceptibility analysis matrix process are each intended to reduce a relatively large number of variables to one manageable rating, i.e., a ‘threat’ and a ‘risk’ rating, which are then intended to be ‘combined’ in this third matrix.  As Table 5.9 now stands, it seems logically inconsistent in that identical combinations of ‘threat’ and ‘risk’ yield different vulnerability ratings.  Admittedly this is because the like ratings really are unlike one another, based on factors accounted for, and supposedly reduced into one rating, in earlier matrices.  In addition to this logical difficulty, the method poses the potential pragmatic one that an uninitiated analyst who lacks prior experience filling in the matrices, and does his or her first in sequence, will get to the vulnerability matrix only to discover that he or she must did most of the entering of ratings in cells for naught and must reconsider most of the variables again that he or she presumably accounted for previously.

The culmination of the CoSWAP susceptibility analysis method seems to weight nonpoint and point PSOCs equivalently.  For example, given this analyst’s understanding of the matrix exercise, agricultural lands and roads are given one tally each, just as one leaky underground storage tank is given one tally.  The same single tally is given regardless of the number of acres and quantity of chemicals applied to the agricultural land, or the miles of roads regardless of traffic loads and actual or potential number and severity of spill incidents.  This method seems to understate the potential threat posed by nonpoint sources.

Use of the evaluative matrix method is susceptible to error.  The use of the spreadsheet format to develop the susceptibility analysis seems to hold high potential for entry of erroneous ratings.  CDPHE should consider developing another method for data entry, ensuring that the desired rating is entered in the appropriate cell.

RECOMMENDATIONS FOR REFINING SWA

General

Develop an assessment of local stakeholders’ needs related to source water information that is comprehensive, in-depth, and thorough.  The Project accommodated the potential needs of local stakeholders as well as it could within the parameters constraining it, and CoSWAP and budget in particular.  Regardless, as the Project progressed, it seemed more apparent that the products, if completed, would meet needs related to public presentation well, but not meet those related to site-specific management of resources and PSOCs.  I t remains to be determined if the ideal CoSWAP requirements, if fulfilled, meet these local needs.

Develop SWA for the entire SWAA delineated.  This includes the SWAA area contributing to the City of Greeley intake, i.e., the CLP North Fork and Hansen Canal – CBT.

Develop the source water assessment and maps at a larger scale (e.g., 1:24,000 or larger).  All maps for the entire SWAA would need to be developed at a larger scale and achieve the appropriate accuracy standard to achieve requirements of the CoSWAP susceptibility analysis method that evolved ultimately.  However, of particular concern, are the areas shown to have a high concentration of point source PSOCs present.  These are primarily areas along the CLP mainstem/Colorado Highway 14, Glacier View Meadows, Pingree Park, area nearly surrounded by the Comanche Peak Wilderness, and Rist Canyon Road/Stove Prairie area in the southeastern portion of the SWAA.

Delineation

Develop maps at larger scale, and the hydrology layer specifically, to comply with CoSWAP susceptibility analysis requirements, and sensitivity requirements specifically. 

Add an overlay for the location of public water system source wells.

Contaminant Inventory

Continue developing the contaminant inventory based on additional data sources.  The Project was not able to develop data for all features and PSOCs, or coverages for all areas within the PSOC even where it did develop data for a particular feature or PSOC for most of the SWAA.  The Project was not able to identify and review all available data sources, electronic or ‘paper’, public or private.

Conduct field site inspection to verify or correct data developed from readily accessible and useable electronic.  Notable is the need to verify if particular PSOCs near SWAA boundaries are in fact located within the SWASS, and the need to verify if particular PSOCs near the ‘near/far’ boundary are rated accurately in the ‘sensitivity’ exercise.

Conduct field survey and/or interview facility owners or managers to develop the facility specific data the CoSWAP method requires.  Examples that seem particularly important are owners or managers of relatively large tracts of land, e.g., CSU regarding facilities and PSOC on its Pingree Park campus; ARNF regarding application and storage of agricultural chemicals, vehicle and equipment storage, storage tanks, etc.  To achieve the ideal CoSWAP requirements, eventually every owner or manager of every PSOC will need to be interviewed.

Susceptibility Analysis

Revise the susceptibility analysis as better data is developed.

Automate the susceptibility analysis methodology to ensure consistency for all PSOCs within the SWAA, and with SWAs throughout Colorado

Review our draft GIS maps in various states of completion...