日本リモートセンシング学会誌
Online ISSN : 1883-1184
Print ISSN : 0289-7911
ISSN-L : 0289-7911
TMデータに基づく10万分の1ランドサット地図の試作
田中 総太郎杉村 俊郎亀田 和昭朝日 守外崎 重厚
著者情報
ジャーナル フリー

1985 年 5 巻 1 号 p. 41-56_13

詳細
抄録

This paper addresses the following three concerns to make clear the knowledge for producing Landsat Maps based on Landsat TM data in an experimental area (Tokyo district):
(1) Projection of TM Image to the UTM system.
(2) Color composite from 7 band images.
(3) Land classification map from 6 band data.
On the projection of TM image data to the UTM coordinate system, an evaluation of the mapping accuracy was carried out at 8 ground control points in the test area. Orientation accuracy of TM image became clear to within one pixel i.e., approximately 30 meters.
On the color composite, single band images were evaluated individually in the first step. Next, an evaluation of five kinds of color composites from three of the sevenn band images was made in which discrimination of ground objectives was discussed.
(1) true color composite.
(2) infra-red color composite.
(3) middle infra-red color composite..Band3 (B), Ban5 (G), Band7 (R)
(4) thematic color composite..Complement of Band4 (B), Band5 (G), Band7 (R)
(5) thermal color cmposite..Band6(continuous color transformation of the digital number)
Among these composites, the networks of roads and/or railways seemed to be expressed best in the 3rd one (i.e., middle infra-red color composite).
In the land classification from 6 band digital numbers, 6 groups were defined in the first level classification, and 28 groups were defined in the second level. Based on the multi-spectral signature of 6 band digital numbers (except for TM 6 band), the above groups were analyzed using multi-variate categorical processor programmed for principal compornent analysis. This processor is part of the Bendix made M-DAS system.
The performance categorized as each land group in the training area was more than 90%among the first level classification groups, an more than 80% among the second level groups.
Investigating the number of pixels of each land group contained in each of the 23 wards of Tokyo metropolitan area seems to be a natural approach. This sort of data may be useful for the monitoring of the physical environment in local government areas like the Tokyo metropolitan wards.
Finally, the authors include two kinds of Landsat map--one a middle infra-red color composite and the other a color-coded land classifiction map, both at a scale of. 1:100, 000.

著者関連情報
© 社団法人 日本リモートセンシング学会
前の記事 次の記事
feedback
Top