WETLAND AND WATER RESOURCE MODELING AND ASSESSMENT: A Watershed Perspective - Chapter 18 potx

However, while these GIS tools are powerful and useful in data management, visualization, and spatial analysis, they usually lack the decision rules or models that can link geospatial da

Part V

Watershed Assessment and Management

Models for Assessing

the Vulnerability of

Wetlands to Potential

Human Impacts

Wei “Wayne” Ji and Jia Ma

18.1 INTRODUCTION

Characterized by a shallow water table (Sharitz and Batzer 1999), wetlands are tran-sitional landscapes between open water systems and terrestrial uplands They provide many crucial ecosystem functions and values, such as flood control, groundwater recharge, sediment and pollutant retention/stabilization, nutrient removal/transfor-mation, and fish and wildlife habitat and diversity (Mitsch and Gosselink 2000) Wetlands are prone to be filled in, drained, or ponded for a variety of human uses including stream channelization and maintenance, urban development, transporta-tion improvement, or conversion to agricultural uses (Dodds 2002) To protect wet-land resources, many laws and regulatory programs have been established Among them, the Clean Water Act (CWA) Section 404 is the primary federal law aiming to maintain and restore the chemical, physical, and biological integrity of the wetlands

in the United States It authorizes U.S federal agencies, mainly the U.S Army Corps

of Engineers, to issue permits for the discharge of dredged or fill material into the navigable waters at specific disposal sites, including wetlands (USACE: http://www usace.army.mil/cw/cecwo/reg/sec404.htm) To comprehensively evaluate individual

or cumulative impacts of human activities on existing wetlands, a regulatory per-mit assessment requires quickly retrievable environmental and socioeconomic data, and more importantly, a scientifically justifiable evaluation framework for analyz-ing those data In recent decades, GIS techniques have been increasanalyz-ingly used to facilitate the data management and visualization in regulatory wetland assessments

or permit reviews, aiming to improve the efficiency of the permit assessment pro-cess A pilot decision supporting GIS for the permit analysis was developed in the early 1990s (Ji and Johnston 1994, 1995) The system was based on a widely used commercial GIS (Arc/Info, ESRI, Inc.), with customized user interfaces for data retrieval, visualization, and analysis Other similar GIS-based technical tools were also developed, such as the Permit Application Management System (PAMS) for

216 Wetland and Water Resource Modeling and Assessment

evaluating and tracking the status of permit applications submitted for approval by the Connecticut Department of Environmental Protection, and ERATools for man-aging permit data and analyzing potential impacts of permitted activities by the Florida Department of Environmental Protection However, while these GIS tools are powerful and useful in data management, visualization, and spatial analysis, they usually lack the decision rules or models that can link geospatial data manipulations

to evaluating how vulnerable wetland functions and values would be under potential human impacts in the context of regulatory assessment Thus, rule-based decision models need to be developed and incorporated with the GIS tools

During recent decades, numerous environment assessment models have been developed, which can be used, at least partially, to evaluate wetland functions and values for various decision-making purposes Examples of these models include the Habitat Evaluation Procedure (HEP) (USFWS [U.S Fish and Wildlife Ser-vice] 1980), the Wetland Evaluation Technique (WET) (Adamus 1983), the Index

of Biological Integrity (IBI) (Karr 1997), the GIS-based Wetland Value Assessment Methodology (Ji and Mitchell 1995), and the Hydrogeomorphic Approach (HGM) (Hollands and Magee 1985, Brinson 1996) However, none of the existing models can be effectively used with GIS data to assess wetlands for regulatory wetland assessments, such as the Section 404 permit review This is because (1) these models were originally developed for other applications (e.g., wetland restoration planning

or wildlife habitat evaluation), and thus do not address all the functions, such as socioeconomic function, of wetlands that need to be assessed in regulatory assess-ments; and (2) all of these models require a great amount of field data collection and specialized expertise for implementation Therefore, it is not effective and efficient

to directly adopt and integrate the existing models with GIS to address the needs in wetland regulatory assessment Clearly, there is a critical need for GIS-based deci-sion models in order to handle increasing volumes of existing geospatial data for rapidly assessing wetland vulnerability in management decision making (USEPA

2004 research solicitation: EPA FRL-OW-7620-6)

To address this objective, our efforts focused on the design of geospatial decision models that generate a ranked wetland vulnerability index (WVI) based on geospa-tial data and analysis In addition, a user-friendly decision support GIS with custom-ized user interfaces was developed to facilitate the implementation of the models The developed decision models were applied to the Little Blue River watershed in the state of Missouri in the United States

18.2 GEOSPATIAL DECISION MODELS

A geospatial decision model is one that generates output for management decision support, such as ranked indices, based on geospatial data and analysis In this study, the models are to be used for assessing wetland vulnerability, which is defined as the degradation likelihood of wetland functions and values under potential anthro-pogenic pressures Certain characteristics of wetlands (e.g., the size or recreational usage) and the spatial occurrence of certain concerned entities (e.g., endangered spe-cies or a historical site) related to particular wetlands may largely determine the degree of vulnerability of the wetland’s functions and values Thus, the geospatial

Geospatial Decision Models for Assessing the Vulnerability of Wetlands 217

decision models were developed so they can be used to identify and evaluate these characteristics and the concerned entities of wetlands with GIS-based data and anal-ysis; they follow a 3-step procedure (Figure 18.1):

18.2.1 DETERMINATION OF INDICATORS AND METRICS

To address the fundamental needs of the regulatory assessment of wetlands, four wet-land functions were selected for our modeling: (1) biological supporting function, (2) hydrological function, (3) physiographic function, and (4) socioeconomic function For each of these functions, three indicators were identified (Table 18.1) The selection

of the indicators for the geospatial decision models follows three considerations: (a) the selected indicators of a particular wetland function should be able to address major concerns of wetland regulatory assessment;

(b) the selected indicators should have been used in wetland assessment by related environmental management agencies or identified in research pub-lications; and

(c) the indicators can be evaluated using GIS-based data and analysis

The indicator selection process involved consultations with related governmen-tal agencies that are responsible for wetland regulatory assessment, including the U.S Army Corps of Engineers (USACOE) and the U.S Environmental Protection Agency We also conducted literature reviews (e.g., USEPA 2002, Stein 1998, Ada-mus 1983, USACOE 1997, Hollands and Magee 1985, Brinson 1996, Sousa 1985, Cook et al 1993, Karr and Chu 1998, Zampella 1994, Hruby et al 1995) As shown

inTable 18.1, the measurement of an indicator is referred to as metrics According

to the value of the metrics, the decision criteria (or decision boundaries) for evaluat-ing a particular indicator are determined Rankevaluat-ing scores are assigned based on the decision criteria When necessary, weights may be determined and applied to certain ranking scores To evaluate a particular indicator, appropriate GIS data and spa-tial analysis operations need to be identified and used, as shown in the last column

of Table 18.1 The metrics are either geospatial or descriptive Geospatial metrics can measure the spatial occurrence and size of a wetland or concerned entities in

a potentially impacted area, or their spatial proximity to concerned human activity locations The descriptive metrics help identify certain characteristics or features of

Determination of indicators

and their metrics

Fuzzy math-based determination

of decision criteria

Generation of wetland

vulnerability indices

Use of appropriate geospatial data and spatial analysis to evaluate the metrics

of indicators of particular wetland

functions

FIGURE 18.1 Wetland vulnerability assessment with geospatial decision models

TABLE 18.1

Structure of geospacial decision models The indicators and metrics are for selected wetland functions The decision criteria

and score/weight columns show the examples of possible metric values and scores The GIS data column illustrates some

typical data sets that can be used to evaluate the metrics.

Biological supporting function

Total area of a target wetland (BV1) Total area of a wetland in a

potentially impacted location

>75 percentile 1.0 Wetland data and maps

≥ 25 and < 75 percentile 0.5

< 25 percentile 0.1 Proximity to species of concern (BV2) The number of species of concern >5 species 1.0 3 Wildlife species data

≥2 and < 5 species 0.5 2

≥ 10% and < 20% 0.5

Hydrological function

Flood risk (HV1) The percentage of floodway >75 percentile 1.0 Stream/river data for high-risk

flooding regions

≥ 25 and < 75 percentile 0.5

< 25 percentile 0.1 Hydrological modification (HV2) The occurrence of dams Occurrence

Not occurrence

0

Pollution potential (HV2) The number of pollution sites >75 percentile 1.0 Landfills and mining wastes

data, etc.

≥ 25 and < 75 percentile 0.5

< 25 percentile 0.1

Physiographic function

Erosion potential (PV1) The percentage of erodible soil > 75 percentile 1.0 Soil data and maps

≥25 and < 75 percentile 0.5

<25 percentile 0.1 Drinking water relevance (PV2) The occurrence of public water

supply facilities

Occurrence Not occurrence

1 0

PWS lakes, tanks, wells, and springs data

Nearby land uses (PV3) The percentage of urban and

agricultural lands

>75 percentile 1.0 Land use/land cover data

≥ 25 and < 75 percentile 0.5

< 25 percentile 0.1

Socioeconomic function

Proximity to important public land (SV1) Percentage of public lands >75 percentile 1.0 Data for public lands, national

wild lands, scenic rivers, etc.

≥ 25 and < 75 percentile 0.5

< 25 percentile 0.1 Recreation potential (SV2) Presence of public parks or

recreation areas

parks, etc.

Proximity to historic and cultural sites (SV3) The number of historic and

cultural sites

≥1 and < 3 percentile 0.5

220 Wetland and Water Resource Modeling and Assessment

the wetlands or related entities under assessment, such as the usage of wetlands (e.g., for recreation), hydrological facilities near wetlands (e.g., a dam), and riparian land use types and ownership

18.2.2 DETERMINATION OFDECISIONCRITERIA

Decision criteria are used to evaluate indicators based on the values of the corre-sponding metrics (Table 18.1) In our study, three methods were employed to deter-mine the decision criteria based on (a) spatial statistics, (b) published results, or (c) professional judgments The spatial statistics, such as the percentile, equal interval, standard deviation, or user-defined interval of metric values, were used to determine the decision criteria for ranking most of the selected indicators of wetlands This statistical approach considers the variation of values of a particular indicator across the study area Some decision criteria were adopted from the findings of published environmental studies or environmental management documents (e.g., ones relat-ing to species density; USEPA 2002) As in many other instances of environmental decision making, professional judgments also played a role in determination of deci-sion criteria for some indicators Taking the indicator “Proximity to species of con-cerns (BV2),” for example, the decision criteria were adopted based on the published guidelines (USEPA 2002):

If less than 2 species are found (“less vulnerable”) near a wetland, the indica-tor receives a score of 0.1;

If 2–5 species (“vulnerable”), the indicator receives a score of 0.5;

If more than 5 species (“highly vulnerable”), the indicator receives a score of 1.0

In addition, different weights were given to the species of concern based on their conservation status as endangered (a weight of 3), threatened (a weight of 2), or at risk (a weight of 1)

The decision criteria with the cutoff thresholds, like those above, may cause imprecise evaluation of metrics, especially when the metric value is close to the thresholds For example, according to the thresholds of the decision criteria used for the above species indicator, a wetland that supports 3 species is ranked the same (a score of 0.5 or “vulnerable”) with a wetland supporting 4 species To take account

of the vagueness and the nonspecificity of certain metrics values, a computational method based on fuzzy math was developed and applied to the evaluation of some indicators With this method, the triangular-shaped fuzzy membership function

(Tran et al 2002) is utilized to determine the degree of certainty ( fuzzification)

of each metric value belonging to a certain vulnerability level, which is calculated

by combining the portioned degree of certainty of the metric values in each vul-nerability level Taking the “Proximity to species of concerns (BV2)” indicator, for example, the fuzzification works as illustrated inFigure 18.2

According to Figure 18.2, the wetland with 3 species of concern in its proximity has a “certainty” value of 0.25 in the “less vulnerable” domain, the value of 0.75 in the “vulnerable” domain, and the value of 0 for the “highly vulnerable” domain For the wetland with 4 species of concern, the certainty values in these domains are 0, 1.0, and 0, respectively When combining these values:

Geospatial Decision Models for Assessing the Vulnerability of Wetlands 221

Metric value = 0.25 * 0.1 + 0.75 * 0.5 + 0.0 * 1.0 = 0.4 (for the 3 species case) Metric value = 0.0 * 0.1 + 1.0 * 0.5 + 0.0 * 1.0 = 0.5 (for the 4 species case) Thus, the fuzzy math−based method is more precise for distinguishing the differ-ence between wetlands with 3 species and 4 species than using the cutoff thresholds that treats the two cases equally

18.2.3 CALCULATION OFWETLAND VULNERABILITY INDEX

We developed calculations that incorporate the ranked scores of the indicators to generate both the vulnerability index for individual wetland functions and the over-all vulnerability index for over-all wetland functions together The vulnerability index for individual wetland functions is calculated using the following equation:

VIk = W1*V1 + W1*V2+ + Wn*Vn (18.1) where:

VIk = vulnerability index for the wetland function k

K = denoting an individual function: b for biological supporting function, h for hydrological function, p for physiographic function, and s for socioeconomic function, respectively

Vn = the ranked score value of the nth indicator of a particular function

Wn = the weight of nth indicator when it applies

Then, the vulnerability index for each wetland function (VIk) is ranked in one of three possible vulnerability levels, “highly vulnerable,” “vulnerable,” or “less vul-nerable,” by evenly dividing the maximum value of VIk for the assessment area into three intervals To calculate the overall wetland vulnerability index, we first normal-ize each VIk using the range of the score value of each wetland function:

NVIk= (VIk– VImin )/ (VImax – VImin) (18.2)

Certainty

Metrics evaluation threshold (the numer of species)

(1) (2) (3)

1.0 0.8 0.6 0.4 0.2 0.0

FIGURE 18.2 The triangular fuzzy membership functions for evaluating “species of con-cern” indicator Membership function (1) denotes the “less vulnerable” level (the metric value

= 0.1); (2) denotes the “vulnerable” level (the metric value = 0.5); (3) denotes the “highly vulnerable” level (the metric value = 1.0)

222 Wetland and Water Resource Modeling and Assessment

where:

NVIk = normalized vulnerability index of the wetland function k,

VImin = the sum of possible minimum score values of all indicators of the wetland

function k,

VImax = the sum of possible maximum score values of all indicators of the wetland

function k

The objective of the normalization is to treat all the functions equally in the index calculation by eliminating the effect of different score ranges among different wetland functions Then the overall wetland vulnerability index (WVI) is calculated

by combining the normalized vulnerability indices of all the wetland functions:

WVI = NVIb + NVIh + NVIp + NVIs (18.3) The overall wetland vulnerability indices are ranked by evenly dividing the maxi-mum value of WVI for the assessment area into three possible vulnerability levels:

“highly vulnerable,” “vulnerable,” and “less vulnerable.”

18.3 DECISION SUPPORT GIS FOR MODEL IMPLEMENTATION

Focusing on the implementation of the geospatial decision models for wetland vulnerability assessment, a decision support GIS (Figure 18.3) was developed with four major functions: (1) geospatial data management, (2) analytical query, (3) vulnerability assessment modeling, and (4) assessment result output This was accomplished by customizing a widely used commercial GIS, ArcView (ESRI, Inc.),

in order to fully utilize its capabilities in geospatial data handling, and increase the model’s applicability and transferability in the community of users Visual Basic

of Application, an object-oriented language, was used to program ArcObject (the customizable components available with ArcView) in creation of the user-friendly graphical user interfaces (GUI) for implementing all the model functions

18.3.1 GEOSPATIALDATA MANAGEMENT FUNCTION

A comprehensive wetland vulnerability assessment relies on efficiently retrieving sufficient data and information that address major concerns of wetland conserva-tion Therefore, a geospatial database management function is fundamental to the decision support GIS This system function (Figure 18.4) is focused on two technical objectives that allow users to efficiently manage geospatial data for modeling:

1 Categorizing and organizing existing data sets The interface shown in Figure 18.4a provides the user a tool for categorizing miscellaneous unor-ganized geospatial data into the classes that address major concerns in the assessment of biological, hydrologic, physiographic, and socioeconomic functions of a wetland

2 Facilitating data retrieval in modeling A dataset for addressing a particular concern or a group of data sets for addressing multiple concerns can be effi-ciently selected from corresponding data categories through the interface shown by Figure 18.4b

Geospatial Decision Models for Assessing the Vulnerability of Wetlands 223

18.3.2 ANALYTICALQUERY FUNCTION

As a GIS operation, the analytical query is referred to the retrieval, visualization, spatial analysis, or modeling of geospatial data in order to evaluate specific criteria

or answer research questions quantitatively or qualitatively A wetland vulnerability assessment usually requires the verification of spatial proximity or other relation-ships between the site of a proposed activity and a potentially impacted wetland

or the locations of other entities of concern, such as historical permit sites, impor-tant habitats, biological resources, and cultural facilities Therefore, the analytical query function of the decision support GIS was developed for the following major capabilities, which can be implemented through several customized interfaces (Figure 18.5):

1 Identifying and displaying the spatial location of a proposed activity in relation to a potentially affected wetland (Figure 18.5a) This is done by searching the proposed activity site with its known geographic coordinates

or using the linguistic description (e.g., the name of a river) of the proposed activity location to identify the site on a background map

Decision Support GIS

Analytical Query

Vulnerability Assessment Modeling

Assessment Result Output

Historical activity documents

Vulnerability index map

Used decision criteria

Assessment Implementation

Biological Data

Geodatabase (Data storage, integration and update)

Socioeconomic Data

Geospatial Data Management

Hydrological Data

Physiographic Data Activity Site

FIGURE 18.3 Key functions and architecture diagram of the decision support GIS described in section 18.3