Airborne Thermal IR Remote Sensing of Lushan Geothermal Field
4. Thermal IR Data Calibration
4.1 Data Quality Evaluation
For the acquired thermal image, the recorded blackbody temperatures and their corresponding DN were checked to evaluate data quality. It was found that the calibration parameters of channel 11 were very unstable during flight mission. Consequently, channel 12 data were selected, in lieu of channel 11, for the following processing and analysis. In channel 12, the recorded calibration parameters were relatively stable, while its gain is 0.5 instead of 1.0 in channel 11.
However, the calibration parameters of some image lines are still unstable in channel 12 data, in such case the average value of the corresponding parameters of the neighboring two lines is adopted to replace the unstable one.
4.2 Radiometric Calibration
According to blackbody temperatures and their corresponding DN, Plank's equation was used to calibrate thermal IR data. The DN value was then transformed to radiance. Signal to noise (S/N) ratio of the thermal image was also calculated and was as high as 17.36, and its noise equivalent delta T (NEDT) was 0.98 °C. These values indicate that the quality of channel 12 was acceptable.
4.3 Brightness Temperature Calculation
A radiance to temperature conversion table obtained from the above steps was adopted to calculate the brightness temperature (BT) of each pixel of the image. Table 2 lists the count of temperature distribution. The statistics are : mean BT 15.56 °C, Standard deviation 2.13°C and mode at 16°C.
| BT (°C)
| # pixel
| BT (°C)
| # pixel
| BT (°C)
| #
pixel |
| 4
| 94
| 14
| 284667
| 24
| 343 |
| 5
| 2390
| 15
| 244170
| 25
| 190 |
| 6
| 1501
| 16
| 341608
| 26
| 55 |
| 7
| 987
| 17
| 228066
| 27
| 34 |
| 8
| 2147
| 18
| 209711
| 28
| 13 |
| 9
| 6294
| 19
| 60446
| 29
| 5 |
| 10
| 11668
| 20
| 16234
| 30
| 2 |
| 11
| 24564
| 21
| 8987
| 31
| 1 |
| 12
| 64178
| 22
| 3088
| 32
| 0 |
| 13
| 103652
| 23
| 1056
| 33
| 1 |
Table 2. BT Distribution in Lushan Area
By field check and comparison with base maps, it was found that those " hot spots" withBT>27°C were all located at the slope lands of Chinjin Farm and NTU Experimental Farm and they were irrelevant to geothermal activity. Instead, those thermal effects were owing to metal roofed houses or residential buildings.
5. Hardcopies Generation
5.1 False Color Composite
After contrast enhancement, images of MSS channel 4, 6, and 9, i. e. green, orange and reflected IR bands, were composite (B/G/R) for hard copy output. Such a false color composite (FCC) simulates a IR photograph for surface feature interpretation..
5.2 Pseudo-color Image of Brightness Temperature
The temperature image was transformed to a pseudo-color image in which each color represents a specific temperature range. However, owing to the atmospheric absorption effect, usually the derived BT is lower than the actual surface temperature ; also, it must be realized that the pixel temperature represents the average temperature in a area covered by that pixel. In this case the pixel coverage is about 10m×10m on the ground.
5.3 Thematic Map of Thermal Anomaly
Although the image generated in 5.2 shows the surface BT of every pixel in Lushan area, it's difficult to identity where are the pixels with higher temperature in that map. In order to distinguish "hot spots" or "hot area", the pixels with temperature greater than 19°C were extracted and overlaid on the MSS 4/6/9 FCC to generate a thermal anomaly image , in which yellow represents 20~21°C, green represents 22~23°C and red for the pixels with temperature great than 23°C (Fig.1).
6. Field Check
Field checks were carried out twice to verify the "ground truth" in Feb. and Apr. 1997. Those hot spots within 1 km along the planned route, if accessible, were checked to see if they were related to hot spring activity.
7. Results and Interpretation
In Fig. 1 the background image displayed is MSS 4/6/9 FCC, while the BT anomalies are shown in yellow, green and red colors which represent 20~21°C, 22~23°C and 24°C BT, respectively.
In MSS 4/6/9 FCC image, deep blue area is the water body of Wanda Reservoir. Light blue represents bare lands including landslides, cultivated land, road and shallow water riverbed. These surface features have similar spectral signature and consequently they appear as similar color. Pinky and reddish color indicate vegetation including forest, grass land, orchards and vegetable patches. Dark area is shadow which enhance the rugged terrain. However within some shadow area some dark greenish spots still can be seen, which were landslides or cultivated land.
Most of the high temperature anomalies were scattered on the sunshine slope, they were roadside buildings, landslides and cultivated land. The local environment is that there are many metal roofed houses in Lushan, Chinjin and Wushe, and many hotels and outdoor swimming pools locate at Lushan Spring Area. All these landuses appeared as hot spots in thermal image. It's found that all hot spots withBT>27°C were located at the dense developed area around Chinjin Farm and NTU Experimental Farm, and they were irrelevant to geothermal activity. As for bare land, the dry bare rocks had higher temperature and appeared as red, while debris had lower temperature and appeared as green or yellow color.
Lushan area is a geothermal field in the metamorphic terrain and is the typical location of Lushan Formation. Lushan is located at the middle member of Lushan Formation, the Yuanfen Member, and is sandstone-dominated. On the other hand, the Chunyang Spring Area to its west is located at the lower Chunyang Member, which is slate-dominated (Chang and Lee, 1980 ; Fan and Lee, 1994). Fractures and joints in the brittle Lushan F. are common and produce important secondary porosity for geothermal water. Geothermal water move upwards along the fractures and usually appear on the riverbed. For this reason, the detection of geothermal activities had been focused on the river valley of Talowanshi river, Mahaipushi River and a creek near Chunyang.
In Fig. 1, to the south of Muansan mountain there are many hot spots scattering and extending more than 1 km along the riverbank of Talowanshi R. This area has become a famous recreation area. However those hot spots were all buildings and outdoor swimming pools and their spring water are all from Spring Head located at 1 km upstream or from private geothermal borehole. The place 1.5 km upstream from here and on the left riverbank near PA18 borehole, there are two yellow hot spots in Fig.1 (S1). After field checks it's confirmed that they were spring outcrops. During field checks upstream along Talowanshi R. we had also found that many small spring water seepages appeared on the left riverbed, however their coverage were too small to detect in thermal image. Yellow spot N1 on the right riverbank there was also a hot spot near PA 19 borehole and below a silt arrester, during field checks we had found it was a big iron-tank. The L1 hot spot was a landslide on a steep slope where did not has any geothermal activity. Another hot area near PA20 borehole was also a landslide (L2), where Highway 14 passes by. Other hot spots such as L3 and L4 in Fig. 1 are too difficult to reach, they are also interpreted as landslide according to their spectral signature.
At 3km downstream of Talowanshi R. from Lushan it is Chungyang. The yellow spot (S2) at the intersection of Talowanshi R. and a creek had been verified as a geothermal outcrop during field check. L5 and L6 were also hot areas located at 700m and 1km upstream of that creek. On the way to check these two areas, more than ten spring seepages were found, however they could not be identified in the thermal image because of their small size. L5 was a landslide having strong geothermal activity, as we could see hot water were tapped and piped outside to hotels. Around L5 some spring seepages were also found. L6 was also a landslide, while some spring seepages appear on the other side of the creek. It is concluded that the riverbed of that creek has strong geothermal activity underlain.
In Fig. 1 there is not a hot spot at all appearing along the riverbed of Mahaipushi R. However hot water and steam seepage points were really existing at 400m upstream of the river. When we lowered the BT of color coding to 19°C and displayed the pixels having this BT as cyanine color, the hot spot S3 of this outcrop appeared. From the results it is clear that a spring outcrop is not necessarily appearing as higher temperature in a thermal image because of mixed pixel effect.
