EPA`s BASINS Model

advertisement
EPA’s BASINS MODEL –GOOD SCIENCE OR SERENDIPITOUS MODELING?
Ray C. Whittemore 1, John A. Beebe 2
National Council for Air and Stream Improvement, Inc.
ABSTRACT
Better Assessment Science Integrating Point and Nonpoint Sources (BASINS) is a geographic
information system based tool developed by EPA's Office of Water to help states more
efficiently target and evaluate water-bodies that are not meeting water quality standards.
BASINS (EPA, 1996a, 1998) brings together data on water quality and quantity, land uses, point
source loadings, and other related spatial data with supporting nonpoint and water quality model
at a quicker and more effective pace. EPA developed BASINS, a geographic-based watershed
tool to better integrate point and nonpoint source water quality assessments for the Nation’s
2100+ watersheds. In its zeal to achieve this endpoint, EPA has initiated a simplistic approach
that was expected to grow through scientific enhancements as TMDL developers become more
familiar with modeling requirements. BASINS builds upon federal databases of water quality
conditions and point source loadings for numerous parameters where quality assurance is suspect
in some cases. Its design allows comprehensive assessments and modeling in typical Total
Maximum Daily Load (TMDL) computations. While the TMDL utility is the primary reason
BASINS was developed, other longer-range water quality assessments will become possible as
the Agency expands the suite of assessment models and databases in future releases. The
simplistic approach to modeling and user-friendly tools gives rise, however, to technical and
philosophical concerns related to default data usage, seamless generation of model input files,
and the failure of some utilities to work properly suggest to NCASI that serious problems may
still exist and prompts the need for a more rigorous peer-review. Furthermore, sustainable
training becomes paramount, as some older modelers will be unfamiliar with (Geographic
Information System) GIS technology and associated computer skills. Overall, however,
BASINS was judged to be an excellent beginning tool to meet the complex environmental
modeling needs in the 21st Century.
Key words – watersheds, modeling, GIS, TMDL, quality assurance
INTRODUCTION
EPA’s Watershed Initiative, started in the early 1990s, provides a regulatory approach to
environmental protection programs different from the ‘command and perform’ scenarios of the
Clean Water Act in 1972. The new approach places the entire watershed as the unit of regulatory
focus rather than individual rivers or stream segments. While this approach has garnered
worldwide attention and support, some still question its wisdom given that some states have
effective nonpoint source control programs in place and water quality in most locations is
1
Principal Research Engineer, National Council of the Paper Industry for Air and Stream Improvement, Inc., Tufts
University, Medford, MA
2
Research Scientist, National Council of the Paper Industry for Air and Stream Improvement, Inc., Tufts University,
Medford, MA
1
gradually improving (per recent EPA 305b report) using wasteload allocation approaches of the
1970s and 1980s (EPA, 1998a).
States are presently required by the Clean Water Act to determine the total maximum daily load
(TMDL) of pollutants in areas where technology-based controls alone are not adequate to
achieve compliance with water-quality standards. The federally mandated requirement is to
develop wasteload allocations specifically for point sources of pollution and load allocations for
the nonpoint (diffuse) sources. If a State fails to meet these requirements of the Act, EPA is
mandated to assume responsibility and to develop TMDLs as necessary to carry out and meet its
goals. In 1996, EPA was defending itself in litigation around the country brought by
environmental groups challenging the slow pace of TMDL development. In response to growing
pressures to develop Total Maximum Daily Loads (TMDLs) assessment tools, EPA created a
new flexible modeling system (BASINS) to more effectively integrate assessment of point and
nonpoint pollution sources. While the events that prompted the BASINS research and
development justified the attention, some TMDL process and modeling related tasks are
currently still controversial and will continue to confound watershed stakeholders for many years
to come. Unquestionably, a set of easy to use and robust water quality assessment and modeling
tools are needed for quantitative watershed assessment on the scale of the present TMDL needs.
Will BASINS ever become this tool? The answer is probably ‘yes’, but EPA needs to more
seriously address quality assurance and provide sustainable technical support and training for
an increasing array of users.
BASINS OVERVIEW AND HISTORY
BASINS v.1.0 (EPA, 1996a) was the inaugural tool that brought together data on water quality
and quantity, land uses, point sources, and a limited suite of water quality models to perform
watershed-scale assessments. The current suite of models is expected to expand as users develop
experience, gain familiarity, and master skills with GIS geographic information system-based
approaches to watershed assessment and management. This is an important and highly
significant trend in environmental management that emerged in the late 20th Century. Extremely
useful utilities in the BASINS’ tools are the water quality assessment tools – TARGET,
ASSESS, and DATA MINING.
As with most GIS-based approaches, powerful display and summary graphic information is
created for visual or written reports. This feature is critical in the watershed assessment arena
and makes BASINS worthy of attention and the time investment to learn its functionality. If
these utilities are to provide the foundation for reliable water-quality assessments, their quality
must be assured through the established peer-review process. To date EPA has not promoted
comprehensive peer-review as a priority and, thus has not forced a critical evaluation of the
databases inherent to BASINS. Instead, the approach has been to rely upon the respective
developing programs such as PCS (Permit Compliance System), STORET (STOrage and
RETrieval database of water quality data), and NSI (National Sediment Inventory) to develop
assurance measures.
In mid-1998, EPA commissioned a minimal peer-review of BASINS, nearly two full years after
its public release and distribution. The assignment to the three reviewers included: (a) a review
2
of the BASINS package consisting of the choice of data layers, the analytical and reporting
functions, and the linkages between the components; and (b) an evaluation of whether BASINS
provided a sound, credible tool to EPA Regions, states, and local governments for developing
TMDLs for a large percentage of their 303(d) listed waterbodies. The objective did NOT include
evaluation of commercial software application (ArcView 3.0a/3.1), the national spatial
databases, the models themselves, or most importantly the array of default parameters meant to
give the new user a ‘quick start’ in model development. Because only three reviewers were
included, robust and unequivocal conclusions were precluded. The sample of three was,
however, supportive of BASINS’ development in the TMDL program.
While good science in modeling practice would dictate that the initial data choices be critically
examined during model calibration, it will not always happen in some priority, time-sensitive
situations. In spite of advances in model sophistication, including complex models such as
HSPF, effective calibration is still an ‘art’ where multiple parameter datasets can achieve the
same degree of simulation matching to monitoring data. In the peer-review report (EPA. 1999),
EPA notes that both Qual2E and HSPF are established models with prior peer reviews and that
other EPA or Federal agencies are responsible for data QA/QC. In effect, EPA has concluded
that a majority of the BASINS users will have sufficient professional judgment and insight to use
the system wisely and to fairly represent the results to the broad array of TMDL stakeholders.
This laudable expectation hopefully will be realized in the coming years and will be subjected to
continued scrutiny by all users, including state and federal agencies, and academia. There are
those who would argue, however, that there are too many TMDLs to be developed for the
universal application of good science and that quality compromises will occur. It is further
interesting to note that despite the perceived widespread acceptance and use of BASINS, few
reports have been published or are available on BASINS’ applications. Publication of varied
technical experiences and problem solutions is an effective means of technology transfer and is
at the core of scientific advancement.
BASINS WATER QUALITY MODELS
The current version of BASINS is built around the models - HSPF, QUAL2E, and
TOXIROUTE. Except for TOXIROUTE, these models represent state-of-the-art efforts from the
EPA Center for Exposure Analysis and Modeling (CEAM) in Athens, Georgia (EPA, 1993) for
nonpoint, point sources, and toxic compounds, respectively. The nonpoint source model is
version 11.0 HSPF (Hydrologic Simulation Program Fortran), and called “NPSM” within the
BASINS interface. It is a critical component of BASINS because it simulates the meteorological
and land-based processes important to the understanding the sources of nonpoint source loadings
required in TMDLs. HSPF is a PC-based adaptation of the Stanford Watershed Model, the
premier hydrologic land-based simulation model of the 1960s and 1970s (Crawford and Lindsey,
1976), and has a substantial following, in spite of the fact that it is hard to understand and
effectively use. QUAL2E (Brown and Barnwell, 1985), on the other hand, has become the
model of choice for point source modeling of organic and nutrient pollutants and used
extensively for wasteload allocation in the past three decades. The remaining model,
TOXIROUTE, is a simplistic toxic fate model with limited chemical fate simulation capability
(GSC, 1993). In spite of the use of its predecessor PROUTE (an older EPA dilution model), it
3
lacks current stature within EPA and will be discontinued in future BASINS releases (Kinerson,
2000).
HSPF is a lumped parameter, continuous simulation nonpoint source model that provides
powerful simulation potential when used by knowledgeable personnel. Achieving this state of
expertise, however, still remains a formidable task even to experienced modelers. Expertise in
this area amongst state modeling staff is highly variable. Maryland, because of its proximity to
EPA’s Chesapeake Bay Program Office in Annapolis where HSPF has been applied many times,
is one of the few states with this expertise. Other states rely upon consultants or still use
empirical washoff and loading functions, common in the decades of the 1960s, 1970s, and 1980s.
Currently, EPA is not aggressively addressing these limitations in its training schedules,
primarily because of funding limitations. Unfortunately, users of BASINS/NPSM are quick to
assume that the problems they encounter are HSPF problems. This is not necessarily a correct
assumption as, in our testing, most problems were frequently correlated with the BASINS
interface itself or user specific setup issues. Those that follow the EPA BASINS ListServer on
the BASINS web site, an interactive user communication medium, have noticed that many
problems that are related to importing new (non-default) or local data and other problems related
to installation. Early NCASI experience echoes these problems and strongly advocates more
complete user guidance, including case studies and tutorials to alleviate many of these problems
by advancing detailed guidance. EPA’s posting of a series of technical notes on the BASINS
web site has addressed some of these issues, but more is needed.
HSPF has undergone a series of updates and code changes since its adaptation from the early
Stanford Watershed Model and predecessor HSP (EPA, 1996b). Numerous applications in the
1970s are significant because they were the driving force behind making HSPF the Agency’s
premier nonpoint source model. Furthermore, these numerous experiences have led to HSPF
becoming a robust watershed analysis tool. Both EPA and the USGS have adopted HSPF in
their watershed analyses and will continue to improve features and promote its widespread use.
In fact, the Army Corps on Engineers at the Waterways Experiment Station in Mississippi has
recently endorsed HSPF and will devote its modeling resources to HSPF development
(Donigian, 1998). Complex code in models such as HSPF effectively precludes a detailed
examination of code integrity, such as that performed on QUAL-II by NCASI (NCASI, 1984).
One should not assume that the code is infallible, as some combinations of model processes
could be incorrectly coded, or otherwise problematic as was found in the NCASI analysis of
QUAL-II and more recently in BASINS/NPSM trials. Stakeholders in contentious TMDLs will
demand that the tools be of unquestionable veracity when faced with expensive treatment
upgrades or restrictive land management BMPs.
A major criticism of HSPF is that it is a non-distributed model and is, therefore, unable to handle
the variety of site variables important for assessing multiple land use management effects at a
meaningful field-scale.
This includes variations in topography, soils, vegetation, and
management activities (including roads). A second important problem is that the model is not
physically based. This means that even though you can accurately calibrate the model (which
can be done by manipulating the numerous coefficients), it can be difficult to understand and
communicate changes in hydrologic or water quality function associated with the management
action. The linkage between the change in site properties and the model coefficients is either
4
poorly defined or not well understood by many, in spite of its 30-year history. Given the
regulatory uses of such models in the years ahead, the Agency needs to foster stronger links
between model parameters and algorithms and management practices (BMPs) so that TMDL
implementation has a more direct physical linkage to the analytical tools used in their
development. This will facilitate better lay-public stakeholder understanding and acceptance.
QUAL2E developed in the 1980s by EPA, Tufts University, and NCASI is an evolutionary
product of a stream water quality model introduced by the Texas Water Development Board with
later modifications by Water Resources Engineers. A Windows version was released in 1995
(EPA, 1995). The core model (and various nearly equivalent state versions) has been used
extensively for waste load allocation studies in the U.S. and for river water-quality assessments
worldwide since it’s development in the late 1970s through the late 1990s. EPA and others
promoted this utility through hands-on stream modeling workshops in the mid- to late-1980s,
which regrettably have stopped. The model is consequently well known in state agencies and
used preferentially over others in waste load allocation and TMDL assessments. This type of
consistent training is needed for BASINS. Although it will no longer be free, NCASI was
encouraged by recent Agency announcements on the BASINS web page regarding training
opportunities in academia through mid-2001. This level and quality of hands-on training is
representative of what’s needed for the BASINS toolkit to achieve its promise. Additional
further outreach to academia may prove productive by developing BASINS skills in waterquality personnel early in their career development and challenge faculty to develop research
enquiries.
Within BASINS the QUAL2E model is parameterized by nontransparent data inherent to EPA’s
river reach file (RF1) and includes data necessary to simulate both hydraulics and water-quality.
Rather than creating the input file by examining the characteristics of each reach separately,
BASINS extracts the necessary information and seamlessly creates this file. While the user can
still edit it on an individual reach basis, the initial effort of configuring the river system,
assigning connectivity, and populating the model parameters provides the modeler with a unique
opportunity to better understanding of the physical system and, therefore, a higher comfort level
with the application. While this utility is of significant help to new HSPF users in dealing with
cumbersome input format, the essence of the criticism is equally valid for both models. The
seamless utility of the BASINS user interface, while easing the burden of data entry into an
exacting file format, is viewed suspiciously by many experienced modelers and judged a poor
idea by others. NCASI supports these viewpoints, as modeling should not be made easy at the
expense of good science. Some industrial stakeholders’ fear that some TMDL modeling
assessments will be formulated with minimal effort and conceivably could be only desktop
activities.
As noted earlier, TOXIROUTE will be discontinued in future releases of BASINS in FY
2000/2001. In its present form TOXIROUTE is too simplistic for all but trivial screening level
applications, primarily because many toxic substances interact with sediment and biota in a
complex way. No interactions between toxic pollutants and other water quality constituents are
allowed in TOXIROUTE.
5
DATABASE QUALITY ASSURANCE
At the cornerstone of quality assurance, BASINS’ users rely upon the completeness and overall
accuracy of background GIS data used in the program as default information. Over the past
several months, EPA’s Office of Water has compiled and made available metadata records for
most of their geographic coverage. Many of these records, unfortunately, are incomplete or
suggest data quality that does not comply with the content of the standard for digital geospatial
data (USGS, 1998a). In a more detailed assessment (NCASI, 1997) evaluated data in one
database (the Permit Compliance System - PCS) and found the following types of errors (see
Table 1) for 50, randomly selected, pulp and paper mill facilities.
While the implications of these errors will vary according to facility, some can lead to erroneous analyses
and unfairly influence regulatory decision-making. It is presently unknown how widespread these errors
are across all industrial categories, but the fact that they may still exist in this one industry’s database is
cause for caution in general. It is known that some paper industry facilities reported problems to
state/federal agencies with unknown results. In every case, the modelers developing the TMDL and the
stakeholders affected by the analysis have a responsibility to acquire accurate information and to note
problems where they occur. Since the BASINS software is available to anyone free-of-charge, rogue
analyses are always possible and even probable in some adversarial situations. The general public is ill
equipped to distinguish the good-science from inappropriate analysis when it comes to technical
watershed assessment and modeling issues. Most water quality professionals would acknowledge that
these resources are not infallible and error free. All of the spatial data products imported into the
BASINS toolkit should be part of an ongoing quality assurance process. In doing so, all spatial data
should retain original metadata records to maintain government compliance standards set forth by the
Federal Geographic Data Committee and should evaluate usability of data by completeness of these
records (USGS, 1998a).
Table 1: Paper Industry PCS Error Summary (NCASI, 1997)
CATEGORY
# of MILLS
No Errors
5
No Typos
9
Minor Typos (defined as those that are less
than one order of magnitude from the
expected value, not easily detected, or have
no impact on the resulting use)
< 5 records containing errors
16
> 5 records containing errors
10
Major Typos (defined as those greater than
one order of magnitude from the expected
value)
< 5 records containing errors
15
> 5 records containing errors
2
Unit Errors (incorrect units)
9
Missing Data (data submitted but not
2
recorded)
Truncated data (data that exceeds the
2
database field width – generally those with
6
qualifiers such as > or <
Limit Errors (those that incorrectly state
permitted limits stated in DMR reports)
3
WATER QUALITY DATA PROVIDED IN BASINS
While any one watershed (also known as a cataloging unit) may contain over one hundred water
quality observation stations, only 15 (or less) per watershed were selected for use in BASINS,
because of space limitations in the CD ROM distribution format. The observation stations
represented were “selected (by EPA) because they have more data than the average water quality
station and thus provide a better starting point for comparisons with simulation model output
(e.g., QUAL2E, NPSM).” It would appear that, based on the location of the stations in the
vicinity of the test watershed examined in this evaluation, that little or no consideration was
made to choose stations that were evenly distributed throughout the watershed.
In the test watershed (Figure 1), the data from the 15 observation stations examined represented
daily observations for many different water quality parameters and the frequency of these
observations were highly variable per station and year. For example, while there were a total of
782 dissolved oxygen observations distributed among nine of the 15 stations, there were no BOD
observations for the entire watershed. Additionally, at one station there were 60 observations in
one year for one parameter and then only 12 the following year for the same parameter. These
inconsistencies would make it difficult for any modeler to calibrate a subset of the water quality
parameters for the watershed in question. Confounding things further, the arrangement of the
water quality stations was such that it did not account for many of the main tributaries. Note the
darker streams that are accounted for by STORET WQ stations (dotted circles) compared with
the lighter streams in. It is also interesting to note that, out of the 15 stations located in this
particular watershed, there is not a single one that is located at the pour point of the watershed.
This would make it very challenging for modelers to calibrate a watershed of similar size and
structure at the scale that was originally intended for BASINS applications.
Figure 1
Utility of STORET Water Quality Stations in BASINS
7
In addition to the potential shortcomings of the PCS database, a number of other limitations have
been revealed by NCASI Staff, EPA and other users of the BASINS modeling package. Table 3
highlights some of the key limitations of several datasets included as default information in
BASINS. Additional information is available at http://www.epa.gov/OST/BASINS/metadata.htm.
Table 3. Key Limitations of Default Datasets in BASINS.
Dataset
Landuse
Watersheds
Elevation
Soils
Rivers
Streams
Segments
Hydrology
Weather
Limitation
Limited spatially to 1:250,000 and temporally from 1977 to the mid-1980’s.
Use with large scale data (i.e., 1:24k hydrography) is beyond resolution
capabilities.
Resolution of data (300 meters) is unsuitable for smaller subwatershed
delineations.
Positional accuracy between physical boundaries and digitized locations is
unknown.
Flow data not updated since 1982 and some RF1 reaches not reported (see
list below).
Stream locations are missing, disconnected or flow direction is incorrect.
Channel geometry data is missing or misrepresentative of real physical
conditions.
Gauging stations with missing data or observations required general
estimates1.
Station locations cannot be averaged and data cannot be written or
manipulated.
Lack positional accuracy information and do not follow National Map
Accuracy Standards. National Map Accuracy Standards are established for
the federal government by the U.S. Geological Survey,
http://mapping.usgs.gov/standards/ .
Dams,
Discharge
Sites, Toxic
Releases, and
Industrial Sites
(1) EPA needed to estimate flows for more than 60,000 reaches because it needed flow data on
all RF1 reaches and the USGS had consistent streamflow data on fewer than 2,000 of them.
Several other limitations and error sources of the RF1 database have been revealed and could be
significant in some cases of modeling for TMDL development. Based on metadata compiled for
the RF1 spatial database by the EPA ,an abbreviated list of spatial and temporal limitations is
presented here:
 Data were needed at the downstream ends of the reaches rather than where the USGS
gauges were typically located. All of the RF1 flow data for these locations are
estimates.
 Vast areas of the country were without USGS data for drainage areas under 500
square miles.
 Approximately 4,000 gauges did not have useful flow data but were used to assign
drainage area to reaches in RF1.
8
 Because of data skew and anomalous conditions at many of these gauges, graphical
techniques were applied to the solution of the mean annual and 7Q10 flow estimates
at the gauged sites.
 Overall flow estimation methodology used in producing the RF1 flow estimates was
not designed to produce accurate results on start reaches or small un-gauged
tributaries, nor in estuaries or un-gauged coastal streams. Thus the accuracy of the
flow estimates in these types of streams is not expected to be adequate for many
applications.
 Factors such as irregular density of reaches in a particular area of a state or data
problems within the period of record for a particular gauge, may introduce local
variations in the accuracy of these flow estimates.
 The USGS Water Resources Division cautions users that since the RF1 flow
estimates are provided at the downstream ends of gauged and un-gauged reaches and
not at the actual gauge sites, these RF1 flow estimates may differ from USGS
records at the gauge sites.
 Furthermore, the USGS Water Resources Division cautions users that these 1982
RF1 flow data are from EPA files and may not accurately represents current records
of the USGS.
Users of BASINS and its internal modeling capabilities would be well advised to seriously
consider each of these limitations. Assessments developed with such limitations may not be
considered reliable and may, therefore, be potentially disapproved by EPA for TMDL
development.
HSPF (BASINS/NPSM) CALIBRATION CAVEATS
With nonpoint source models such as HSPF, simulated responses can have minor and major
discrepancies to the observed records that are caused by slight “shifts” in the hydrologic response
to precipitation events. These differences, by themselves, do not affect the efficacy of the
calibration, but would result in large percentage differences if made in the same mode as a steady
state model. Example HSPF calibrations are displayed in the two panels of Figure 2. The left
panel depicts a near perfect calibration for a 145 day period. Simulated and observed output
coincides for a wide range of conditions, with the model simulating nearly perfectly every storm
event. The example in the right panel, however, represents a time period where individual storm
events seemingly are captured in the simulation but appear a few days earlier than expected,
suggesting an adjustment is needed in HSPF hydrology parameters that affect flow timing, rather
than flow distribution. In many calibrations, however, both magnitude and timing differences
occur that further confounds the calibration.
Little formal guidance is available for calibrating complex water quality models such as HSPF.
An expert system-style manual exists for calibrating the hydrology routines in HSPF (USGS,
1998b) and is based upon a logical progression from the annual average water balance to
tracking individual storm events in the hydrographs and ascertaining that the partitioning
between runoff, baseflow, and interflow are correctly modeled. Some discrepancies are to be
expected since available precipitation as rainfall or snowfall may not be exactly captured by a
weather-station gauge. The prediction of the timing of snowmelt is another equally uncertain
9
phenomenon where precise matching of simulated with observed runoff/quality may not be
achievable.
Figure 2
HSPF Hydrology Calibrations
HSPF Calibration - Fitted
HSPF Calibration - Problematic
350
Simulated
300
Volumetric Flow (CFS)
Observed
250
200
150
100
50
Observed
250
200
150
100
50
Time (Days)
145
133
121
97
109
85
73
61
49
37
25
1
145
133
121
109
97
85
73
61
49
37
25
13
0
1
0
Simulated
300
13
Volumetric Flow (CFS)
350
Time (Days)
FUTURE BASINS ENHANCEMENTS AND ADDITIONS
The following (Table 2) represents this author’s best understanding of BASINS’ future inferred
from technical meeting summaries, web page listings, and through direct email correspondence.
Funding limitations and shifting agency priorities could alter these plans and preliminary
schedules. Comments are provided as necessary that promotes the general constructive criticism
theme of this paper. These future plans highlight the importance of the BASINS toolkit in
watershed-scale assessments and specific plans the Agency has for improvements.
Table 2. Features Provided in Current and Future Releases of BASINS.
Version (Date)
Possible Addition/ Change
1.0 (1996)
Inaugural release of GIS-based assessment tool and modeling.
2.0 (1998)
Updated point source loadings, STORET water observations, river reaches
files (RF3), STATSGO soils, topography (digital elevation), ecoregions,
fish and wildlife consumption advisories, shellfish contamination, Clean
Water Needs Survey, watershed delineation and landuse reclassification
utilities, HSPF postprocessing tools, full RCHRES module capabilities, and
watershed report generation utility.
2.1 (2000)
Replacing current NPSM interface with GenScn (an interactive USGS
developed watershed modeling interface that facilitates generation and
analysis of watershed management scenarios)(USGS, 1998b), and HSPF
12.0 that includes Special Actions, yield-based nutrient modeling, new
10
3.0 (2000)
4.0 (2001-)
degree-day Snow module, and air deposition simulation. Also, several
water quality linkages will be corrected from the NPSM interface.
Discontinuing TOXIROUTE model option in lieu of full HSPF
implementation.
Choice of nonpoint model by adding SWAT (Soil Water Assessment
Tool)(Arnold, et al, 1995), requiring users to purchase ArcView’s Spatial
Analyst Extension to deal with raster data formats. Watershed delineation
based upon 8-, 11-, and 14-digit predetermined watershed boundaries with
the aid of a digital elevation model (DEM). HSPF 12.0 will be included
with all the caveats noted for BASINS 2.1.
Investigations underway on AQUATOX (Park, et al, 1999) to provide
simulation capabilities for lake eutrophication, and river bioaccumulation
processes. Work is also underway on the EFDC (Environmental Fluid
Dynamic Code) model to deal with time varying 2D and 3D estuarine
waterbodies and sediment fate and transport behavior.
BASINS v3.0, expected in FY 2000, will provide the user with a choice of watershed models for
nonpoint source assessment; the HSPF, version 12 and the Soil Water Assessment Tool
(SWAT). For some the SWAT model (Arnold, et al, 1995) is ideally suited to predict the effects
of land management and arose from the Department of Agriculture’s (USDA) substantial
research centered at Texas A&M University. The ability to use USDA developed and supported
modeling approaches in parallel with the EPA and USGS supported HSPF model should increase
the comfort level of the agricultural community as the agency deals with agricultural nonpoint
source dominated TMDLs. Enhancements to the watershed delineation tool will rely upon raster
data and will require additional commercial software to be resident with BASINS. ArcView’s
Spatial Analyst Extension (version 1.1), which currently sells for about $2000, will be required if
use of SWAT is desired.
Recognizing that some users may not be interested (or cannot afford) to purchase this extension,
EPA also will be releasing BASINS v2.1, which will provide full functionality of HSPF v12.0
through the use of the USGS GenScn (Generation and Analysis of Model Simulation Scenarios),
which will retain the look and feel of the current NPSM (USGS, 1998b). The features of NPSM
are being coded into the GenScn to facilitate simulation scenarios and to analyze and compare
results. The GenScn interface will provide full functionality of HSPF v12.0 to include: a
simplified snow melt algorithm (i.e., degree day approach); modeling of land to land transfers,
high water tables and surface ponding (wetlands), yield-based nutrient modeling, air deposition,
and a BMP and reporting module. In addition, a file utility, called WDM Utility, will be added
into GenScn to facilitate creation and modification of weather data and other time series data and
a parameter database of values used in previously documented HSPF studies (HSParm).
In versions beyond BASINS 3.0, the Agency is considering a lentic process model to deal with
sediment contamination and associated tools to simulate sediment source loadings important to
many TMDL assessments. This model should be capable of being configured in 1, 2, or 3dimensions and deal with tidal flows and circulation. EFDC (Hamrick, 1992), strongly linked to
the Chesapeake Bay water quality model (CH3D) is a candidate because it can be used in these
11
complex hydraulic and water quality situations situations. EFDC is also being used to develop
the Sediment Modeling Toolkit that may ultimately be incorporated into BASINS. A field scale
model in envisioned that will allow natural resource managers to explore the effects of specific
combinations of BMPs and land management practices on controlling NPS loadings to water
bodies. In addition, the mixing zone model CORMIX (Cornell Mixing Zone Model and Expert
System) (EPA, 1990) and the Storm Water Management Model (SWMM) (EPA, 1988) also may
be added in future years to meet identified needs of a broad user community. These models
(especially EFDC and SWMM) also have a steep-learning curve if they are to be used reliably.
Similarly, AQUATOX is being evaluated as a general ecological risk model for assessing
exposure to nutrients and toxic chemicals in aquatic ecosystems at several tropic levels. It
evolved from the CLEAN model at RPI in the 1970s (Park, et al, 1999).
Longer term still, EPA has noted that ESRI’s ArcView upgrade to version 4.0, due out in 2000,
will dramatically change the structure of ArcView by removing support of ArcView’s scripting
language, AVENUE, and amending it with the ability to interface with code written in Visual
Basic or C. While BASINS 2.1/3.0 will continue to use AVENUE scripts, and thus require
ArcView 3.x, the Agency is aware of ESRI’s commitment to ArcView 3.x until 2002/3. This
move to a more fully integrated GIS and database system will provide much more flexibility to
BASINS’ development in the long run. The watershed tool will evolve to a more component
based architecture, so that users will need to install only the components required for their
specific analysis needs, including their own legacy models.
The Agency has indicated that the spatial database will also be continuously updated to the
BASINS GIS download web page, as new information becomes available. While these data and
other data may be packaged for public distribution, access via the Internet will continue to be a
prime distribution mechanism.
SUMMARY
EPA is lauded for its foresight in creating the BASINS tool, but the work should not end with the
creation. While the Agency should be responsible for model support issues, the affected private
and public entities must also be critical and demand the continued application by all of good
science and reasoned modeling judgment. Modeling, for all the new tools emerging in the
1990s, is still an ‘art’ that is done well by only a select few. This user base needs to expand if
the challenges and visions of the Clean Water Act are to be achieved in the early 21 st Century.
TMDL development need not be a contentious process when stakeholders meet early and often
to resolve their differences toward a common good – the health of the watershed.
Common sense, professional judgment, and caution must temper user zest and enthusiasm in
using the tool. Early NCASI experience has revealed the temptation to become enthralled by the
GIS technology because it is new and exciting and provides some powerful analysis and
mapping capability. Many potential users are not experienced with GIS technology, but the
philosophy of BASINS makes it readily usable by both inexperienced modelers and spatial
analysts. This is good in the sense that more can be accomplished quicker with fewer personnel
resources, but bad in the sense that modeling quality assurance may not be maintained without
strong EPA oversight. Consistent, long-range system support and training must be maintained
12
by the Agency to insure that the laudable goals of BASINS are still achievable beyond a five or
ten year time horizon. Readers are reminded that the QUAL2E model was the subject of
frequent Agency training in the decades of the 1970s and 1980s through the CEAM in Athens,
Georgia, but virtually disappeared in the 1990s. Senior Agency management needs to ensure
that BASINS does not meet a similar fate. One can project that the Agency’s Watershed
Initiative will preclude this unfortunate fate, but unforeseen congressional budget cuts could
have a devastating effect. The Agency needs to facilitate public dialogue to apply political
pressure to mandate consistent and sustainable support.
Webster’s defines serendipity, as the ‘faculty of unexpectedly finding valuable or agreeable
things not sought for’. While the Agency’s long-range goal for the BASINS tool may ultimately
fit into this definition, there is an element of wishful thinking, or even naivete, in the current
philosophy. In its zeal to create a new and useful tool for the overly litigious TMDL domain,
EPA has overlooked some basic tenets of good science and modeling practice and may have
placed a premium on speed at the expense of quality and scientific truth. James Greenfield of
EPA succinctly discussed TMDLs at a recent conference by stating they can be cheap, fast, or
accurate – pick any two (Georgia, 1999). This paper has suggested that ‘fast’ or ‘cheap’ at the
expense of accuracy should not be acceptable.
REFERENCES
Arnold, J.G., J.R. Williams, and D.R. Maidment. 1995. A continuous-time watershed sediment
routing model for large basins. ASCE J. Hydr. Eng. 142:47-69.
Brown. L.C. and T.O. Barnwell, 1987. The Enhanced Stream Quality Models QUAL2E and
QUAL2E-UNCAS: Documentation and Users Manual. EPA/600/3-87/007.
Crawford, N. H. and R.K. Lindsey, 1976. Digital Simulation in Hydrology: The Stanford
Watershed Model IV, Technical Report No. 39, Stanford University, Palo Alto, CA.
Donigian, A. S., Personal Communication at EPA HSPF Training Workshop, March 3-6, 1998,
Boston, MA.
EPA, 1988. Storm Water Management Model Version 4 User’s Manual. EPA/600/3-88/001a
(NTIS PB88-236641).
EPA, 1990. CORMIX1: An Expert System for Mixing Zone Analysis of Conventional amd
Toxic Port Aquatic Discharges, EPA/600/3-90/012.
EPA, 1993. Exposure Models Library and Integrated Model Evaluation System. EPA/600/C92/002 (revised September 1996).
EPA, 1995. QUAL2E Windows Interface User’s Guide. EPA/823/B/95/003.
EPA, 1996a. Better Assessment Science Integrating Point and Nonpoint Sources, EPA-823-R96-001.
13
_____. 1996b. Hydrological Simulation Program - FORTRAN. Users Manual for Release 11.0,
EPA, Athens, GA and USGS, Reston, VA.
EPA, 1998a. The Quality of Our Nation’s Waters National Water Quality Inventory: 1996
Report to Congress, EPA-841-R-97-008..
EPA, 1998. Better Assessment Science Integrating Point and Nonpoint Sources, BASINS 2.0,
EPA-823-B-98-006.
EPA, 1999. Peer Review of BASINS v2.0, Internal EPA Memorandum signed by Betsy
Southerland, Acting Director OST, October 12, 1999.
Kinerson, Russell, 2000. Personal communication to NCASI, (January, 2000)
Georgia DNR, 1999 Annual Water Resources Conference (unpublished proceedings)
GSC (General Sciences Corporation), 1993. PC-ROUTE User’s Manual DRAFT, GSC-TR-EI93-005, EPA OST, Washington, DC.
Hamrick, J.M., 1992. A Three Dimensional Environmental Fluid Dynamics Computer Code:
Theoretical and Computational Aspects. The College of William and Mary, Virginia Institute of
Marine Science. Special Report 317.
NCASI, 1984. Modifications to the QUAL-2 (SEMCOG) Water Quality Model and User’s
Manual, Technical Bulletin No. 437, National Council of the Paper Industry for Air and Stream
Improvement, Inc., Research Triangle Park, NCASI.
NCASI, 1997. Evaluation of EPA’s Permit Compliance System (PCS) Database, Technical
Bulletin No. 735, National Council of the Paper Industry for Air and Stream Improvement, Inc.,
Research Triangle Park, NC
Park, R., et al, 1999. AQUATOX for Windows: A Modular Toxic Effects model for Aquatic
Ecosystems, EPA Contract 68-C4-0051, US EPA, Washington, DC.
USGS, 1998a. Content Standard for Geospatial Metadata.
Spatial Data Committee, Washington, DC.
FGDC-STD-001-1998, Federal
USGS, 1998b. A Tool for the Generation and Analysis of Model Simulation Scenarios for
Watershed (GenScn)”, Water Resources Investigations Report 98-4134, 1998.
14
Download