The Canopy Continuum Project:

Ecosystem Services from urban canopy structure on maternal and child health


In the US, approximately 72% of the population lives within a metropolitan region. This dramatic rise in urbanization and changes in the patterns of urban development have given rise to several significant environmental and public health concerns, including air pollution and heat stress, and consequent increases in urban pollution-induced morbidity and mortality.

The Canopy Continuum

We refer to the described dual challenges as the canopy continuum, which is to say:

We are addressing the historic inequitable distribution of canopy in our cities to provide equitable benefits to all in the future.

Ecosystem services provided by tree cover are an essential part of every neighborhood, understanding how the urban canopy helps to create a safe, clean, and healthy urban living environment is a grand challenge facing society, and we aim to understand the role of the canopy continuum in creating thriving cities for all residents.

Research Goals

Through our ongoing research we have identified several critical questions, which remain unanswered, and are at the core of the present proposal. They include:

How do human vulnerabilities, environmental stressors, and land use conditions interact to create inhospitable living environments?

To what extent can micro-scale changes to the urban canopy improve neighborhood ecosystem services and living conditions?

In what ways can expansion of tree canopy reduce long-term vulnerabilities to the most sensitive urban populations, including pregnant women and children?

Addressing these questions using recent development in high-resolution tree canopy assessments, and air quality and urban heat sensors can offer immediate and timely support to organizations for addressing pressing urban living challenges in the U.S. and across the urbanizing world.


This project focuses on five cities in the Western U.S. The datasets for our project include measurements of:

Measuring Air Quality and Urban Heat

We assessed both temperature and PM2.5 at sub-neighborhood scales by developing a new mobile environmental sensing platform that integrates our previously designed temperature sensor system with newly available low-cost optical particle counters (OPC) from Alphasense, Inc. Unlike our previous temperature surveys, which we conducted solely with vehicle-mounted transects, for this study we conducted transects with a combination of vehicle and bicycle traverses. Bicycle traverses enabled us to capture variation in areas not accessible to vehicles (e.g. parks, school grounds, bicycle paths). Based on our previous work and data from the Oregon Department of Environmental Quality, the summer season measurements were selected to capture the highest temperatures and potential for heat stress, while the winter season captures the highest PM2.5 levels. Both vehicles and bicycles were simultaneously deployed to capture three replicates of early morning, midday and evening measurements on three summer days and three winter days. As with our previous measurement campaigns, students engaged in undergraduate and graduate research in Environmental Science, Biology and Urban Planning were recruited to conduct the transects.

See related publications:

Characterizing Canopy Within and Across Cities

Beginning at a landscape level, we used a high-resolution Light and Detection and Ranging (LiDaR) and a geographic information system (i.e. ArcGIS 10.x) to characterize urban canopy in terms of biomass, functional type (e.g. deciduous v coniferous), and distribution. LiDaR offers a three-dimensional point based measure of structural attributes of urban vegetation that is was available for all our study sites. LiDaR was combined with spatial analytical techniques and programming language, R (statistical analysis software), through the application of high performance computing and other computer languages (e.g. Python and IDL) that are suitable for analyzing big datasets. Using R, we characterized the 2014 Portland LiDaR dataset to compute biomass by multiplying the density of canopy points with the height of the structure. Finally, we applied focal and zonal statistics (ArcGIS 10.x) to quantify the distribution – fragmentation, nearest-neighbor, and contagion of vegetation at a one-meter resolution. By testing for multi-collinearity across each of these measures we could reduce bias by constructing a statically orthogonal subset of vegetation classifiers. In order to establish ground-truth data for LiDAR calibration and validation, we used the method outlined in Schreyer et al., 2014 and  performed independent tree surveys along a land-use transect in each of our study cities. In each case, validation trees were selected along the transect and field measures of crown height, crown width, diameter breast height (DBH), and species/functional type were determined and used to calibrate and validate the LiDAR extracted values. Each of the representative vegetation structures were also photographed, which assisted in creating a visual resource for populating our website, presentation, and for comparison across the multiple cities in our study.

Evaluating the relationship among environmental stressors, canopy amount, and measures of maternal and child health

Three birth outcomes were constructed for analysis. First, we used preterm birth, which we defined as birth at less than 37 weeks’ gestation to an infant weighing less than 3,888g [1}. A second measure, term birth weight, is infant birth weight within live births delivered between 37-42 weeks gestation. Small for gestational age was defined as being born with a weight below the 10th percentile for gestational age. Finally, gestational age was calculated by subtracting maternal reported last menstrual period (LMP) date from the delivery date. When LMP data are missing or biologically implausible, a clinical estimate for gestational age was substituted. These birth outcomes were chosen because they can be constructed from data that are reliably reported on birth records [1}, and are relevant for maternal and child health disparities. These data were provided by working with the Oregon Health Authority’s division of Maternal and Child health through our collaboration with the four other State health departments – Washington (through Pierce County Public Health), California, Idaho, and New Mexico – and our local partners to procure, geocode, and analyze maternal and child health records in relation to environmental stressors and canopy structure.

The HESI is an attempt to integrate the multiple factors that impact human health. The index, as a result, offers a mechanism for understanding how individuals are impacted by environmental conditions, and opportunities to reduce exposures. The highly resolved datasets of air quality, urban heat, and LiDAR derived tree canopy data offers a precise means for describing each household with a specific score. The HESI takes advantage of the abundance of the fact that environmental exposure data of this resolution has never before existed. We note that the HESI cannot account for all of the environmental stressors that a household faces, although it offers a framework for including additional data as they become available.

Click Here to See Interactive Maps of Each City

Back to Top

Back to Top

%d bloggers like this: