IR thermal imaging does not directly measure surface temperature. Surface temperatures are calculated from the longwave radiation emanating from structures. The radiation temperature, as it is called, is measured. Radiation temperature is a transport of energy by electromagnetic waves. Radiation is then defined as the flow of electromagnetic waves per given area during a given period of time. The radiation and temperature of an object’s immediate surface are in a functional relationship to each other, as expressed in the Stefan-Boltzmann equation. This relationship occurs when the surface approaches its full emissivity (emissive capability) (theoretical emissive value = 1). Values are known for all important surface elements within the imaged wavelength spectrums of 10.4 to 12.5 mm, so that the influence of the atmosphere on measured emissive behavior remains minimal. The difference between the radiation temperature measured by satellite and the calculated surface temperature is then usually negligible. Only metal surfaces, such as used in flat roofs, deviate significantly with emissive values of 0.1. They must be given a special category during interpretation.
Of much greater significance is the degree of spatial resolution of the image elements in pixels of 120 m x 120 m. These are transposed into pixels of 30 m x 30 m before delivery by the ESA ground station. However, the larger initial size often means the registration of mixed signals which impedes the determination of traffic areas, smaller urban squares, or various vegetal structures. Each grid of approximately 14,000 m2 can initially only be given an average radiation temperature integrating all the surface elements within the grid.
The initial material was digitally processed by ERDAS, an image processing system. It was necessary to first perform a geometric distortion correction before both images could be superimposed to generate a Differential Map. The low resolution of the initial pixels produced a large degree of mixed signals. A control reference value was needed. Surface waters were found to be suitable as reference values, for they are large homogenous areas, and the temperature difference makes it easier to distinguish them from their surrounding areas.
The conversion of the radiation amounts measured by satellite into temperature values resulted initially in a total of 53 gray scale values; each represented a temperature interval of about 0.5 °C. The lowest value was 4.3 °C at night, and the highest value was 28.9 °C during the day. Deviations for selected structures from those temperatures determined in the ground measuring program conducted at the same time led to a temperature classification in stages of 1°C. An easier overview of the amounts of data was made, without any great loss of information, by condensing the lowest and highest values into the open categories of < 8 and > 26 °C.
A combined qualitative gradation of temperature differences into 5 categories from “low” to “high” was made for the Differential Map. This scale was chosen in order to take into account the non-optimal time period of survey. Optimal overflight times would have given a stronger differentiation of the radiation behavior of surface areas during the day and at night, especially in the city center area. City center green spaces in particular would then, as more quickly cooling areas, be more intensely differentiated from densely built-up areas and industrial areas with high heat potential.
The map does not present statements regarding the level at which the temperature differences range, i.e., whether at relatively high or at lower surface temperatures. Figure 2 provides information about this.
The possibilities and limits of the survey technology and time periods described above should be mentioned again as a basis for the interpretation and comparative analysis of day and night levels:
- Small areas of differentiated horizontal and vertical structures, such as interior courtyards, street areas, and city squares, could be recorded only as mixed pixel images.
- The overflight times during the morning and early evening did not record the time periods of greater heating or greater cooling. Material-dependent heat conduction and heat storage exert a special influence. The dry sand soil, with high air content, of farmlands and dried-off vacant areas has a poor heat conductivity, particularly on sunny weather days with weak winds. This produces a quick morning heating and a quick evening cooling in the map image. Inversely, the high heat storage properties of building materials such as concrete, asphalt, and stone lead to a slower heating and cooling, and thus to a limited representation of “the urban heat-island”.
- Seasonal changes in open areas are of great importance. Critical modifications in temperature behavior sometimes occur during harvest or large-scale dying of surface stocks, particularly in field areas and rough meadows.