Southern Piedmont Region Executive Summary

May 16, 2012 at 5:59 pm | 0 comments

Introduction

For the last few decades, much of the Southern Piedmont has experienced
dramatic population growth. Attracted by the region’s mild climate and
strong economy, people from other parts of the country have been pouring in,
changing the region in significant ways. The most visible consequence of
this growth has been the increase in development. Narrow country roads winding through pastoral scenes of small farms and rolling hills have been replaced by multi-lane expressways connecting countless subdivisions and strip centers. And every indication suggests this trend will continue.

This study aims to improve understanding of the region’s growth. It examines a 24-county region of the Southern Piedmont (consisting of Alexander, Anson, Burke, Cabarrus, Caldwell, Catawba, Cleveland, Davidson, Davie, Gaston, Iredell, Lincoln, McDowell, Mecklenburg, Montgomery, Randolph, Richmond, Rowan, Rutherford, Stanly, and Union counties in North Carolina and Chester, Lancaster, and York counties in South Carolina). The study makes use of satellite imagery to measure development patterns with more precision than traditional measures of urban growth. The study tracks increasing development from 1976 (the year for which complete satellite data were first available) through 2006. The study then offers forecasts of future growth in the region, showing what the region’s land-patterns are expected to look like in 2030.

Methods and Imagery

Previous studies have examined urban growth using data tied to census tracts. That is, the studies have calculated population density using total population and total acreage figures for the census tracts as a whole. While these studies have offered important insights about urban growth, the nature of the source data limits the kinds of analysis that can be done. Let us consider two census tracts with identical population densities. The actual land use patterns within those census tracts may vary dramatically. The first census tract may hold its entire population within one-fourth of its area, leaving the rest of the land as undeveloped open space. The second census tract, in contrast, may have the population evenly distributed across its entire land area, leaving no undeveloped open space. These two census tracts obviously use land very differently. Nevertheless, for traditional studies using census tract data, these differences would be invisible, causing visual displays of the results of these studies to be somewhat misleading.

This study is unique in its use of satellite images for its data source. The satellite imagery comes from Landsat MSS and TM sensors. The Landsat program has been running since the early 1970s and is the only continuous data source of its type. Five image scenes were required for the study extent. The imagery provides information for areas of land much smaller than census tracts. In fact, the resolution of the satellite imagery is such that parcels of 30 meters by 30 meters (900 square meters) can be identified as developed or undeveloped. In other words, the satellite imagery allows for much more precise analysis of land-use patterns than is possible using census tract data. If we return to our hypothetical census tracts – this time using satellite imagery – the differences in development patterns would become obvious, and visual representations of the results would accurately portray these differences.

Historic Mapping

The study uses satellite imagery for the 24-county Southern Piedmont region for four time periods: 1976, 1985, 1996 and 2006. All imagery was acquired in midsummer and was largely cloud-free. Analysis was conducted at a 900 square meter spatial resolution. Subpixel mixture analysis was used to measure the fractional components representing impervious surfaces, soil and vegetation present in the reflectance data of each pixel. Units were classified into “developed” and “undeveloped” by thresholding levels of impervious surface and soil fraction components based on validation data obtained from high-resolution aerial imagery. In all, more than 500 million 900 square-meter parcels were coded.

Forecasts

Using trends identified in the historical imagery the study forecasts future urban growth using logistic regression modeling. Units that had transitioned from undeveloped to developed categories during the period 1996-2006 supplied the dependent variable. Independent variables found statistically significant were per capita land consumption, proximity to existing development, road density, distance to highway interchange, distance to urban centers (Charlotte, the Triad, and the Triangle), and topographical slope. The results were used to map the transition probability of each undeveloped pixel. Population-based urbanization trend models were made for each county based on the relationship between population and development as mapped 1976-2006. These “urban demand” projections were then cast onto the transition probability surface with highest likelihoods transitioning first. This process was repeated for timesteps 2010, 2015, 2020, 2025 and 2030.

Results

The spread of urbanization across the Southern Piedmont is visualized in a series of animated and static maps. The maps indicate dramatic increases in historic and projected urbanization in the Charlotte metropolitan region with less change in the predominantly agricultural counties. In 1976, only 1.8% of the study extent was developed, leaving the vast majority of the land classified as undeveloped. By 2006, 17.2% of the land was developed, a loss of over 1 million acres.

The study also shows the changing rate of development. That is, the study shows that the rate at which “undeveloped” parcels become “developed” has accelerated through the years. From 1976 to 1985, the rate of development for the 24 counties was 30 acres per day. By 2006 the rate had reached more than 140 acres per day, and is anticipated to average more than 100 acres per day through 2030.

The data can also be viewed at the sub-regional or county level, allowing for deeper analysis of land-use and development patterns across the region. For instance, a comparison of Cabarrus and Chester counties show that Cabarrus has lost over 60% of its undeveloped land since 1976 while Chester has experienced only modest development (19% loss). Different counties within the region have experienced different rates and kinds of development at different times, and this study permits exploration of those differences.

Moreover, when used in combination with population figures, the study offers a valuable look at the development “footprint” of population growth in the region. In other words, the study helps to show how much undeveloped land has been developed for each new person migrating to the region. With regard to Mecklenburg County, this analysis shows that from 1976 to 2006 the amount of land used per person increased from 0.11 acres per person to 0.23, meaning that newcomers to the region were using more land per capita than existing residents.

-Interactive County Maps

-County Profile Data

-Interactive County Maps for Full Study Area

Tags: Southern Piedmont Region