TEMPERATURE, COLOR AND DEFORMATION MONITORING OF VOLCANIC REGIONS IN NEW ZEALAND K.E. Joyce, S. Samsonov and G.Jolly GNS Science, PO Box 30368, Lower Hutt 5040, New Zealand. Email:
[email protected] ABSTRACT There are many examples around the world where satellite based remote sensing has been used to successfully monitor different stages of volcanic activity. This paper describes some of the methods used and their results for monitoring two active volcanoes in New Zealand - Mt Ruapehu and Raoul Island. A time series of ASTER night-time thermal images has been successfully used to assess crater lake temperature variability; Hyperion hyperspectral imagery was tested for crater lake color monitoring with inconclusive results; and interferograms using ALOS PALSAR data were generated of Mt Ruapehu for the purpose of mapping deformation patterns. These data provide information for baseline monitoring, as no major volcanic activity was evidenced over the duration of the study. Index Terms— Remote sensing, image processing, image analysis, synthetic aperture radar, infrared image sensors
measurements using synthetic aperture radar are frequently undertaken and well understood [1]. Global thermal anomaly identification using MODIS and GOES is also operational [2]. On a local scale, Landsat and ASTER have been successful for providing relatively high spatial resolution thermal imagery of volcanic activity [3, 4]. Ash and gas clouds can be detected and their concentrations estimated using a number of sensors [5]. Several of these techniques are used in New Zealand for monitoring volcanic regions both within the country in the greater region covered by the Wellington Volcanic Ash Advisory Centre (VAAC). The work contained herein discusses the results of three aspects of satellite based remote sensing for monitoring Mt Ruapehu and Raoul Island – crater lake thermal monitoring using ASTER night time thermal imagery; testing Hyperion hyperspectral imagery for its potential in crater lake color and geochemistry evaluation; and patterns of deformation at Mt Ruapehu using ALOS PALSAR interferometry.
2. METHODS
1. INTRODUCTION New Zealand is home to many active and potentially active volcanoes. Mt Ruapehu (central North Island) is the most active onshore volcano and is also the site of two major commercial ski-fields, including the largest in the country, Whakapapa. The site is economically valuable to the region and provides a popular recreation facility for New Zealanders and international visitors. The most recent eruption was in September 2007, although there was also a lahar following the Crater Lake dam wall burst in March 2007. Raoul Island is a scientific reserve and is inhabited only by a small number of Department of Conservation (DOC) staff. The most recent eruption was in March 2006 and resulted in the death of one DOC staff member. As two of the most recently active volcanoes in the country with potentially hazardous consequences, Mt Ruapehu and Raoul Island were selected to develop an on-going satellite based remote sensing program to complement the numerous installed in-situ monitoring tools. Currently, both sites are continuously monitored in-situ for seismic activity, ground deformation (GPS) and temperature variability of their crater lakes. Additionally, visual observations are made by staff and web cameras, and water chemistry and airborne gas measurements are regularly made at Mt Ruapehu. Satellite remote sensing offers the potential for monitoring several aspects of volcanic activity. Ground deformation
978-1-4244-3395-7/09/$25.00 ©2009 IEEE
2.1. Thermal monitoring ASTER thermal images have been acquired over Mt Ruapehu and Raoul Island on a routine basis since September 2007, whenever a cloud free image is available (Fig 1). The data are purchased as raw level 1A data, and all atmospheric correction, orthorectification and conversion to emissivity and temperature (emissivity normalisation method) was conducted in ENVI/ITTIVS. Only the nighttime images are presented for analysis here. Field data were used for calibration and validation of the extracted temperature values [6]. All night-time images were stacked, and an average image created for both sites for comparative purposes in the time series. Mt Ruapehu S
O
N
D
J
F
M
A
M
J
J
A
S
O
N
D
J
F
M
A
M
Raoul Island 2007
2008 Night-time TIR
2009 Day time TIR
Fig 1. Availability of ASTER TIR image data 2.2. Color monitoring In order to assess the possibility of using satellite remote sensing for measuring color variations as an indication of crater lake chemistry, Hyperion images were obtained for both Mt Ruapehu and Raoul Island. Mt Ruapehu has a single crater lake that varies in color according to the level of upwelling, and Raoul Island has two lakes of blue and
I - 17
IGARSS 2009
Authorized licensed use limited to: UNIVERSITY OF WESTERN ONTARIO. Downloaded on March 29,2010 at 11:44:24 EDT from IEEE Xplore. Restrictions apply.
green in color. Atmospheric correction using ENVI/IDL FLAASH was applied to both data sets and the data were spectrally subset to remove bad bands, resulting in 158 individual spectral bands over the visible, NIR and SWIR regions. Spectra from the Mt Ruapehu Crater Lake were compared with a nearby dam in the absence of any open water in the scene. Spectra from the Raoul Island lakes were compared with surrounding sea water. First and second derivatives of the average spectra were also calculated to analyse differences in spectral form of the target regions. 2.3. Deformation monitoring GNS Science collects all available L-band ALOS PALSAR data of Mt Ruapehu and the Taupo Volcanic Zone. The study of ground deformation at Mt Ruapehu is particularly important because a large eruption at this active volcano is imminent in the future, and early detection of deformation can be used for its forecasting. Until now, four ALOS PALSAR images have been collected (acquired on 1 January, 16 February 2007 and 4 January, 5 April 2008) and seven differential interferograms created.
With the exception of a minor explosive eruption at Mt Ruapehu on 25 September 2007, there has been no significant volcanic activity in this region. Raoul Island has also remained at a low level of alert. The data obtained here do not demonstrate any significant anomalous temperature values at either site over the study period, although a rise in temperature was evident at Mt Ruapehu post-eruption (Fig 2, Fig 3). These data are well calibrated to in-situ temperatures [6]. Average temperature values at Mt Ruapehu between September 2007 and September 2008 were higher than any surrounding features (Fig 2), though were not considered extreme. The minor heating post eruption was not even enough to trigger a thermal alert from MODVOLC due to the small area and relatively low temperature increase [6]. This supports the continual use of ASTER imagery for monitoring low level, small feature temperature variations. Little variation in temperature was observed between the Green and Blue Lakes at Raoul Island, even with respect to the open ocean (Fig 3). These data remain a valuable source of information about the heating and cooling cycles of these volcanoes, and are part of a continuing monitoring program at these sites.
3. RESULTS AND DISCUSSION A
3.1. Thermal monitoring Since the commencement of this project in September 2007, 24 ASTER images have been obtained of Mt Ruapehu, and 14 of Raoul Island (correct as at June 2009).
Temperature (oC)
C
A Auckland
Raoul Island 3 Sep 07
Mt Ruapehu
Green Lake Blue Lake Ocean
24 22
Blue Lake
20 Green Lake
18
5 Apr 08
16 Wellington
B
Christchurch
N B 22 Nov 07
40
Mt Ruapehu Crater Lake
16 May 08
Temperature (oC)
26 Sep 07
20 Aug 08
35 30 25 20 15
31 Dec 07
5 Sep 08
Aug
Dec 2007
5km
-10
25o C
26 Feb 08 0
28 Sep 08 40oC
Fig 2. Mt Ruapehu TIR imagery Sept 07- Sept 08. (a) Location of study sites; (b) 12 month Average temperature image; (c) Crater Lake subset images
Apr
Aug 2008
Dec
Apr 2009
Fig 3. Temperature values extracted from ASTER night time thermal imagery of (a) Raoul Island (inset is Hyperion true color image); and (b) Mt Ruapehu. Note the scale change on the y-axis.
I - 18 Authorized licensed use limited to: UNIVERSITY OF WESTERN ONTARIO. Downloaded on March 29,2010 at 11:44:24 EDT from IEEE Xplore. Restrictions apply.
3.2. Color monitoring Hyperspectral imagery has been used for monitoring the color of oceans, rivers, and inland lakes, where variations are attributed to organic matter and suspended sediment [7]. Yet to date, little has been done to test its ability for assessing crater lake color and associated chemical constituents. In the absence of a time series of hyperspectral imagery, it was not possible to analyse spectral changes at individual locations. Instead, the variability in reflectance spectra and their derivatives between the Green Lake, Blue Lake and open ocean at Raoul Island is demonstrated here (Fig 4). The largest degree of variability is apparent in the shorter wavelengths, up to 1000 nm. In the longer wavelengths it appears that water absorption is the dominant feature, causing significant noise in all samples. The reflectance spectrum of the Green Lake shows several peaks that can also be seen as the zero crossing point in the derivative spectrum (579, 660, 711, and 762 nm). In particular, the peak at 711 nm is distinctively different to the other spectra. More work needs to be done to determine if this is diagnostic of any specific chemical constituent in the water, and if this, or other spectral regions can be used to monitor change. The accuracy and effectiveness of multi-date image calibration over Raoul Island and Mt Ruapehu is also unknown. There is a considerable amount of noise evident in the data shown here, and retrieving calibrated signals may prove difficult. 8
A
Green Lake Blue Lake Ocean
7 Reflectance (%)
6
Due to the lack of data and low temporal resolution, it is too early to ascertain whether the spatial, spectral, and radiometric resolution of Hyperion is sufficient to provide a viable means of monitoring either of these sites. In particular, only a single cloud free scene over each of these volcanoes was available in the archive, which is clearly inadequate for monitoring purposes. Additional scenes will be required to better evaluate the potential of this type of monitoring. 3.3. Deformation monitoring The coherence of the processed interferogram was high except in regions covered by snow at the summit of Mt Ruapehu. Differential interferograms were successfully unwrapped and a variety of localized signal was observed (Fig. 5). The large scale gradient on the image (pink to orange) is probably due to orbital and atmospheric errors. Signal on the slopes of the volcano is probably caused by topographic errors. Black holes at the summit are due to the lack of DEM coverage. It is anticipated that coherent signal with a magnitude of 10 cm and larger can be easily detectable with this technique. No deformation due to volcanic activity has yet been detected, and the work will continue when a high resolution and high accuracy DEM is available or when more PALSAR images are acquired. In spite the fact that large ground deformation were not observed on these interferograms it was confirmed that L-band interferometry can be successfully used for monitoring of deformation because of its superior quality (in comparison to C-band SAR data) of being coherent in densely vegetated regions.
5
N
4 3 2 1 0
B Magnitude of First Derivative
0.2 0.15 0.1 0.05 0 -0.05 0
-0.1 -0.15 -0.2 500
700 900 1100 1300 1500 1700 1900 2100 2300 Wavelength (nm)
Fig 4. Spectral signatures and first derivatives of spectra taken from a region of interest within the Green Lake, Blue Lake and open ocean at Raoul Island
5km
Fig 5. Wrapped differential interferogram for two ALOS PALSAR images acquired on 1 January 2007 and 5 April 2008 (perpendicular baseline 2127 meters). The full color scale on this image corresponds to about 13 cm of line-ofsight deformations.
I - 19 Authorized licensed use limited to: UNIVERSITY OF WESTERN ONTARIO. Downloaded on March 29,2010 at 11:44:24 EDT from IEEE Xplore. Restrictions apply.
4. CONCLUSIONS It is important to continue monitoring over short and long time scales to better understand the phenomena of temperature and color variations as well as deformations as an indicator of potentially hazardous volcanic activity. Thermal monitoring with ASTER of both Mt Ruapehu and Raoul Island has been successfully demonstrated to complement in-situ data. While color variations were apparent in the Hyperion imagery, more data are required to determine if this information could also supplement our understanding of crater lake geochemistry and potential precursors to an eruption. The utility of PALSAR in the New Zealand region has yet to be fully realised due to the lack of significant volcanic activity experienced at the study sites over the duration of this work. 5. ACKNOWLEDGEMENTS
This manuscript incorporates data which is © Japan Aerospace Exploration Agency ("JAXA") (2008). The data has been used in this manuscript with the permission of JAXA and the Commonwealth of Australia (Geoscience Australia) ("the Commonwealth"). JAXA and the Commonwealth have not evaluated the data as altered and incorporated within the manuscript, and therefore give no warranty regarding its accuracy, completeness, currency or suitability for any particular purpose. 6. REFERENCES
measurements of Kilauea Volcano, Hawaii, from SIR-C radar interferometry," Journal of Geophysical Research, vol. 101, pp. 23109 - 23125, 1996. [2] R. Wright, L. P. Flynn, H. Garbeil, A. J. L. Harris, and E. Pilger, "MODVOLC: near-real-time thermal monitoring of global volcanism," Journal of Volcanology and Geothermal Research, vol. 135, pp. 29-49, 2004. [3] C. Oppenheimer, "Infrared surveillance of crater lakes using satellite data," Journal of Volcanology and Geothermal Research, vol. 55, pp. 117 - 128, 1993. [4] D. Pieri and M. Abrams, "ASTER watches the world's volcanoes: a new paradigm for volcanological observations from orbit," Journal of Volcanology and Geothermal Research, vol. 135, pp. 13 - 28, 2004. [5] G. J. S. Bluth, T. J. Casdevall, C. C. Schnetzler, S. D. Doiron, L. S. Walter, A. J. Krueger, and M. Badruddin, "Evaluation of sulfur dioxide emissions from explosive volcanism: the 1982-1983 eruptions of Galunggung, Java, Indonesia," Journal of Volcanology and Geothermal Research, vol. 63, pp. 243 - 256, 1994. [6] K. E. Joyce, S. Samonsov, and G. Jolly, "Satellite remote sensing of volcanic activity in New Zealand," presented at 2nd Workshop on the use of remote sensing techniques for monitoring volcanoes and seismogenic areas, Naples, Italy, 2008. [7] A. Dekker, V. E. A. Brando, J.M, N. Pinnel, T. Kutser, E. J. Hoogenboom, S. Peters, R. Pasterkamp, R. Vos, C. Olbert, and T. J. M. Malthus, "Chapter 11: Imaging spectrometry of water," in Imaging Spectrometry: Basic principles and prospective applications: Remote Sensing and Digital Image Processing, F. van der Meer and S. M. de Jong, Eds. Boston: Kluwer Academic Publishers, 2001, pp. 307 - 359.
[1] P. A. Rosen, S. Hensley, H. Zebker, F. Webb, and E. Fielding, "Surface deformation and coherence
I - 20 Authorized licensed use limited to: UNIVERSITY OF WESTERN ONTARIO. Downloaded on March 29,2010 at 11:44:24 EDT from IEEE Xplore. Restrictions apply.
