Bruce C. MitchellA thesis proposal submitted in partial fulfillment of the requirements for a degree of Master of Arts Department of Geography College of Arts and Sciences University of South Florida
Introduction – Urbanization and UHI Literature Review Research Questions Study Area - Pinellas County Methods Results Mitigation Strategies - cool roofs/urban forestry Conclusions
• Half of the world’s population now live in urban areas and this is projected to increase to 61% by 2035. Tropical regions show greatest increase.• Urbanization: decreased vegetation, increased impervious surface, growing population• Environmental consequences: greater storm water run-off, increased air pollution and reduced CO2 filtration.2 Also changes to urban micro-climate, including the Urban Heat Island (UHI)phenomenon which has direct and indirect effects.• Several studies have correlated the elements of urbanization with increases in land surface temperature (LST), a key factor in the urban heat island (UHI) 1 http://esa.un.org/unpd/wup/index.htm Oct., 30, 2010, U.N. Department of Economic and Social Affairs 2 http://nrs.fs.fed.us/units/urban/ Oct., 30, 2010, USDA, Forest Service
Think of a square meter of grass, and ofasphalt in the summer sun.Which would you prefer to stand on?Why?
The radiative properties of a substancedetermine what happens to the Sun’s energy.Is it reflected? Albedo grass – more reflective asphalt – less reflectiveIs it transmitted? EmissivityIs it absorbed? grass – higher emissivity asphalt - slightly lower emissivity
What is the heat capacity? grass – low asphalt – highWhat is the thermal conductivity? grass – low asphalt - high
Heat balance equation -Rn + F = H + G + A + LERn is net all-wave radiationF is artificial and anthropogenic heat generated within the urban areaH is the convective sensible heat transferG is net heat storage within the urban fabric(buildings, roads, soil, etc.)A is net advected energyLE is the latent heat transfer From Chandler, T.J., (1976) Urban Climatology and its Relevance to Urban Design, WMO publication
Urban Heat Island (UHI)– General term for the difference in air temperature between rural and urban areas. Usually measured at “screen-level” Urban Canopy Layer Heat Island – Increased air temperature between the ground to about building height Urban Boundary Layer Heat Island – Increased urban air temperature of the planetary boundary layer above the canopy layer Surface Urban Heat Island (SUHI) – Urban to rural difference in the land surface temperature. This is the focus of the thesis Micro Urban Heat Island (MUHI) – Small urban heat islands which exist below the local scale. Associated with individual structures or groups of structures
Luke Howard,1833 The Climate of London: Deduced fromMeteorological Observations Made in the Metropolis andat Various Places Around It. In Three Volumes.Quantified temperature differences betweenmetropolitan London and surrounding rural areas.Describing the basic mechanisms of the UHI, he noted: Royal Meteorological Society http://www.rmets.org/cloudb• Differences in materials of built urban areas which ank/detail.php?ID=104 retain and reradiate thermal energy more slowly than vegetated rural areas• Absorption and reflection of thermal energy by vertical surfaces of the city• Domestic and industrial processes in urban areas produce heat• Diminished evapotranspiration in urban areas
Wilhelm Schmidt – first use of thermometers attached to automobiles 1920’s Austria Middleton & Millar – 1936 Automobile measurements to do transects of rural to urban temperature differences in Toronto Ake Sundborg – 1950 Automobile transects with point measurements and isoline mapping. Use of statistics. Uppsala Sweden J.M. Mitchell and then T.J. Chandler 1950’s & early 60’s
Automobile transects, and point measurement on a large scale. Comprehensive statistical analysis.
Columbia, MD studydocumented growth ofan urban heat island asa rural landscape wasdeveloped. Used point 1968 population 1,000data collection.The Urban Climate, 1981 1974 population 20,000
T.R. Oke, 1968 – present Boundary Layer Climates, 1978• Population dynamics and the UHI• Energy dynamics of the UHI http://www.geog.ubc.ca/~toke/• Describes its relation to land surface temperature (LST) through the surface urban heat island (SUHI)• Remote sensing of LST ( Voogt & Oke. 2003, Thermal Remote Sensing of Urban Climates). Use of satellite imagery to assess the SUHI
• LST is an indicator of the SUHI• Synoptic view – Captures data over a large area simultaneously• Satellite remote sensing data is comprehensive- extensive archive of images• LANDSAT 5 TM and TERRA ASTER have good enough resolution for urban studies at 120 m2 and 90 m2
Deficient analysis of (sub)tropical regions Methodology has traditionally relied on transects and point data collection. RS is coming into its own in this area, however it can only evaluate LST Need for enhanced surveying, efficient, low- cost methods of evaluating the SUHI at small scales (MUHI) – link to mitigation and urban planning
1. Is there a discernable LST pattern in Pinellas? If so, what are its spatio-temporal characteristics?2. How do the spatio-temporal characteristics of the LST pattern in Pinellas correlate with impervious surface area (ISA), vegetation (NDVI), and land use/land cover (LULC)?
3. How effective are remote sensing techniques at assessing the LST pattern within the study area, and can they provide an efficient method of analyzing spatial patterns indicative of the surface urban heat island (SUHI)?
• Subtropical climate (Koppen Cfa) areas with this climate type have been understudied. (Roth, 2008)• Densely populated - underwent a process of rapid Madeira Beach urbanization in the last century• With its flat local terrain and urbanized area, Pinellas County is an ideal subject for remote sensing techniques. Downtown St. Petersburg
Usedremote sensing data to create LST images using mono-window algorithm Validated with water temperature data and normalized Created NDVI, ISA, and LULC images Statistical analysis Comparative analysis
Remote Sensing and Land Surface Temperature• One of the most extensive archives of remote sensing imagery, Landsat Thematic Mapper or TM has not used more due to the difficulty in completing atmospheric correction with a single thermal band.• Technique used by Qin, Karnieli, & Berliner. 2001, A mono-window algorithm for retrieving land surface temperature from Landsat TM data an its application to the Israeli-Egypt border region• Utilizes Landsat at-sensor radiance image and parameters of land cover emissivity, atmospheric transmittance, and mean atmospheric temperature to calculate a LST image Ts = [a6(1- C6- D6)+(b6(1- C6- D6)+C6+D6)T 6- D6 Ta] Ts is surface temperature C6 is ε6 τ6 and D6 is (1 - τ6)[1 + (1 - ε6) τ6] where ε6 is Emissivity of band 6 τ6 is Atmospheric transmittance of band 6 a6 is -67.355351 (coefficient of temperature range 0 - 70˚C) (Qin et al., p. 3726) b6 is 0.458606 (coefficient of temperature range 0 - 70˚C) (Qin et al., p.3726) T6 is brightness temperature at sensor) Ta is effective mean atmospheric temperature (calculated using LOWTRAN 7 model)
T6 ε6 Atmospheric TransmittanceAt-Sensor Radiance Emissivity based calculated by MODTRAN 4 on NDVI using atmospheric data from NWS Ruskin office Radiosonde image from NOAA website for Ruskin: http://www.srh.noaa.gov/tbw/?n=tampabayofficetour
1. RS image acquisition2. Atmospheric data collection3. Construct emissivity image4. Landsat thermal band (6)5. Run MWA program6. Text file for display in GIS7. Validation8. Normalization of multi- temporal images
Land surface temperatures(excludes water)Mean = 30.14˚CMin = 16.83˚CMax = 50.99˚CSD = 4.2076Validated within 0.423˚C ofthe water sample sites
Land surface temperatures(excludes water)Mean = 27.46˚CMin = 12.87˚CMax = 50.57˚CSD = 3.8163Normalized to 27.76˚C using alinear regression of the threeimages
Land surface temperatures(excludes water)Mean = 32.40˚CMin = 18.03˚CMax = 61.399˚CSD = 3.8345Normalized to 28.72˚C using alinear regression of the threeimages
• Dependent Variable – LST as derived from remote sensing images• Independent Variables - 1)ISA 2)NDVI 3)LULC Impervious Surface Area Normalized Difference Land use land 2009 data 2002 USGS Vegetation Index 2009 cover 2008 data
Stratified random sample • Exclude water and land outside the study area • 3000 pixels randomly chosen • LST • NDVI • Impervious/not impervious 2009 image or actual impervious percentage for the 2001 image • LULC based on FLUCCS level 2 coding • Divide LULC into rural/urban types
LST NDVI IMPERVIOUSLST 1 -0.580** 0.468**NDVI -0.580** 1 -0.678**IMPERVIOUS 0.468** -0.678** 1** significant at the α= .01 level 0 = not impervious M=29.23˚C LST to NDVI R2 = 0.337 1 = impervious M=32.49˚C
Mean Rural Temperature25.03˚CMean Urban Temperature31.62˚CLST ΔT = 6.59˚Cat 11:48 EDT on 4/8/2009
LST NDVI Impervious LST 1 -0.714** 0.628** NDVI -0.714** 1 -0.748** Impervious 0.628** -0.734** 1 ** significant at the α= .01 levelRelationship of LST to NDVI, 2001 dataset Relationship of LST to Imperviousness, 2001 and 2002 (R2=.510) datasets (R2 =.395)
In 2009 and 2001 image statistically significant negative linear correlation of LST and NDVI In 2009 and 2001 image statistically significant positive linear correlation of LST and Imperviousness Mean LST varies by LULC types, with rural land cover having generally lower temperature than urban land cover types at the time of image capture.
LST North Pinellas Transect (South of Lake Tarpon) 50 Commercial Temperature °C 40 Brooker Creek 30 LST 20 Water 10 Barrier--Gulf of Mexico----------LD--HD Resid------------------------------Wetland-Upland-------> Island Resid Forest Forest 0 LAND COVER LST Gulf to Bay, Clearwater Transect 50 Transportation Recreational TransportationTemperature ° C 40 30 LST 20 10 Gulf---------------HD<Water>--HD-Rec-------HD------<-Comm-HD--WetlandHDWetlandBay Resid Resid Resid Resid Forest Resid Forest 0 LAND COVER LST Central Ave., St. Petersburg Transect 50 Gulf Beaches Downtown Waterfront Temperature ° C 40 30 LST 20 Gulf------<-HD & Water--------------HD Residential-----------------------------------Recreational 10 0 LAND COVER
Descriptive mapping: LocalScale 1km up to 50 km•Generally cooler water andcoastal temperatures.•Temperature increases withdistance from the coast•Southern portion of thepeninsula shows evidence ofa pronounced SUHI
Descriptive mapping –Local scale•Central Plaza in thecenter of the lower portionof the peninsula.•Temperatures 28 – 40˚C•Area 4 x 5kmHighly urbanized withCommercial and high-Density residential
Descriptive mapping – Micro-Scale.•A series of “hot” islands and coolerpark areas which create an “oasiseffect” appear across the landscape•“Hot” islands are MUHIs asdescribed by Aniello et al. (1995)Temperature gradientLand Use TemperatureWater/Parks 22-28˚CResidential 28-32˚CCommercial 32-36˚CInstitutionalHigh-densityResidentialMUHIs 36-50˚C(structures)
The park is 4˚C cooler than thesurrounding land cover types.This creates an “oasis effect”Cannot tell how far this mayextend to the surroundingarea. Rosenzweig et al. (2007)found that cooling of CentralPark extended no more than 60meters. Cannot extrapolate LSTto near-surface air temp.Temperature gradientLand Use TemperatureWater/Parks 22-28˚CResidential 28-32˚CCommercial 32-36˚CInstitutionalHigh-densityResidentialMUHIs 36-50˚C(structures)
14 1312 1010 8864 2 22 10 industrial Institutional Power Plant Services Shopping Shopping Mall Plaza
While urbanization is at too small a scale to directly impact global climate change, the UHI acts to compound broader regional heating patterns intensifying them at the local level (Grimmond, 2007) Public health – intense heat and higher mortality rates for vulnerable segments of the population: the elderly, children under 5, people with medical conditions Vector-borne diseases – malaria, encephalitis, dengue fever
Personal discomfort causing increased use of air conditioning. This is a counterproductive adaptation strategy. (Richardson, Otero, Lebedeva, Chan, 2009) Increases use of electricity 1˚C increase above 15-20˚C threshold results in 2-4% increase in electricity demand (Akbari et al., 2001) Increased electrical consumption results in burning of more fossil-fuels More fossil-fuel use results in increased Carbon emissions, intensifying the problem of global climate change
IncreasedA/C use is maladaptive, though it may be necessary for vulnerable individuals (Richardson, Otero, Lebedeva, Chan, 2009) Mitigation should be carbon neutral Sincechange in land cover is a primary factor of the UHI, modifying land cover to increase albedo and emissivity, and increase vegetation can mitigate the UHI
Cool and green roofs Increase albedo (reflectivity) and emissivity (ability to reradiate thermal energy) Increase vegetation and insulation Increased vegetation – urban forestry Increase shade Increase evapotranspiration Decrease thermal energy storage Increase permeable surfaces Increase evapotranspiration Decrease thermal energy storage
Cool roof Green roofStructure Coating or roofing Structure to hold material growing medium and underlying membraneCost $ .50 to $6.00 ft2 $10.00 ft2 and upMaintenance Cleaning and sealing VariesAdvantages Prevents absorption of Prevents absorption of heat heat, adds benefits of vegetation, Provides winter insulationPromoters New York City (street Chicago and Toronto trees)
Tropicana field –Structure is atbackground temperaturelevels of 29˚C which is12˚C cooler thanadjacent parking lot and14˚C cooler than nearbyschool.
Urban forest already comprises 20-40% of the average North American city (Oke,1989) Parks appear to have limited temperature moderating impact (Rosenzweig et al., 2007) Street trees may have more impact since they shade the pavement and structures and increase evapotranspiration (Richardson et al., 2009) Quantification of energy savings. Strategic placement can effect 25-50% reduction in cooling (Parker, 1983; Meier, 1991; Akbari, 2001) Studies emphasize in careful placement and a neighborhood level approach (Richardson et al., 2009)
Low-cost with extensive archive Efficient in surveying large areas Has sufficient resolution to locate MUHIs for remediation When used with aerial photography can be effective in neighborhood level surveys of urban forestry by evaluating NDVI levels.
Is there a discernable LST pattern in Pinellas? Ifso, what are its spatio-temporal patterns?Yes – There are patterns at both a local andmicro-scale level. A gradient of cool coastalareas with temperature increases toward theinterior. A pattern of MUHIs (greater than 40˚C)and cool park areas which create an “oasiseffect” exist across the landscape. This is wellresolved at the time of satellite over-flight(˜15:30 UTC) and appears in all images.
How do the spatio-temporal characteristicsof the LST pattern in Pinellas correlate withimpervious surface area (ISA), vegetation(NDVI), and land use/land cover?Statistically significant correlation of LSTand both NDVI and Impervious surfaces.LULC also appears to be associated withsignificantly different mean temperaturelevels between rural and urban land covertypes. Transects and mapping visuallyconfirm spatial relationship.
How effective are remote sensing techniques atassessing the LST pattern within the study area,and can they provide an efficient method ofanalyzing spatial patterns indicative of thesurface urban heat island (SUHI)?This thesis demonstrates the ability of LANDSATTM sensor imagery, when processed using theMWA to provide accurate (within 0.432˚C) LSTimages. They provide sufficient resolution toidentify MUHIs for possible remediation. It isan efficient, low-cost surveying technique whencombined with aerial photography.
Since human modification of land cover is responsible for the UHI, it can be mitigated. Mitigation is worthwhile due to its effects on health, comfort, and energy use. Direct benefits of mitigation are reduction in air conditioning, and energy use. There are also indirect benefits in reduced fossil-fuel use and carbon emissions These changes can be made at the neighborhood level and remote sensing provides an efficient, low- cost method of identifying MUHIs for mitigation