GeoJournal 52: 203212, 2001. 2001 Kluwer Academic Publishers. Printed in the Netherlands. 203
A GIS-based study toward forecast of suburban forest change
Yasuhiro Suzuki1,, Keiji Kimura2 and Tatsuto Aoki31Aichi prefectural University, Aichi, Japan; 2Tokyo Metropolitan University, Tokyo, Japan: 3HokkaidoUniversity, Hokkaido, Japan; Author for correspondence (Tel: +81-561-64-1111 (ext. 3302); Fax: +81-561-64-1108;E-mail: email@example.com)
Received 31 March 2001; accepted 5 October 2001
Key words: carbon fixation, environmental assessment, forest growth, GIS, global environment, photogrammetry, remotesensing
At a time when the concept of human and environmental symbiosis has taken on much significance, protection of suburbanforests (i.e. forests adjacent to or near developed areas) is a topic that has drawn much attention. Suburban forests have, sinceancient times, been places where people have gathered firewood and cultured trees. As a result, the vegetation of suburbanforests is only partially natural and continues to change as the forms of human activity in and around them changes. Accurateforecasts of how suburban forests will change are, therefore, an important element in the debate over how to protect them.In this study, a suburban forest was analyzed with laser radar sensing, multi-spectrum scanning, digital photogrammetryanalysis, aerial photograph interpretation, and a field survey. Data gathered using these techniques were compiled on a GISto forecast future changes in the forest. Aerial photographs taken over the past 50 years were analyzed to illuminate changesin the forest over that period. Specifically, comparisons of precise Digital Elevation Models (DEMs) measured by usingdigital photogrammetry workstations made it possible to estimate growth in forest height. The possible future conversion ofsuch results to estimates of amounts of carbon dioxide consolidated by forests should be very significant for discussions ofglobal environmental problems.
Introduction - What is required in an epoch of humanand environmental symbiosis
In recent years, the concept of environmental symbiosis hasgained wider acceptance, and greater attention has been paidto the changes that human activity has brought about in thenatural environment. One result of this increased concernwas the 1997 passage of a new law requiring environmentalimpact studies. This law does much to correct the shortcom-ings of previous laws, and among its highlights are that it(1) leaves no room for exceptions to the requirement that en-vironmental assessments be performed before developmentbegins, (2) requires that assessment results be directly re-flected in decisions regarding the granting of approvals, and(3) provides for more opportunities for citizen involvementin environmental assessments (Harashina, 1998). Neverthe-less, this law has received strong criticism for requiringno more than qualitative studies in most cases. Qualita-tive studies often use the words impact will be minimal.With only this to go on, however, verifying the proprietyof forecasts is difficult and comparing forecasts with actualfeedback is problematic. The consequence of this is that itis difficult to improve methods of forecasting environmentaleffects as we go into the future (Matsuda, 1998). New envi-ronmental forecasting methods that have clearly understooddegrees of precision (or imprecision) and that use explicitlogic (hypotheses) are needed.
Under the current state of affairs, the role that a GIS(Geographic Information System) can play is enormous.Environmental data that have been organized using a GIS fa-cilitate both public disclosure and common use. It is obviousthat GIS data simplify the work of understanding the currentstatus of the environment, and people in many quarters havepointed this out. However, it should also be possible to usea GIS as the basis for analyzing not only current data butalso historical data to look at how environmental conditionshave changed over time and at differential change due toparticular environmental elements. This in turn would makeit possible to develop forecasts based on explicit forecastmodels. There is also a strong possibility that environmentalassessments based on detailed GIS data could contribute sig-nificantly to efforts to deal with global environmental issues.Though it may be seen that societys need for such researchhas grown rapidly in recent years, it would be difficult to saythat it is being pursued with sufficient vigor. Though GISgraphics functions have gained strong attention, it seemsto be the case, and regrettably so, that very little attentionhas been paid to the advanced analysis capabilities that GISoffers.
The 2005 World Exposition is scheduled to take placein Aichi, Japan. The theme of the Exposition is BeyondDevelopments Rediscovering Natures Wisdom. However,symbolizing the depth of the issues at hand, how human
Figure 1a. A birds-eye view of the Kaisho Forest in 1995. Based on a 5 m-mesh DSM (Digital Surface Model), which is a kind of DEM (Digital ElevationModel) reflecting treetop elevations. DSM was created with a digital photogrammetry workstation.
Figure 1b. A birds-eye view of the Kaisho Forest in 1949. Based on a 5 m-mesh DSM reflecting treetop elevations. DSM was created with a digitalphotogrammetry workstation.
Figure 2. Change in forest height between 1949 and 1995, based on the 5 m-mesh DSMs.
activity and nature should coexist has become a problemfor the Exposition itself. This problem is not just one ofprotecting nature it requires a detailed analysis of the in-teraction of human activity with nature and the developmentof a method for forecasting future changes in a forest.
To respond to this need, the Research Group fora Regional Environmental GIS, an industry-academia-government joint research group headquartered at the De-partment of Information Science and Technology of AichiPrefectural University and organized by the authors in 1998,began to research the suburban forest known as the KaishoForest. This forest is partially overlapped by the plannedExposition site, which is located in Seto City and NagakuteTown, in Aichi Prefecture. One goal of this research is to es-timate the growth of this suburban forests biomass, togetherwith the amount of carbon dioxide that could be consolidatedby it. Another is the development of a method for estimatingfuture forest growth. This research has only recently got-ten underway and methodologies are still being worked out,but the authors would like to venture a presentation of theconcepts they are working with and the progress they haveachieved to date. It is their intent that in doing so, they mayalso discuss the effectiveness of, and issues related to the useof, GIS in future research of environmental change forecast.
Conceptual description of suburban forest researchintegrating digital photogrammetry analysis, remotesensing, and a GIS
Suburban forests, which have been significantly affected byhuman activities over the ages, exhibit great complexity interms of the species and ages of trees that comprise them.The application of digital photogrammetry analysis (per-formed with a digital photogrammetry workstation (DPW)),aerial photograph interpretation, multi-spectrum scanning(MSS), laser radar sensing, and other remote sensing tech-niques to a suburban forest; the use of a GIS to manage andintegrate the various types of resulting data; and illuminationof the change experienced by the subject suburban forestmake up the primary elements of this research. In morespecific terms, this research involves the following:
Creation of 5m-mesh DEMsFor subject areas with topographies characterized by a highconcentration of hills and valleys, it is impossible to performprecise topographical and runoff analyses using the tradi-tional 50 m-mesh DEM. These types of analyses require ahigh-resolution 510 m-mesh DEM (Oguchi et al., 1999).However, collecting the data necessary for producing DEMsof this resolution requires that the subject land be surveyed.Therefore, except for cases addressing land of very limitedarea, data collection has been practically impossible. An al-ternative approach involves the use of DPWs, developed inrecent years. DPWs use stereo matching to process aerialphotographs and make it possible to automatically survey
Table 1. The members of the Research Group as of April 1, 2001.
Yasuhiro SUZUKI, Nobuhiko HANDA, Hiroki YOSHIOKA (Aichi prefectural University),Takashi OGUCHI, Hiroaki SUGIMORI, Yasushi TANAKA, Keiichi KATSUBE,Shu LIN (University of Tokyo), Keiji KIMURA (Tokyo metropolitan university),Kazukiyo YAMAMOTO, Chisato TAKENAKA, Sachiko KIYONAGA (Nagoya University),Yuuichi ONDA, Tsuyoshi WAKATSUKI, Tetsu ITOKAZU (Tsukuba University),Takashi KUMAMOTO (Okayama University). Daichi NAKAYAMA,Daisake KAWABATA (Kyoto University), Tatsuto AOKI (Hokkaido University),Shigeki SANO, Tatsujiro NOZAWA, Naoki KATSUNO, Masayuki YUHARA, Tetsuro NOMURA,Masahiko HIROSE, Masaru NAKAJIMA (TAMANO Consultants Co.,Ltd.),Naoaki MURATE, Satoshi MIYASAKA, Kimiaki TOKUMURA, Satoru KATO,Tomomi TSUBOI (Nakanihon Air Service Co., Ltd),Nobuyuki TSUTSUI, Yasunari SEKIHARA, Tsuyoshi ITO, Keiji NAGATA, Miha UENO,Toshiro HASHIMOTO, Hisayuki SUGAUCHI, Yuuichi NOMURA (SOKEN, INC.),Yuuichi FURUSE, Kiyoshi TAKEJIMA (Falcon Corporation)
the elevation of points in a mesh of designated spacing. Forthe purpose of discussing tree height, two kinds of DEMs one reflecting the height of treetops and the other, the groundsurface were produced. The former is referred to specifi-cally as a Digital Surface Model (DSM), while the latter issimply referred to as a DEM. Figures 1a and 1b show birds-eye views of a surveyed area based on DSMs created using aDPW. The DSM for Figure 1a is based on aerial photographstaken in 1995, while the DSM for Figure 1b is based onaerial photographs taken in 1949.
Using DSMs and DEMs to survey forest growthTo examine a progression of changes in the forest, DSMsreflecting treetops and DEMs reflecting the ground surfacewere created for three points in time 1949, 1977, and 1995.Data for these years is maintained in the form of land surveycoordinates in a GIS, so it is possible to examine marginaldifferences between them (Suzuki et al., 2000). In particular,when comparing DSMs, it is now possible to determine theamount of change in forest height. Figure 2 presents resultsof measurements performed with the method described here.Though aerial photographs have been used to evaluate theheight of trees, they have not previously been applied inevaluating the height of forests extending over several squarekilometers.
Creating a detailed vegetation map
To clarify changes in vegetation, aerial photograph interpre-tation was used to produce detailed vegetation classificationmaps for the years, 1949, 1977, and 1995. This work wasperformed by a forestry researcher with many years of prac-tical experience at District Forest Offices. Because thesemaps were intended for use in discussing forest growthcharted using a 5 m mesh, the vegetation boundaries wereoverlaid on positive film and reprocessed using a photogram-metry system to create a set of GIS data. This work wasreported on in detail by Nomura and Nakajima (2000).
Also being examined are vegetation classifications de-veloped by processing MSS images taken from an aircraft.
Table 2. Simplified description of vegeta-tion and land-use categories. (Revised fromNomura and Nakajima, 2000.)
A Fir and Hemlock forestB Evergreen Broad Leave forestC Deciduous Broad Leave forestD Pine tree forestE Bamboo tree forestF Afforested Cedar and CypressC Bare GroundH Residential land and Artificial pondI Cultivate field and Cleared section
Using a 5 m-mesh DSM to perform detailed geometriccorrections, and taking measurements of the same areaover several time periods, the volume of data for analysiswas increased. Research team members aim to use knowl-edge gained from the vegetation classification results of theaerial photograph interpretation referred to above, to im-prove the effectiveness of image-analysis-based vegetationclassification (Miyasaka and Tokumura, 2000).
Forest measurements at the level of individual trees andassessments of carbon dioxide consolidationThe use of changes in forest height to estimate carbondioxide consolidation requires the construction of a growthmodel that addresses individual trees and, therefore, themeasurement of individual trees. Those working in the fieldof forest mensuration continue to accumulate forest mea-surement data and are striving to develop an examinationmethod that is based on a growth model that applies remotesensing data (Yamamoto, 2000). They are also aiming to de-velop a system in which helicopter-based laser radar sensingwould be used to take measurements with a resolution of50 cm. It is their hope that such measurements could then beused to develop estimates of carbon dioxide consolidationover a wide area (Tsuboi and Murate, 2000).
Figure 3. A 50 cm-mesh DSM of the Kaisho Forest created using a helicopter-mounted laser-radar system. Shown is a north-facing view of a 400 m 400 m section of the Kaisho Forest, known as the Kaisho District.
Forest growth simulation
The ultimate goal of this research project is the simulationof the future growth of a forest based on analyses of changesit has undergone over time. This would involve the accumu-lation in a GIS of topographical data, such as that indicatingslope direction and incline, together with geologic, soil, andother types of data, all of which would be accumulated basedon a 5 m-mesh DEM. These data would then be compared todata on patterns and speed of vegetation change to build anempirical equation. Because the subject is a suburban forest,many issues, such as the effect of human activity over time,will have to be resolved.
Contributions of suburban forest research using GISs toglobal environmental research and geography
The significance of this research, which aims to develop away to forecast suburban forest change by using data fromthe monitoring of the subject suburban forest, is not limitedto the preservation of suburban forests. It has the potentialof making important contributions to the problem of fore-casting global climate change. Because an increase in forestbiomass is intimately related to the amount of carbon diox-ide that is absorbed, there is potential for estimates of forestbiomass to be applied in the assessment of the role of terres-trial life forms in the carbon cycle, of which carbon dioxideresulting from human activities is a part.
There is much debate over how man-made carbon diox-ide, which is said to be 60% responsible for causing thegreenhouse effect, is consolidated in the atmosphere, in theoceans, and in terrestrial plant life. Currently, based on dataprovided by surveys and models, it is estimated that 55%of carbon dioxide is absorbed into the atmosphere and an-other 2528% is absorbed b...