Keywords

1 Introduction

The Chernobyl nuclear disaster report of the International Atomic Energy Agency [1] states that it is academically and socially important to conduct ecological studies regarding the levels and effects of radiation exposure on wild animal populations over several generations. Immediately following the Fukushima Daiichi nuclear power plant disaster, remnants of which are shown in Fig. 1, Ishida (a research collaborator at the University of Tokyo) started conducting regular ecological studies of wild animals in the northern Abukuma Mountains near the Fukushima Daiichi nuclear power plant, where high levels of radiation were detected. Ishida aims to place automatic recording devices at over 500 locations and has been collecting and analyzing vocalizations of target wild animals. The long-term and wide-range monitoring is required to understand the effects of the nuclear accident because he yet has little evidence of the direct effects of radioactivity on wildlife at Fukushima [2]. For monitoring such species, counting the recorded calls of animals is often an effective method because acoustic communication is used by various types of animals, including mammals, birds, amphibians, fish and insects [3, 4]. In particular, as well as using visual counts, the method is commonly used to investigate birds and amphibians [5]. An observer manually listens to calls and identifies the species from the recorded data. Therefore, the method has a disadvantage as the result is affected by the concentration of the observer to identify the species of calls.

Fig. 1.
figure 1

Wildlife near the Fukushima Daiichi Nuclear Power Plant [6].

Full size image

This paper will discuss the design, development and evaluation of a system to tackle the previously mentioned issue. First, on the basis of the related studies, the author designs a new experiment system based on related observation methodologies. In addition, the spatial-temporal process of nonhuman-centric interactions of users is evaluated by quantitative content analysis. Finally, on the basis of the experimental results, the overall findings are discussed, including an answer to the possible applications of the system. The structure of this paper is as follows. Section 2 details background, Sect. 3 presents the proposed method, Sect. 4 details the result and Sect. 5 presents the discussions in detail. Section 6 summarizes future directions. Section 7 offers conclusions.

2 Background

As introduced above, in ecological studies, it is desirable to develop a technology that most effectively supports a study with minimal resources. More specifically, we aim to establish a long-term continuously operating ubiquitous system that delivers, in real time, environmental information, such as sound. Researchers worldwide are conducting ecological studies by recording and analyzing the spatial information of wild animal vocalizations [7]. Furthermore, ecological studies of the environment close to urban areas are being conducted using cell phones [8]; however, it is difficult to confirm the behavior of wild animals using cell phones. To record vocalizations of wild animals whose behaviors are difficult to predict, it is necessary to continuously operate a monitoring system. As it is difficult to conduct system maintenance due to the severe environmental conditions of wild animal habitats (e.g., when out of infrastructure service areas and in high-temperature and high-humidity environments), system redundancy becomes crucial.

In a previous study, we have researched and developed a proprietary system that delivers and records environmental sounds in real-time [9]. This system has been almost continuously operational on Iriomote Island in Okinawa since 1996, using equipment such as that shown in Fig. 2. To date, the basic research on Iriomote Island has been expanded to include 18 domestic and international sites, including Los Angeles and the San Francisco Bay area in the United States, Sanshiro Pond at the Hongo Campus of the University of Tokyo in Japan, Kyoto Shokokuji Mizuharu in Suikinkutsu, Mumbai City in India, the Antarctica Syowa Station (under construction), Morotsuka Village in Miyazaki and Fukushima University in Japan. We have worked with project collaborators and have introduced our system to the University of Tokyo Chichibu Forest, Otsuchi in Iwate, Shinshu University, the University of Tokyo Fuji Forest, the University of Tokyo Hokkaido Forest (under construction), the University of Tokyo International Coastal Research Center and on an uninhabited island in Iwate (also under construction). We have also been conducting research on ubiquitous interfaces for ecology studies of wild animals since April 1997. However, there is a disadvantage that the result would be affected by the concentration of the observer to identify the species of calls from the streamed live sound data.

Fig. 2.
figure 2

Ubiquitous systems for the real-time delivery of environmental sounds in long-term and continuous operation.

Full size image

3 Proposed Method

On the basis of the mentioned problem, this study has attempted to understand the processes of nonhuman-centric interaction between users and remote uninhabited environments through the use of information technologies and reveal new knowledge regarding such interactivity [9] by observing people’s reaction to the interactions for ecological studies as in Fig. 3. In doing so, this study hopes to discover spatial-temporal processes of our imagination mechanism. Such a discovery would help us design a system that leverages the boundary of the real and virtual worlds by engaging a large number of participants to perform a specific Internet-based scientific task without knowing its purpose for ecological studies. If users were found to pay attention to the bio-acoustic information present in natural sounds, it would help us design a system that leverages the boundary of the real and virtual worlds by engaging a large number of participants to perform a specific Internet-based scientific task without knowing its purpose.

Fig. 3.
figure 3

Query presence of proposed method (right)

Full size image

On the basis of this requirement and construction and operating experiences from previous studies [10], a new observation tool for monitoring and processing user feedback was created in the northern part of Iriomote Island (24°20′N, 123°55′E) in the southern Ryukyu Islands, Japan, seen in Fig. 4 (left). The system allows users to listen to live stream audio from a location near a tree in an uninhabited forest as in Fig. 4 (right). The system also collects audio feedback from users in real time. Furthermore, the system can separate individual words from the collected feedback data, categorize them and count their usage frequency. Figure 5 (left) shows a diagram of the system. A weatherproof microphone placed in the ecosystem collects environmental sounds 24 h a day, 365 days a year.

Fig. 4.
figure 4

(left) Map of cellular service cover (red) on Iriomote Island, Japan. Service coverage areas are highlighted in red by NTT Docomo. (right) Audio digitized system: A remote microphone system installed on Iriomote Island. (Color figure online)

Full size image
Fig. 5.
figure 5

Audio digitized system: a remote microphone system installed on Iriomote Island

Full size image

The microphones are attached in pairs to trees with elastic bands, as shown in Fig. 4 (right). Next, the processed audio signal from the microphones is sent to the encoding/recording system and encoded to MP3 live stream and WAVE sound format files. The MP3 live stream is sent to a server in the data archive system for direct streaming over the Internet and is played on various MP3-based audio software formats at different locations simultaneously around the world. The storage/analysis system stores the WAVE sound files to its hard disks. Users can listen to the MP3 live stream through the Sound Bum interface, which uses the Apache web server with PHP and a PostgreSQL database.

Each comment is logged with a user defined name and location and is saved with a time stamp. The collected comment data are processed and analyzed by the KH Coder [11] to determine word frequency. On the basis of these comments, user feedback regarding the bioacoustic information is collected, and it is possible to evaluate which sounds receive the most attention. The system has been presented on the Sound Bum [12] Project web site since 1997 as in Fig. 5 (right). Since 1997, real-time environmental sounds from Iriomote Island’s subtropical forests have been monitored by networked microphones and transmitted via the website as “Live-Sounds from Iriomote Island” 24 h a day, 365 days a year. The real-time streaming system has been upgraded several times over the years to improve long-term stability under unmanned operating conditions (Fig. 6).

Fig. 6.
figure 6

Web Interface for live sounds from Iriomote Island

Full size image

4 Result

The field trial was conducted 24 h per day for 365 days from April 1, 2000 to April 1, 2010 in the northern part of Iriomote Island (24°20′N, 123°55′E) in the southern Ryukyu Islands, Japan. Electric power and information infrastructures required for monitoring the sound of a tree falling are nonexistent in this location. Figure 2 (left) shows the limited distribution of cellular service in the area. The climate is primarily seasonal with summer occurring from June to September and winter occurring from December to March. The average monthly temperature is highest in July (28.3 °C) and lowest in January (18.0 °C). The average annual rainfall is 2342 mm (Iriomote Meteorological Station). Iriomote Island consists of highly folded mountains, and its highest peak (Mt. Komi) is 469 m above sea level. Most of Iriomote Island is covered by a subtropical evergreen broadleaved forest (83 % of the island). During the 11 years of observation period, feedback from user reactions was collected and archived, and a total of 2831 comments were recorded through the WEB interface as in Fig. 7. To analyze the semantic content of the comments, a kh_coder was employed to process the data and subtract “nouns”, as shown in Table 1.

Fig. 7.
figure 7

Web interface used to collect feedback from users (in Japanese)

Full size image
Table 1. Frequency of words in feedback comments
Full size table

5 Discussion

We successfully extracted and analyzed the comments posted by users. The phrase that appeared most frequently was “animal voices.” No specific names of animals plants or words associated with the sound of falling trees or of wind or ocean tides appeared. From the recorded data, we identified the top 18 frequently appearing noun phrases from the comments, as shown in Table 1. The most frequent was “Singing Voice,” which appears in 7.44 % (89 instances) of the 2831 comments [13].

From the standpoint of informatics, the amount and duration of bioacoustic information contained in sounds that could be perceived as a “Singing Voice,” such as bird and animal calls, which comprise a tiny fraction of the total soundscape, are small when compared to other continuously present environmental sounds, such as wind and running water. This result indicates that listeners do not pay attention to the dominant component of the soundscape. Some listeners tried to identify callers by their specific names: “Frog” (42, 3.51 %), “Bell Cricket” (25, 2.09 %), “Cicada” (19, 1.559 %), “Brown-eared Bulbul,” “Crow” (10, 0.84 %), and “Animal” (13, 1.09 %). However, results from our previous ecological study indicate that at least six types of frogs inhabited the island: Rana limnocharis, Rana psaltes, Rana supranarina, Chirixalus eiffingeri, Rhacophorus owstoni and Microhyla ornata [13]. The recognition gap indicated by comparing this study and the previous study indicates that the vast majority of users only perform sweeping recognition without paying attention to the detail of the sounds. Landscape related comments, such as “Typhoon” (33, 2.76 %) and “Airplane” (13, 1.09 %) were also identified. Also, sounds that were described as “A Sign of Presence” (10, 0.84 %) were detected by the users [13].

The type of nonhuman-centric interaction described in this paper is reflected in the semiotic theories of Jakob von Uexküll [14]. Von Uexküll established the concept of Umwelt, from the German word meaning “environment” or “surrounding world,” and suggested that all animals, from the simplest to the most complex, fit into unique worlds with equal completeness. “Singing Voice” corresponds to a single animal and a simple world; “A sign of presence” corresponds to a complex, well-articulated human world. Thus, a user’s reaction to the sounds can be explained by the Umwelt theory. When users listen to live sounds from the invisible world through the Internet, they tend to pay attention to both a simple singing voice and a complex presence.

The author and his associates initially introduced the concept of HCBI [15] at HCI venues focused on environmental sustainability from 2009. The theory, method and evaluation of human and wildlife interaction were not discussed in detail because the research was not sufficiently well developed. However, the future direction of this study has been suggested by several researchers. The author and his associates, ecology scientists, have developed a new type of bird census method coined “audio census,” using the same live stream audio system and social media (Internet Relay Chat and Twitter) [16]. A total of 36 bird species were recorded by several ornithologists in separate places using the audio census method during the 3-month-long breeding seasons of the 3-year study period. We showed that the Live Sound System and social media could contribute to raising public interest in and realizing the auditory real-time experience of a remote natural forest.

6 Future Direction

During the recovery from the Fukushima Nuclear Power Plant disaster, the journal Nature [17] pointed out the importance of ecological studies from the very beginning of a disaster. Ishida (research collaborator) of the University of Tokyo, who has conducted ecological studies since immediately after the disaster, has stated that it would be extremely difficult to continue conducting ecological studies on an ongoing basis [2]. On the basis of this requirement, the development of a new monitoring tool is in progress by the author in the “difficult-to-return zone” [18], a zone whose annual radiation exposure level exceeds 50 millisieverts around the Fukushima Daiichi Nuclear Power Plant.

7 Conclusion

This paper presents an evaluation of nonhuman-centric and spatial-temporal interactions to observe people’s reaction to interactions through ecological studies. The system operated 24 h a day, 365 days a year from April 1, 2000 to April 1, 2010 in the northern part of Iriomote Island (24°20′N, 123°55′E) in the southern Ryukyu Islands, Japan. In doing so, this study hopes to discover spatial-temporal processes of our imagination mechanism. Such a discovery would help us design a system that leverages the boundary of the real and virtual worlds by engaging a large number of participants to perform a specific Internet-based scientific task without knowing its purpose for ecological studies.