de Jong, Steven 
  
1. Minimum Noise Fraction transformation: In order to remove image noise from the six thermal bands, the MNF 
transformation was used. Four MNF bands are of good quality, these bands are used for the inverse MNF 
transformation. 
2. Field reference data into blackbody temperature: In order to recalculate the thermal DAIS bands to a 
temperature map in degrees Celsius, field measurements are used. During the 1998 DAIS over flight, 
temperature measurements were taken of all land cover types. The measurements were taken with a radiant 
thermometer. This radiant thermometer measures the radiant temperature in degrees Celsius in the atmospheric 
window between 8 and 14 um. During the measurements, an emissivity of 1 was applied. The resulting 
blackbody temperatures are shown in table 2. 
Water Vailhan 24.2 
Bare soil 115 43.2 
i 30.5 
Grassland 145 34.7 
Grassland 1111 42.1 
i 32.6 
Dam 51.6 
Pine forest 27.4 
Table 2: Field measurements of the average blackbody temperature in the 8-14 pm atmospheric window 
  
3. Calculation of mean radiance temperature in the 8 - 14 um region: In order to compare the field measurements 
with the radiant temperature in the DAIS image, the mean radiant temperature of the six thermal bands has to 
be calculated. In this way, the information content of the six thermal bands is being reduced to one radiant 
temperature band. 
4. Empirical Line correction: With the previous data it is now possible to calculate a linear regression function. 
The empirical line function calculates a mathematical function between the field temperature measurements 
and the thermal DAIS data at the same location with a satisfying correlation (R^—0.92). The resulting map is an 
image containing for each pixel the blackbody temperature in degrees Celsius. 
To convert the blackbody temperatures into absolute temperatures, emissivity data are required. In the Peyne area, there 
is a large variation in land cover types and hence, a large variance of emissivity values. Therefore, it is not correct to 
apply a method such as the reference channel method, as this method assumes a constant emissivity in the reference 
channel. This emissivity value is an average value for silicate rocks and not applicable for the Peyne situation. 
An alternative approach for emissivity correction is applied in this study. The unique combination of optical- and 
thermal bands with the same resolution in the DAIS scanner allows us to use the information content of the optical 
bands for correction of the thermal bands. The optical DAIS bands are used to classify land cover (paragraph 3.3). Next, 
the resulting land cover map is used for converting the blackbody temperature map to an absolute temperature map by 
assigning to each land cover type the corresponding emissivity value. These emissivity values are derived from 
laboratory and field emissivity measurements by Rubio et al., (1997). Table 3 shows the emissivity value for each land 
cover type: 
Water Vailhan 0.990 
i 0.986 Mean of several deciduous trees 
0.984 Mean of and j 
Pine forest 0.982 
Vi 0.995 
Grassland 0.990 
Bare soil 0.954 Calcareous soil 
Buildi effies 0.950 Mean of and limestone 
Table 3: emissivity values of the different land cover types, derived from Rubio et al.(1997). 
  
Next, the absolute temperature is calculated by multiplying the blackbody temperature with the emissivity. This map is 
used for further study of temperature differences and processes within the study area. 
  
350 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.