DISCUSSION

Landsat digital image data are used to investigate the intraurban temperature patterns within the metropolitan area of Phoenix, AZ. (Figures 4 and 5).  Landsat TM thermal data (Figures 5, 7, and 8) have a resolution of 120 meters, while the six spectral bands of reflected sunlight have 30 meter resolution.  Schmid and Oke (1992), Brazel, et al. (1993) and Hubble (1993) have all shown that the spatial pattern of climatic variability within a city manifests itself at a resolution of less than 200 meters.  Landsat TM data provide information well within this scale.

At the time of Landsat image acquisition for this study, weather observations two meters above an irrigated grass surface revealed an ambient air temperature of 37 degrees Celsius, with 24 percent relative humidity and wind velocity of less than four meters per second.  Air temperatures two meters above a desert plot and an asphalt parking lot were 41 and 38.9 degrees Celsius respectively.  Horizontal insolation values observed at all locations were 806 watts per square meter.  [A parallel discussion of this study is presented in Lougeay, et. al., 1996.]

Surface temperature is closely related to the energy available at the earth's surface which, in turn, warms the near surface atmosphere.  Observations associated with this work in Phoenix, Arizona found a significant correlation between surface temperature and near surface air temperature (Hubble, 1993 and Brazel, et al., 1993).  Comparison of insitu observations of near surface air temperature with remotely sensed radiometric surface temperatures yielded an R² of .40, significant at the .05 level.  Similarly, when data from mobile transects were compared with remotely sensed radiometric surface temperatures the R² equaled .39, significant at the .05 level (Hubble 1993).  The surface temperature is actually a result of the net radiant energy budget of the surface and the various sinks for this energy.  Since the incoming solar and atmospheric radiant energy flux is approximately equal for all surfaces within an imaged scene, the net radiant energy is a function of the surface reflectance and emittance.

Net radiant energy of various environmental surfaces is used, primarily, to warm the air, warm the ground, or is used in the process of evapotranspiration.  This is especially true in the mid morning of a cloud free day, which is the case for scenes imaged with Landsat TM data.  In general, an evapotranspirative heat sink is the most significant use of net energy.  Thus, the presence of open water or transpiring vegetation is a strong independent variable controlling local surface temperature.  This is especially important in arid environments such as those experienced during June in Phoenix, Arizona USA.  On June 24, 1992, potential evapotranspiration rates in the Phoenix area were very high due to high net radiation values and low relative humidity.  However, actual evapotranspiration rates were very low where moisture was unavailable, but high where moisture was available to the atmosphere.  This difference in rates of evapotranspiration proved to be a very important variable in explaining surface temperature and near surface air temperatures within the urban environment. 

Modern demands for electrical service and amounts of potable and irrigation water are directly related to ambient air temperature.  In Phoenix, urban development tends to reduce the spatial extent of evapotranspiring surfaces in this extensively irrigated valley, modify surface albedo values, and significantly increase the surface temperature.  The magnitude of climatic modification associated with human activity varies with different types of land use (e.g., commercial, parkland, and various densities and landscaping of residential neighborhoods).

Surface temperature values were extracted from the Landsat data following the procedures discussed in Lougeay, et al. (1994).  In brief, the digital values of the Landsat thermal data were transformed to observed ground surface temperatures by first calculating the spectral radiance and then calculating the "at satellite" temperature of each pixel.  The "at satellite" temperature is a calculated ground surface temperature of the pixel, assuming the surface is radiating as a black body (i.e., emissivity = 1.0).

A more detailed discussion of this study, with ground temperatures adjusted for emissivity, can be seen in Lougeay, et. al. (1996).  Much of this discussion is taken from that article.  However, the total image area analyzed is different, and in the case of images presented with this lecture, surface temperatures have not been adjusted for emissivity characteristics.

Ground level observations of climatic parameters were made at the time the landsat satellite acquired these images.  Although the relative temperature difference among various categories of land use and land cover were of primary interest, ancillary ground observations were conducted to insure accuracy of radiometric data extracted from the remotely sensed data.  Temperature values extracted from Landsat remotely sensed data were compared with ground observations at the time of satellite overpass to assess effects of atmospheric attenuation and to insure accuracy of extracted surface radiometric temperature values.