The Relationship Between Urban Leaf Area and Summertime Household Energy Usage in Provo

by

Print Friendly, PDF & Email

Jacob Adams and Dr. Ryan Jensen, Department of Geography

The purpose of our study is to investigate the relationship between urban leaf area, estimated through remote sensing, and household energy usage in the city of Provo, Utah. Anecdotal evidence indicates that trees provide cooling in the summer by shading homes, which reduces the need for air conditioning and the amount of electricity used. This study will attempt to determine this decrease of energy usage based on varying amounts of tree cover as measured by the Leaf Area Index (LAI) obtained from a satellite image. We expect that LAI and household energy consumption will be negatively correlated— as LAI increases, energy consumption decreases. The results of this study can be useful to city officials and energy production companies, who often sponsor urban tree-planting programs. This study is a confirmation of the methods and results of a previous study conducted by Jensen, et al. (2003).

This study makes use of the science of remote sensing, which is the process of observing the earth with satellites that measure the electromagnetic radiation (EMR) reflected from the earth. This data is measured in much the same way as a digital camera measures the reflected light (or EMR) from an object. However, unlike a digital camera, these sensors measures EMR in several distinct regions of the electromagnetic spectrum referred to as bands. The ASTER sensor used for this study captures EMR in the green, red, and near infrared portions of the spectrum at a resolution of 15 meters per pixel (Jensen 2007). For this study we used an image acquired on July 27th, 2010, which provided relatively cloud-free data at the proper point of the year.

Because satellite imagery covers a large swath of the landscape multiple times a year, indices showing the approximate amount of vegetation over an entire area can be calculated using the unique reflectance patterns of vegetation in different bands. The two most common indices are the simple ratio and Normalized Difference Vegetation Index (NDVI) (Jensen 2007). However, these indices are not direct estimations of leaf area, and so the Leaf Area Index must be calculated. LAI is a simple ratio of leaf area in square meters per ground area in square meters (Jensen 2007). An LAI value of zero indicates no leaves present, while our experience in the field shows that a lone mature tree generally has an LAI value of around one and half to two and that a dense grove of trees has a value of four or greater.

One caveat of LAI is that it cannot be directly calculated from satellite imagery; instead, it must be derived using a statistical correlation between “on the ground” (or in-situ) measurements and the previously mentioned vegetation indices. During the first two weeks of August these in-situ measurements were taken at 100 points across the city of Provo, Utah, between the hours of 11:00 A.M. and 2:00 P.M. to coincide with the time of day and year that the satellite image was acquired. These data were collected with a handheld ceptometer that calculates LAI by comparing above- and below-canopy sunlight readings (Decagon Devices 2006). At each site an above canopy measurement was taken by measuring direct, un-shaded sunlight followed by below canopy measurements taken in the four cardinal directions at the center and each corner of a ten-meter box. This produced an average LAI value for that 10-meter area that more closely matches the 15-meter resolution of the ASTER sensor than a single point LAI value.

Following the collection of these in-situ measurements, we attempted to create a model for LAI based on data from the atmospherically- and geographically-corrected ASTER image. Following the work of Shin et al. (2010), we ran a regression analysis between NDVI calculated from the image and our in-situ LAI measurements to determine a prediction for LAI. While our results were statistically significant (p = .01), NDVI only explained a small amount of the variation of the LAI values (r = 0.27) (see Figure 1). Despite its statistical significance, this weak correlation has led us to look to different methods of estimating LAI from the ASTER image.

This is our current progress on this project, and this progress was presented in a poster at the annual meeting of the Great Plains/Rocky Mountain Division of the Association of American Geographers on October 12th, 2012. Our plan for the future is to create an artificial neural network as suggested in Jensen and Hardin (2005) to create a model for LAI across Provo based on various data from the ASTER image. Once this model is created we will acquire household energy usage data from Provo City Power for random residential addresses. We will then run another regression analysis to determine the correlation between LAI and household energy usage. Based on the previous study (Jensen et al. 2003), we expect to see decreased household energy usage at addresses with higher LAI values. We expect to finish this analysis by February and submit a paper to a journal such as the Journal of Arboriculture no later than April 2013.

References

  • Decagon Devices, Inc. 2006. How the LP80 Measures Leaf Area Index. http://www.decagon.com/assets/Uploads/HowtheLP80MeasuresLAI.pdf. Web. 19 Nov 2012.
  • Jensen, John R. Remote Sensing of the Environment: An Earth Resource Perspective. 2nd ed. New Jersey: Pearson-Prentice Hall. 2007. Print.
  • Jensen, Ryan R.; Bouton, James R.; Harper, Bruce T. 2003. The relationship between urban leaf area and household energy usage in Terre Haute, Indiana, U.S. Journal of Arboriculture 29(4): 226-229.
  • Jensen, Ryan R. and Hardin, Perry J. 2005. Estimating urban leaf area using field measurements and satellite remote sensing data. Journal of Arboriculture 31(1): 21-27.
  • Shin, Yonghee; Seguchi, Masahiro; Koriyama, Masumi; Isnansetyo, Alim. 2010. Estimation of LAI in the forested watershed using ASTER data based on Price’s model in summer and winter. European Journal of Forest Research 129(6): 1237-1245.

We acknowledge the BYU Office of Research and Creative Activities for funding this project and Andrew Hardin for assistance with correcting the satellite data.