Assessing variations in urban heat island effects within Roanoke, Virginia

Tags: data collection, landscape characteristics, tree canopy, mobile mesonet, temperature, National Weather Service, microclimates, weather stations, Roanoke, Virginia, urban area, readings, urban heat island, collection, Virginia Polytechnic Institute, Department of Environmental Quality, Stanger Street Major Williams Hall Blacksburg, International Journal, urban areas, temperature variations, Virginia Western Community College, Davis Vantage, Roanoke City Public Schools, Virginia Heights Elementary School, mobile mesonets, Virginia Department of Environmental Quality, Professor David Carroll, Virginia, urban climate, local news stations, urban agriculture, James B. Campbell, Thomas Fitzpatrick, tree canopy cover, impervious surface, impervious surfaces, percent, mobile unit, fixed station, temperature readings, differences in temperature, air temperature, Virginia Tech, Virginia Tech Graduate Research and Development Program, Virginia Department of Forestry, College of Natural Resources, Urban Greening, office administrators, RCPS Science Coordinator, U.S. Geospatial Intelligence Foundation Doctoral Scholarship, Department of Forest and Environmental Conservation, Sidman Poole Endowment Fund, City of Roanoke, Cabell Brand Center First Freedom Scholarship, Dr. John McGee
Content: ASSESSING VARIATIONS IN URBAN HEAT ISLAND EFFECTS WITHIN ROANOKE, VIRGINIA Tammy E. Parece, Ph.D. Candidate James B. Campbell, Professor David Carroll, Instructor Virginia Polytechnic Institute and State University 220 Stanger Street Major Williams Hall Blacksburg, Virginia 24061 [email protected] [email protected] [email protected] ABSTRACT Within urban areas, variations within the built environment create unique microclimates because of diversity in thermal properties of surface materials and alterations of the hydrologic cycle. Resolving intra-urban microclimate variability presents an opportunity to evaluate spatial dimensions of urban heat island effects, including daily air temperature fluctuations and local variations in start and end of growing seasons. Observations from National Weather Service (NWS) stations are often used to characterize regional conditions, yet such data are widely spaced and can only indicate conditions specific to that site. To effectively represent the fabric of temperature variations within an urban area, a finer network of data collection points is required. We report on a weather data collection campaign within Roanoke, Virginia using mobile weather units and weather stations newly installed at local public schools. We describe these data collection programs, outline methods developed for our collection pattern, and our preliminary analyses. We discuss our results and how they relate to the variation in Roanoke's built environment. This research forms the first phase of dissertation research evaluating urban social and environmental patterns to facilitate optimal placement of urban agriculture. It provides the basis for understanding the spatial context for urban agriculture, and for ameliorating social and environmental difficulties inherent to modern urban systems. It fills a gap in current strategies, which largely have lacked spatial perspectives, and uses the power of geospatial technologies to identify relationships between the environmental and social dimensions of urban systems, and the spatial nature of their synergies. KEY WORDS: urban heat island effect, mobile mesonet units, fixed weather stations, infrared thermometers, urban microclimates INTRODUCTION Urban areas are inherently warmer than their rural surroundings. A plethora of studies document the urban heat island (UHI) effect and its origins in landcover/landuse changes (Hedquist and Brazel 2006, Weng 2012). More specifically, areas with higher amounts of impervious surfaces and lower vegetative cover tend to be warmer. Within an urban area, variations in the built environment create unique microclimates generated by alterations in thermal properties of surface materials, by absence of vegetative cover, and by alteration of the hydrologic cycle (Arnfield 2003, Geiger, Aron et al. 2003). Understanding microclimates requires evaluation of the spatial variability of air temperatures in the context of precipitation and humidity, across an urban area (Oke 2006). Urban remote sensing analyses documenting microclimatic variations of the UHI are numerous but use differing resolutions (i.e. Landsat vs. SPOT) and differing techniques (Voogt and Oke 2003). Some studies have attempted to identify specific temperature values using various equations, and then to validate the temperature in the field at a limited number of sites (Weng 2012). Such observations are typically obtained either from widely-spaced ASPRS 2014 annual conference Louisville, Kentucky March 23-28, 2014
National Weather Service (NWS) stations, local news stations, or from mobile mesonets driven across specified transects (Arnfield 2003). Other researchers have used mobile units to document differences between urban and rural temperatures. Although mobile units do provide data across many locations, they are limited to a specific time frame, usually either one day or a series of days across specified transects (e.g. Hedquist and Brazel 2006, Stabler, et al 2005). Fixed weather stations can provide a continuous stream of weather data and many studies on UHI use a network of fixed stations across an urban area (e.g. Graffin et al 2008, Bourbia 2010, Yahia 2013). However these data only reflect lower atmospheric conditions specific to site characteristics of each station (Arnfield 2003). To effectively evaluate the differing precipitation, humidity, and air temperature across an urban area, finer network of data collection points are clearly required (Geiger, Aron et al. 2003; Oke 2006; Gaffin, Rosenzweig et al. 2008). STUDY AREA The City of Roanoke, Virginia is located in a valley at between the Blue Ridge Mountains and the Alleghany Highlands (Figure 1). Roanoke, the largest metropolitan region in southwestern Virginia, is characterized by a variety of urban land uses. The city's history is largely based upon its role as a regional transportation hub for rail and road traffic with services and industries supporting the rail system, as well as finance, distribution, trade, manufacturing, and health care businesses. Figure 1. Roanoke, Virginia Reference Map Over recent decades, Roanoke has formed a focus for substantial urbanization, economic stress, and landuse change. In addition, although it is a small urban area (110 km2), Roanoke is intensely urbanized ­ both population density (880 persons per square kilometer) and land extent. These changes have resulted in many environmental issues, e.g. CO2 emissions were estimated at 2.3 million tons in 2009 (Roanoke 2011). Roanoke also has substantial drainage problems and experiences frequent flooding due to its proximity to the Roanoke River, the river's tributaries, and urban stormwater runoff from impervious surfaces. Roanoke's impervious surfaces range from ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
13.3% to 89.5% of total area by census tract (Parece and Campbell, 2013). Many segments of the Roanoke River system within the city are on the Virginia Department of Environmental Quality's impaired waters list, due to contaminants such as Escherichia coli, high water temperatures, and heavy metals (Virginia 2010). For our analysis, Roanoke offers the advantage of a compact urban region with a range of land uses and urban settings that permit evaluation of urban microclimates.
METHODS
Our data collection period extended from April through November of 2013, the extent of the 2013 growing season in the City of Roanoke. At the beginning of our research, weather stations existed at four K-12 schools, the airport (National Weather Service location), Virginia Western Community College, a few private residences, and the two local TV stations. Many of these locations are WeatherBug® sites (Virginia Western Community College, Virginia Heights Elementary School, Roanoke Valley Governor's School, Roanoke Catholic School, WDBJ CBS TV). The private residences' weather stations were identified as either Davis Vantage Vue® Pros or with a MADIS Identification Number. The additional weather stations described here were purchased with grant monies, faculty research funds, scholarship monies, and by the Roanoke City Public Schools. We purchased Davis Vantage Vue® Wireless Weather Stations (Model No. 6250), and Davis WeatherLink Dataloggers and Software to add additional locations to the network of fixed weather stations in Roanoke. This model has an Integrated Sensor Suite (ISS) to collect outside weather data every 2.5 seconds, and it transmits data wirelessly to an indoor console via FCC-certified radio transmitter. The stations are equipped with a rain collector, temperature/humidity sensor (mounted within a passive solar radiation shield), anemometer, and wind vane. The transmission limit from outdoor unit to indoor console is 1000 feet (Davis 2012). These eleven new weather stations were installed at public schools by Roanoke City Public Schools (RCPS) facilities personnel and located in regions of the city without weather stations. RCPS personnel mounted the outdoor equipment on school roofs in locations not easily accessible by unauthorized people. Virginia Tech meteorology students installed indoor equipment and computer software. All weather stations were set up to transmit on-line to WeatherUnderground (www.wunderground.com). As part of this internet set up process, the latitude, longitude, and elevation were obtained for each outdoor unit. During the summer of 2013, while we were installing the new weather stations, three of the prior existing weather stations went off-line. In early November, near the end of our of our study period, two additional private residents acquired weather stations ­ one a Davis Vantage Vue® Wireless Weather Station and the other a Davis Vantage Vue® Pro2. Virginia Tech's Geography Department's fleet of mobile mesonet units were used to collect additional temperature data across the study site, providing local snapshots of temperatures for specific times and dates. Table 1 provides the mesonet collection dates, times, the number of mesonet units that were used on that particular date, and number of readings that were collected during a specific time period.
Table 1. Date, times, number of mesonet units, and infrared thermometers readings per collection campaign
Date
Times in Roanoke
Number of Mesonet Units
Number of Mesonet Readings
Number of IR Readings
April 21, 2013
9:32 a.m. ­ 10:50 a.m. 3:36 p.m. ­ 4:52 p.m.
3 3
5,843
18
5,099
16
April 22, 2013
9:37 a.m. ­ 10:56 a.m. 2:54 p.m. ­ 3:50 p.m.
3 3
5,801
19
5,335
18
April 23, 2013
9:21 a.m. ­ 10:15 a.m. 2:46 p.m. ­ 4:00 p.m.
3 3
4,758
18
5,774
19
July 23, 2013
9:10 a.m. ­ 12:06 p.m. 2:30 p.m. ­ 3:48 p.m.
2 2
6,766
33
4,497
32
6:57 a.m. ­ 9:10 .am.
1
3830
27
August 5, 2013
11:17 a.m. ­ 11:48 a.m. 1:28 p.m. ­ 1:59 p.m.
1 1
803 932
0 0
3:39 p.m. ­ 5:36 p.m.
1
3808
30
August 14, 2013 9:00 a.m. ­ 11:30 a.m.
3
9,868
55
ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
1:28 p.m. ­ 4:23 p.m.
3
September 6, 10:20 a.m. ­ 1:10 p.m.
3
2013
2:03 pm. ­ 4:14 p.m.
3
October 26, 2013
1:04 a.m. ­ 2:56 a.m. 4:00 a.m. ­ 4:56 a.m.
1 1
November 4, 2013
3:08 a.m. - 6:12 a.m.
1
November 8, 2013
10:27 p.m. ­ 12:30 a.m.
1
November 9, 2013
1:01 a.m. ­ 2:11 a.m.
1
November 25, 2013
6:39 a.m. - 8:29 a.m. 11:07 a.m.- 12:24 p.m. 1:22 p.m. ­ 2:39 p.m.
1 1 1
November 28, 2013
11:20 a.m. ­ 12:07 p.m.
1
6,610
33
3,688
0
3,033
0
6,983
0
1.704
0
5,533
0
3,716
0
2,118
0
3,291
0
2,104
0
2,300
0
1,430
0
These units were mounted on Chevrolet Cobalts and driven into Roanoke during each data collection run (Figure 2). While we did not drive the exact same vehicles on each date, we did use the same make and model. The meteorological equipment is Campbell Scientific mobile metrological units: RM Young wind monitor, CSL Temperature/RH probe, Sentra 278 Barometer, Garmin GPS receiver, CR800-ST-SW-NC Measurement & Control Datalogger. The temperature and humidity sensor is shielded and aspirated, all sensors are programmable for different sampling rates, and the unit registers latitude and longitude in WGS1984.
Figure 2. Two of the mobile mesonet units, mounted on Chevrolet Cobalts, driving off I-81 into Roanoke, Virginia on April 21, 2013 We set the sensors to record the climate data every two seconds, registering temperature in degrees Fahrenheit. After each run, we downloaded the data file and then created a point shapefile for use in GIS. Each shapefile contained the latitude and longitude, barometric pressure, temperature, relative humidity, windspeed and direction, and time of collection for each point. During daytime data collection with the mobile mesonet units, each vehicle periodically stopped and obtained surface temperature using infrared thermometers (Figure 3). The infrared thermometers used in this portion of our data collection were: Fluke 574 Precision Infrared Thermometers, temperature range of -25 to +900 degrees Fahrenheit, digitally adjustable emissivity, distance to spot size (close) 50:1, and spectral range 8 ­ 14µm. For each surface temperature reading, we noted the date, time and location. Table 1 summarizes infrared thermometer readings obtained during each mesonet run. Readings were not obtained during every single mesonet collection date, but all readings were obtained from asphalt surfaces. ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
Figure 3. Using an infrared thermometer to capture the surface temperature of asphalt In addition to temperature data collected with the mesonet units, fixed weather stations and infrared thermometers, we downloaded the digital elevation model (DEM), 10 x 10 meter resolution, from the U.S.G.S. Seamless Server (http://nationalmap.gov/viewer.html) and the National Land Cover Database 2006 Percent Developed Imperviousness (NLCD IS), 30 x 30 meter resolution, file from the Multi-Resolution Land Characteristics Consortium (http://www.mrlc.gov/nlcd06_data.php). We utilized a Tree Canopy Cover (TCC)1, a 1 x 1 meter resolution binary raster dataset, for the City of Roanoke that was provided by Virginia Tech's Geospatial Extension Specialist. Using ArcGIS®, we derived aspect and percent slope raster files from the DEM, and we aggregated the TCC raster file to 30 x 30 meter resolution and calculated percent TCC cover for each grid cell. We created a point-shapefile for each fixed weather station using latitudes and longitudes from WeatherUnderground and schools shapefile downloaded from the City of Roanoke GIS Portal (ftp://ftp.roanokeva.gov/GIS/). We obtained the exact location of the National Weather Station unit from the National Weather Service in Blacksburg, Virginia. WDBJ and Virginia Western Community College personnel provided us with temperature readings for the dates of our mesonet runs. Temperature data from other fixed weather station data was downloaded from WeatherUnderground as.csv files, separately for each station, and corresponding to each mesonet run. We created separate fixed weather station shapefiles for each date and time, also corresponding to each mesonet run and joined the .csv file. We created a point-shapefile for the infrared temperature collection, a separate shapefile for each mesonet run. We also aggregated the shapefiles into one file for locations where we obtained both morning and afternoon temperature readings. We analyzed the temperature data in the following ways: 1. In GIS, we compared and contrasted temperature data collected by the mesonet units to each fixed weather station temperature data collected at the equivalent time, to verify the accuracy of the readings; 2. In GIS, we extracted values from the Percent TCC, DEM, NLCD IS, aspect, and percent slope raster and shapefiles for each data point collected by the mesonet units, for each fixed weather station, and for each IR temperature location to determine the specific characteristics of the adjacent landscape; 3. We performed a backwards stepwise regression analysis on fixed stations to identify which landscape characteristics influenced temperatures; 4. In GIS, we extracted the mesonet data that corresponded to the infrared temperature data, and compared and contrasted it with the infrared temperature data - calculating the differences between the air temperature and surface temperature, the changes in surface temperature due to time change, and performed a stepwise regression to determine what characteristics of the landscape influenced surface temperature; and 5. We performed a stepwise regression analysis separately for each mobile unit for each run, and also jointly for all probes on each run to determine which characteristic of that specific site was related to temperature. 1 City of Roanoke Tree Canopy Cover was completed using 1 x 1 meter resolution National Agriculture Imagery Program (NAIP) 2008 imagery, with a resultant overall accuracy of 93% (Pugh 2010). ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
We employ these methods to take a first look at our data, and are now, with little guidance offered in the literature, experimenting to define strategies to analyze these data. Our goals are to define simple relationships between our observed temperature and the proximate landscape variables, as a basis for estimating temperatures within a zone of about thirty meters from the points of observation.
RESULTS
We compared each mobile mesonet data to each fixed station data for every date and time of our mesonet campaign. We determined that fixed stations were registering temperature data when the mobile unit was near the station 96 times. In most cases, the mobile unit was on the street driving by the fixed station at that specific time. Seventeen (17) of these 96 times, the mesonet was between 200 and 500 meters of the fixed stations. Many other fixed stations were registering temperatures during the mobile collection dates and times but they were at least 1 kilometer or more from the mobile unit. Differences in readings between fixed stations and the mobile units (N = 96) varied from 0 and 10.1 degrees Fahrenheit, with a mean difference in temperature of +0.20 degrees. The fixed station that was furthest distance (500 meters) from the mobile unit registered the greatest difference at 10.1 degrees. Table 2 provides the distribution of the differences in temperature (N = 96).
Table 2. Distribution of temperature difference between mobile and fixed stations
Absolute value of temperature difference, °F
Number of readings
0 ­ 0.5
20
0.6 ­ 1.0
14
1.1 ­ 1.5
16
1.6 ­ 2.0
13
2.1 ­ 3.0
18
3.1 and above
15
We then plotted the two readings to compare mobile mesonet temperature on x-axis and fixed station temperature on y-axis, and the correlation between the two readings (Figure 4). The readings are highly correlated at the 99% level.
Figure 4. Correlation between mobile mesonet temperature reading and nearby fixed weather stations on same date and time (N = 96, ellipse @ 99.0% level). From our larger collection of records, we chose five different times to complete a regression analysis to determine which landscape characteristics neighboring the fixed weather stations influenced temperatures. We choose times for which the greatest number of fixed stations were recording data ­ 5:00 p.m. on August 14, 5:00 p.m. on September 6, and 6 a.m., Noon, and 5:00 p.m. on October 10. Table 3 shows the results of this analysis. ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
Elevation and slope had no influence on our temperatures. Aspect had the greatest influence in most cases, however, when removing either percent tree canopy cover or percent impervious, our R2 decreased 20+ points every analysis.
Table 3. Fixed station regression results for identification of landscape characteristics influencing air
Date and Time
N
R2
temperature Landscape characteristics influencing temperature
August 14, 5:00 p.m. 18 0.97
Aspect, percent tree canopy cover
September 6, 5:00 p.m. 17 0.96
Aspect, percent impervious, percent tree canopy cover
October 10, 6:00 a.m. 16 0.83
Aspect, percent tree canopy cover
October 10, Noon
10 0.54
Aspect, percent tree canopy cover, percent impervious
October 10, 5:00 p.m. 10 0.67
Aspect, percent tree canopy cover, percent impervious
As noted from Table 1, we obtained 315 infrared temperature readings. However in some instances, readings were either not taken in the same location or readings were missed, different drivers in the afternoon did not follow the routes taken by the morning drivers, or the data was misplaced. So we only used 218 readings for this portion of the analysis (109 locations with morning and afternoon readings). We calculated differences between morning and afternoon infrared temperature readings. We then performed a backwards stepwise regression with the difference between morning and afternoon surface temperatures as our dependent variable, and change in air temperature, percent slope, aspect, percent tree canopy cover, percent impervious, and elevation as our independent variables. Table 4 shows the results of this process. For the six dates, our R2 ranged from 0.30 to 0.79. The landscape characteristics that influenced the surface temperature varied. Aspect, for all dates, was the most significant of variables influencing the difference between the morning and afternoon surface temperatures. For the April dates, the next characteristic of most significance was percent impervious. Slope and elevation had no influence on any of the dates. Percent tree canopy cover had no influence on the April dates, not surprising as in 2013, Roanoke and southwestern Virginia experienced a late start of growing season, thus trees had not yet fully leafed out. As we progressed to a warmer time of year, the summer months -- July and August -- change in air temperature between morning and afternoon did have a significant influence on change in surface temperature, but it was less influential than aspect, percent tree canopy cover and percent impervious.
Table 4. R2 values and significant landscape characteristics influencing change in asphalt surface
temperature from morning to afternoon for 109 locations in Roanoke
Date
N R2
Factors in order of significance
April 21, 2013
10 0.74 aspect, percent impervious
April 22, 2013
14 0.62 aspect, percent impervious
April 23, 2013
6 0.40 aspect, percent impervious
July 23, 2013
23
0.30
aspect, percent tree canopy cover, percent impervious, change in air temperature from morning to afternoon
August 5, 2013
25
0.79
aspect, percent tree canopy cover, change in air temperature from morning to afternoon
August 14, 2013
31
0.30
aspect, percent tree canopy cover, percent impervious, change in air temperature from morning to afternoon
The results of the backwards stepwise regression for the mobile mesonet units was extremely variable with both the R2 and the landscape characteristics impacting temperature. The R2 was the lowest for two runs ­ midday on November 25 and afternoon of July 23 (0.08 for both), and the highest at 0.54 on August 5 early morning. The August 5 early morning was the only R2 that exceeded 0.40. Table 5 shows the range of R2 values and the number of mesonet runs that fall within that range. Additionally, no one specific landscape characteristic affected all R2; aspect and elevation were contributing factors in 22 of 25 runs, percent impervious in 18 of 25 runs, percent tree canopy cover only 7 out of 25, and percent slope was never a contributing factor (Table 6).
Table 5. Range of R2 and the number of mesonet runs within that range
R2 Range
Number of Mesonet Runs
0.08 ­ 0.10
3
0.11 ­ 0.20
7
ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
0.21 ­ 0.30
7
0.30 ­ 0.54
8
Table 6. Role of landscape characteristics in R2
Landscape characteristic
Frequency of occurrence out of 25 mesonet
data collection campaigns
Aspect
22
Percent Impervious
18
Percent Tree Canopy Cover
7
Elevation
22
Slope
0
Since aspect is a landscape characteristic that we know has an effect on temperature (south facing slopes are warmer and drier than north facing slopes), we performed the backwards stepwise regression again, this time with aspect as a blocking factor. Since slope was not a contributor in any of our prior regressions, we eliminated that characteristic from any further analysis. We chose four dates from the previous regression, the two with the highest and the two with the lowest R2, to perform this step. Again, our results were variable. For the lowest R2, blocking only made a slight difference and only changing the results for some of the north facing slopes. For November 25 for north facing slopes, it increased by 16 points. For July 23, north facing slopes increased by 26 points, northeast facing slopes by 18 points, and northwest increased by 14 points. For our highest R2 ­ August 5 early morning - blocking decreased most results, with the exception of the north facing slopes which increased to 0.75. For the August 5 midday run (our second highest R2 from the prior regression), we had 5 out of 8 increase. Again, our contributing landscape characteristics were variable; elevation and percent impervious were the most frequent contributing characteristic (Table 7).
Table 7. Regression analysis results for selected dates with aspect as the block, mobile units
Date Time Original R2 Characteristics impacting R2
August 5 6:57 ­ 9:10 a.m. R2 = 0.54 Percent Impervious Elevation
August 5 1:28 ­ 1:59 p.m. R2 = 0.38 Elevation Aspect
November 25 11:07 a.m. ­ 12:24 p.m. R2 = 0.08 Aspect Percent Impervious
July 23 9:10 a.m. ­ 12:06 p.m. R2 = 0.08 Percent Impervious Elevation
Blocking Factor East Flat North Northeast Northwest South
0.59 (N=415) Elevation Percent impervious 0.24 (N=17) Percent impervious Percent tree canopy cover 0.75 (N=536) Elevation 0.49 (N=646) Elevation Percent impervious 0.52 (N=134) Elevation Percent tree canopy cover 0.31 (N=571)
R2 (N) Landscape characteristics, in order of impact on R2
0.56 (N=65) Elevation Percent impervious
0.05 (N=199) Percent impervious, Percent tree canopy cover
0.09 (N=672) Elevation Percent impervious,
No points (N=0)
No points (N=0)
0.06 (N=19) Percent impervious
0.40 (N=173) Elevation, Percent impervious, Percent tree canopy cover 0.84 (N=136) Elevation, Percent tree canopy cover, Percent impervious 0.17 (N=111) Elevation, Percent impervious, Percent tree canopy cover 0.04 (N=181)
0.21 (N=294) Percent impervious, Percent tree canopy cover 0.07 (N=249) Percent Impervious 0.19 (N=215) Elevation, Percent impervious 0.00 (N=596)
0.34 (N=372) Elevation, Percent impervious 0.26 (N=979) Elevation, Percent tree canopy cover, Percent impervious 0.22 (N=1086) Percent impervious, Percent tree canopy cover, Elevation 0.05 (N=1174)
ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
Southeast Southwest West
Elevation, Percent tree canopy cover 0.53 (N=496) Elevation 0.38 (N=769) Elevation 0.47 (N=246) Percent tree canopy cover, Elevation
Percent impervious, Percent tree canopy cover 0.79 (N=84) Elevation, Percent tree canopy cover 0.12 (N=49) Percent impervious, Percent tree canopy cover 0.54 (N=133) Elevation, Percent tree canopy cover, Percent impervious
None
Percent impervious
0.08 (N=341) Percent tree canopy cover, Percent impervious, Elevation 0.02 (N=82) Percent tree canopy cover, Percent impervious, Elevation 0.09 (N=128) Elevation Percent impervious
0.06 (N=890) Percent impervious, Elevation 0.04 (N=1063) Elevation, Percent tree canopy cover 0.05 (N=511) Elevation
CONCLUSIONS Despite our low R2 when evaluating the influence different landscape characteristics had on temperature, clearly percent impervious, and to a lesser extent, percent tree canopy cover, did have an impact. We anticipated percent tree canopy cover would have a larger impact. Upon reflection, when driving mobile mesonet units, we are mostly in areas of impervious surfaces and less tree canopy cover, so these results are reasonable. In addition, southwestern Virginia experienced a milder summer than in most past years, as such we feel that our results would be different had the air temperatures been higher. We are also in the process of completing an impervious surface layer, hand delineated from high resolution aerial photos. This file will provide us with a greater accuracy and precision for the extent and locations of impervious surfaces. As we continue to analyze the data we have thus far collected, this new data layer will assist us in identifying the temperature variations within Roanoke, and can further guide us as to the most appropriate locations to compare temperature differences due to landscape characteristics. We will slightly change our methods for any future data collection campaigns. When traveling near the fixed stations, we will drive around the station multiple times, gradually increasing our distance outward, so as to determine at what distance the air temperature actually changes relative to the data recorded by the fixed stations. We need to assure that the same routes are followed if our intentions are to compare morning to afternoon changes, and that surface temperature readings are taken in the same locations, even if our drivers vary. In many instances when we stopped to obtain the surface temperature reading, the mesonet was only stopped for a few seconds, we should let the mesonet stay immobile for a length of about to see if the air temperature rises more quickly over the asphalt surfaces. We suspect that this is the case as we were traveling to the RCPS schools, we observed distinct differences in the mesonet readings from when we drove into the school parking lot to when we left the school parking lot. For most mobile mesonet dates, we timed our data collection with the Landsat overpass schedule because we intended to overlay our mobile mesonet data on Landsat imagery for analysis. However, no scenes with sufficiently low cloud cover, matched to, or close to, our collection dates during the time of our surveys. As a general statement, we note that our numerous meteorological and logistical constraints limited our ability to match our collection efforts to dates of Landsat overpasses for our area. We also invite suggestions and critique to guide our future investigations. ACKNOWLEDGEMENTS We acknowledge the following people who assisted us in our mobile data collection campaign: Virginia Tech Students and Faculty -- Thomas Tutchings, Ash Elmelick, Erika Cropp, Bonnie Long, Michael Marston, Paul Miller, Hans VanBenschoten, Sam Freeman, Mario Garza, and Dr. Mike Hyer; and private citizen - Chris White. We would also like to thank Brent Watts and Robin Reed of WBDJ CBS Channel 6, Roanoke for their advice on our fixed stations and working with the public schools; David Webb of Virginia Western Community College for providing the school's temperature data; Dr, Andrew Ellis of Virginia Tech Department of Geography and Steve Keighton of the National Weather Service in Blacksburg for their advice on evaluating the results of our collection ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014
campaign; Thomas Fitzpatrick, RCPS Science Coordinator and the many RCPS science teachers, librarians, office administrators and facilities managers who accommodated our repeated visits to different school buildings. The City of Roanoke Tree Canopy Cover file was provided by Dr. John McGee of Virginia Tech's College of Natural Resources' Department of Forest and Environmental Conservation. We would also like to thank the follow funding sources: Sidman Poole Endowment Fund, Virginia Tech Graduate Research and Development Program, U.S. Geospatial Intelligence Foundation Doctoral Scholarship, Sigma Xi Graduate Award for Doctoral Degree Students; and Cabell Brand Center First Freedom Scholarship. REFERENCES Arnfield, A.J. 2003. Two decades of urban climate research: A review of turbulence, exchanges of energy and water, and the urban heat island. International journal of Climatology, 23, 1-26. Bourbia, F., and Boucheriba, F. 2010. Impact of street design on urban microclimate for semi arid climate (Constantine). Renewable Energy, 35, 343-347. Gaffin, S.R., Rosenzweig, C., Khanbilvardi, R., Parshall, L., Mahani, S., Glickman, H., Goldberg, R., Blake, R., Slosberg, R.B., and Hillel, D. 2008. Variations in New York City's urban heat island strength over time and space. Theoretical and Applied Climatology, 94, 1-11. Geiger, R., Aron, R.H., and Todhunter, P. 2003. The Climate Near the Ground, 6th Edition. Rowman & Littlefield, Lanham, MD. Hedquist, B.C., and Brazel, A.J. 2006. Urban, residential, and rural climate comparisons from mobile transects and fixed stations: Phoenix, AZ. Journal of the Arizona-Nevada Academy of Science, 38, 77-87. Oke, T.R. 2006. Initial Guidance to obtain representative meteorological observations at urban sites. In, Instruments and Observing Methods. World Meteorological Organization, Vancouver, British Columbia, Canada. Parece, T.E., and Campbell, J.B. 2013. Comparing Urban Impervious Surface Identification Using Landsat and High Resolution aerial photography. Remote Sensing, 5, 4942-4960. Pugh, J. 2010. A report on the City of Roanoke's Existing and possible urban tree canopy. Virginia Department of Forestry, Charlottesville, VA. Available at: http://gep.frec.vt.edu/UTC/UTC_Report_RoanokeCity.pdf Roanoke. 2011. City of Roanoke 2009 carbon emissions and Energy Summary. Available online: http://www.roanokeva.gov/85256A8D0062AF37/CurrentBaseLink/433EBD6594B462838525788B00685 A4A/$File/carbon_emissions.pdf (accessed on 30 January 2012). Stabler, L.B., Martin, C.A., and Brazel, A.J. 2005. Microclimates in a desert city were related to landuse and vegetation index. Urban Forestry and Urban Greening, 3, 137-147. Virginia. 2010. Department of Environmental Quality. GIS Data Sets. Accessed on June 13, 2012 at http://www.deq.virginia.gov/ConnectWithDEQ/VEGIS/VEGISDatasets.aspx. Voogt, J.A., and Oke, T.R. 2003. Thermal remote sensing of urban climates. Remote Sensing of Environment, 86, 370-384. Weng, Q. 2012. Remote sensing of impervious surfaces in the urban areas: Requirements, methods, and trends. Remote Sensing of Environment, 117, 34-49. Yahia, M., and Johansson, E. 2013. Evaluating the behaviour of different thermal indices by investigating various outdoor urban environments in the hot dry city of Damascus, Syria. International Journal of Biometeorology, 57, 615-630. ASPRS 2014 Annual Conference Louisville, Kentucky March 23-28, 2014

File: assessing-variations-in-urban-heat-island-effects-within-roanoke.pdf
Author: Tammy
Published: Mon Feb 24 16:17:00 2014
Pages: 10
File size: 0.43 Mb


Research Methods, 1 pages, 0.37 Mb

Human space, 45 pages, 0.23 Mb
Copyright © 2018 doc.uments.com