Remote Sensing of Environment 99 (2005) 95 – 104 www.elsevier.com/locate/rse

Discolored detection using ASTER reflectance products: A case study of Satsuma-Iwojima, Japan

Minoru Urai a,*, Shoichi Machida b aGeological Survey of Japan, National Institute of Advanced Industrial Science and Technology, Central 7, Higashi 1-1-1, Tsukuba, Ibaraki 305-8567, Japan bNittetsu Mining Co., Ltd., Marunouchi 2-3-2, Chiyoda, Tokyo 100-8377, Japan

Received 15 August 2004; received in revised form 19 April 2005; accepted 28 April 2005

Abstract

Discolored seawater caused by volcanic activity can be detected with satellite remote sensing at some submarine volcanoes and volcanoes near coastlines. We demonstrate that two kinds of discolored can be discriminated by their reflectance patterns using ASTER Reflectance Products at Satsuma-Iwojima, Japan. White to green seawater that is due to hydrous oxides formed when Al- and Fe-rich acidic waters mix with seawater shows high, medium and low reflectances at band 1 (central wavelength: 0.56 Am), band 2 (central wavelength: 0.66 Am), and band 3 (central wavelength: 0.81 Am), respectively. On the other hand, reddish-brown seawater that is due to the outflow of neutral- pH, iron-bicarbonate water shows high reflectance at band 2, medium reflectance at band 1 and low reflectance at band 3. The chemical composition of discolored seawater, which is one of the characteristics of volcanic activity, can be estimated from remote-sensing data. The ASTER Reflectance Product is a standard validated product generated by the ASTER Ground Data System using an atmospheric correction. D 2005 Elsevier Inc. All rights reserved.

Keywords: Discolored seawater; ASTER; Satsuma-Iwojima; Reflectance spectra; Submarine ; Active volcano; Remote sensing

1. Introduction ored seawater at Hukutoku-Oka-no-Ba and Hukuzin Kaizan using Landsat MSS images. Discolored seawater is one of the indicators of volcanic Satsuma-Iwojima is a volcanic island located 90 km activity at submarine volcanoes or around volcanic islands south of Kagoshima City, Kyushu, Japan. This island is a (e.g., Nogami et al., 1993; Ossaka et al., 2000). Ossaka good target for discolored seawater monitoring using (1975) showed that at the Nishinoshima submarine volcano, satellite remote sensing because two different shades of Japan, discolored seawater is caused by very fine SiO2 – discolored seawater have been observed continuously Al2O3–Fe2O3–H2O particles that are precipitated from the around the island. In this paper, we demonstrate discolored mixture of hot-spring water and seawater. The color of the seawater detection at Satsuma-Iwojima, Japan, using the seawater changes from milky white to reddish-brown ASTER Reflectance Products. corresponding to the Fe2O3 content of the seawater (Ossaka, 1975). Ossaka et al. (1996) found that the color of the seawater changed from milky white to yellow, yellow- 2. Geologic setting of Satsuma-Iwojima brown, or yellow-green depending on the Fe and Al contents that increase with volcanic activity. Because Satsuma-Iwojima is located on the northwest rim of the remotely sensed images are sensitive to these color changes, Kikai , which extends about 20 km east–west and they used to evaluate the nature and intensity of volcanic 17 km north–south (Fig. 1). Most of the Kikai Caldera is activity. Indeed, Tsuchide and Ohtani (1983) found discol- covered by the (Ono et al., 1982). Shin-Iwojima, which is located 2 km east of Satsuma-Iwojima, is a new island * Corresponding author. Tel.: +81 29 861 3843; fax: +81 29 861 3788. formed by volcanic activity during 1934–1935. There are E-mail address: [email protected] (M. Urai). two post-caldera cones in Satsuma-Iwojima. One is the

0034-4257/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.rse.2005.04.028 96 M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104

Kagoshima Heikenojo

Satsuma-Iwojima 31N

Kikai Caldera Tanegashima Ketsunohama

130E 131E 4 Iwo dake 5 Inamuradake

Yunotaki Higashi Nagahama 2 1 Isomatsuzaki Akayu 3

0 1 2km

Fig. 1. Topographic map of the study area. Elevation in meters. Numbered dots are the spectral measurement and seawater sampling points. The thick broken lines are the rim of the Kikai Caldera. Rectangles surrounded by thin broken lines are areas where average reflectance values are calculated as shown in 2.

Inamuradake volcano, which is the only basaltic post- combined characteristics of water, vegetation, and soil. It is caldera cone presently known in the Kikai Caldera. The thus necessary to remove irradiance and atmospheric effects other is the Iwodake volcano, which is a steep-sided from the integrated observed radiance to extract surface about 700 m high with a large crater at its reflectance. In the past, ground data processors typically summit. The volcano is a pile of thick flows and bedded provided only observed total radiance products. However, coarse volcanic rocks, mostly talus deposits probably the ASTER Ground Data System provides a surface formed at the front of the lava flow or . Many reflectance product as one of its Standard Products. The fumaroles are active around the summit crater and flank Standard Products are validated by field measurements (e.g., area. Some fumaroles have very high temperatures, over 800 Thome et al., 1999) and are especially convenient for users -C(Shinohara et al., 2002). Silicic rocks formed by not familiar with remote sensing. fumarolic alteration are found around the summit crater The ASTER instrument, which was launched on the and other fumarole areas. Two different discrete hues of Terra platform on December 1999, is a high spatial discolored seawater related to volcanic activity have been resolution imaging spectro-radiometer having a total of 14 observed continuously around the island. One is reddish- bands in Visible to Near-Infrared (VNIR), ShortWave- brown and is due to the outflow of neutral-pH, iron- Infrared (SWIR) and Thermal-Infrared (TIR) wavelengths. bicarbonate water (Hedenquist et al., 1994; Kamada, 1964) The ground resolutions of the VNIR, SWIR and TIR images around the seacoast of Inamuradake (Fig. 2a). The other is are 15, 30 and 90 m, respectively. Detailed descriptions of white to green and is due to hydrous oxides formed when the ASTER are found in Fujisada et al. (1998) and Yamaguchi Al- and Fe-rich acidic waters mix with seawater (Nogami et et al. (1998). al., 1993) around the seacoast of Iwodake (Fig. 2b). Kamada ASTER Level 2B05V data present surface reflectances (1964) estimated that 10–20Â106 tons of acidic hot-spring generated from ASTER Level 1B radiance data measured on water per year are discharged from Satsuma-Iwojima. orbit, and an atmospheric profile generated by a radiative- transfer code (Thome et al., 1999). U.S. National Center for Environmental Prediction (NCEP) Global Data Assimilation 3. Overview of the ASTER Level 2B05V—Surface System (GDAS) data (Kalnay et al., 1990) are used as the Reflectance Product atmospheric profiles for the current 2B05V products. The current version of the processing software is 2.5. The 2B05V Radiance observed at a sensor is a function of surface consists of Surface Reflectance and Quality Assessment reflectance, irradiance from the sun, and atmospheric (QA) data planes plus metadata. The metadata include an conditions. Surface reflectance itself further represents the inventory of the observation conditions and data processing M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104 97

different colors. One appears as a ‘‘greenish area’’ (as represented in ASTER images) that is seen at Nagahama, Isomatsuzaki and Akayu. This seawater appears visually as reddish-brown and was attributed to the outflow of neutral- pH, iron-bicarbonate water by Kamada (1964). [Because ASTER does not have a band corresponding to the blue spectral range (0.45–0.52 Am), the colors of ASTER image are different from colors detected by the naked eye.] The other is a ‘‘bluish area’’ that is seen at Higashi, Yunotaki, Ketsunohama and Heikenojo, which corresponds to the naked eye white to green seawaters attributed to hydrous oxides formed when the Al- and Fe-rich acidic waters mix with seawater (Kamada, 1964; Nogami et al., 1993). The discolored seawater drifts with and currents, generally within 3 km of the seacoast. Fig. 4 shows the typical reflectance spectra of discolored and clear seawater around Satsuma-Iwojima. The reflec- tance spectrum represents a single pixel of the L2B05V image that is in the densest discolored seawater of each area, and not the same location at each observation time. Reflectance of the discolored seawater tends to be high when the reflectance of the clear sea is high. The reflectance of clear sea is proportional to the wind speeds (Table 1 and Fig. 5) at Tanegashima Meteorological Observatory, which is 70 km east of Satsuma-Iwojima. Winds may cause sea waves and associated high-reflec- tance spectra in clear sea. We assume that intrinsic reflectance of clear sea is constant with time in this area, and sea waves cause this apparent change of the Fig. 2. Photo of the discolored seawater. (a) Reddish-brown seawater at reflectance. We subtract the reflectance of clear sea from Nagahama viewed from the west, (b) white to green seawater at Heikenojo reflectance of discolored seawater of the same observation viewed from the Iwodake summit. date to remove the apparent spectral change and obtain corrected reflectance as shown in Fig. 6.Corrected reflectance of Heikenojo and Yunotaki (‘‘bluish colored’’ parameters. The Surface Reflectance data consist of three seawater) can be distinguished from other seawater in Fig. bands corresponding to the VNIR bands 1 to 3 of the Level 6 for its high reflectance (>11%) at band 1 (central 1B radiance data. The spatial resolution of the Surface wavelength: 0.56 Am) and medium reflectance at band 2 Reflectance Product is 15 m, the same as VNIR of Level 1B. (central wavelength: 0.66 Am), and low reflectance at band The QA data are intended to indicate data quality and 3 (central wavelength: 0.81 Am). Corrected reflectances of accuracy of the Surface Reflectance Product on a pixel-by- Higashi and Ketsunohama are lower than those of pixel basis. However, the current version of the products Heikenojo and Yunotaki, but they show a similar trend as does not have valid data quality information. those of Heikenojo and Yunotaki. These areas are seen in ‘‘bluish color’’ in Fig. 3 because band 1 is assigned to blue in Fig. 3. However, absolute reflectance values of every 4. Satellite image analyses observation are not the same for the same discolored sea areas because the discolored seawater density changes due Five largely cloud-free images were acquired at to the , sea current and amount of hot water supplied by Satsuma-Iwojima from August 2000 to May 2003 by volcanoes. Corrected reflectance of Nagahama and Akayu ASTER. Fig. 3 shows the five ASTER surface reflectance can be distinguished from others by their high, medium and images (L2B05V) of Satsuma-Iwojima displayed in false low reflectances at band 2, band 1 and band 3, respectively color with bands 1, 2 and 3 represented by blue, green and (Fig. 6). These areas are seen in ‘‘greenish’’ color in Fig. 3 red, respectively. Discolored seawater is seen at the because band 2 is assigned to green in Fig. 3. However, southern and eastern coastlines from Nagahama to this feature is not always seen at Isomatuszaki because Heikenojo. No discolored seawater is found at the north- corrected reflectance of band 2 is more declined than band western coastline, which is outside of the Kikai Caldera 1 in weaker discolored seawater. No discolored seawater rim. The discolored seawaters can be divided into two was detected at Isomatsuzaki, Akayu, Higashi and Yunotaki 98 M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104

Fig. 3. ASTER reflectance image (2B05V) around Satsuma-Iwojima. ASTER bands 1, 2 and 3 are assigned to blue, green and red, respectively. on August 23, 2000, even in the cloud-free image. proportional to the product of band 1 reflectance by Generally, an area of discolored seawater expands on the discolored seawater area (right most column of Table 2). ebb tide and reduces on the flood tide. The ASTER image of August 23, 2000 was taken on the flood tide and other images were taken on ebb tides (Table 1). No discolored 5. Reflectance spectral measurements in the field seawater was detected at Ketsunohama on March 10, 2001, because of cloud cover. Corrected reflectance can be used Reflectance spectral measurements of discolored sea- to map the extent and flow of discolored seawater as shown water were conducted using a small boat at four in Fig. 7. Table 2 shows average corrected reflectance of discolored sea areas and a clear sea area around each discolored seawater area. The corrected reflectance Satsuma-Iwojima shown in Fig. 1 and listed in Table 3 values of the five ASTER observations are averaged within at 14:00–15:40 local time on November 18, 2000. The the rectangular areas surrounded by broken lines indicated weather was fine with few clouds. Spectral radiance from in Fig. 1. All discolored seawater has similar reflectance discolored seawater was measured from 0.35 to 1.0 Am pattern that is high reflectance at band 1, medium using the Personal Spectrometer II manufactured by reflectance at band 2 and low reflectance at band 3 in Analytical Spectral Devices, Inc. (Table 4). Spectral Table 2. It is because corrected reflectance of band 2 is radiance was collected by an attached foreoptic, which more declined than band 1 in weaker discolored seawater. has 18- field of view, pointed straight down about several Since the corrected reflectance of band 1 at Heikenojo is tens of centimeters above the sea surface. Just after the the highest, we surmise that it represents that high measurement of discolored seawater, radiance from a discharge and low diffusion rates of Al- and Fe-rich acidic Spectralon pure white reference target (Jackson et al., waters at Heikenojo. Ketsunohama is the largest source of 1992) was collected on the deck of the boat. Spectral Al- and Fe-rich acidic water if the acidic water volume is reflectance was calculated from the seawater and refer- M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104 99

35 Nagahama 35 Nagahama Isomatsuzaki Isomatsuzaki Akayu Akayu 30 Higashi 30 Higashi Yunotaki Yunotaki 25 Ketsunohama 25 Ketsunohama Heikenojo Heikenojo Clean sea Clean sea 20 20

15 15 Reflectance (%) Reflectance (%) 10 10

5 5

0 0 0.5 0.6 0.7 0.8 0.9 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) Wavelength (micrometer) 14AUG2000 19MAR2001

35 Nagahama 35 Nagahama Isomatsuzaki Ketsunohama Akayu 30 30 Higashi Yunotaki Heikenojo 25 25 Ketsunohama Heikenojo Clean sea Clean sea 20 20

15 15

10 Reflectance (%) 10

5 5

0 0 0.5 0.6 0.7 0.8 0.9 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) Wavelength (micrometer) 23AUG2000 12MAY2003

35 Nagahama Isomatsuzaki Akayu 30 Higashi Yunotaki 25 Heikenojo Clean sea 20

15

Reflectance (%)10 Reflectance (%)

5

0 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) 10MAR2001

Fig. 4. Reflectance spectra of discolored and clear seawaters derived from ASTER. ence radiances by software provided by Analytical seawater areas, have a broad peak from 0.58 Amto Spectral Devices, Inc. The spectral reflectances of the 0.66 Am. On the other hand, reflectances of Higashi and discolored and clear sea are shown in Fig. 8. Reflectances Ketsunohama, which are white to green seawater areas, of Nagahama and Akayu, which are reddish-brown have a peak at 0.56 Am and a sharp drop at 0.60 Am. 100 M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104

These spectral reflectance data were converted to the 12 reflectance values corresponding to the ASTER bands 1 Band 1 to 3 using the spectral response profiles of ASTER 10 Band 2 (Fujisada et al., 1998). Corrected reflectance values were Band 3 calculated from the reflectance values of clear and 8 discolored seawater, as described above (‘‘satellite image analyses’’ section) and shown in Fig. 9. Corrected 6 reflectance values of Higashi and Ketsunohama show high, medium and low reflectances at band 1, band 2 and

Reflectance (%) 4 band 3, respectively. On the other hand, corrected reflectance values of Nagahama and Akayu show high reflectance at band 2, medium reflectance at band 1 and 2 low reflectance at band 3. These reflectance features are 0 the same as the satellite image analysis (Fig. 6). The 02468 water depths of the measuring points are less than 10 m Wind speed (m/s) except for the clear sea (No. 5). Reflection from the sea bottom may affect the spectral measurement. However, Fig. 5. Reflectance of the clear sea around Satsuma-Iwojima and wind sea bottoms were not visible to the naked eye at any of speed at Tanegashima. the measuring points.

ber 18, 2000, when the reflectance spectral measurements 6. Chemical composition of the discolored seawater were conducted. The Si, Fe and Al concentration were determined from the seawater samples using the method Ossaka (1975) showed that discoloration of seawater developed by Ossaka et al. (1996) as following: (1) 100-ml was caused by very fine SiO2 –Al2O3 –Fe2O3–H2O pre- sample was transferred to a beaker. (2) The sampling bottle cipitation. Since the concentration of the suspended was washed with redistilled water, and the washings were precipitates is typically low, large samples of the discol- added to the sample. (3) The solution including the ored seawater are necessary for chemical analysis. Nogami precipitate was warmed to 80 -C on a hot plate for 24 h et al. (1993) developed a new method for estimating after the addition of 1-ml of 6 N HCl in order to dissolve the chemical compositions of the precipitates using the differ- precipitate. (4) It was transferred to a 200-ml measuring ences of the chemical composition between two splits of flask and diluted to the mark with redistilled water. (5) seawater samples that were treated differently. One method The Si, Fe and Al concentrations were determined by an is to dissolve the precipitates in HCl. In the other ICP emission spectrometry. Table 3 shows the chemical treatment, the precipitates are removed with a filter paper. composition of the discolored and clear seawaters. In this study, Nogami et al. (1993) found that most of the Fe Chemical compositions of the seawaters sampled at and Al are concentrated in the precipitations in neutral-pH Nagahama and Akayu show high Fe compared with Al. discolored seawater. In this study, chemical analysis is On the other hand, chemical compositions of seawaters conducted using the seawater in which precipitations are sampled at Higashi and Ketsunohama show high Al dissolved with HCl to estimate chemical compositions of compared with Fe. These results agree with the existing the precipitates because this method needs only a small chemical analysis (Kamada, 1964; Nogami et al., 1993). amount of sweater sample and originally dissolved Fe and Al are negligible. A set of discolored and clear seawater samples was collected around Satsuma-Iwojima on Novem- 7. Discussion

Some submarine volcanoes and volcanic islands are Table 1 located in remote areas that are not easy to access. Seismic Wind data at Tanegashima and tidal data at Nagahama on the ASTER observation is the easiest and most useful method for observation dates volcano monitoring on land, and difficult when used in the a Date Wind speed Wind Time of Time of ocean. Only a few submarine volcanoes are monitored (m/s) directiona flood tideb ebb tideb periodically, visually or photographically, by aircraft. Most 14 August 2000 6.4 W 6:15, 19:24 0:43, 12:52 are not be monitored. Eruptions of these volcanoes may be 23 August 2000 1.7 NNW 13:27 6:44, 19:00 10 March 2001 3.9 W 7:44, 19:43 1:28, 13:44 reported from ships and aircraft, but only by chance. For 19 March 2001 6.0 WNW 4:44, 15:26 10:28, 22:22 these volcanoes, fluxes of discolored seawater may be the 12 May 2003 5.0 NE 4:22, 16:28 10:29, 22:38 best indicator of activity, and satellite remote sensing is the a 11 AM at Tanegashima Metrological Observatory. best available tool for detecting these fluxes. Since most b Flood and ebb tide times at Nagahama. patches of discolored seawater are small (less than a few M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104 101

25 Nagahama 25 Nagahama Isomatsuzaki Isomatsuzaki Akayu Akayu 20 Higashi 20 Higashi Yunotaki Yunotaki Ketsunohama Ketsunohama 15 Heikenojo 15 Heikenojo

10 10 Reflectance (%) Reflectance (%)

5 5

0 0 0.5 0.6 0.7 0.8 0.9 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) Wavelength (micrometer) 14AUG2000 19MAR2001

25 Nagahama 25 Nagahama Isomatsuzaki Akayu 20 Ketsunohama 20 Higashi Yunotaki Ketsunohama 15 Heikenojo 15 Heikenojo

10 10 Reflectance (%)

5 5

0 0 0.5 0.6 0.7 0.8 0.9 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) Wavelength (micrometer) 23AUG2000 12MAY2003

Nagahama 25 Isomatsuzaki Akayu Higashi 20 Yunotaki Heikenojo 15

10 Reflectance (%) Reflectance (%)

5

0 0.5 0.6 0.7 0.8 0.9 Wavelength (micrometer) 10MAR2001

Fig. 6. Corrected reflectance of discolored seawater derived from ASTER. kilometers) and have irregular and elongate shapes, a high- because the peak reflectance depends on the contaminants in resolution imaging sensor is required to observe them. the seawater. In addition, the near-infrared band can be Green (0.55 Am), red (0.65 Am) and near-infrared (¨0.8 useful in discriminating seawater from clouds and land in Am) bands are needed for discolored seawater monitoring the images. 102 M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104

Table 4 Specifications of Personal Spectrometer II Spectral range 350–1050 nm 1.4 nm Spectral resolution 3 nm at 750 nm Signal-to-noise 200:1 at 400 nm, 1500:1 at 700 nm, 200:1 at 1000 nm Wavelength accuracy 1 nm Integration time 44 ms to 5.6 s Weight 3.3 kg Battery life 3 h with fully charged battery pack

ing of discolored seawater. Furthermore, its blue band (TM Band 1), which ASTER lacks, is especially useful for ocean studies. However, most Landsat-type sensors (including ASTER and SPOT HRV as well as TM) are suitable for monitoring discolored seawater near submarine volcanoes.

Nagahama 20 Akayu Fig. 7. Corrected reflectance of ASTER band 1 near Heikenojo. Reflectance Higashi in percent. Land area is masked using ASTER band 4. The image size is 1.3 Ketsunohama by 1.2 km. 15 Clean sea Landsat TM, which one of the most widely used of the remote-sensing imaging systems, has a long history of data acquisition and is therefore suitable for long-term monitor- 10 Reflectance (%) Table 2 5 Average reflectance and extension of discolored seawaters Location Averaged reflectancea (%) Averaged Band 1 reflectance extensionb by extension Band 1 Band 2 Band 3 2 2 0 (km ) (% km ) 0.4 0.5 0.6 0.7 0.8 0.9 Nagahama 4.6 4.0 2.6 0.20 0.91 Wavelength (micrometer) Isomatsuzaki 3.5 2.3 1.8 0.33 1.15 Akayuc 2.6 2.1 1.3 0.03 0.08 Fig. 8. Reflectance spectra of discolored and clear seawaters collected in the Higashic 3.5 1.9 1.7 0.05 0.20 field. Yunotakic 4.6 2.7 2.0 0.08 0.37 Ketsunohamad 5.4 3.0 0.9 1.10 5.88 25 Heikenojo 7.4 3.7 2.5 0.37 2.77 Nagahama a Reflectance values are averaged within the area and five ASTER Akayu observations. 20 b Extension is averaged area within the five ASTER observations that has Higashi a band 1 reflectance value more than 2% within the rectangle indicated in Ketsunohama Fig. 1. c ASTER observation on August 23, 2000 was excluded because 15 discolored seawater was not observed. d ASTER observations on March 10, 2001 and March 19, 2001 were excluded because of clouds and volcanic plume. 10 Reflectance (%) Table 3 Spectral measurement, seawater sampling points and chemical composi- tions of the seawater 5 No. Spectral measurement and seawater Si Fe Al sampling points (WGS84) (mg/l) (mg/l) (mg/l) 1 Nagahama (30-46V48WN, 130-16V43WE) 2.8 0.6 <0.1 - - 0 2 Akayu (30 46V47WN, 130 17V31WE) 2.4 0.4 <0.1 0.5 0.6 0.7 0.8 0.9 3 Higashi (30-46V52WN, 130-17V50WE) 6.2 4.4 15.6 4 Ketsunohama (30-47V40WE, 130-19V20WE) 2.6 0.5 1.1 Wavelength (micrometer) - V W - V W 5 Clear sea (30 47 25 N, 130 19 54 E) 2.2 <0.1 <0.1 Fig. 9. Corrected reflectance of discolored seawaters collected in the field Sampling positions are measured by a portable GPS receiver. and resampled to the ASTER bands. M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104 103

Table 5 Fe. It is possible to estimate the chemical composition of Summary of the discolored seawater characteristics around Satsuma- discolored seawater from a corrected ASTER reflectance. Iwojima Naked-eye observations, satellite image analysis, reflectance Location Nagahama Higashi Yunotaki spectral measurements in the field and chemical composi- Isomatsuzaki Akayu Ketsunohama Heikenojo tion analysis of the discolored seawater are summarized in Type of discolored Iron-bicabonate type Hydrate oxide type Table 5. seawater Naked eye Reddish brown color White to green color observation Acknowledgements Corrected ASTER Band 1: medium Band 1: high Band 2: reflectance Band 2: high Band 3: medium Band 3: low low The authors are grateful to Drs. Motoaki Kishino, Reflectance in the A peak at 0.56 Am A broad peak from Hiroyuki Fujisada and Alan Gillespie with the ASTER fielda and sharp drop at 0.58 Am to 0.66 Am Science Team, and Hiroshi Shinohara with Geological 0.60 Am Survey of Japan, AIST, and ASTER Science Team members Chemical Fe: high Al/Fe: low Fe: high Al/Fe: high compositionb for their helpful comments on the manuscript. This paper was improved greatly by three anonymous reviewers. The a This feature is sometimes not obvious for Isomatsuzaki, Higashi and Ketsunohama. authors thank the Ministry of Economy, Trade and Industry b No data for Isomatsuzaki, Yunotaki and Heikenojo. (METI) and Remote Sensing Data Analysis Center (ERSDAC) for providing ASTER data. The data source for the wind speed is the Monthly Report published by the The high spatial resolution (15 m) of the ASTER VNIR Japan Metrological Agency. images make them especially useful for near-shore inves- tigations and finding the small, irregular bodies of discol- ored seawater. References

Fujisada, H., Sakuma, F., Ono, A., & Kudoh, M. (1998). Design and 8. Conclusions preflight performance of ASTER Instrument protoflight model. IEEE Transactions on Geoscience and Remote Sensing, 36, 1152–1160. Hedenquist, J. W., Aoki, M., & Shinohara, H. (1994). Flux of volatiles and We can see two kinds of discolored seawater around ore-forming metals from the magmatic-hydrothermal system of Satsuma Satsuma-Iwojima, Japan using ASTER Reflectance Prod- Iwojima volcano. Geology, 22, 585–588. ucts. The reflectance of the clear sea is proportional to the Jackson, R. D., Clarke, T. R., & Moran, M. S. (1992). Bidirectional wind speeds. We introduce corrected reflectance that is the calibration results for 11 spectralon and 16 BaSO4 reference reflectance difference between the reflectance of discolored and clear panels. Remote Sensing of Environment, 40, 231–239. Kalnay, E., Kanamitsu, M., & Baker, W. E. (1990). Global numerical seawater. Corrected reflectances of Heikenojo and Yunotaki, weather prediction at the National Meteorological Center. Bulletin of the which correspond to the white to green seawater that is American Meteorological Society, 71, 1410–1428. colored by the hydrous oxides formed when the Al- and Fe- Kamada, M. (1964). Volcanoes and geothermy of Satsuma-Iwojima, rich acidic waters mix with the seawater, can be distin- Kagoshima prefecture. Journal of the Japan Geothermal Energy guished from other seawater because of its high reflectance Association, 3, 1–23 (in Japanese). Nogami, K., Yoshida, M., & Ossaka, J. (1993). Chemical composition of in band 1, medium reflectance in band 2, and low discolored seawater around Satsuma-Iwojima, Kagoshima, Japan. reflectance in band 3. Corrected reflectances of Nagahama Bulletin of the Volcanological Society of Japan, Series 2, 38, 71–77. and Akayu, which correspond to reddish-brown seawater Ono, K., Soya, T., & Hosono, T. (1982). Geology of the Satsuma-Io-Jima due to the outflow of neutral-pH, iron-bicarbonate water, District. Quad-rangle Series: Scale 1:50,000. (80p.), Tsukuba: Geo- can be distinguished from other seawater by its high, logical Survey of Japan (in Japanese with English abstract). Ossaka, J. (1975). The eruption of Nishinoshima submarine volcano and medium and low reflectances in bands 2, 1, and 3, Geochemical study of the composition of the ejecta and the volcanic respectively. Reflectance spectra of discolored seawater activity. Chemistry Today (Gendai Kagaku), 55, 12–20. were measured in the field at four sites and one clear sea Ossaka, J., Adachi, N., Tsuchide, M., & Nogami, K. (2000). Chemical area around Satsuma-Iwojima for control. Reflectances of compositions of discolored sea water around Izu-Oshima at the 1986 Nagahama and Akayu, with reddish-brown seawater areas, Eruption. Bulletin of the Volcanological Society of Japan, Series 2, 45, A 271–280. have a broad peak from 0.58 to 0.66 m. On the other hand, Ossaka, J., Hirabayashi, J., Nogami, K., Tsuchide, M., & Adachi, N. (1996). reflectances of Higashi and Ketsunohama, with white to Chemical composition of discolored seawater corresponding to activity green seawater, have a peak at 0.56 Am and a sharp drop at of fukutoku-Oka-no-Ba submarine volcano—as an index of submarine 0.60 Am. These reflectance features agree with the satellite volcanism. Bulletin of the Volcanological Society of Japan, Series 2, 41, image analysis. Chemical compositions of seawater sampled 107–113 (in Japanese with English abstract). Shinohara, H., Kazahaya, K., Saito, G., Matsushima, N., & Kawanabe, Y. at Nagahama and Akayu show high Fe compared with Al, (2002). Degassing activity from Iwodake rhyolitic cone, Satsuma- while chemical compositions of seawater sampled at Iwojima volcano, Japan: Formation of a new degassing vent, 1990– Higashi and Ketsunohama show high Al compared with 1999. Earth, Planets and Space, 54, 175–185. 104 M. Urai, S. Machida / Remote Sensing of Environment 99 (2005) 95–104

Thome, K., Biggar, S., & Takashima, T. (1999). Algorithm theoretical basis Yamaguchi, Y., Kahle, A. B., Tsu, H., Kawakami, T., & Pniel, M. (1998). document for ASTER Level 2B1- surface radiance and ASTER Level Overview of Advanced Spaceborne Thermal Emission and Reflection 2B5- surface reflectance. Tucson’ University of Arizona (45p.). Radiometer (ASTER). IEEE Transactions on Geoscience and Remote Tsuchide, M., & Ohtani, Y. (1983). Surveillance of submarine volcanoes— Sensing, 36, 1062–1071. the possibility of manmade satellites and . Bulletin of the Volcanological Society of Japan, Series 2, 28, 375–394 (in Japanese with English abstract).