Several years ago, MOMRA (Ministry of Municipal and Rural Affairs) – DMTP (Deputy Ministry of Town Planning) General Directorate of Urban Planning launched a national project to convert structural and local plans of the Kingdom of Saudi Arabia into digital geo-format, and established a Planning Information

Remote Sensing satellite imagery has been in use since 1972. The first remote sensing satellite was LANDSAT-1 MSS (Multi Spectral Scanner). The sensor technology has improved tremendously since then. Nowadays, a great deal of information covering large areas of land, such as land cover and land use, can be extracted from the satellite data.

Regarding the project mentioned above, for the classification and detection of land use and land cover data, we used LANDSAT-7 ETM and approximately 100 frames to cover all of Saudi Arabia. Figure 1 shows the coverage of LANDSAT-7 ETM frames over the whole country.

LANDSAT-7 ETM has 8 bands in varying resolutions from 15-meter to 60-meter. One of these bands is 15-meter resolution in panchromatic, one band is 60-meter resolution in thermal channel, and the others are 30-meter resolution. Detailed information for these bands is given in Table 1.

 

Band	Wavelength (µm)	Resolution (meter)	Spectrum
Pan	0.45-0.52	15	Visual
1	0.45-0.52	30	Blue-Green
2	0.52-0.60	30	Green
3	0.63-0.69	30	Red
4	0.76-0.90	30	Near IR
5	1.55-1.75	30	Mid IR
6	10.42-12.50	60	Thermal IR
7	2.08-2.35	30	Mid IR

The satellite spends 14 days collecting data for each location and a single frame covers an area of 180km x 180km on the ground.

Images taken in the Spring are used to identify suitable areas for agricultural and vegetation activities. In some pilot areas, two sets of images – one Spring, one Autumn - have been obtained. Having two sets of images enables us to more accurately classify the vegetative areas. All the images must be taken during clear weather conditions.

In order to make geo-registration and rectification of these images, ground control points (GCPs) are taken from available geo-registered maps or from GPS surveys. We chose simple and distinct points, such as bridges and crossroads, which can be seen easily on the image display. GCP locations like these are entered into the image as pixel coordinates.

LANDSAT-7 image gives us a chance to determine a lot of characteristics of the Earth’s surface such as urban, vegetation, grass, agricultural area, forest, water, marsh, arid area, geological structure, etc. LANDSAT multi-spectral imagery can support these classes because the satellite has 7 different types of sensors; and these sensors are sending waves in different wavelengths which are reflected from the Earth’s surface in different ways. Each image, which is provided by a different sensor, is called a band. 3 band combinations of the imagery - band 5 4 3 combinations - will be enough for this project to extract the information for land cover and land use classes. These classes are given in the following table in detail.

LANDSAT-7 provides an excellent foundation regionally for small scale applications. For large scale applications (urban area), we used very high resolution satellite images (such as IKONOS and QuickBird) and orthophotos.
 

#	Class Name	Class Description
1	Dense Urban -Tall Buildings	Dense areas within the urban perimeter (essentially urban areas with dense development) where built-up features do not appear distinct from each other in LANDSAT TM satellite imagery. Average Building Height - greater than 36m
2	Dense Urban	Dense areas within the urban perimeter (essentially urban areas with dense development) where built-up features do not appear distinct from each other in LANDSAT TM satellite imagery. Average Building Height - less than 36m
3	Urban	Moderately dense areas within the urban perimeter where major built-up features are distinct from each other in LANDSAT TM satellite imagery. Urban should have a mean street density with no pattern; however major streets are visible.
4	Commercial/Industrial	Areas that include buildings with large footprints (i.e., warehouses, shopping centers), obvious industrial activities (i.e., manufacturing, facilities, ports), or extractive land uses (i.e., quarries, strip mines).
5	Suburban - Low Density Trees	Houses in a suburban / residential environment; essentially houses with yards including vegetation with less than 50% crown canopy measured by LANDSAT TM satellite sensors.
6	Suburban - High Density Trees	Houses in a suburban / residential environment; essentially houses with yards including vegetation with greater than 50% crown canopy measured by LANDSAT TM satellite sensors.
7	Village	Small built-up areas in a mostly rural setting (i.e. ~75% surrounded by non built-up classes).
8	Open in Urban	Predominantly areas of open land surrounded by Dense Urban, Urban or Commercial/Industrial (e.g., parking lots, cemeteries, wide roads, parks etc).
9	Airport	Airport runways, terminals, associated buildings.
10	Agriculture/Grass	Vegetated areas of active farming/cultivation.
11	Open - Vegetated	Areas of open land with vegetation (i.e., grass fields, parks or golf courses).
12	Open- Bare	Areas of open land with little or no vegetation (e.g., desert, rocks, quarries, beaches).
13	Forest	Vegetated land in any environment that has a crown canopy closure of more than 50%.
14	Marsh	Areas of open vegetated lands or frequently inundated with water or containing shallow standing water.
15	Water	All water. Includes lakes, rivers, streams, oceans, canals, etc.

Next issue: Detailed explanations about classification