Jiayi Li, Penn State University Penn State Logo

Jiayi LiI am currently a graduate student in the Department of Agricultural Economics and Rural Sociology at Penn State University. I received my undergraduate degrees in environmental sciences and economics at Peking University in China.

- Link to poster presented at Clean Energy Expo , April 2-3, 2004, State College, PA

Designing Wetland Conservation Strategies under Climate Change
Jiayi Li, Elizabeth Marshall, James Shortle, Richard Ready, and Carl Hershner*
Agricultural Economics and Rural Sociology, Pennsylvania State University
* Center for Coastal Resources Management, Virginia Institute of Marine Science

Short Summary: A methodology for evaluating public wetlands conservation investments that considers climate change is developed and applied to Virginia's Elizabeth River watershed. A revised cellular automaton (CA) model is applied to project future land use change. Discrete stochastic sequential programming (DSSP) is used to model a parcel-based discrete-time decision process.

Extended Abstract:
Wetland conservation is a major environmental concern in the Chesapeake Bay region. Substantial losses due to land development and other factors have had profound impacts on the Bay's aquatic resources. Current conservation efforts fail to account for the impacts of climate change on sea level, which can affect the success of conservation efforts. Land use controls are essential to effective wetlands conservation. This study develops a methodology for evaluating public wetlands conservation investments that takes climate change into account, and demonstrates the methodology for the Elizabeth River watershed in Virginia under plausible sea-level rise and land use scenarios.

Given the large uncertainty about the non-market values of wetlands, we use a cost-effectiveness analysis framework as the fundamental structure of our study. Two measures of effectiveness are considered in our study. One is the total amount of wetlands. The other is related to the wetland functions. We use a tool for wetlands identification and planning that was developed by the Chesapeake Bay Program. The tool uses information on wetland type, surrounding land use, and external influences to generate scores for five major wetland functions: habitat provision, water quality improvement, flood protection, bank stabilization, and sediment control.

Because it is essentially impossible to confidently predict the future sea level rise and land use, we develop scenarios that establish probable upper and lower bounds on future conditions. Sea-level rise scenarios are constructed using projections for the southern Chesapeake Bay region and local information. We use 4 to 12 inches sea-level rise for 2030. We also construct land use scenarios for the area by considering land use change drivers and comprehensive plans for the region. The current landscape is represented as a regular grid of cells of 25 acres each. A revised cellular automaton (CA) model is used to generate development vulnerability indexes for all the cells within the watershed. Strict CA articulate the growth (or change) process in terms of highly localized neighborhoods where change takes place purely as a function of what happens in the immediate vicinity of any particular cell. But in our model, we identify four major drivers that influence the development possibility for each undeveloped land cell. The four drivers are immediate vicinity (8-cell neighborhood) land use, distance to shoreline, distance to primary roads, and distance to population centers. Three land use scenarios developed in our study, compact, dispersed, and nodal, are based on the development concepts used in the 2026 Comprehensive Plan of the City of Chesapeake, Virginia. We assign different weights to the four drivers to reflect the three land use scenarios. A random term is also added for each cell in order to capture influences other than the four major drivers. Based on literature and local information, we assign Markov transition probabilities to the land uses within the watershed. We rank all the available undeveloped land cells based on their development index and convert a certain percentage of them into developed land cells according to the transition probability we set. Because of the existence of the random term, we can run the Monte Carlo simulation to generate future land use scenarios.

CA Model

Figure 1: CA Model Illustration

Three management strategies are considered. The first allows private landowners to erect protection structures at the landward existing wetlands and has higher elevation than the wetlands. In this case, when sea-level rises, the wetlands can migrate inland to survive. The third relocates wetlands strategically by public acquisition of low value and low elevation lands for restoration. Candidate restoration sites are identified based on whether the present landscape still retains features that allowed it to support wetlands in the past. We use the results of the protocols for implementation of a GIS-based model for the selection of potential wetlands restoration sites in southeastern Virginia.

The decision-making process we consider for wetland management strategy 2 and 3 is a parcel-based discrete-time process. At the beginning of each five-year time period, for each undeveloped land parcel, decision-makers need to decide whether to buy, not buy, keep, or sell. Under different scenario combinations, we use discrete stochastic sequential programming (DSSP) to model this process and compare different wetland management strategies. The objective of our DSSP model is to minimize the costs of the wetland management strategy. We consider the costs of buying the undeveloped land and the wetland restoration costs. In our study, the prices of undeveloped land are modeled as a function of the development indexes. One important advantage of DSSP is that it allows for explicit consideration of the a priori known probabilities of uncertain events. In the DSSP framework, we consider two types of uncertain events that may affect the decisions. One is the acquisition of new information about sea-level rise. We assume that new climate information will become available every five years. We simplify the information as indicating low or high sea-level rise and arbitrarily assign probabilities for them. The other type of uncertainty arises from the development probability of each undeveloped land parcel. It is necessary to consider this uncertainty, because when decision-makers consider whether to buy an undeveloped land parcel during any future time frame, they need to consider information about the likelihood that parcel will still be available. These probabilities are derived from the Monte Carlo simulations described earlier. Sensitivity analysis is conducted for important parameters of the model.

2-stage model

Figure 2: DSSP 2-stage decision model.

Our study provides a methodology for assessing wetland conservation strategies that takes climate change into consideration. We believe that sea-level rise is an important issue which affects the success and effectiveness of wetland protection efforts because of the low-lying feature of wetlands. This methodology can be applied to other areas with some adjustments based on local situations. Value of information (VOI) estimates can also be easily extracted from the framework presented in our study.