Ashraf, S. 2018 Measuring volcanic crater lakes' temperature from space-borne thermal imagery. October 2018. Lower Hutt, N.Z.: GNS Science. GNS Science report 2018/10. 52 p.; doi: 10.21420/G2J34M
Abstract: This report aims to use the latest Landsat-8 Thermal Infra-Red Sensor (TIRS) data to determine the surface temperatures of selected crater lakes and compare the results with in-situ measurements where applicable. The TIRS is capable of capturing thermal images in two discrete spectral bands. This dual channel capability can enable the application of split-window methods for surface temperature evaluations. This report compares the implementation of a Split-Window (S-W) method with a Mono-Window (M-W) method. It is believed that S-W methods account for atmospheric effects better than the surface temperatures calculated from a single thermal band (M-W method). In this report, cloud-free satellite images of five crater lakes at three different sites between August 2013 and August 2017 are analysed. These sites are Green Lake and Blue Lake at Raoul Caldera, Raoul Island, Crater Lake at Mt Ruapehu and Frying Pan Lake and Inferno Crater Lake at Waimangu Thermal Valley. The results show no significant anomalous heating in either of the crater lakes of Raoul Caldera. Conversely, the Crater Lake at Mt. Ruapehu shows significant convective heating, and lake surface temperatures inferred from the TIRS images were in close correlation to in-situ water temperatures. Moreover, the TIRS inferred surface temperatures were in general only a few degrees cooler than the in-situ measurements at Crater Lake. For the lakes at Waimangu Thermal Valley, significant spatial aliasing (i.e. mixing of cooler surrounding temperatures) was observed which resulted in lower than measured surface temperatures for both lakes. The TIRS inferred surface temperature results for Frying Pan Lake show good correlation with in-situ temperature measurements; however, the observed surface temperature values have much larger off-set compared to in-situ measurements than the observations at Mt. Ruapehu Crater Lake. It is due to the surface area of Inferno Crater Lake, which is quite small to obtain any meaningful observation using TIRS images. In general, results obtained using the S-W method showed better correlation to ground observations than the M-W method. Moreover, S-W method was easy to apply to the selected satellite imagery. (auth)
