**3. Results and discussion**

#### **3.1 Stationary measurements**

Stationary measurements were commenced past the semester-long student project to evaluate whether there was a seasonality to the UHI in BCS. As shown in **Figure 3**, no clear seasonal variation was observed. However, the five highest UHI intensities, all above 4°C, were all measured during spring-time (March and April). They occurred on clear or mostly clear-sky days after recent cold front passages, and possible reasons for high values under those conditions are discussed below. Lowest values, down to −1.3°C, indicative of a local urban cool island, occurred dominantly during morning measurements. Comparing sites and measurement times, the median morning UHI intensity was slightly, but statistically significantly lower than the median night/evening UHI intensity. Site differences were statistically insignificant.

**Figure 4** shows that the UHI intensity was slightly dependent on large-scale wind directions. Northerly winds, as occur behind cold front passages, were responsible for the spring-time maxima in UHI intensity at the Trails site, but no such significant difference could be found for the Quad site. This could be due to the fact that a mid-size mall, a 0.4 km2 large area of impervious surface area, lies just beyond a narrow green corridor north of the Trails site, but no such prominent heat source lies near the Quad site.

**33**

**Figure 4.**

**Figure 3.**

from 0 to 7 m s<sup>−</sup><sup>1</sup>

*95% confidence interval estimates.*

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect*

*Seasonal changes of UHI intensity for the stationary sites and measurement times in 2015.*

We used a multi-linear regression to evaluate whether and which meteorological factors may have contributed to the UHI intensity at each site. Eliminating wind direction, we found that lower temperature, and lower relative humidity days were significantly (p > 0.95) correlated with the UHI intensity, a result strongly driven by the high spring values though. In addition, days with higher pressures were significantly correlated with the observed UHI intensity at the Quad site (p > 0.95). However, no statistically significant relationship with (rural) wind speed, ranging

*Boxplot of all 2015 UHI intensity measurements as a function of cardinal wind direction during the observation period. Thick horizontal bars mark medians, box edges mark the interquartile range (IQR), and whiskers are* 

than wind direction (**Figure 4**) stood out in explaining the observations.

, was found. In summary, no single meteorological parameter other

*DOI: http://dx.doi.org/10.5772/intechopen.89514*

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect DOI: http://dx.doi.org/10.5772/intechopen.89514*

**Figure 3.** *Seasonal changes of UHI intensity for the stationary sites and measurement times in 2015.*

#### **Figure 4.**

*Spatial Variability in Environmental Science - Patterns, Processes, and Analyses*

*2.3.3 Quantitative land-cover/land-use classification accuracy assessment*

Using standard methods [70, 71], we performed a quantitative thematic/classification accuracy assessment on the finalized land-cover/land-use (LCLU) classified map, edited post-classification. We employed a stratified random sampling approach for generating the classification accuracy-assessment points/sample locations, and we used 100 such points per class [70, 71]. We used the DOQQ aerial photograph mosaic as the reference data; we evaluated the accuracy of the postclassification-edited LCLU map via manual/visual interpretation of the DOQQs at the stratified random point locations. This analysis enabled the generation of an error matrix from which we computed statistics, including the overall accuracy [72, 73]. The overall classification accuracy of the post-classification edited LCLU

Stationary measurements were commenced past the semester-long student project to evaluate whether there was a seasonality to the UHI in BCS. As shown in **Figure 3**, no clear seasonal variation was observed. However, the five highest UHI intensities, all above 4°C, were all measured during spring-time (March and April). They occurred on clear or mostly clear-sky days after recent cold front passages, and possible reasons for high values under those conditions are discussed below. Lowest values, down to −1.3°C, indicative of a local urban cool island, occurred dominantly during morning measurements. Comparing sites and measurement times, the median morning UHI intensity was slightly, but statistically significantly lower than the median night/eve-

**Figure 4** shows that the UHI intensity was slightly dependent on large-scale wind directions. Northerly winds, as occur behind cold front passages, were responsible for the spring-time maxima in UHI intensity at the Trails site, but no such significant difference could be found for the Quad site. This could be due to

just beyond a narrow green corridor north of the Trails site, but no such prominent

large area of impervious surface area, lies

ning UHI intensity. Site differences were statistically insignificant.

depending upon road type.

map was 76.5%.

**3. Results and discussion**

**3.1 Stationary measurements**

the fact that a mid-size mall, a 0.4 km2

heat source lies near the Quad site.

footprint data from the City of College Station, we only applied this operation to areas within College Station, and not the City of Bryan; however, a large portion of the study area lies within the City of College Station boundary. For the other impervious class, based on visual interpretation of the aerial photography, we digitized impervious areas—especially parking lots—where misclassifications were evident. After rasterization of the digitized polygons, we then overwrote misclassified pixels in the original LCLU raster using these newly digitized data. Based on visual inspection, these post-classification GIS operations generally markedly improved classification accuracy. One caveat though is that although the DOQQs and LiDAR point clouds were collected close in time, the data in the vector GIS layers used in this post-segmentation/classification editing phase were acquired some time prior to the DOQQ and LiDAR acquisitions, and this temporal disjunction may have resulted in some errors in the edited LCLU map, particularly given the relatively high rate of urbanization in the BCS area in recent years. Furthermore, future analysis may entail using variable buffer distances for the road centerlines,

**32**

*Boxplot of all 2015 UHI intensity measurements as a function of cardinal wind direction during the observation period. Thick horizontal bars mark medians, box edges mark the interquartile range (IQR), and whiskers are 95% confidence interval estimates.*

We used a multi-linear regression to evaluate whether and which meteorological factors may have contributed to the UHI intensity at each site. Eliminating wind direction, we found that lower temperature, and lower relative humidity days were significantly (p > 0.95) correlated with the UHI intensity, a result strongly driven by the high spring values though. In addition, days with higher pressures were significantly correlated with the observed UHI intensity at the Quad site (p > 0.95). However, no statistically significant relationship with (rural) wind speed, ranging from 0 to 7 m s<sup>−</sup><sup>1</sup> , was found. In summary, no single meteorological parameter other than wind direction (**Figure 4**) stood out in explaining the observations.

The observed UHI intensity range is comparable with similar measurements in other mid-size urban areas (e.g. [33, 39, 74]). However, parameters other than population (or city) size and meteorological conditions are typically more relevant, particularly impervious area fraction, anthropogenic heat release, and urban morphology [29]. Therefore, we can surmise that because both our sites have a higher than average canyon aspect ratio (building height, *H*, to canyon width, *W*, ratio), and higher than average population and thus energy use density, morning cool island findings could be related to shading and strong heat admittance into the urban fabric [75–77], while the highest UHI intensities may have been related to high local anthropogenic heat emissions, respectively. This will be discussed in more detail with respect to LCZ classifications elsewhere.

#### **3.2 Mobile measurements**

The selected driving routes covered central areas north and south of the city border between Bryan and College Station, where impervious surface areas maximize (LCZ 3 and 5). They also covered residential areas in south College Station, where LCZ 6 dominates.

**Figure 5** shows an overview of results from three representative late evening drives, depicting three of the four routes. These drives occurred on (from north to south) September 16, August 28, and October 14, 2015. Rural wind directions and speeds for those drives were SE at 3–4 m s<sup>−</sup><sup>1</sup> , ESE at 1–2 m s<sup>−</sup><sup>1</sup> , and ESE at 3–4 m s<sup>−</sup><sup>1</sup> , respectively.

In total, 15 traverses were completed at night and 16 during daytime (12 afternoon and 4 mornings). The nighttime traverses showed UHI intensities of 3.2 ± 1.9 K (mean + − 1 sd), while the daytime UHI intensities were lower at

#### **Figure 5.**

*Roadmap based view of the BCS urban metro area overlaid with temperature anomalies (in Kelvin) during three representative late evening traverses. All traverses occurred under weak southeasterly winds. The central traverse on August 26, 2015 had the lowest wind speeds and a UHI intensity of 6.1 K.*

**35**

**Figure 6.**

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect*

2.3 ± 1.2 K. Only the nighttime UHI intensities displayed clear relationships to the prevailing meteorological conditions, namely atmospheric pressure and wind speed (p > 0.99), but also air temperature itself (p > 0.95). In addition, a weak dependence on the day's solar radiation totals was observed (p > 0.77). The daytime traverses' UHI intensities were not significantly related to the meteorological data; only air temperature (p > 0.87) and solar radiation (p > 0.78) were weakly related to

*Sectional GE overlay of mobile air temperature measurements during the late evening traverse on August 26, 2015. The white arrow shows wind speed and direction during the traverse. Driving direction is indicated by the light blue arrows, and relative driving speed can be deduced from symbol spacing. Symbols designate temperatures ranging from 29.5° C (red) in the road sections downwind densely build-up urban areas 1–2 km to the west of highway 6, to 26° C (yellow) downwind the irrigated outdoor sports complex near the center of the map, to 24° C (green) downwind of the then (2015) undeveloped area southeast of the sports complex.*

The combined data maintained dependencies on wind speed (p > 0.95) and atmospheric pressure (p > 0.9), and, together with the above, these results indicate that the UHI intensity of this midsize urban area was significantly more pronounced at night- as compared with daytime conditions. The day-night difference, as well as the relationships with wind speed, atmospheric pressure, and related

The results from the mobile measurements will be used further to analyze the major urban drivers of UHI intensity in terms of LCZ type [8, 52, 57, 63, 78–84],

Here, we show two case studies: one using a typical Google Earth (GE) view, and one using remotely sensed urban land cover. **Figure 6** shows a GE view of a part of the eastern BCS urban area, where the local highway separated a densely build-up area from a large outdoor sports facility, featuring frequent lawn irrigation and

cloud cover, reproduced findings from previous UHI studies [1].

and associated impervious area fractions and aspect ratios [29, 85–87].

UHI intensity for the afternoon traverses.

open, mostly undeveloped land to its southeast.

*DOI: http://dx.doi.org/10.5772/intechopen.89514*

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect DOI: http://dx.doi.org/10.5772/intechopen.89514*

#### **Figure 6.**

*Spatial Variability in Environmental Science - Patterns, Processes, and Analyses*

more detail with respect to LCZ classifications elsewhere.

**3.2 Mobile measurements**

where LCZ 6 dominates.

respectively.

speeds for those drives were SE at 3–4 m s<sup>−</sup><sup>1</sup>

The observed UHI intensity range is comparable with similar measurements in other mid-size urban areas (e.g. [33, 39, 74]). However, parameters other than population (or city) size and meteorological conditions are typically more relevant, particularly impervious area fraction, anthropogenic heat release, and urban morphology [29]. Therefore, we can surmise that because both our sites have a higher than average canyon aspect ratio (building height, *H*, to canyon width, *W*, ratio), and higher than average population and thus energy use density, morning cool island findings could be related to shading and strong heat admittance into the urban fabric [75–77], while the highest UHI intensities may have been related to high local anthropogenic heat emissions, respectively. This will be discussed in

The selected driving routes covered central areas north and south of the city border between Bryan and College Station, where impervious surface areas maximize (LCZ 3 and 5). They also covered residential areas in south College Station,

**Figure 5** shows an overview of results from three representative late evening drives, depicting three of the four routes. These drives occurred on (from north to south) September 16, August 28, and October 14, 2015. Rural wind directions and

In total, 15 traverses were completed at night and 16 during daytime (12 afternoon and 4 mornings). The nighttime traverses showed UHI intensities of 3.2 ± 1.9 K (mean + − 1 sd), while the daytime UHI intensities were lower at

*Roadmap based view of the BCS urban metro area overlaid with temperature anomalies (in Kelvin) during three representative late evening traverses. All traverses occurred under weak southeasterly winds. The central* 

*traverse on August 26, 2015 had the lowest wind speeds and a UHI intensity of 6.1 K.*

, ESE at 1–2 m s<sup>−</sup><sup>1</sup>

, and ESE at 3–4 m s<sup>−</sup><sup>1</sup>

,

**34**

**Figure 5.**

*Sectional GE overlay of mobile air temperature measurements during the late evening traverse on August 26, 2015. The white arrow shows wind speed and direction during the traverse. Driving direction is indicated by the light blue arrows, and relative driving speed can be deduced from symbol spacing. Symbols designate temperatures ranging from 29.5° C (red) in the road sections downwind densely build-up urban areas 1–2 km to the west of highway 6, to 26° C (yellow) downwind the irrigated outdoor sports complex near the center of the map, to 24° C (green) downwind of the then (2015) undeveloped area southeast of the sports complex.*

2.3 ± 1.2 K. Only the nighttime UHI intensities displayed clear relationships to the prevailing meteorological conditions, namely atmospheric pressure and wind speed (p > 0.99), but also air temperature itself (p > 0.95). In addition, a weak dependence on the day's solar radiation totals was observed (p > 0.77). The daytime traverses' UHI intensities were not significantly related to the meteorological data; only air temperature (p > 0.87) and solar radiation (p > 0.78) were weakly related to UHI intensity for the afternoon traverses.

The combined data maintained dependencies on wind speed (p > 0.95) and atmospheric pressure (p > 0.9), and, together with the above, these results indicate that the UHI intensity of this midsize urban area was significantly more pronounced at night- as compared with daytime conditions. The day-night difference, as well as the relationships with wind speed, atmospheric pressure, and related cloud cover, reproduced findings from previous UHI studies [1].

The results from the mobile measurements will be used further to analyze the major urban drivers of UHI intensity in terms of LCZ type [8, 52, 57, 63, 78–84], and associated impervious area fractions and aspect ratios [29, 85–87].

Here, we show two case studies: one using a typical Google Earth (GE) view, and one using remotely sensed urban land cover. **Figure 6** shows a GE view of a part of the eastern BCS urban area, where the local highway separated a densely build-up area from a large outdoor sports facility, featuring frequent lawn irrigation and open, mostly undeveloped land to its southeast.

#### **Figure 7.**

*Temperature anomalies from several daytime (afternoon) traverses, overlaid on the LCLU map of the BCS area. The depicted routes were driven on September 2, 2015 (central SW–NE extension; winds E at 4 m s<sup>−</sup><sup>1</sup> ), October 14, 2015 (southern extension; winds SE at 4 m s<sup>−</sup><sup>1</sup> ), and August 26, 2015 (northern extension; winds S at 3.5 m s<sup>−</sup><sup>1</sup> ). X and y-axis units are in meters and reflect* easting *and* northing *of UTM zone R14 (i.e., distance, respectively, from the central meridian of the zone and from the equator).*

Temperatures dropped 3°C crossing the highway toward the east into a less developed area, and another 2°C once downwind of a largely undeveloped area at the time (LCZ B/C/D). Driving direction and symbol spacing highlighted in this close-up illustrate the delayed response of the temperature sensor at higher driving speeds on the southern leg as compared with the northern leg, which included two short stops at traffic lights, one before crossing the highway, and another north of the sports complex near the center of the map.

Case study 2 is shown in **Figure 7**, which depicts temperature anomalies from three independent daytime traverses overlaid on the high resolution LCLU map. The daytime traverses showed a much more sophisticated, smaller scale pattern of the UHI as compared with the nighttime traverses. Wind speed during all these drives was south to southeasterly, and upon closer inspection the data revealed impacts of impervious surfaces areas near and upwind of the driving location [38, 88] as an important factor in determining UHI intensity in the BCS metro area. Research is ongoing to improve the LCLU map, and quantify UHI intensity as a function of the footprint's impervious area fraction, as well as other LCZ properties such as canyon aspect ratio and surface albedo.

## **4. Conclusions**

We have translated an undergraduate course project of stationary measurements of the urban heat island in a mid-size urban area in east Texas into a more detailed undergraduate study of the cities' UHI intensity as a function of meteorological conditions and urban morphology. Our study was able to reproduce the

**37**

**Author details**

and Burak Güneralp3,4

College Station, USA

, Gunnar W. Schade2

A&M University, College Station, USA

provided the original work is properly cited.

\*Address all correspondence to: gws@geos.tamu.edu

1 Environmental Programs, Texas A&M University, College Station, USA

3 Department of Geography, Texas A&M University, College Station, USA

4 Center for Geospatial Sciences, Applications and Technology (GEOSAT), Texas

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,

2 Department of Atmospheric Sciences, Texas A&M University,

ing the two undergraduate students involved in this project.

\*, Anthony M. Filippi3,4, Garrison Goessler3,4

Kristen Koch1

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect*

expected UHI intensity relationships with wind speed and atmospheric pressure, emphasizing synoptic high pressure conditions with intense solar radiation as a major driver of the UHI. In addition, we may have also found effects of local anthropogenic heat release and building shading. The latter, urban morphology and its local to regional impacts, is an ongoing focus of UHI research. We find that not only on larger scales is the UHI effect advected downwind, but local temperatures during daytime are advected downwind on a sub-kilometer scale [8]. Hence, the fraction of impervious area immediately upwind may more strongly affect local air temperature than the local impervious area fraction at the point of measurement. Future work is intended to better quantify this effect for the urban area studied. If confirmed, and of significant size, this opens avenues for local area UHI effect mitigation planning. It suggests that previous calls for (i) more surface unsealing to reduce urban runoff and increase surface moisture and (ii) increased tree cover for additional shading and latent heat cooling, may indeed be the two most efficient pathways to reduce the UHI locally, at least in mid-size metropolitan

We are grateful to the College of Geosciences, Texas A&M University, for fund-

*DOI: http://dx.doi.org/10.5772/intechopen.89514*

areas, where LCZs 5 and 6 dominate.

**Acknowledgements**

*Low-Key Stationary and Mobile Tools for Probing the Atmospheric UHI Effect DOI: http://dx.doi.org/10.5772/intechopen.89514*

expected UHI intensity relationships with wind speed and atmospheric pressure, emphasizing synoptic high pressure conditions with intense solar radiation as a major driver of the UHI. In addition, we may have also found effects of local anthropogenic heat release and building shading. The latter, urban morphology and its local to regional impacts, is an ongoing focus of UHI research. We find that not only on larger scales is the UHI effect advected downwind, but local temperatures during daytime are advected downwind on a sub-kilometer scale [8]. Hence, the fraction of impervious area immediately upwind may more strongly affect local air temperature than the local impervious area fraction at the point of measurement. Future work is intended to better quantify this effect for the urban area studied. If confirmed, and of significant size, this opens avenues for local area UHI effect mitigation planning. It suggests that previous calls for (i) more surface unsealing to reduce urban runoff and increase surface moisture and (ii) increased tree cover for additional shading and latent heat cooling, may indeed be the two most efficient pathways to reduce the UHI locally, at least in mid-size metropolitan areas, where LCZs 5 and 6 dominate.
