a systematic framework for prioritizing farmland preservation (pdf)

A systematic framework for prioritizing farmland preservation

Elia A. Machado

David M. Stoms

Frank W. Davis

University Of California, Santa Barbara

National Center For Ecological Analysis And Synthesis

Report to The Resources Agency Of California

August 2003

2

EXECUTIVE SUMMARY

The California Legacy Project (CLP) mission is “to enable the state and its partners in conservation to develop and implement a strategic and inclusive approach to conserving and restoring California’s lands and natural resources.” In 2001 The Resources Agency of California contracted with the National Center for Ecological Analysis and Synthesis at UC Santa Barbara to convene a working group to help bring systematic conservation planning theory and methods to bear on the design and implementation of CLP. The conservation planning framework for farmland described in this report for is one of the products from that working group.

The framework is intended to serve the dual purpose of helping decision makers to evaluate current opportunities (e.g., current proposal applications for State conservation funds) and to help planners develop longer term conservation strategies that highlight areas for more focused analysis and collaborative planning. We do not present a plan or “blueprint” for future conservation activities. Instead, we offer an analytical, data-driven planning process that could be applied to ongoing conservation assessments and evaluations by State conservation planning staff and collaborating organizations over the State or regions of the State.

We organize the planning framework based on a hierarchy of conservation goals and objectives, each of which is further elaborated in terms of specific objectives, criteria, and sources of evidence. For farmland preservation, we summarize these goals as retaining farmlands: 1) with the greatest sustained production capacity, 2) that provide high amenity values (e.g., habitat, open space, floodplain management, and scenic values), and 3) whose location reduces the risk of urban sprawl. The framework applies GIS technology to map farmland conservation value and investment priorities based on available spatial data, derived indices and simple algebraic functions. A planning region is divided into sites, and each site is scored in terms of its marginal conservation value, that is, the incremental value added to the current system of conservation lands by making the next conservation investment in that site. Site prioritization depends on the farmland resources the site contains, the threat to those resources, and the conservation cost of mitigating that threat. The strategic objective is to allocate conservation funds among a set of candidate sites such that there is the greatest possible farmland value remaining at the end of the planning period.

We demonstrate the framework for preservation of farmlands in the Bay Area/Delta Bioregion. Because the criteria for measuring objectives 2 and 3 require spatial and nonspatial data that are not readily available statewide or even for a bioregion, we only develop and demonstrate the framework for objective 1. Existing data are used to map resource values and threats to arrive at maps of marginal conservation value without consideration of site cost. We use a crude estimate of the cost of conservation easements to demonstrate how the framework could then be used to prioritize conservation investments subject to a fixed budget.

3

ACKNOWLEDGEMENTS

We gratefully acknowledge the financial and logistical support of The Resources Agency of California and the National Center for Ecological Analysis and Synthesis, a Center funded by NSF (Grant #DEB-0072909), the University of California, and the Santa Barbara campus.

We thank Secretary of Resources Mary Nichols, California Legacy Project Director Madelyn Glickfeld, Resources Agency Science Science Advisor Greg Greenwood, California Fish and Game Senior Biologist Marc Hoshovsky, and Bill Stewart, Chief of the CDF Fire and Resource Assessment Program, for their guidance and advice throughout the project. Thanks also to California Legacy Project staff members Rainer Hoenicke and Mike Byrne for their collaboration and technical support. Chris Costello of the UCSB Bren School was especially helpful in guiding the development of our framework within an economic context.

We would like to recognize and thank the participants in the NCEAS workshops who contributed a great deal to the development of the framework for farmland conservation:

Sandy Andelman National Center for Ecological Analysis and Synthesis

Pete Dangermond Dangermond and Associates

Greg Greenwood The Resources Agency of California

David Kelley K and AES, Inc.

Jeff Loux University of California, Davis

Wendy Rash University of California, Davis

Helen Regan San Diego State University

Richard Standiford University of California, Berkeley

Bill Stewart California Department of Forestry and Fire Protection

Erik Vink California Department of Conservation

Contract administration and accounting were provided by the UCSB Marine Science Insititute. We are especially grateful to Deb Owens, Arlene Phillips and Marie Cilhuaga for their excellent support

4

1. INTRODUCTION

The very qualities that make some lands the most productive for agricultural crops also make them highly suitable for urban growth (Solomon 1984). As human population expands around the world, market forces prompt the conversion of prime farmland into new suburban and urban uses. According to Daniels (1991), about one-fifth of the best US farmland is located within metropolitan counties; this proportion increases to one-third when the adjacent counties are included (Daniels 1991). Fragmentation of remaining farmland may lead to declining productivity and profitability (Brabec and Smith 2002; Levia 1998), promoting additional sales of agricultural land as conflicts between farmers and new residents increase (Bradshaw 1998; Nelson 1992). In addition, the “impermanence syndrome” -belief among farmers that agriculture has no future in their area due to urban pressures- results in disinvestments, selling of properties, shifting to less labor intensive crops, etc. that may prompt the conversion of parcels in the nearby vicinity (Conklin and Lesher 1997; Nelson 1992).

The problem of farmland conservation is intensified when marginal areas of lower agricultural quality are put into production to compensate for the conversion of prime farmlands to urban use. These new farmlands require greater inputs of chemicals and/or water to be viable, which result in an overall greater environmental cost (Greene and Stager 2001).

Although there is general social awareness about farmland conversion, the severity of the phenomenon and whether there should be public intervention into the market to regulate it is controversial (Greene and Stager 2001; Nelson 1992; Tulloch et al. 2003).Some policy analysts argue that markets are imperfect due to their inability to account for non-production farmland values and land price inflation due to environmental regulations and other economic forces (American Farmland Trust 1997). The imperfection of the market causes the undervaluation of farmland and, as a result, urban uses typically outbid agriculture wherever urban suitability is high (Nelson 1992).

Many farmers and the broader public advocate farmland preservation for non-market public benefits such as open space, maintenance of traditional lifestyles and provision of locally grown foodstuffs. Other cited reasons to preserve farmland include flood absorption, air cleansing and water filtration, spatial definition of urban areas, and growth management (American Farmland Trust 1997; Kline and Wichelns 1998; Nelson 1992). On the other hand, farm practices also produce water pollution and other negative environmental impacts. Poe (1999), citing Gardner (1977), claims that market failure only justifies public intervention in the case of environmental and open space concerns; that farming viability and agrarian lifestyle concerns are equity issues and do not correspond to standard welfare economic justifications.

5

Despite the debate, many measures and policies have been implemented by public agencies and organizations in the United States to prevent farmland conversion, e.g. purchase of development rights (PDRs), transfer of development rights (TDRs), tax incentives, zoning, and growth management programs (Alterman 1997; American Farmland Trust 1997; Lynch and Musser 2001).

Regardless of the policy option used, agencies and institutions face a common problem of developing a strategic process for prioritizing farmland for preservation. The challenge in designing such a strategy is to make it explicit, objective, fair, understandable to all parties, feasible to implement with available information, and most importantly, directly related to the goals of conservation.

With its combination of fertile soils, benign climate, and heavily capitalized infrastructure, California is one of the premier agricultural areas of the world. Agricultural ecosystem services from 10.5 million ha (27% of the land area) generate over $30 billion income per year (California Agricultural Statistics Service 2002). California’s farmland also provides non-market amenities to its large population. Nevertheless, human population is currently growing by roughly 0.5 million per year and is projected to reach 58.7 million by 2040 (California State Department of Finance 1998). Much of the new development to meet this demand is slated to occur on prime farmlands near existing urban development. In response to this threatened loss, significant funds are being invested to preserve farmland throughout the state. The process of prioritizing and preserving new conservation lands is conducted by dozens of different public and private organizations (California Environmental Dialogue 1999; The Resources Agency 2001). These include the U. S. Department of Agriculture, the California Department of Conservation, county and local governments, and more than 100 local and regional land trusts and conservancies. State government funding for conservation easements comes from a variety of sources, including special funds and public bond initiatives bonds. Although these represent considerable funding for conservation they fall far short of what most agencies and conservation groups believe is required to meet even short-term demands for farmland, open space and habitat conservation (California Environmental Dialogue 1999). Thus there is intense competition for these public funds and the need on the part of the State funding agencies to make decisions in what are often acrimonious public forums.

In its 1996 analysis of land conservation activities by State agencies, the California Legislative Analysts Office found that the State was unable to set clear conservation priorities because it lacked a comprehensive and cohesive statewide land conservation plan, suffered from poor coordination among departments, and had limited ability to formally evaluate conservation opportunities as they arose (California Legislative Analyst's Office, 1996). In response the California legislature mandated the creation of a new conservation planning program known as the California Legacy Project (formerly named CCRISP, the “Continuing California Resource Investment Strategy Project”) under The Resources Agency. The

6

California Legacy Project (CLP) mission is “to enable the state and its partners in conservation to develop and implement a strategic and inclusive approach to conserving and restoring California’s lands and natural resources”.

The purpose of CLP is not to serve as a substitute for these existing efforts. Instead CLP is envisioned as a distinctive new strategic planning process to provide consistent, statewide information and analyses that will help guide the State’s financial investments in resource conservation and restoration and will also promote effective conservation actions through partnerships with non-State organizations.

In 2001 The Resources Agency contracted with the National Center for Ecological Analysis and Synthesis (www.nceas.ucsb.edu) to convene a working group to help bring systematic conservation planning theory and methods to bear on the design and implementation of CLP. A series of workshops and other interactions were designed to elucidate the specific objectives of the program and current methods of priority setting by state agencies through extensive deliberations over the pros and cons of different methods and approaches.

Here we present one of the products from that working group, a top-down analytical procedure to support strategic planning for farmland conservation over large areas using relatively coarse, generally available information. We review existing approaches and lay out guiding principles for our framework. We define a hierarchy of farmland conservation objectives and associated criteria and assess the marginal conservation value of candidate farmland areas based on resource quality and threat. We allocate available conservation funds to maximize the overall utility of farmland preservation based on assessed conservation value relative to cost. We demonstrate these ideas through an application in the Bay Area/Delta bioregion of California, USA.

2. EXISTING APPROACHES

Farmland preservation is a land use issue and, as such, is strongly influenced by culture, traditions, and laws. Stakeholders have different perceptions of the farmland conversion phenomenon, different objectives for preserving farmland, and different criteria for assessing conservation value. As a result, there is a wide diversity of programs to protect farmland with significant differences in the way they operate (Alterman 1997). We will briefly review the approaches and criteria used by these programs to preserve farmland.

Prioritizing farmlands for conservation has been addressed in two distinct modes: 1) a “bottom-up” approach in which properties are offered to the agency by a willing landowner and the agency must evaluate and rank the offers, and 2) a “top-down” strategic approach in which all farmland is evaluated and prioritized simultaneously (Tulloch et al. 2003). The difference between the two approaches reflects the asymmetry of the information possessed by the landowner and the planner. Landowners have greater knowledge about the site-specific practices of

7

their farm, especially about the financial status and therefore what economic incentive would induce them to preserve their farmland. In contrast, the planner tends to have better information about the larger context such as the pattern of development pressures and of previously preserved farms. The more common bottom-up approach responds to preservation opportunities as they arise, which may result in an inefficient use of funds towards meeting program goals (Lynch and Musser 2001). Furthermore, local agencies often lack the spatial context, resources and power of a regional scale plan. Regional perspective is especially important in formulating strategies to limit urban sprawl and maintain an economically viable agricultural industry.

In the case of top-down approaches such as Agricultural Protection Zoning (APZ), the objectives tend to focus on agricultural values, e.g. to preserve high quality farmland. Implementation entails identifying which farmland will be included (or in some cases excluded from the agricultural district) and where to prohibit non-farm development in the agricultural district. Criteria for measuring farmland conservation value are typically related to land quality, soil productivity and current land use. Several agricultural land quality ratings have been used with the purpose of measuring land quality and productivity. For instance, San Mateo County in California uses Soil Capability ratings and the Storie Index. In Scott and Black Hawk counties in Iowa, Corn Suitability Ratings are preferred (American Farmland Trust 1997).

In bottom-up policies the scale and site specificity of these programs, along with the need to discriminate among applications, results in the consideration of non-agricultural criteria such as the scenic values of the property. Soil quality and agricultural productivity continue to be an important factor in these programs, but other factors related to the farm viability such as farm size and farm management practices are also considered. In addition, the spatial context of the site (e.g., proximity to urban centers, surrounding land uses) is a factor. This factor is important not only to the viability of the farm, but also because it addresses threat and the use of farm protection as a growth management tool. Regardless of the approach taken, threat is considered as a separate criterion (e.g., urban pressure), and generally assessed through a static measure of the current resource (e.g., distance from urban center) rather than dynamically embedded in each considered criterion. In general, spatially explicit scenarios of urban sprawl and other threatening processes are not used in this process.

Lynch and Musser (2001) describe the characteristics of a desirable parcel for the Maryland Farmland Preservation Programs as “large number of acres, high percentage of prime soil and of crop land, near another preserved parcel, near a metropolitan area, near the closest town, and with a low percentage of pasture and forest acres”. They also describe how goals of the programs may compete and trade-off. For instance, threat correlates with higher cost, which reduces the area of land that can be acquired for a given budget.

8

Budget constraints are certainly a concern for all land preservation strategies. However, there is no uniformity in the way cost is considered. For instance, some PDRs programs operate in a first come first served basis; in others the farmers bid to enter the program once their parcel meets some minimum requirements (e.g. Delaware and Maryland), still others (e.g. New Jersey) combine a scoring system with an option for the farmer to offer a discount for ceding his or her development rights. In Connecticut, applications are penalized if the estimated cost of purchasing the development rights exceeds $10,000 per acre (McCarthy and Frisman 2000).

The Land Evaluation and Site Analysis system (LESA) has been widely used by local and State governments to guide agricultural zoning and to implement farmland protection through PDR and TDR programs (Ferguson, Bowen, and Khan 1991; Pease et al. 1994). It was developed by the Soil Conservation Service in 1981 as a tool to address farmland conversion to non-agricultural uses (Wright 1994). Soon after its design, it was adopted as a federal tool to evaluate the effects of federal programs on farmland protection and to ensure their compatibility with state, local and private policies to protect farmland as mandated by the Farmland Protection Policy Act of 1981.

LESA consists of two parts, land evaluation (LE) and site assessment (SA). The LE part rates the land for crop production and is designed by SCS and local Soil and Water Conservation Districts. The SA component, which is designed by local officials, includes factors other than agricultural productivity. For instance, the California LESA system includes two LE measures, Land Capability Classification and Storie Index ratings, and five SA factors including project size, water availability, surrounding agricultural land and surrounding protected resource land ratings (California Department of Conservation 1997). A final composite score on a 100-point scale is calculated as a weighted sum of the criteria scores.

Although LESA is considered a valuable tool to help decision making in protecting farmlands, it has shortcomings for prioritizing farmland preservation. Criteria weighting tends to be fixed and implicit rather than flexible for sensitivity analysis and generation of alternative conservation scenarios. Threat is often measured indirectly using variables related to potential for development. Also, conservation costs are generally not explicitly included as a scoring factor. Some agencies consider the willingness of the landowner to discount part of the price of the parcel as a decision criterion besides the LESA score (Delaware Department of Agriculture). Some criteria can only be assessed by direct observation through a field visit.

Coughlin (1994), in an evaluation of several LESA systems, concluded that often two or more SA factors were correlated. He also referred to score ambiguity issues and the concern that the effects of changing factors weights may not be very intuitive. Ferguson, Bowen, and Khan (1991) concluded that LESA could be susceptible to manipulation and was difficult to update or adapt over time.

9

Prior to the advent of modern database and computerized mapping systems, manual methods were used to identify and establish farmland preservation priorities based on land use and land capability classification maps. The process was tedious, subjective and prone to errors (Ferguson, Bowen, and Khan 1991; Tulloch et al. 2003). Although manual point scoring systems are still being used to prioritize farmland preservation, the trend is to automate the system.

The LE part of LESA was automated in the early 1980s through a Computer-Assisted Land Evaluation System (CALES) program designed originally by the Corps of Engineers Research Laboratory to use soil survey data from the NRCS NASIS database (US Department of Agriculture 1991). More recent efforts employ Geographic Information System (GIS) technology. The LESA system has been automated using GIS software for different purposes in Hawaii, Illinois, Kansas, Oregon, Vermont, Delaware, and Pennsylvania (Pease et al. 1994). For example, researchers at Pennsylvania State University developed a LESA ArcView GIS interface for prioritizing counties’ applications to the Pennsylvania’s Agricultural Conservation Easements Purchase Program. The LE of parcels is based on its soils productivity whereas the SA includes three factors, development potential, farmland potential and clustering potential. Each of these is characterized by several subcriteria specific to each county (National Consortium for Rural Geospatial Innovations 2000).

Tulloch et al. (2003) describe their attempt to automate a manual bottom-up process of farmland parcel prioritization and to extend it to a top-down evaluation for Hunterdon County, New Jersey, USA. The authors used the criteria currently in practice—soil productivity, compatibility of adjacent land use, local commitment to preservation, farm size, density of easements in the neighborhood, and farm management practices. These criteria emphasize viability of farming but do not address threat or cost-effectiveness, although the state of New Jersey lists those as general guidelines (Tulloch et al. 2003). Each of the five criteria is based on a prescribed point scoring system. For instance, the soil productivity score for a parcel is a weighted sum of the relative proportions of a farm in various categories of importance (i.e., prime, statewide importance, unique, or local importance). The maximum points for each of the five criteria carries an implicit weighting of their relative importance in addition to the weighting of the subcriteria such as that mentioned for soil categories. Total conservation value is based on a simple sum of the individual criteria scores. Tulloch et al. (2003), encountered difficulties automating some of the criteria, specifically management practices, illustrating the asymmetry of information held by planners and landowners. On the other hand, they found GIS automation reduced the opportunities for error and subjectivity inherent in the manual processing.

In summary, there are a wide variety of tools and approaches to farmland preservation that we have considered in developing our framework. Our model,

10

however, constitutes a distinctive approach to the farmland preservation problem in several key ways.

• The emphasis of most approaches has been bottom-up and opportunistic. Our strategic top-down approach is proactive and prioritizes conservation investment priorities within the state or region as required for CLP.

• Most existing approaches are not oriented towards an exploratory collaborative decision making system. We propose a framework with variable weights for each conservation objective, so that different stakeholder preferences could be used to generate and evaluate alternative farmland preservation scenarios.

• Threat is generally treated as one criterion on par with criteria such as soil capability or compatibility of farming with local land use. In our framework, threat is embedded in each objective as a measure of the potential loss of the corresponding resource; each kind of threat pertains to a specific resource value. Our assessment of conservation value is based on expected future condition rather than current condition.

• We integrate the costs of the conservation action within the strategic allocation model, which results in a more efficient and transparent way of allocating funds.

• We take an economics perspective that argues that the marginal benefit of conserving a unit of farmland diminishes as the amount of farmland preserved in a region increases (Hughey, Cullen, and Moran 2003).

3. CONCEPTUAL FRAMEWORK

GUIDING PRINCIPLES

In designing a framework for prioritizing farmland for preservation, we adopted certain guiding principles. Foremost of these is that it should be process-oriented, not product-oriented. Land conservation is contentious and expensive. The purpose of the framework is to provide a structured way of thinking about the problem and to support collaborative processes and negotiation among competing interest groups. To be successful, the framework must possess the following characteristics:

• Simplicity--Avoid overly complex rules and formulae. Any stakeholder should be able to understand and “play the game.”

• Explicitness--Be clear how a value is arrived at.

• Flexibility--Make it easy to change the weights or rules used to calculate a value to facilitate policy and sensitivity analyses.

11

• Modularity--Keep different components of value separate so that one can decompose or reconstruct the overall value of a place using various subsets of criteria; easy to substitute new data.

• Feasibility--Use widely available map and monitoring data.

SPATIAL ANALYTICAL FRAMEWORK

We distinguish four different kinds of spatial units of analysis:

The planning region encompasses the entire area under consideration for conservation investments. This could be a country, an ecological region, or the entire state.

Observations (o) are data pertaining to a particular resource concern that are available across the entire planning region at some minimum spatial resolution. This resolution could be the minimum mapping unit of a map of irregular polygons or the cell size of a regular grid such as those produced by classification of remotely sensed imagery.

Sites (i) are discrete spatial units that are the candidate areas being prioritized for conservation investments. In our formulation these areas have fixed boundaries (Cocks and Baird 1989). Observations are perfectly nested within sites and sites are perfectly nested within the planning region (Davis and Stoms 1996).

A reference region (r) is the area that is considered when evaluating a site with respect to a particular conservation concern. For example, a county or identifiable agricultural areas might be the reference area for assessing farmland conservation value. Observations and sites are nested within reference regions. A reference region is nested within the planning region (or could be the region itself) but, reference regions for different concerns can be non-nested and based on very different criteria.

SETTING CONSERVATION PRIORITIES

The problem of prioritizing farmland for conservation is one instance of the more general problem of multicriteria land use planning, and thus it is appropriate to approach it with a similar planning framework. Adapting from Steinitz (1990), we restate the framework as a series of six steps.

��Define goals and measurement criteria�

2. Describe current resource amount by the measurement criteria

3. Generate scenario(s) of threatening processes

4. Predict future resource amount

12

5. Measure the conservation value of candidate areas based on resource quality, condition, and threats.

6. Allocate funds based on the maximum conservation value for the budget

The planning goal in this case can be defined as maximizing the farmland conservation value in a planning region that remains at a given time in the future by investing a fixed level of conservation funds to protect a set of sites. Formally,

[1] ∑i

it

i XMaximize V

[2] BCXtoSubjecti

ii ≤∑

where Vi is the farmland conservation value of site i at time t in the future, Xi is the decision variable that identifies whether site i was selected (1) or not (0), Ci is the conservation cost of preserving farmland in site i, and B is a budget constraint.

The conservation value of a site is defined through a weighted combination of general objectives that reflect stakeholder preferences in the planning region. Specifically,

[3] ∑∑

=

=

= J

j N

i ij

ijji M

MwV 11

)(

where wj is the weight assigned to objective j and Mij is the marginal value of conserving site i for objective j. The marginal value is a function of the site’s farmland value or resource amount and the conservation goal set for the reference region.

Stakeholders within a single area are likely to have different opinions about these criteria and their relative importance. For instance, environmentally-oriented stakeholders may wish to preserve farmland for their amenity values such as habitat and groundwater benefits or for scenic quality and pastoral landscapes, whereas agricultural interests typically value agrarian lifestyle criteria and soil productivity more highly (Kline and Wichelns 1998). Some see farmland preservation as a tool for controlling the rate or direction of urban growth and therefore would favor criteria related to negative impacts of urban development. Each general objective could be stated as “to retain threatened farmlands with the greatest farmland value or resource amount”; the type of resource will be different in each case. From a review of the literature and existing farmland preservation programs, and the conclusions of an expert workshop held at the National Center

13

for Ecological Analysis and Synthesis, (Regan 2001) we have identified three general objectives that are commonly applied in prioritizing sites for the preservation of farmland.

1. Retain farmlands with the greatest sustained production capacity.

2. Retain farmlands that provide high amenity values (e.g., habitat, open space, floodplain management, scenic values).

3. Retain farmlands whose location reduces the risk of urban sprawl.

Appropriate criteria must be identified to determine the current site and reference region value for each objective. In bottom-up approaches, this may be done through forms filled out by landowners (and the appropriate agency), but in the top-down approach it is accomplished through spatial analysis with GIS technology (Tulloch et al. 2003).

We define resource amount or farmland value of a site as a quality-weighted term. We do not attempt to define it in detail for all three objectives here, but illustrate for Objective #1. We define production capacity or production value at present time (to), to be the capability- and-condition weighted farmland area in site i. In particular,

[4] oio

oti cpa ∑

∈

=0

where 0tia is the current production capacity or value for site i expressed as a

product of resource capability (po) and condition (co) of each observation cell, summed over all observation cells in the site.

We refer to capability as an intrinsic measure of the land’s ability to grow sustainable yields based on its biophysical and chemical soil properties, climate and water supply. Condition refers to other factors such as current land use that modify such ability.

We assume that conversion of farmland to urban and suburban use poses the greatest threat of “irreversible” loss of farmland in California, and that areas will remain farmland unless they are converted to urban and suburban use (Greene and Stager 2001). Conservation efforts would logically focus on areas where high production capacity will be lost or greatly reduced by land conversion unless conservation actions are taken. Consequently, it is necessary to predict where urban growth will occur (step 3) to estimate the future resource amount and the potential loss of farmland benefits (step 4) before integrating the criteria into the overall conservation value (step 5).

We calculate the projected loss of the site’s farmland value if no conservation is implemented. We assume that the future value with conservation is equal to its present value at time (t0) as follows,

14

[5] ti

tii aav −= 0

where tia is the future resource amount in site i, and vi is the potential reduction in

amount without conservation intervention.

We assume that the total “utility” or social value of farmland in a reference region increases as a function of total resource amount in that area, but that the marginal utility of each increment of resource amount decreases as more farmland is secured. For simplicity we assume a linearly decreasing marginal value, although other functional forms could be substituted to address non-linear cumulative effects (e.g., Wu and Boggess 1999). We derive the site conservation value based on its marginal value expressed as follows,

[6]

+−=

GvAvM

r

itr

ii

)*5.0(1*

where Gr is the goal for farmland preservation in reference region r and trA is the

total future amount of farmland value in that region, specifically,

[7] ∑∈

=ri

ti

tr aA

The demand for the public non-market benefits of farmland is not fixed, but diminishes as the supply of prime farmland relative to the extent of urban development increases (Poe 1999). We assume that the total utility of protected farmland increases but at a diminishing rate as the secured amount increases in a reference region (illustrated with hypothetical values in Figure 1.a). The shape of this utility function could take many forms. For simplicity let us assume a quadratic utility function and thus a linearly diminishing marginal utility that equals zero at the conservation goal (Fig. 1.b). The triangle on the right end of the curve represents farmland amount that has already been converted to urban use. The point on the X-axis labeled ot

rA denotes the total amount of farmland value remaining today. Without any additional farmland preservation, the amount of farmland value remaining at the end of the planning period is the area under the curve between the origin and t

rA . A conservation action will shift the predicted

future amount to a revised future amount, 't

rA . The marginal conservation value of

site i, Mi, is the area under the curve between trA and ( t

rA + vi) (Equation [6]).

Step 6 in the planning process is the prioritization of farmland sites for a specific development scenario. This process is driven by two factors, the marginal

15

conservation value of the candidate planning unit and the cost of implementing the conservation policy, as formulated in equations [1] and [2]. The cost of conservation intervention is a key factor in our conceptual framework. Ignoring cost could lead to decisions to preserve the highest valued farmland in the short-term but would not necessarily maximize the overall farmland value remaining in the future. As mentioned in the Introduction, several conservation policies can be used to preserve farmland, and the costs will vary depending on the policy.

It should be kept in mind that planning is a social and political process, involving extensive iteration, exploration of alternatives, and negotiation before a plan is adopted. The technical framework outline here is intended to facilitate and guide that process.

4. FRAMEWORK APPLICATION IN THE BAY AREA/DELTA BIOREGION

In this section we apply our framework to examine conservation priorities in the Bay Area/ Delta Region of California (Fig.2). This bioregion covers an area of about 1.3 million hectares (3.1 million acres) and essentially encompasses the San Francisco Bay Area and the Sacramento-San Joaquin River Delta. It includes Marin, Contra Costa, Santa Clara, Alameda, Solano, San Mateo, San Francisco, Sonoma Napa, San Joaquin counties and part of Yolo and Sacramento counties. This is the second most populous bioregion of the state with about 6.8 million people in 2000. The population of the region is projected to increase by 15% to more than 8 million people by 2020 (California Resources Agency, 1998). According to projections, the greatest growth will occur in the eastern counties with larger irrigated agriculture areas (Fig.6), whereas the coastal counties will experience a slower growth rate (California Dept of Forestry & Fire Protection 1997; Landis et al. 1998).

GOALS AND MEASUREMENT CRITERIA

The overall goal for farmland preservation is to maximize the farmland conservation value in the bioregion that remains at a given time in the future for a specified economic cost. Here we simplify the set of objectives and only measure conservation value for agricultural productivity. The second objective requires subjective data such as farmland practices and size and amenity values that typically can only be compiled at a very local scale. The minimizing sprawl objective similarly requires local data on infrastructure, policies, zoning, and costs of services. Therefore, we submit that these two objectives may be more appropriate for farmland prioritization at more local scales. Because CLP focuses on lands of statewide significance, we opt for the objectives and measurement criteria that can be feasibly assessed with comprehensive statewide data.

Within the planning region, we use counties as reference regions for computing the marginal conservation value (we combine the portions of Sacramento and Yolo counties). Sites in counties with less remaining farmland will tend to have higher marginal conservation value, Mi, than sites with the same vi in a county with more

16

farmland. Within counties, we delineated sites, for which we measure marginal conservation value, based on 3 x 3 mile quarter townships, derived from a modified version of the Public Land Survey (California Spatial Information Library 1999). We believe these units are at an appropriate resolution for state and regional prioritization, and they generally correspond to patterns of land ownership and land use, particularly in agricultural areas. We measure our criteria using 100-meter grid cells (observations), aggregating these to the quarter townships.

ASSESSING CURRENT RESOURCE LEVEL

Site agricultural capability and condition determine the useable production capacity of farmland. The former is based on the inherent characteristics of the land, climate, and water supply to grow crops, whereas the latter reflects the modification of that capability by the social context of the observations.

Capability (p)

The California Department of Conservation conducts the Farmland Mapping and Monitoring Program (FMMP) to survey and monitor land use change of farmland throughout the state. Since 1984 and biannually, the program publishes GIS data and maps showing the location and extent of important and interim farmland classes (Table1). The minimum mapping unit is 4 hectares (10 acres). Agricultural land is rated according to soil quality and irrigation status and important farmland classes are mapped where USDA-Natural Resource Conservation Service (USDA-NRCS) “modern soil surveys” exist. If a county lacks modern soil surveys and there is “expressed local concern on the status of farmland”, then the interim farmland classes are used. These categories are a temporary substitute for the important farmland categories, and designed to be upgraded as the surveys become available (California Department of Conservation 1994).

We used these categories as a measure of agricultural capability. That process involved the assignment of capability weights to each one of the farmland categories. Because there is not a direct equivalency between FMMP categories and quantitative capability scores, we used the technical definitions of each FMMP category (California Department of Conservation 1994) to derive numeric scores. Scoring is based on soil physical and chemical properties that sustain long term production, cropping and irrigation status.

The lack of more structured and detailed information about certain categories (e.g. local importance farmland) increased the subjectivity involved in weighting qualities, making it a challenging task. We would stress that the proposed weights are for demonstration purposes only. Table 1 defines briefly each farmland category and summarizes our scoring rationale. A recent report by McCoy et al. (2002) used similar values.

17

We converted the 2000 version of FMMP data to a 100m grid and recoded all cells from the original farmland classes to corresponding capability weights (Fig.3).

Condition (c):

The intrinsic capability of farmland may be constrained by social factors, primarily objections regarding management practices from residential neighbors (Bradshaw and Muller 1998). We assessed site condition as a function of the degree of urbanization in neighboring areas. FMMP defines urban and built-up land as “land occupied by structures with a building density of at least 1 unit to 1.5 acres, or approximately 6 structures to a 10 acre parcel” (California Department of Conservation 1994). Sites in good condition are surrounded by extensive contiguous farmland, whereas those in poorer condition occur in areas with highly intermingled urban and farmland use, analogous to habitat fragmentation.

To measure condition we determined the number of urban cells within a 500m radius of each 100 m agricultural cell and then subtracted the proportion of urban cells in the 78.5 ha (194 ac) buffer area from 1. Thus the index (c) ranges from 0 (poor condition) to 1 (good condition) (Fig.4).

Next, we calculated the production capacity for each site, 0tia by multiplying the

current capability and condition of each observation cell o, and summing the values over all the farmland cells in each site (Fig.5). In particular,

[9] 000 to

io

to

tia cp∑

∈

=

ESTIMATING FUTURE RESOURCE LOSS AND SITE CONSERVATION VALUE

Future urban growth impacts farmland capability and condition in two ways. First, urbanized cells have no farmland capability (p=0). Second, the advancing urban perimeter reduces the condition term for remaining farmland. We used the California Urban and Biodiversity Analysis (CURBA) model (Landis et al. 1998) as a scenario of future urban growth. This model forecasts the spatial pattern of future urbanization based on statistical models of the relationship between recent historical urbanization and predictor variables such as physical site characteristics, locational and economic factors, characteristics of nearby areas, and policy and administrative controls on land use. For this demonstration we used the predictions for 2050 (Fig.6) and overlaid this on the FMMP-derived capability map to identify where capability will be lost. Condition was also re-evaluated based on this scenario of urban development.

As before, the future production capacity for each site is calculated as the product of future capability and future condition, summed over all cells in a site (Fig.5.b). In particular,

18

[10] tc

ic

tc

ti cpa ∑

∈

=

The potential loss of farmland value, vi, is calculated as the difference between current and future production capacities, as in equation [5] (Fig.7). The marginal conservation value of site i, is then calculated using Equation 6 (Fig.8). We set the farmland preservation goal for each county, Gr, at 100% of current production capacity. Finally, the conservation value is calculated using Equation [3] (in this case wj=1 because our demonstration only has a single objective).

ALLOCATION OF THE FUNDS BASED ON A FIXED BUDGET

Estimating conservation costs

The California Farmland Conservancy Program employs conservation easements (Purchase of Development Rights or PDR) as its primary tool for preserving farmland of statewide significance. In theory, the easement cost is the difference between the maximum value (i.e., current market value as developed land, minus any conversion costs) and its value in permanent agricultural use (Plantinga and Miller 2001). Both of these terms vary somewhat independently, so that the cost of easements is not a simple proportion of current market value. Thus easements should cost the most where urban potential is high but farmland productivity is relatively low, and will cost less where farmland productivity is high but urban potential is low. Determining the easement value by appraisal methods is problematic, because neither the time of development nor the rents to be obtained from it in the future can be observed. An alternative approach is to model easement value from observable data (Lynch and Lovell 2002; Plantinga and Miller 2001).

For demonstration purposes, we modeled easement value based on 31 recent farmland conservation easements in or near the Bay Area/Delta bioregion for which the location and easement price were known. These data are not exhaustive but represent a sample throughout the planning region based on readily available information (California Department of Conservation 2003; Marin Agricultural Land Trust 2003; Midpeninsula Regional Open Space District 2003; Peninsula Open Space Trust (POST) 2003; Sonoma County's Coalition for the Outdoor Recreation Plan; The Bay Area Open Space Council 2003; The Land Trust of Napa County 2003; Yolo Land Trust). Predictor variables suggested by Lynch and Lovell (2002) were derived by GIS processing for the distance in miles from either Sacramento or the San Francisco-San Jose-Oakland urban area, the minimum distance from all other populated places with greater than 5000 people in the 1990 census, the distance from ocean or bay shoreline (as a scenic amenity that would potentially add to the development pressure), and the proportion of the sites occupied by

19

Prime or Statewide Significant farmland in the 2000 FMMP mapping (or of the farm itself if known). Regression tree analysis was used to determine the form of a statistical model that predicted easement values in the training data. The predicted values range from $582 to $8,888 per acre, with a mean in the bioregion of $2,636. Actual easement prices in our sample ranged from $600 per acre for dryland grains to over $15,000 per acre for vineyards. The conservation cost of a site was then computed as the product of its per acre value and its acres of farmland (Fig.9).

Prioritization of sites for farmland preservation

Using the predicted PDR costs and the measure of overall marginal value for farmland preservation, we applied a greedy heuristic algorithm to solve the optimization problem in Equation [1]. The heuristic modifies the formulation into Equation [11] that selects the site with the highest ratio of benefits to costs (i.e., the biggest bang for the buck). The Ar

t and Mij terms are updated with each selection as threat is removed, and then the site that now has the highest benefit-cost ratio is selected until the budget, B, is exceeded.

[11] ii

ii CXVMaximize ∑ /*

We ran the algorithm until an arbitrary 20 planning units were selected at a predicted total easement cost of $113 million for 31,000 acres of farmland (Fig. 10). This represents an average of $3,650 per acre, or $1,000 higher than the average price predicted in the bioregion. This result should be interpreted as one possible scenario, which is very sensitive to our very crude estimates of easement values, to the choice of reference regions and goals, and to the future model of urbanization.

Rather than focusing on the locations, it is more useful to examine what the allocation reveals. Figure 10 shows that the 20 planning units selected occur where marginal value divided by easement price is highest. These higher values occur in the eastern half of the bioregion, particularly in Alameda County. This result is largely reflects the small amount of farmland in this county and relatively low predicted easement values. In other words, each planning unit in Alameda County contributes a relatively large share of the total amount of marginal farmland value. In contrast, Sonoma County has a relatively larger amount of farmland value, so that each planning unit contributes relatively little, even if the vi values were similar. Moreover, predicted farmland loss is much greater in Alameda County than in Sonoma County (Fig.1.c).

20

5. CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS

We have presented a simple framework for prioritizing farmland areas for conservation based on cost-effectiveness. The framework evaluates a hierarchical set of weighted concerns and associated criteria, and allows stakeholders to explore scenarios by assigning a choice of weights to the objectives. It is explicit, flexible, modular, and feasible to implement. The evaluation and prioritization process could be applied to consider a range of policy options for implementing farmland preservation (e.g., purchase or transfer of development rights, tax incentives, zoning, or growth management programs), assuming that costs of different conservation tactics can be estimated reliably.

We applied the framework to evaluate conservation priorities in a rapidly urbanizing region of California. Because of the coarse scale of this analysis, only the farmland productivity objective could be quantified with available GIS data. Thus conservation value was based on site capability, condition, and threat of conversion to urban use. Easement cost was estimated from a sample of recent transactions in the study region. We used a greedy selection algorithm to select 20 sites for farmland preservation in the region with a total predicted easement value of $113 million.

The primary input data for measuring farmland productive capability was the FMMP GIS coverage for 2000 (California Department of Conservation 2002). Not all important farmland areas of California have been mapped by FMMP. In the interim, we were able to use other data layers to develop a predictive model to identify important farmlands in these unmapped locations (see Appendix B), but more complete mapping would benefit future applications of this model.

FMMP farmland importance classes were assigned weights reflecting their perceived productive capability and therefore conservation value (Table 1). We were unable in our review of the literature and discussions with agency staff to arrive at a firm basis for estimating relative capability of the different classes (e.g. profit or the average yield per acre). McCoy et al. (2002) recently proposed a scoring very similar to our initial attempt. Other authors have weighted Important Farmland classifications (Tulloch et al. 2003) but we could not find a one-to-one equivalency between their categories and the ones used in the FMMP. Additionally, selecting the appropriate land evaluation system is not easy. There are many classifications available to rate soil quality.

The measure of threat to farmland resources was derived from a version of the CURBA model of urban growth. For simplicity, we used a binary version of a CURBA output. However, CURBA generates probabilities of development, which could be used as a stochastic representation of threat rather than the deterministic (i.e., threatened or not threatened) method used here. In addition, the model is undergoing continuous refinements that may alter the patterns of growth and

21

threat, and therefore is likely to change the prioritization of farmland for preservation.

In this implementation we did not consider the interaction between conservation decisions and future development. We made the simplifying assumption that when a site is selected for preservation, the threat of urban growth at that site is removed rather than displaced to other locations in the region. To implement the latter would require feedback from the allocation model to the urban growth model and then a re-calculation of the marginal values in light of shifting demand and threat. In principle this could be accomplished by tighter integration of the conservation model and the urban growth model, which we consider an interesting and important research direction.

We applied a 500-meter window to determine the potential impact of urban areas on farming practices. This 500-m value was based on a statement in Bradshaw and Muller (1998) about this being the dimension of a zone-of-conflict between urban and agricultural uses that affects the ability to farm. However, Ferguson and Khan (1992) used 800m (0.5 mile) distance from the sites to compute potential conflict with urban areas. Other authors use the shortest distance of each pixel to constructions as a measure of fragmentation (Gulinck and Wagendorp 2002). The length of the urban-agriculture border has also been used as a proxy to predict farmland conversion pressures (Kuminoff and Summer 2001). The width of the zone of conflict may vary depending on the size and type of farming practices.

We focused our valuation of farmland on the productivity objective because the criteria for measuring it were most readily available. The objectives for amenity values of farmland and controlling urban sprawl were identified from the literature and the NCEAS workshop but were not developed in detail in this report. We recommend that further consideration be given to their role in the framework, especially as they are often the most important objectives for many stakeholders (Kline and Wichelns 1996; Kline and Wichelns 1998). We expect that these latter two objectives will be challenging to evaluate at a regional or state scale and acknowledge that their role in defining farmland of "statewide significance" is less straightforward than the productivity objective.

In our framework, goals for farmland preservation, and the level of goal achievement, are based on user-defined reference regions. For this application, we opted for counties as reference regions. Agricultural land trusts are often organized at a county level and so that seems to be a reasonable spatial unit for the framework. However, there may be some other land divisions that reflect the agricultural landscape that citizens identify with better than counties do (e.g., Farmland Priority Zones or FPZs, Land Information and Computer Graphics Facility 2003). Alternatively, a bioclimatic classification may be the best system for delineating regions of different agricultural potential, especially in the face of climate change. This issue of reference regions should be investigated further, as it could significantly affect the prioritization.

22

Finally, uncertainty is not addressed in this version of the framework, and yet it clearly exists throughout the process. There are various forms of map error in the farmland class and ownership maps. As mentioned, urban growth is a forecast of an unknown future, and could be treated more appropriately as probabilities. Likewise, the value of development rights is modeled and undoubtedly contains a large degree of uncertainty.

The framework was designed for top-down strategic planning, but we believe it can be adapted to bottom-up prioritizing of individual farms that are offered by landowners. This could be accomplished by extending the criteria to include information available to the landowner or a county (but not the state or regional planner) or adapting the general criteria based on higher resolution and quality information about soil capability, fragmentation, and threats. For instance, the landowner may offer an easement for much less than the predicted value of their development rights, which would dramatically alter the cost effectiveness of the property. Data such as local zoning might be used in evaluating the amenity values and controlling sprawl objectives as well. Nevertheless, the regional scale prioritization demonstrated here can provide a context for these opportunistic, local decisions. “The effectiveness of farmland preservation efforts is potentially higher if state and local agencies work together in crafting comprehensive strategies adapted to the uniqueness of the planning region” (American Farmland Trust 1997). Additional research is needed to design a framework that links the two modes and perhaps provides feedback from the information content of farm level proposals to improve the estimates of regional scale values.

6. REFERENCES

Alterman, R. 1997. The Challenge of Farmland Preservation. Lessons from a Six-Nations Comparison. Journal of the American Planning Association 63 (2).

American Farmland Trust, (AFT). 1997. Saving American Farmland: What works. Northampton: American Farmland Trust.

Brabec, E., and C. Smith. 2002. Agricultural land fragmentation: the spatial effects of three land protection strategies in the eastern United States. Landscape and Urban Planning 58 (2-4):255-268.

Bradshaw, J. K. and Muller, B. 1998. Impacts of rapid urban growth on farmland conversion: Application of new regional land use policy models and geographical information systems. Rural Sociology 63:1-25.

California Agricultural Statistics Service. 2002. Summary of County Agricultural Commissioners Reports, 2001. Sacramento, California.

23

California Department of Conservation. 1994. A Guide to the Farmland Mapping and Monitoring Program. Publication FM 94-02. [accessed 2002]. Available from http://www.consrv.ca.gov/DLRP/fmmp/pubs/FMMP_GUIDE.pdf.

California Department of Conservation. 2003. California Agricultural Land Evaluation and Site Assessment Model Instruction Manual. California Department of Conservation.1997. [accessed March, 24 2003]. Available from http://www.consrv.ca.gov/DLRP/LESA/lesamodl.pdf.

California Department of Conservation. 2003. Easement Projects. March 24, 2003 [accessed March 21 2003]. Available from http://consrv.ca.gov/DLRP/cfcp/stories/easement_projects.htm.

California Department of Conservation. 2002. Farmland Mapping and Monitoring Program (FMMP). March 26, 2002 [accessed July 6 2002]. Available from http://www.consrv.ca.gov/DLRP/fmmp/index.htm.

California Department of Conservation. 2002. Farmland Mapping and Monitoring Program (FMMP). Important Farmland 2000 GIS data. Available from http://www.consrv.ca.gov/DLRP/fmmp/map_products/download_gis_data.htm.

California Department of Forestry and Fire Protection (CDF/FRAP). 2003. California Bioregions (INACGREG). 1992 [accessed August, 8 2003]. Available from http://www.frap.cdf.ca.gov/data/html.

California Dept of Forestry & Fire Protection (CDF/FRAP). 2003. Bioregional Demographic Trends and Implications for Biodiversity. Fire and Resource Assessment Program.1997 [accessed March 2003]. Available from http://frap.cdf.ca.gov/projects/bioregional_trends/index.html.

California Environmental Dialogue. 1999. Land conservation in California: needs for the next decade. Planning and Conservation League. Sacramento, California.

California Legislative Analyst's Office. 1996. LAO Analysis of the 1996-97 Budget Bill Resources, Part I. California Legislative Analyst's Office. Sacramento.

California Spatial Information Library. 2002. US Geological Survey/CDF-FRAP Filled Public Land Survey. 1999. Modified by UCSB Biogeography Lab. [accessed 2002]. Available from http://www.gis.ca.gov.

California Resources Agency. 2003. Preserving California's Natural Heritage. A Bioregional Guide to Land and Water Conservation. Revised Edition. Dec 1998 [accessed Feb 2003]. Available from

24

http://ceres.ca.gov/planning/conservation_guidebook/Bioregions.html.

California State Department of Finance. 1998. County population projections with age, sex and race/ethnic detail: July 1990-2040 in 10-year increments. Demographic Research Unit. Sacramento, California.

Conklin, H.E., and Lesher W.G. 1997. Farm value assessment as means for reducing premature and excessive agricultural disinvestment in urban fringes. American Journal of Agricultural Economics 59 (47):755-59.

Coughlin, Robert E. 1994. Sensitivity, Ambiguity and Redundancy in LESA. In A Decade with LESA. The evolution of Land Evaluation and Site Assessment., edited by F. R. Steiner, J. R. Pease and R. E. Coughlin. Ankeny, Iowa: Soil and Water Conservation Society.

Daniels, T.L. 1991. The Purchase of Development Rights. Preserving agricultural lands and open space. Journal of the American Planning Association 57 (4):421-431.

Davis, F. W., and D. Stoms. 1996. A spatial analytical hierarchy for Gap Analysis. Pages 15-24 in Gap Analysis: A Landscape Approach to Conservation Planning. J. M. Scott, T. Tear and F. W. Davis. Washington D.C., American Society for Photogrammetry and Remote Sensing.

Delaware Department of Agriculture. 2003. Delaware Agricultural Lands Preservation Act. Title III. Chapter 9. [accessed March 26 2003]. Available from http://www.state.de.us/deptagri/law-code/Chapter_9.htm#P129_24632.

Ferguson, C.A, R.L. Bowen, and M.A. Khan. 1991. A statewide LESA system for Hawaii. Journal of Soil and Water Conservation (July-August, 1991).

Ferguson, C.A , and A. Khan. 1992. Protecting farmland near cities. Trade-offs with affordable housing in Hawaii. Land Use Policy (October, 1992).

Geoghegan, J., L. A. Wainger, and N. E. Bockstael. 1997. Spatial landscape indices in a hedonic framework: an ecological economics analysis using GIS. Ecological Economics 23:251-264.

Greene, R.P., and J. Stager. 2001. Rangeland to cropland conversions as replacement land for prime farmland lost to urban development. Social Science Journal 38 (4):543-555.

Guikema, S., and M. Milke. 1999. Quantitative decision tools for conservation programs planning: Practice, theory and potential. Environmental

25

Conservation 26:179-189.

Gulinck, H., and T. Wagendorp. 2002. References for fragmentation analysis of the rural matrix in cultural landscapes. Landscape and Urban Planning 58 (2-4):137-146.

Harlin, J.M. 1995. Rangeland to Cropland conversions in the Western States: Implications for Soil Loss and Sustainability. Paper read at Forum of the Association fro Arid Land Studies XI, at International Center for Arid And Semiarid Land Studies, Lubbock, Texas.

Heinzel, W. M. 2000. Using the Spatial Analyst Model Builder to support citizen-based land use planning. Paper read at ESRI User Conference, at San Diego.

Hughey, K. F. D., R. Cullen, and E. Moran. 2003. Integrating economics into priority setting and evaluation in conservation management. Conservation Biology 17:93-103.

Kline, J., and D. Wichelns. 1996. Public preferences regarding the goals of farmland preservation programs. Land Economics 72 (4):538-549.

Kline, J., and D. Wichelns. 1998. Measuring heterogeneous preferences for preserving farmland and open space. Ecological Economics 26:211-224.

Kline, J. D., and R. J. Alig. 1999. Does land use planning slow the conversion of forest and farmlands? Growth and Change 30: 3-22.

Kuminoff, N.V., and D. A. Summer. 2001. Modeling Farmland Conversion with New GIS Data. Paper read at Annual Meeting of the American Agricultural Economics Association, August 5-8, at Chicago. Available from http://aic.ucdavis.edu/research1/FarmGIS1.pdf.

Land Information and Computer Graphics Facility. 2003. Farmland preservation and GIS: A model for deriving farmland priority zones. Technical paper no. 3 University of Wisconsin-Madison, 2000 [accessed February 16 2003]. Available from http://www.lic.wisc.edu/pubs/FPZ.pdf.

Landis, J.D., J.P. Monzon, M. Reilly, and C. Cogan. 1998. Development and Pilot Application of the California Urban and Biodiversity Analysis (CURBA) Model. University of California at Berkeley, Institute of Urban and Regional Planning.

Levia, D.F. 1998. Farmland conversion and residential development in north central Massachusetts. Land Degradation and Development . 9:123-130.

Lynch, L., and N.W. Musser. 2001. A Relative Efficiency Analysis of Farmland

26

Preservation Programs. Land Economics. 77 (4) Nov.2001.

Lynch, L., and S. J. Lovell. 2002. Hedonic Price Analysis of Easement Payments in Agricultural Land Preservation Programs. Department of Agricultural and Resource Economics. College Park, MD.: University of Maryland.

Marin Agricultural Land Trust, (MALT). 2003.[accessed April 2003]. Available from http://www.malt.org.

McCarthy, Kevin E., and Paul Frisman. 2003. OLR Research Report. Farmland Preservation Programs in Other States. Connecticut General Assembly, 2000 [accessed January 2003].

McCoy, M. C., J. F. Quinn, and N. Kalman. 2002. Identifying Environmental and Agricultural Values and Opportunities for Regional Planning: A GIS Approach. Information Center for the Environment, University of California Davis. Report prepared for the Sacramento Council of Governments. Accessed June 18, 2003. Available online at http://www.ice.ucdavis.edu/sacog/mccoy_environmental_paper.doc.

Midpeninsula Regional Open Space District. 2003. [accessed April 2003]. Available from http://www.openspace.org/.

National Consortium for Rural Geospatial Innovations. 2003. Farmland Protection and GIS: GIS Interface Helps Pennsylvania Counties Prioritize Farmland for Preservation. Land Information Bulletin. Pennsylvania State University, December 2000 [accessed March 2003]. Available from http://www.lic.wisc.edu/pubs/Penn1.pdf.

Nelson, A.C.1990b. Blazing New Planning Trails in Oregon, Urban Land. Washington DC: Urban Land Institute.

Nelson, A.C. 1992. Preserving Prime Farmland in the Face of Urbanization. Lessons from Oregon. Journal of the American Planning Association 58 (4):467-488.

Pease, James R., Robert E. Coughlin, Frederick R. Steiner, Adam P. Sussman, Lyssa Papazian, Ann Pressley Joice, and C. Leach John. 1994. State and Local LESA Systems: Status and Evaluation. In A Decade with LESA.The evolution of Land Evaluation and Site Assessment. Edited by F. R. Steiner, J. R. Pease and R. E. Coughlin. Ankeny, Iowa: Soil and Water Conservation Society.

Peninsula Open Space Trust (POST). 2003. [accessed March 2003]. Available from http://www.openspacetrust.org/

27

Plantinga, A. J., and D.J. Miller. 2001. Agricultural land values and the value of rights to future land development. Land Economics. 77:56-67.

Platt, R.H. 1985. The Farmland Conversion Debate. Professional Geographer 37 (4):433-42.

Poe, G.L. 1999. Maximizing the environmental benefits per dollar expended: An economic interpretation and review of agricultural environmental benefits and costs. Society & Natural Resources 12:571-598.

Regan, H., S. Andelman, P. Dangermond, F. Davis, G. Greenwood, D. Kelley, J. Loux, W. Rash, R. Standiford, B. Stewart and E. Vink. 2001. Report of NCEAS/CCRISP Workshop on Agricultural Lands. Santa Barbara, California, National Center for Ecological Analysis and Synthesis.

Rosenberg, D. 2003. Agricultural conservation easements: Protecting farmland in the 21st century. February 9 2000. [accessed March 24 2003]. Available from http://www.yolocounty.org/org/bos/Rosenberg/easement.htm.

Solomon, Barry. 1984. Farmland protection: a case of quality Not Quantity. Land Use Policy 1 (4):357-66.

Sonoma County's Coalition for the Outdoor Recreation Plan, (CORP). 2003. Acquisitions of the Sonoma County Agricultural reservation and Open Space District [accessed April 2003]. Available from http://www.sonomatrails.org/docs/acqtabd2.htm.

Steiner, F., R. Dunford, and N. Dosdall. 1987. The use of the agricultural land evaluation and site assessment system in the United States. Landscape and Urban Planning 14:183-199.

Steinitz, C. 1990. A framework for theory applicable to the education of landscape architects (and other design professionals). Landscape Journal 9:136-143.

The Resources Agency. 2001. California Continuing Resources Investment Strategy Project .Legal mandates related to the conservation of land and natural resources. Sacramento, California.

The Bay Area Open Space Council. 2003. Annual Reports. [accessed April 2003]. Available from http://www.openspacecouncil.org/Documents/.

The Land Trust of Napa County. 2003. Land-Saving Accomplishments 2000. [accessed April 2003]. Available from http://members.napanet.net/~nclt/Accomplishments2000.htm.

28

Tulloch, D. L., J. R. Myers, J. E. Haase, P. J. Parks, and R. G. Lathrop. 2003. Integrating GIS into farmland preservation policy and decision making. Landscape and Urban Planning 63: 33-48.

US Department of Agriculture. 2003. User Manual for LESA-CALES: Computer-Assisted Land Evaluation System. Natural Resources Conservation Service,1991 [accessed April 2003]. Available from http://nasis.usda.gov/cales/htdocs/usrman.pdf.

US Department of Agriculture. 2003. Land use, value and management: Maryland case study. Economic Research Service. Briefing Room, November 4, 2002 [accessed March 26 2003]. Available from http://www.ers.usda.gov/Briefing/LandUse/CaseStudies/Maryland.htm.

Wright, L.E. 1994. The Development and Status of LESA . In A Decade with LESA. The evolution of Land Evaluation and Site Assessment, edited by F. R. Steiner, J. R. Pease and R. E. Coughlin. Ankeny, Iowa: Soil and Water Conservation Society.

Wu, J.J., and W.G. Boggess. 1999. The optimal allocation of conservation funds. Journal of Environmental Economics and Management. 38 (3):302-21.

Yolo Land Trust. 2003. [accessed April 2003]. Available from http://www.yololandtrust.org/.

29

7. TABLES

Farmland Category Score Type Brief Definition

Culti-vated Irrigated

Prime 1.0 Important

Farmland with the best combination of physical and chemical features able to sustain long-term agricultural production. This land has the soil quality, growing season, and moisture supply needed to produce sustained high yields.

Yes Yes

Statewide 0.8 Important

Farmland similar to Prime Farmland but with minor shortcomings, such as greater slopes or less ability to store soil moisture.

Yes Yes

Unique 0.6 Important

Farmland of lesser quality soils used for the production of the state's leading agricultural crops. This land is usually irrigated, but may include no irrigated orchards or vineyards as found in some climatic zones in California.

Yes Usually Irrigated

Local Importan

ce 0.2 Important

Land of importance to the local agricultural economy as determined by each county's board of supervisors and a local advisory committee. Each county has it own definition. This farmland is either currently producing crops, has the capability of production, or is used for the production of confined livestock.

varies varies

Local Potential

0.2 Important

Defined by a local advisory committee. Land that includes soils which qualify for prime or Farmland of Statewide Importance.

Generally Not

Generally Not

Irrigated 0.5 Interim Cropped land with a developed irrigation water supply that is dependable and of adequate quality.

Yes Yes

30

Farmland Category Score Type Brief Definition

Culti-vated Irrigated

Non-Irrigated 0.2 Interim

Land on which agricultural commodities are produced on a continuing or cyclic basis utilizing stored soil moisture.

? No

Table 1. FMMP Categories Scoring (Classes and definitions from California

Department of Conservation 1994, 2002).

31

8. FIGURES

Figure-1.a Total conservation value.

Figure-1.b Marginal conservation value and conservation goal.

Figure-1.c Marginal conservation value and conservation goal, Alameda and

Sonoma counties.

Figure-2 Study Area

Figure-3 Agricultural Capability 2000.

Figure-4.a Agricultural Condition Index 2000.

Figure-4.b Agricultural Condition Index 2050.

Figure-5.a Agricultural Production Capacity 2000.

Figure-5.b Agricultural Production Capacity 2050.

Figure-6 Projected Development.

Figure-7 Potential Loss of Production Capacity (vi)

Figure-8 Site Marginal Value.

Figure-9 Modeled Conservation Cost

Figure-10 Model Solution for 20 sites

32

Fig. 1.a

Total Conservation Value

0 20 40 60 80 100% of resource amount

Futureamount, A r

t

Currentamount, A r

to

Past conversion

Secure

33

Fig. 1.b

Marginal Conservation Value

0

0.2

0.4

0.6

0.8

1

0 20 40 60 80 100

% of conservation goal, G r

v i

Past conversion

Secure

Original futureamount, A r

t

Currentamount, A r

to

Revised futureamount, A r

t'

M i

34

Fig. 1.c

35

��

YOLO

SONOMANAPA

SAN JOAQUIN

SOLANO

SANTA CLARA

MARIN

ALAMEDA

SACRAMENTO

CONTRA COSTA

SAN MATEO

SAN FRANCISCO

��

��

��

��

��

��

²� ��

��

��

Pa cif ic Ocean

36

��

��

��

��

��

�� !"#$��%&&&'�

��

��

��

��

��

��

�� !��"��

��#$��%��&'�

��

� ��

²� ��

Pa ci f ic Ocean

37

��

��

��

Paci f ic Ocean

��

�� !��"� �#��"�"�$��%�#&'

²� ��

��

��

��

��

��

��

38

��

��

��

��

²� ��

Pac if ic Ocean

�� !��"� �#��"�"�$��%�#&'

��

��

��

��

��

��

39

��

��

��

��

²� ��

Paci f ic Ocean

��

0

��

��

��

��

��

��

��

��

��

40

��

��

��

��

²� ��

Paci f ic Ocean

��

��

��

��

��

��

��

�

��

��

��

41

��

��

��

��

��

��

�� !��

Urban Development 2000

"�#��$�!��

�� !"�#��$%��!�&��'��&��(��)��(�(� ��*�)$+

��

��

²� ��

Paci f ic Ocean

42

��

��

��

��

²� ��

��

��

��

��

0

Low

Medium

Moderate

High

Very High

Very Low

Paci f ic Ocean

43

��

��

��

��

Paci fic Ocean

��

²� ��

��

�

��

� ��

��

��

��

��

��

44

��

��

Paci fic Ocean

��

²

��

��

��

��

��

��

��

��

��

� ��

��

!��"��!��

45

9. APPENDIX

MODEL FOR PREDICTED HIGH VALUE FARMLAND GIS LAYER

Purpose:

This dataset was developed to model high value farmland where current data from the Farmland Mapping and Monitoring Program (FMMP, 2000) are unavailable.

“High value farmland” refers in this case to either Prime Farmland or Farmland of Statewide Importance as defined by the U.S Department of Agriculture (USDA).

Extent:

FMMP focuses on private important irrigated agricultural lands. The mapping program rarely extends into National Forests. Because the Important Farmland maps are based on USDA-Natural Resource Conservation Service (USDA-NRCS), “modern soil surveys”, the extent of the FMMP mapped area is limited to the contents of the survey: 48 out of California's 58 counties, some of them have been mapped only partially.

Although it would be ideal to cover the whole state, the extent of this model is restricted by the availability of Land Use data from the Department of Water Resources (DWR). Specifically it covers eastern Fresno, western Stanislaus, Modoc, Del Norte, Humboldt, Lassen, and western Shasta counties.

Data sources:

Farmland Mapping and Monitoring Program, FMMP (2000)

The goal of the FMMP is to map agricultural land use and to track changes in land use on prime farmland and farmland of state and local importance. Agricultural land is rated according to soil quality and irrigation status into eight classes. This model, however focuses only in the upper two categories, Prime Farmland and Farmland of Statewide Importance:

Prime Farmland (P)

Irrigated land with the best combination of physical and chemical features able to sustain long-term production of agricultural crops. This land has the soil quality, growing season, and moisture supply needed to produce sustained high yields. Land must have been used for production of irrigated crops at some time during the four years prior to the mapping date.

Farmland of Statewide Importance (S)

46

Irrigated land similar to Prime Farmland that has a good combination of physical and chemical characteristics for the production of agricultural crops. This land has minor shortcomings, such as greater slopes or less ability to store soil moisture than Prime Farmland. Land must have been used for production of irrigated crops at some time during the four years prior to the mapping date.

The maps are updated every two years with current land use information. The minimum mapping unit is 10 acres and the projection Albers Equal Area (NAD 87).

High value farmland is modeled by extracting the soil mapping units from STATSGO that meet the requirements stated on the technical definitions of Prime Farmland and Farmland of Statewide importance. Such technical specifications are detailed on "A Guide for the Farmland Mapping and Monitoring Program" included in the appendix of this document.

The State Soil Geographic Database (STATSGO) and derived STATSGO aggregated data sets:

STATSGO is a soil survey database developed by the USDA-NRCS. The mapping scale for STATSGO map is 1:250000 (with the exception of Alaska, which is 1:1,000,000) and the minimum area mapped is about 1,544 acres.

Created in 1994, the soil surveys include agricultural information and over 25 physical and chemical soil properties contained in 15 attribute tables that could be liked to spatial data through the soil mapping unit. Among these tables the Map Unit, Component, and Layer tables provide the information used to model high value farmland.

The attribute database describes the proportion of the component soils and their properties for each mapping unit. There are 1 to 21 components in each STATSGO mapping unit and each component can have multiple layers. Thus, in order to represent spatially the information, the data should be handled from the lowest level to the highest (the order is: layer, comp and mapping unit) resulting in a rather cumbersome process. Fortunately there are some STATSGO aggregated data sets versions available that made the use of the STATSGO information much easier.

Shirazi et al. (2001) used STATSGO as a research tool for linking the soil properties to water quality monitoring data. For this purpose they selected 27 soil properties that influenced water properties and aggregated them to the map unit level.

Their methodology is based in two arithmetic rules derived from Ohm’s or Darcy’s law. The model described in this paper uses the slope and K factor from their study to measure the erodibility of each STATSGO mapping unit.

CONUS-SOIL Dataset (Miller, D.A. and R.A. White, 1998)

47

CONUS-SOIL is defined by its authors as “a multi-layer soil characteristics data set for the conterminous United States which has been specifically designed for regional and continental-scale climate, hydrology, and ecosystem modeling”.

The coverages were created using Geographic Information Systems (GIS) and Perl scripts to process exported attribute tables from the original STATSGO data set. The processed information was then transferred back to the GIS. Publicly available, this data set includes soil texture and rock fragment classes, depth-to-bedrock, bulk density, porosity, rock fragment volume, particle-size fractions, water capacity, and hydrologic soil group.

With the exception of the erodibility property, all the other soil characteristics used in the model were derived from this data set.

California Department of Forestry and Fire Protection (CDF-FRAP) Multi-source Land Cover Data (2002 v1)

This dataset was created to support the required analysis for the 2002 Forest and Range Assessment. It is a 100m grid projected in Albers detailing habitat classes from the California Habitat Wildlife Relationships System (CWHR ).

DWR Division of Planning and Local Assistance Land Use Data:

This data set was developed by the DWR through it’s Division of Planning and Local Assistance to aid monitoring land use for the main purpose of determining the amount of and changes in the use of water. The land survey data are based on aerial photography and field visits to identify the land use, the land use boundaries were then digitized onto USGS 1:24,00 quads. Quality control procedures were completed before finalizing each file. The data is in a transverse mercator projection and available in shape files for each quad and aggregated by the whole survey area (by county). The attributes offer a detailed description of agricultural land uses (irrigation system type, type of water source used, etc.) and coarser information on urban and native vegetation.

According to the definitions adopted in the FMMP, Prime Farmland or Farmland of Statewide Importance should be irrigated. This condition was determined in the high value farmland model using this dataset.

Data Development:

The development of this dataset encompasses basically two parts: the elaboration of a preliminary prediction of high value farmland derived from soil properties and the reduction of its extent to irrigated agricultural areas. This involves:

1. Identify the specific Prime Farmland and Farmland of Statewide Importance criteria requirements from the guide to FMMP (see Appendix I).

48

2. Retrieve the corresponding soil properties within the aggregated STATSGO databases and extract the STATSGO soil mapping units that meet the soil properties requirements.

3. Create a “soil based predicted high value farmland layer” by sequentially intersecting (the order is specified by the guide to FMMP) the extracted mapping units.

4. Select the agricultural areas from the FRAP land use data and restrict the extent of the layer obtained in 3 to the selected areas.

5. Clip out the predicted high value farmland that falls within non-mapped areas of the FMMP map.

6. Select the predicted high value farmland polygons that fall out of the FMMP survey area.

7. Reproject the DWR Land Use data into Albers Equal area.

8. Select the irrigated polygons from the DWR dataset.

9. Intersect the resulting polygons from step 5 and 6 with the selected DWR features.

10.Merge the polygons resulting from both intersections.

Important points about using this dataset:

1. Although the model aims to replicate the technical specifications for Prime Farmland and Farmland of Statewide Importance, not all the soil properties specified on the FMMP guide were available from the aggregated STATSGO datasets. Therefore, the degree of fitness of the predicted high value farmland to the important farmland classes’ definitions is limited. The table 1 summarizes the soil properties contained in the guide to FMMP and the criterion applied to model the high value farmland.

2. The soils properties specifications that could not be included in the model due to the lack of data, where checked after the model was completed:

Temperature: None of the soils included in the predicted high value farmland have a temperature regime excluded from the technical definition of Prime Farmland and Farmland of Statewide Importance. In others words, none of them have a pergelic or cryic temperature regime.

Soil Sodium Content: A soil should have a conductivity less than 16 mmhos/cm to qualify for Farmland of Statewide Importance. The maximum conductivity of the predicted soils is 16 mmhos/cm.

49

Flooding: The flooding in the Farmland of Statewide Importance during the growing season occurs infrequently (less often than once every two years). However, about 5% of the area of the modeled soils falls within soil mapping units with frequent flooding (>50% chance of flooding).(see attached map).

3. The STATSGO soil mapping units were sequentially filtered through the USDA-NRCS soil properties criteria following the order established on the FMMP guide. Some of the ranges specified in this document have been adjusted to better reflect the original FMMP data according to the data availability.

4. In cases where the DWR data were unavailable, the modeling process could not be completed. Thus, in some counties (Mendocino, Ventura, Sierra, Mono and Alpine) it was impossible to check the irrigation status of soils polygons that a priori meet all the other soil properties requirements. The area of these polygons, however, is not significant.

What follows is a list of the rules and the criteria applied to create this dataset:

Soil Property

Important Farmland

Category

Data

Source Criterion Applied

Water

Statewide

Importance CONUS-SOIL

water capacity ≥ 9 cm. within a

depth of 150 cm.

Soil

Temperature

Range

Non

Available *

Acid-Alkali

Balance Prime Farmland CONUS-SOIL

pH between 4.5 and 8.4 in all

horizons within a depth of

100cm.

Water Table

Non

Available **

50

Soil Sodium

Content

Non

Available ***

Flooding

Non

Available ****

Erodibility

Statewide

Importance

Shirazi M.A

et al Product K*prctg slope < 3

Permeability

Statewide

Importance

Shirazi M.A

et al No restrictions

Rock

Fragment

Content

Prime/Statewide

Importance CONUS-SOIL

Mean rock vol prctg <10 of the

upper 20 cm.

Rooting

Depth

Statewide

Importance CONUS-SOIL No restrictions

* None of the soils included on the prediction model have pergelic or cyric temperature regimes.

** The water table depth requirements are specific of the type of crop grown. The assessment of this soil property would require a case-by-case study that requires a larger scale data set than the one used in this model. However it must been said that in STATSGO, the water table values vary from 0 to 6 feet statewide and thus it is probable that some of the high value predicted farmland has a water table value higher than desirable.

*** The maximum salinity of the predicted high value farmland soils is 16 mmhos/cm

51

**** About 5% of the high value predicted farmland area falls completely within a mapping unit with at least one component with an annual flooding frequency >50%. However, this is an annual probability and thus an overestimation of the probability of flooding during the growth season. The flooding frequency varies from component to component within the same mapping unit.

References:

California Department of Conservation. 2002. Farmland Mapping and Monitoring Program (FMMP) data, 2000. Available at: http://www.consrv.ca.gov/DLRP/fmmp/index.htm

California Department of Conservation. 2002. Farmland Mapping and Monitoring Program. A guide to the Farmland Mapping and Monitoring Program, Publication FM 94-02. November 1994. Available at: http://www.consrv.ca.gov/dlrp/fmmp/pubs/FMMP_GUIDE.pdf

California Department of Forestry and Fire Protection. 2002. Statewide multi-source land cover data (WHR) compiled for Forest and Range 2002 Assessment (2002 v1). Available at: http://frap.cdf.ca.gov/data/frapgisdata/select.asp

California Department of Forestry and Fire Protection. 2002. Methods for development of habitat data: Forest and Range 2002 Assessment. Technical Working Paper 1-02-02. Fire and Resource Assessment Program.

California Department of Water Resources. Division of Planning and Local Assistance. Land use data. 2001. Available at: http://www.waterplan.water.ca.gov/landwateruse/landuse/ludataindex.htm

Miller, D.A. and R.A. White. 1998. A Conterminous United States Multi-Layer Soil Characteristics Data Set for Regional Climate and Hydrology Modeling. Earth Interactions, Vol 2-Paper 2. Available at: http://www.essc.psu.edu/soil_info/index.cgi?soil_data&conus&data_cov

Mostafa A. Shirazi, Larry Boersma, Patricia K. Haggerty, and Colleen Burch Johnson 2001. Spatial Extrapolation of Soil Characteristics Using Whole-Soil Particle Size Distributions J. Environ. Qual. 30: 101-111

Mostafa A. Shirazi, Larry Boersma, Colleen Burch Johnson, and Patricia K. Haggerty. 2001. Predicting Physical and Chemical Water Properties from Relationships with Watershed Soil Characteristics. J. Environ. Qual. 30: 112-120.

52

GIS Processing Flowchart for Model of Predicted High Value Farmland