| GISdevelopment.net ---> AARS ---> ACRS 2000 ---> Poster Session 2 |
Application of GIS and Remote Sensing to Analyses Landscape Structures
Kazuyuki Takahasi1 and Keitarou Hara2
Graduate School of Businees Administration and Information Science,
Tokyo University of Information Sciences,
1200-2 Yatoh-cho,Wakaba-ku, Chiba 265-8501 Japan
Email: 1kazuyuki@rsch.tuis.ac.jp, 2hara@rsch.tuis.ac.jp Tel: +81-4-236-4628 Fax:+81-236-2601
Keywords: landscape classification, regional planning, remote sensing, GISGraduate School of Businees Administration and Information Science,
Tokyo University of Information Sciences,
1200-2 Yatoh-cho,Wakaba-ku, Chiba 265-8501 Japan
Email: 1kazuyuki@rsch.tuis.ac.jp, 2hara@rsch.tuis.ac.jp Tel: +81-4-236-4628 Fax:+81-236-2601
Abstract
Landscape structure of Sakura-city, central Japan was analysed using GIS and remotely sensed data in order to examine the methodology of landscape classification and e®ectiveness of utilization of GIS and remote sensing. Sakura-city is situated on the fringe of the Tokyo metropolis and urban development has been done in the suburb/rural areas. Following -ve landscape types were classi-edon the basis of the topographic features and land cover types derived from Landsat TM data: 1. Marsh landscape, 2. Lowland rural landscape, 3. Upland rural landscape, 4. Yatsuda rural landscape, and 5. Urban landscape. Ecotopes (the minimum units of landscape) were de-ned by topographic features (altitude and inclination of slope) and land cover types. Then the characteristics of each landscape type were examined according to the composition of ecotopes. The e®ectiveness in the use of GIS and remotely sensed data was veri-ed by this data. Secondly, the resolution of the TM data coincides with the scale of the landscape analysis. Thirdly, the latest information of land cover could be obtained from the area concerned where the rapid urbainization is promoting.
1.Introduction
Rising public concerns on natural environment have accelerated needs in recent years to implement, as part of any development projects, speci-c measures to preserve our environment. The conservation of natural environment should not stop at simple protecting one ecosystem. As pointed out by Forman & Godron (1986) and by Forman (1995), it is essential to tackle this issue from a wider perspective - to conserve an overall landscape that represents composites of many ecosystem.
Many researchers have classi-ed the natural environment by landscape types to come up with indicator to evaluate environment (Blankson & Green, 1991; Bunce, 1996). Land-scape classi-cation systems proposed so far tried to extract landscape elements based on topographic features and land use. These previous systems tended to have unclearly de-ned judgment criteria, which resulted in inconsistent evaluation between researchers. Reasons for such inconsistency included: 1) the ambiguity of the class-cation themselves, and 2) di®erences between individual researchers on de_ning topographic boundaries. Meanwhile, with the rapid progress in accumulation and improvement of digitized geographic data, the level of availability of topographic ad land use information has improved substantially in Japan. For more rational land use and for implementing suburban and urban planning in a manner following natural landscapes more closely, it is necessary to introduce a standardized landscape classi?cation method which yields consistent classi-cation results independent of a person doing the job.
We examine for use in landscape classi-cation methods the generation of standard algorithms such as for handing digital terrain models (DTM) and Landsat TM data in order to attain the standardization mentioned above.
2. Study Area
Landscape structure of Sakura City in Chiba prefecture in central Japan was analyzed in this study. Sakura is located at the center of the shimousa upland in the northern part of Chiba Prefecture. Around Lake Inbanuma located in the northwest part of Sakura, a marshy lowland spreads and this lowland together with the Shimousa upland forms an inter-winding complicated topography. At the head of rivers, a rural landscape speci-c to this region, called the Yatsuda paddy-elds, spreads in tree-branch shapes. Being only 40 kilometers away from the center of Tokyo, there is a rapid housing land development around Sakura Station and other commuter train stations along JR and Keisei commuter train lines, Needs are therefore high for development harmonious with the nature.
3.Method
3.1 Analysis System
The Environmental Data Analysis System at the Tokyo University of Information Sciences was used in the landscape analyses. The hardware and software composition is given below:
| Hardware: | Software |
| *SUN S-7/300 U | * ARC/INFO Ver7.2.1 |
| *Gateway GP6-450, 700 | * ERDAS IMAGINE Ver8.3.1 |
| * ArcView Ver3.2 |
3.2 Data for Analyses
Data used for analyses are listed below:
- Digital Map 50m Grid (Geographical Survey Institute)
- Landsat TM data (14th April 1997)
- Urban Planning Map
- Actual Vegetation Map
- Surface Geology Map
The basic data were scrutinized and well examined before landscape analysis. The landscape types for use in the classi-cation were then de-ned, taking into consideration the topographic condtions and land use types in sakura. As discussed by Hara(2000), the following landscape types were chosen to describe the landcape, Rural landscape (Lowland rural landscape, Upland rural landscape, Yatsuda rural landscape), Urban landscape.
1) Marsh Landscape
Cover Lake Inbanuma and surrouding areas. This landscape, which consists of the open water area of the lake and its shore areas covered by reed (Phragmites communis) and other plant communities. This landscape maintains a high level of natural environment.
2) Rural Landsacpe
- Lowland rural landscape:
Mostly oat and marshy lowlands with an altitude of less than 10 meters. The scenery is mostly paddy-elds and there is no housing. Rice paddies are transformed to dry paddies in winter for cultivation of other crops. Some -elds are also laid abandoned. - Upland rural landscape:
Typical upland farm village landscape on upland with altitudes of 10 to 40 meters. Mainly farmhouses and and dry farmlands, mixed with oak forests and other secondary forests. - Yatsuda rural landscape:
At upstream of small rivers, paddy -elds with narrow widths spread along the streams, forming paddy shapes called yatsuda. In contrast with the paddy -elds around Lake lnbanuma, some paddy -elds in this region remain wet throughout the year. These Yatsuda paddy -elds and the slope forests surrounding these paddies are class-ed as the Yatsuda rural landscape.
This landscape covers both those residential areas going back many years and the relatively newly developed housing areas. In terms of composition, the majority was developed in years after the mid-1960s.
3.4 Method of Classi-cation
Figure 1 shows the decision tree followed in executing the landscape classi-cation. In this classi-cation method, independent topographic elements were extracted and used for classi-cation. For the di®erentiation of lowlands and uplands, the altitude of 20 meters from the sea level was used as threshold borderline.
4 Results and Discussions
4.1 Landscape Map
Figure 2 shows the results executed by the landscape classi-cation method described above. Digital data are treated automatically on the GIS system in accordance with the analyses algorithm, so whoever executes the classi?cation, the outcome should be the same. In this way, the standardization of the classi-cation is achieved. `Results for each landscape type were examined by using the units called ecotopes.
4.2 Extraction of Ecotopes
There are di®erent views on what the minimum units of landscapes should be. For the purpose of our discussions, we will use the concept "ecotope" which is minimum unit combining topographic features and land use.
Remote sensed satellite data and DTM were used for the extraction of ecotopes. Based on the ISODATA method in accordance with procedures of Hara (1997), Landstat TM data covering the entire area of Sakura, detected on April 14th 1997 (referred to herein as TM970414), Marsh (marsh), paddy (paddy-eld), Grass(grassland), Scrub (scrub), Forest-b (broad-leaved froest), Forest-c (conifer forest), Urban-g (urban district with many trees), Urban-o (urban district), Urban-f (factory and industrial area), Field (-eld), Bare-o (bare land), and Bare-c (bare land covered concrete).
From DTM for Sakura, altitudes and inclination of slopes were calculated and land-scapes of the entie Sakura City area were classi-ed into four broad types including lowland, upland and slope areas by topographic features. From these two sets of information, a total of 52 ecotopes were de-ned. Table 1 shows extracted ecotopes.
4.3 Method of Evaluation
A map of areas representing these 52 ecotopes was placed as an over-layer on the land-scape classi-cation map, and percentage of each ecotope in respective landscape areas were calculated. Tables 2-1 to 2-5 summarize the results for each landscape.
4.4 Discussions
For each landscape classi-ed in this study and the percentages of ecotopes which constitute such landscape, the following observations were made: In the marsh landscape, lowland open water area representing a wetland element accounts for more than 90% of the area. In the urban landscape, the urban element accounts for 70% of the area whereas there are topographic element variations in the relevant area. The reclaiming of the lowland by land-lling for urban development is considered a factor behind such variations. In the lowland rural landscape, a number of grasslands are found among the lowland paddy -eld elements and among the vegetation elements, caused by abondament of paddy -eld. The upland rural landscape is made up of diversi-ed elemens, including vegetation elements such as broad-leaved forests and conifer forests, as well as -eld and urban districts with lots of trees. It is interesting to note that Yatsuda which comprises of paddy -eld and vegetation elements are in fact a landscape made up of wet paddies and secondary forests remaining along the slopes. As shown, the landscape classi-cation method which is based on the classi-cation algorithm under this study proves to give rational classi-cations.
5. Conclusion
This study examined a method of landscape classi-cation which uses remote sensed data and GIS. We have shown that the classi-cation method based on our classi-cation algorithms can produce standardized classi-cation results. Our next step will be a comparision of this method with the existing methods including the ones utilizing the multi-variate analysis, with a view to develop a landscape classi-cation adaptable for use at di®erent localities.
Acknowledgement
This study was conducted as part of Sakura City's Natural Environment Survey Project. We owe a lot to the sta® members of the Department of Conservation of Natural Environment at Sakura City, the Chiba prefecture Environemtn Foundation and to the members of the Survey Team. Hereby we would like to make acknowledgement of their contribution and express our sincere thanks.
References
- Blankson, E.J. & B.H. Green. 1991. Use of landscape classi-cation as an essential prerequisite to landscape evaluation. Landscape and Urban Planning, 21: 149-162.
- Burnce R.G.H. 1996. ITE Merlewood Land classi-cation of Great Britain. Journal of Bio-geography, 23: 625-634.
- Forman, R.T.T. 1995. Landscape Mosaics, The Ecology of Landscape and Regions, 632 pp. Cambridge University Press, Cambridge.
- Forman, R.T..T. & M. Godron. 1986. Landscape Ecology, 619 pp. John Wiley & Sons Inc, New York.
- Hara, K. 1997. Macroscale analysis of landscape pattern in Chiba City, In "Conservation of Regional Biodiversity - Surveys of Communities and Ecosystems in chiba City -" (ed M. Numata), 193-206. Shinzansha, Tokyo. (In Japanese with English summary Hara, K. 2000. Landscape of Sakura-city. In " Natural Environment ofSakura city" (ed. Sakura Natural Environment Survey Group), 585-591, +7pp Sakura City. (In Japanese)
| Ecotope Group | Landform | |||
| I | II | III | IV | |
| Marsh element type | ||||
| Open Water | water-1 | water-2 | water-3 | water-4 |
| Marsh | marsh-1 | marsh-2 | marsh-3 | marsh-4 |
| Paddy element type | ||||
| Paddy | paddy-1 | paddy-2 | paddy-3 | paddy-4 |
| Vegetation element type | ||||
| Grass | grass-1 | grass-2 | grass-3 | grass-4 |
| Scrub | scrub-1 | scrub-2 | scrub-3 | scrub-4 |
| Broadleaf tree forest | forest_b-1 | forest_b-2 | forest_b-3 | forest_b-4 |
| Conifer forest | forest_c-1 | forest_c-2 | forest_c-3 | forest_c-4 |
| Urban element type | ||||
| Green Urban | urban_g-1 | urban_g-2 | urban_g-3 | urban_g-4 |
| Ordinary Urban | urban_o-1 | urban_o-2 | urban_o-3 | urban_o-4 |
| Factory | urban_r-1 | urban_r-2 | urban_r-3 | urban_r-4 |
| Field element type | ||||
| Field | `eld-1 | `eld-2 | `eld-3 | `eld-4 |
| Bare element type | ||||
| Ordinary Bare | bare_o-1 | bare_o-2 | bare_o-3 | bare_o-4 |
| Concrete | bare_c-1 | bare_c-2 | bare_c-3 | bare_c-4 |
| I. Altitude(m) (0<=a<=10), Inclination(dgree)(0<=s<5) II. Altitude(m) (10<=a<=20), Inclination(dgree) (0<=s<5) III. Altitude(m) (20<=a<=45), Inclination(dgree) (0<=s<5) IV. Altitude(m) (0<=a<=45), Inclination(dgree) (s=>5) | ||||
| Ecotope Group | Area(ha) | % |
| Marsh element type | 239:58 | 96:67 |
| water-1 | 239:58 | 96:57 |
| Paddy element type | 4:25 | 1:71 |
| Vegetation element type | 1:58 | 0:64 |
| Urban element type | 0:91 | 0:37 |
| Field element type | 0:34 | 0:14 |
| Bare element type | 0:65 | 0:26 |
| Total | 247:85 | 100:00 |
| Ecotope Group | Area (ha) | % |
| Marsh element type | 13:63 | 0:59 |
| Paddy element type | 0:36 | 0.02 |
| Vegetation element type | 480:49 | 20:67 |
| grass-3 | 130.52 | 5.62 |
| others | 349.97 | 15.05 |
| Urban element type | 1586:53 | 68:26 |
| urban_g-3 | 255.63 | 11.00 |
| urban_o-1 | 179.84 | 7.74 |
| urban_o-2 | 266.40 | 11.46 |
| urban_o-3 | 531.48 | 22.87 |
| urban_o-4 | 127.03 | 5.47 |
| others | 226.14 | 9.73 |
| Field element type | 79:01 | 3:40 |
| Bare element type | 163:68 | 7:04 |
| Total | 2324:24 | 100:00 |
| Ecotope Group | Area(ha) | % |
| Marsh element type | 44:30 | 2:12 |
| Paddy element type | 1121:58 | 53:72 |
| paddy-1 | 1110.82 | 53.20 |
| 10.76 | 0.52 | |
| Vegetation element type | 594:83 | 28:49 |
| grass-1 | 35.40 | 16.88 |
| scrub-1 | 123.58 | 5.92 |
| others | 118.85 | 5.69 |
| Urban element type | 130:33 | 6:24 |
| Field element type | 42:30 | 2:03 |
| Bare element type | 153:93 | 7:36 |
| Total | 2087:84 | 100:00 |
| Ecotope Group | Area (ha) | % |
| Marsh element type | 38:88 | 0:88 |
| Paddy element type | 0:95 | 0:02 |
| Vegetatiion element type | 2555:57 | 57:68 |
| grass-3 | 560.42 | 12.65 |
| forest_b-3 | 331.99 | 7.49 |
| forest_c-3 | 498.00 | 11.24 |
| others | 1165316 | 26.30 |
| Urban element type | 1079:85 | 24:37 |
| urban_g-2 | 511.43 | 11.54 |
| urban_o-3 | 214.53 | 4.84 |
| others | 353.89 | 7.99 |
| Field element type | 75:82 | 1:71 |
| Bare element type | 676:38 | 15:27 |
| bare_o-3 | 479.95 | 10.83 |
| others | 196.43 | 4.44 |
| Total | 4430:71 | 100:00 |
| Ecotope Group | Area (ha) | % |
| Marsh element type | 19:68 | 1:54 |
| Paddy element type | 278:64 | 21:83 |
| paddy-2 | 143.82 | 11.27 |
| others | 134.82 | 10.56 |
| Vegetation element type | 836:17 | 65:51 |
| grass-2 | 126.05 | 9.88 |
| grass-4 | 105.31 | 8.25 |
| forest_b-4 | 73.22 | 5.74 |
| forest_c-4 | 117.11 | 9.18 |
| others | 414.48 | 32.46 |
| Urban element type | 53:27 | 4:18 |
| Field element type | 8:51 | 0:67 |
| Bare element type | 77:86 | 6:10 |
| Total | 1276:32 | 100:00 |
Figure 1. Landscape classification method (modified from Hara, 2000)
Figure 2. Landscape classification map (modified from Hara, 2000)