Remote Sensing (Sep 2018)
Monitoring of Thermal Activity at the Hatchobaru–Otake Geothermal Area in Japan Using Multi-Source Satellite Images—With Comparisons of Methods, and Solar and Seasonal Effects
Abstract
The Hatchobaru–Otake (HO) geothermal field is proximal to the Kuju volcano on Kyushu, Japan. There are currently three geothermal power plants operating within this geothermal field. Herein, we explore the thermal status of the HO geothermal area using ASTER thermal infrared data to monitor heat losses from 2009 to 2017. We assessed the solar effects and seasonal variation on heat losses based on day- and night-time Landsat thermal infrared images, and compared three conventional methods of land surface temperature (LST) measurements. The normalized difference vegetation index threshold method of emissivity, the split window algorithm for LST, and the Stefan–Boltzmann equation for radiative heat flux (RHF) were used to determine the heat loss within the study area. The radiative heat loss (RHL) was 0.36 MW, 38.61 MW, and 29.14 MW in 2009, 2013, and 2017, respectively, from the HO geothermal field. The highest anomaly in RHF was recorded in 2013, while the lowest was in 2009. The RHLs were higher from Otake than from the Hatchobaru thermal area in the year of 2013 (~31%) and 2017 (~78%). The seasonal variation in the RHLs based on all three LST estimation methods had a similar pattern, with the highest RHL (about 383–451 MW) in spring and the lowest (about 10–222 MW) in autumn for the daytime images from the HO geothermal field. In the nighttime images, the highest RHL was about 35–67 MW in autumn and the lowest was about 1–3 MW in spring, based on the three LST methods for RHFs. The highest RHL was about 35–42 MW in spring (day) and 3–7 MW in autumn (night) from the Hatchobaru thermal area, analyzed separately. Similarly, the highest RHL was about 22–25 MW in spring (day) and 4–5 MW in winter (night) from the Otake thermal area. The seasonal variation was greatly influenced by the regional ambient temperature. We also observed that clouds had a huge effect, with the highest values for both LST and RHF recorded below clouds on an autumn day. Overall, we obtained higher LSTs at nighttime and lower LSTs during the day from the improved mono-window algorithm than the split window algorithms for all of the seasons. The heat losses were also higher for the improved mono-window algorithm than the split window algorithms, based on the LST nighttime thermal infrared data. Considering the error level of the LST methods and Landsat 8 band 11, this study recommends the IWM method for LST using the Landsat 8 band 10 data. This study also suggests that both the nighttime ASTER and Landsat 8 thermal infrared data could be effective for monitoring the thermal status of the HO geothermal area, given that data is available for the entire period.
Keywords
