Coupled Human & Natural Systems

Biocomplexity II

Landscape Signatures Along Urban-to-Rural Gradients

1.  Research Objectives:

Urban development is dramatically changing landscape pattern and ecological function worldwide. Yet we are just beginning to understand the interactions between patterns and processes in human dominated landscapes. We do not know how interactions of human and biophysical processes affect the landscape patterns of urbanizing regions and how these patterns affect ecological processes and ecosystem function. While important progress has been made in modeling coupled human and ecological systems in urbanizing regions, most current studies have not formally tested hypotheses on emergent landscape patterns and their impact on socioeconomic and biophysical systems.

To develop a mechanistic urban ecology, a formal approach to detecting landscape patterns across multiple regions is necessary. Since landscape pattern is spatially correlated and scale-dependent, understanding landscape structure and functioning requires a multiscale approach. Building on the work of scholars of landscape ecology and the advances in spatial statistics and modeling, we propose to
systematically apply and compare a series of techniques specifically aimed at discriminating patterns of urban landscapes that we hypothesize are relevant for both human and ecological functions.

We hypothesize that distinctive hybrid landscape patterns can be represented by alternate ecological states of the landscape on an urban-to-rural gradient. We describe the urban landscape as hierarchical, dynamic mosaics of patches generated and maintained by processes of patch formation, development, and disappearance. We will expand on previous analyses of urban landscape patterns and urban growth to identify “landscape signatures” in selected metropolitan regions by extracting spatial characteristics of urban landscape change. We hypothesize that emerging patterns can be described along multiple dimensions including form, intensity, heterogeneity, and connectivity. In addition, these patterns may be dependent on initial conditions and random events may lead to very different outcomes. Thus history is an important dimension of landscape complexity. Building on existing research in Seattle, Phoenix, and Baltimore, a classification of urban landscape forms will be developed to describe and quantify the emergent landscape patterns.

2.  Approach/Method:

We aim to quantify emergent patterns of urban ecosystems through analysis of landscapes along an urban-to-rural gradient (Figure 1). Given the hierarchical relationships, interactions among system (both human and biophysical) components, and that ecosystem processes typically operate at characteristic spatial scales, we will assess the use of various analytical techniques for identifying scale dependence in spatial and temporal patterns. This analytic approach offers the possibility of determining whether specific signatures of landscape patterns can be used to draw inferences about coupled human-biophysical processes.

We have partitioned the central Puget Sound area using a gradient index based on three primary variables: slope, distance from the urban core, and percent impervious surface (Figure 1). The partitioning was done using gradient analysis by combining three variables in a common index using principal component analysis (PCA). PCA gives a linear combination of the vector that captures the maximum possible variability of the three variables (Figures 1).

Preliminary results of our pattern analysis show complex relationships between the urban gradient and several metrics including percent cover, mean patch size and fractal dimension, aggregation index, adjacency measures, contagion, connectivity, dominance, and Shannon diversity. We plan to develop a longitudinal and cross-sectional approach in our examination of potential landscape signatures of urban development in both Seattle and Phoenix.

We are also exploring effects of urbanization on landscape pattern and ecosystem processes both in the Seattle and Phoenix metropolitan regions. Reciprocal effects of landscape patterns on ecosystem processes are exemplified in the dynamics of forest function.  We hypothesize that variability of forest function is directly related to the composition and configuration of urbanization.  We are working in the Puget Sound region to test this hypothesis by combining measured patterns of urban development and metrics of forest function at varying spatial extents.  

In Phoenix our research objectives are to compare urban land cover classes with undisturbed Sonoran desert ecosystems in terms of patterns of primary production at broad spatial scales using remotely sensed Normalized Difference Vegetation Index (NDVI) data.  We address the following questions: 1) What are the characteristic features of MODIS NDVI temporal signatures that distinguish different land cover classes in the rapidly urbanizing arid region? 2) Does urbanization lead to a greater potential of carbon sequestration compared to undisturbed ecosystems in the area? 3) What urban/agricultural land cover classes have the highest production in the area and what are their inter-annual variations? 4) What area temporal lags and duration of precipitation events that trigger vegetation growth in different land covers?

Current findings from this study include:

• Land transformations in Central Arizona create more land covers with rates of primary production considerably higher than those of natural vegetation, especially during dry years.
• Vegetation growth and primary production in urban and agricultural land covers is far less variable than in the outside desert due to mainly human ameliorations.
• NDVI relationships with precipitation in Sonoran desert are characterized by larger time lags and rainfall aggregation periods. Such relationships are affected by soil texture characteristics.
• Our analyses of MODIS NDVI and climate data provide important insights into the interactions between vegetation patterns, climate variability, and urbanization in the area.

This project aims to develop a reliable set of metrics to detect and measure signatures of urban landscape patterns revealing emergent properties of urban ecosystem dynamics. Our goal is to develop signatures that can be used to test formal hypotheses about relationships between patterns and processes in urbanizing regions. We will explore a set of analytical techniques to detect and measure patterns and determine the level of spatial correlation and scale dependence. The most promising analytical techniques include geostatistical, wavelet, and lacunarity analyses. We aim to explicitly address issues of spatial non-stationarity and scale-dependency inherent in urban landscapes. First we
will conduct a pilot analysis using a subset of our urban landscape and spatial data that will inform the development of a general methodology for the larger multiple-cities analysis. We will then apply the methodology to eight cities by integrating longitudinal and cross-sectional approaches to take advantage of the diversity of available data sets. We build on existing data sets developed for several urbanizing regions by Urban LTER and Biocomplexity research teams. Our analysis will result in a series of papers including a methodological paper evaluating the strengths and weaknesses of alternative approaches to detecting urban landscape patterns, a synthesis paper presenting a set of metrics useful for quantifying urban landscapes, and an applied paper testing the metrics on multiple metropolitan regions.

 

3. Deliverables/Products for First Year:

• Analysis of structural patterns in land use/land cover along an urban-to-rural gradient
• Gradient signatures of forest function in Puget Sound
• Analysis of interactions between gradient characteristics and ecosystem function (e.g., forest function, benthic index of biotic integrity)
• Signatures of vegetation structure, as measured via remotely sensed imagery, associated with urban development in Phoenix

 

4. Participants:

UW-Seattle:   Marina Alberti, Lucy Hutyra, Steven Walters, Karis Puruncajas, Yan Jiang
ASU-Phoenix:  Jianguo Wu, Chuck Redman, Guangjin Tian, Weijun Shen, Alexander Buyantuyev


5. Resources:

We will build on data sets developed for several urbanizing regions by Urban LTER and Biocomplexity Research Teams. Rich regional spatial data sets are readily available for the above cities, and will constitute our analytical data layers. All the cities have good data sets for land cover, parcel and land use, topography, and hydrography. An initial survey, conducted to determine boundaries and scale mismatches for Seattle and Phoenix, will form the basis for developing a template to assess data consistency and comparability and to determine the appropriate spatial and temporal resolution for analysis.