香港特別行政區政府 環境保護署
香港品牌形象
搜尋 網頁指南 聯絡我們
圖像
主頁
環評與規劃
環境影響評估條例
網上環評
關於環評與規劃
策略性環境評估
環境管理工具
指引與參考資料
搜尋此部份
環境保護互動中心 部門刊物 環保標準及統計資料
空氣 廢物 環評與規劃
噪音 水質 環境保育
空白 主目錄 空白

研究報告

Chapter 4 PRELIMINARY HABITAT MAP

The preliminary habitat map is produced based on a combination of remote sensing and GIS techniques, using satellite imageries, aerial photos and existing/digitised data. This section describes how the preliminary habitat map, consisting of 25 habitat categories (Section 2), is produced. Section 4.1 describes the data sources, hardware and software used; Section 4.2 describes the methodology used and the processing steps; and Section 4.3 assesses the mapping confidence of the resulting map.

The preliminary habitat map was used to help identify data gaps and the design of the ecological surveys. Results of the ecological surveys are incorporated in the preliminary habitat map to produce the final map (Section 9).

   
4.1 Data Sources, Hardware and Software
   
4.1.1 Satellite Imagery
   

Two sources of satellite imagery were used: Landsat (4) Thematic Mapper (TM) and SPOT (5) Panchromatic (PAN). Landsat TM was selected as it has seven multi-spectral bands, thus allowing good land cover discrimination. Landsat TM has a spatial resolution of 30 metres, with reflectance values (stored in separate bands) in the visible, near infrared, middle infrared, and thermal infrared spectral regions. A second source of imagery, SPOT PAN, was acquired to improve the spatial resolution of the Landsat TM imagery through merging. SPOT PAN has a spatial resolution of 10 metres and has a single band, containing reflectance values in the visible (green) and near infrared spectrum. Details of the merging technique applied are provided in Section 4.2.1.

Although multi-spectral satellite data with a 20 metres resolution were available (SPOT XS), this imagery contains data in only three bands and over a narrower spectrum than Landsat TM. Several studies comparing classification accuracy of Landsat TM versus SPOT XS report that Landsat TM consistently outperforms the other (Green et al 1998; Brockhaus and Khorram 1992).

The criteria for data selection were that the imagery had to be cloud free and relatively recent. The process of selecting suitable imagery was made more difficult by the fact that:

  • the swath width of the Landsat TM satellite did not cover the entire study area in one pass, which meant that two suitable scenes would have to be acquired. The Landsat TM satellite passes over the same location only once every two weeks;

  • the dates of the two sources of satellite data (Landsat TM and SPOT) had to be as near as possible to ensure that land features/conditions had not changed;

  • the SPOT satellite can collect data with and angle of incidence of up to 278 from the vertical position, whereas the Landsat TM satellite can collect data with an angle of incidence of up to 58 from the vertical position. To produce a good quality merge, the angle of incidence of the SPOT image should be close to that of the Landsat TM image; and

  • Hong Kong is frequently blanketed by cloud.

Image data meeting the above criteria were identified and acquired as described in Table 4.1a

Table 4.1a Satellite Imagery Used
Imagery Type  Date Path/Row Time Full/Part Scene
Landsat TM  26 Sep 1998 121/45 02:25:10 Part
Landsat TM  3 Oct 1998 22/44 and 45 02:30:57 Part
SPOT PAN  22 Dec 1998 286/305 and 306  03:13:32 Full

(4 ) The Landsat satellites were launched by the Aeronautics and Space Administration (NASA) of USA. The imagery used in this study came from the Landsat 5 satellite launched on 5 March 1984. The new Landsat 7 satellite was launched in 15 April 1999, after this study has started.

(5 ) The SPOT satellites were launched by Centre National d'Etudes Spatiales (CNES) of France. The imagery used in this study came from the SPOT 5 satellite launched on 24 March 1998

   
Back to topTable of ContentNext

 

   
 
2005 版權標誌| 重要告示

最近修訂日期: 二零零五年十二月二十二日