A Planning Tool for Conservationists:
Spatial Modeling of Past and Future Land Use in Vermont Towns
Over the past thirty years, the demand for Vermonts rural land for residential and commercial development has increased markedly . Since the 1960s, Vermont has been experiencing population growth of at least 10% per decade (Bureau of the Census).
Several factors account for such rapid growth in this traditionally rural state. First, the completion of interstate highways brought improved access for tourists. Many tourists were enamored by Vermont's beauty and charm and decided to relocate permanently. Advances in agricultural technology enabled fewer people to do the same amount of work, thereby reducing the demand for farm workers. In 1977, the agricultural industries accounted for 8% of Vermonts total employment. Less than 20 years later, in 1996, only 1.3% of the total employment in Vermont was in agricultural industries (Vermont Annual Planning Reports 1977 and 1996). With the decline of the agricultural job market, Vermonts traditionally resource-based economy diversified and jobs in new fields such as technology and manufacturing arose (Brown and Beale 1981). Such market shifts brought services to outlying communities, thereby making rural communities more attractive, easier places to live. The recent surge in telecommunications also aided in the spread of the population by enabling those who normally work in the city to reside in the country.
In addition, new technologies enabled intensive farming requiring less land to produce a given amount of agricultural goods. Yet the demand for agricultural goods did not increase commensurate with increases in farming efficiency. With less land needed for agricultural production, more land sat idle and vulnerable to market forces. As economic activities were becoming decentralized and rural populations were increasing, Vermonts agricultural land became prime residential property.
A national trend toward an increase in the number of households relative to the population size fueled demand for new residential property (Brown and Beale 1981). As young adults became more apt to live alone and the incidence of divorce rose, so did the number of housing units in Vermont. From 1970 to 1980, there was a 35.2% increase in the number of housing units, but the population grew only by 15%. Likewise, from 1980 to 1990, there was a 21.5% increase in the number of housing units, while the population increased by only 10% (U.S. Bureau of the Census).
As agricultural fields were convereted to residential properties, Vermonts landscape took on a new appearance. In 1974, the acres of Vermonts land in agricultural use were 1,915,520. Less than 30 years later, in 1992, that number dropped by 34% to 1, 276, 525 acres (USDA Census of Agriculture, 1974 and 1992). The new pattern of landscape settlement is a movement away from the urban centers toward the urban fringe and outlying rural areas of the state.
sources: U.S. Census of Housing and Population, 1950-1990
In many instances, rolling pastures that were traditionally used to graze the Holsteins were replaced by residential subdivisions. The residential development pattern that has occurred over the past 30 years is commonly referred to as sprawl. This refers to the scattered development of purely residential subdivisions on isolated tracts surrounded by undeveloped land. The Vermont Forum on Sprawl has defined this leapfrog development pattern as the pattern of low-density development that spreads from urban and market centers along the highways into the countryside (Jon Ewing, Director, Vermont Forum on Sprawl).
Sprawl is a contributing factor in several forms of environmental degradation. Freemark and Merriam (1986) have shown the negative consequences of sprawl to wildlife habitat through habitat fragmentation. Such development has also been correlated with increased nutrient concentrations in bodies of water. The extensive road systems needed to connect the isolated residential developments to market centers has led to an increased surface area from which water runoff can carry nutrients into the water (Peterjohn and Correll 1984).
Numerous studies demonstrate the negative social externalities associated with discontinuous development (see Harvey and Clark 1965, Clawson 1962, Sargent 1976, Council on Environmental Quality 1974, and Maine State Planning Office 1997). As a brief example, consider the impacts of separating the residence from the market center. Taken from an environmental perspective, separation of residence and market center requires residents to travel farther to engage in most activities, thereby using more fuel and producing more air pollution (Ottensmann 1977). Taken from the perspective of an individual, this separation equates to higher car maintenance costs and increased insurance premiums (Maine State Planning Office 1997). Taken from the perspective of the local government, the extension of the road system and its maintenance increases a municipalitys financial burden (Ottensmann 1977). Taken from a social perspective, the scattered development of residential subdivisions and the pavement traversing the landscape brings the formless identity of the urban center to the countryside, thereby upsetting the aesthetic quality of the rural landscape.
As the population moves to the countryside, the landscape has taken on a new look and the social characteristics of the populous has changed. For instance, most Vermont's rural residents no longer work the landscape, rather the new rural citizen resides far away from his/her office and commutes to work each day.
sources: U.S. Census of Housing and Population, 1970-1990
In a project funded by the Orton Family Foundation, researchers at the University of Vermont's Spatial Analysis Laboratory are developing a predictive model of landscape change. By visualizing the pattern and pace at which the landscape is being settled, communities can plan the layout of their town while ensuring for the future needs and services of their townspeople.
The objective of the project is to provide planners and policy-makers with a predictive model of development patterns in Vermont towns. The final product will be a statistical model with estimated coefficients for those socioeconomic and physical parameters found to most significantly influence land use conversion. Understanding the demographic forces that act as causal factors to land use conversion will allow planners and policy-makers to craft policy that guides social trends so as to induce a desired landscape pattern.
Research has shown that social and economic factors are drivers of land use change and forest fragmentation. Significant drivers include population density, road density, and distance from urban centers. Socio-economic drivers interact with biophysical factors such as soil, elevation, slope and vegetation interact to determine the extent of land use change. In Vermont, the interaction between socio-economic and biophysical factors and their relation to land use change is uncertain. To date, no empirical studies in Vermont have examined spatial patterns of land use change as a function of socio-economic factors and land use.
The development of the predictive model will occur in three stages. The first stage characterizes recent trends of land cover type distributions in Vermont towns. A change detection analysis using satellite imagery will provide the baseline spatial data needed to quantitatively discern socioeconomic and physiographic forces driving the patterns of land use.
Landscape change dectection aims to discern differences in land use patterns over time by comparing images of the same location on different dates. Landsat Multispectral Scanner (MSS) images from the early 1970s, mid 1980s, and early 1990s were the primary data source. Vermont spans portions of four MSS scenes (1329, 1330, 1429, 1430) therefore at least three images per scene were compared. Additional imagery was required to fill in clouded areas for some scenes and periods. Ancillary data incorporated into this analysis include a land cover map generated from 1992-93 LANDSAT TM imagery and slope data derived from 1:24,000 and 1:100,000 scale USGS contour maps. The type and location of change targeted in this analysis are:
To obtain a complete land cover distribution for each period, area classified as no change is characterized using the 1992-93 LANDSAT TM derived land cover product. All image processing was conducted in ERDAS IMAGINE.
Classification of the change areas relies on decision rule algorithms (DRA), or conditional logic statements, in IMAGINE's spatial modeler module. DRA statement integrate differencing, image band ratios, and ancillary data to define change classes. Clouds and haze effects were minimized by eclectic use of multiple images. An accuracy validation of the classification will be conducted using reference points acquired from aerial photography and satellite imagery from the three time periods.
One element in modeling the change in the distribution of land uses is to understand the social and economic trends occurring in Vermont towns. Having an understanding for the demographic underpinnings of Vermont's communities will allow for the visualization of the future Vermont landscape. Therefore, concurrently, ArcView-compatible databases of socioeconomic indicators are being developed for each town in Vermont. Data-formatting is being conducted in ArcInfo 7.1.2. Each town will have access to several coverages that exhibit town-level social and economic trends. Socioeconomic variables visualized in these coverages include (but are not limited to): population change, population density, number of housing units, percent of housing units held for occasional use, distribution of labor force, educational attainment, property tax rates, traffic volume, per capita income, employment status and travel distance the capital.
The demographic information used in the Spatial Model of Land Use are provided here for use by the Vermont community.
Click on a county to obtain county-level and town-level socioeconomic information
In the final stage of this project, the socioeconomic and demographic data listed above will be related to the observed changes in the distribution of land uses. This analysis will estimate the correlation of social and physical variables the conversion of land from undeveloped to developed uses and vice versa. Factors found to significantly influence land use conversion will serve as parameters in a predictive model of development for Vermont's towns.
Last update: 25 April 2000