• Tidak ada hasil yang ditemukan

MATERIALS AND METHODS

4.8. CREATING LAND USE MAPS

63

To create choropleth maps and graphically illustrate temporal aerial divisions of different land use classes between September 1990 and June 2001 in Zulti South, common boundaries between the classes from the corrected orthophotos had to be identified through photo-interpretation techniques, mapped and classified through a supervised classification technique.

4.8.1. Digitising

Digitising is the process of converting hardcopy geographic information into digital vectors images1. Traditionally digitising is done on a digitising tablet, a large board with an attached mouse (or 'puck'); the digitising table has a fine grid of wires embedded in it that acts as a Cartesian coordinate system, this underlying circuitry converted points where you clicked the puck into x and y coordinates. (Haddock, 2001). However, high-resolution digital imagery can be linked into GIS software packages allowing the capture, edits and delineation of features directly on the computer screen. This method of digitising is often referred to as 'on-screen' or 'heads-up' digitising.

The basis of heads-up digitising is to use the mouse pointer to trace around the image to be digitised directly on the computer screen using the scanned raster image as backdrop, recording the coordinates as it moves - in much the same manner as moving the puck on a digitising table or tablet (Gillings et.a/.,1998; Wu 1999). This method of 'heads-up' digitising is preferable as it obviates the need for a digitising tablet, and in most instances allows for greater precision because the digital image can be magnified for more exact digitising that is hardly possible on the digitising tablet. In addition to the ability to zoom in on an area of complexity, (Gillings et. a/. , 1998) suggest that it is much easier to undertake the digitising in smaller time-periods, and therefore minimise error resulting from fatigue. This is because the digitised data can be

1A vector image is composed of lines and curves that are mathematically calculated between points

Chapter 4:Materials and Methods 64

displayed and visualised directly on the screen allowing a user to see where they have or have not digitised.

The corrected digital orthophoto's of the study area were opened and displayed in Arcview, here it was possible to zoom in to identify locations of the boundaries between land use types through photo-interpretation. These boundaries were digitised directly on the screen in ArcView at an area threshold determined by the resolution of the aerial imagery, here the boundaries were then traced with the mouse pointer and saved as vector line shapefiles.

4.8.2. Editing and creating polygons

Once all the relevant data in the orthophotos were digitised, it was necessary to edit the shape files to eliminate errors such as undershoots and overshoots to create polygons. This process involves manually moving nodes at the beginning or end of arcs, which either overshot or fell short of other lines, to join and close arcs to form polygons that identify areas of specific land use.

Once these shape files had the original parallax removed by manually snapping dangling nodes to their appropriate positions within ArcView, the polygon topology can be built by using the clean and build operations in ESRl's Arclnfo. Here the shape file is converted into a coverage and the clean operation is run, the automatic snap tolerance is set at zero as the errors were removed from the original shapefile. The final polygon coverage is then converted back to a shapefile containing polygons depicting the topology of land use in the study area.

Chapter 4:Materials and Methods 65

4.8.3.Adding attributes

As the polygons do not contain attribute information pertaining land use information, the corresponding attribute table has to be manually edited to include the land use associated with each individual polygon. This is done via a supervised classification procedure using the land use classes previously defined in the research (see section 3.3.2.).

The resulting datasets represent geo-referenced choropleth land use maps of Zulti South. Each dataset represents the spatial land use patterns present at the time the aerial photographs were taken for each of the orthophotos between September 1990 and June 2001.

4.8.4. Ground Truthing

Assessing the accuracy of the final choropleth land use maps is essential in identifying and correcting sources of error. Congalton and Green (1999) suggest that "if the information derived from the remotely sensed data is to be used in some decision-making process, then it is critical that some measure of its quality be known" (p.4). Possible error could occur from the data capturing stage where incorrect boundaries were digitised, or the classification stage where mapped regions are incorrectly classified or represented on the land use maps. Accuracy of the land use maps were therefore assessed through an error matrix.

An error matrix is a square array of numbers set out in rows and columns that express the number of sample units assigned to a particular category in one classification relative to the number of sample units assigned to a particular category in another classification. In this research 50 random points of ground observations of land use were computed against what was mapped from the orthophotos on the first of June 2001. An error matrix is an effective way to represent map accuracy in that the individual accuracies of each category are plainly described along with both the error of inclusion and errors of exclusion (Congalton and Green, 1999).

Chapter 4:Materials and Methods 66

In their study Fung and LeDrew (1988), found that the Kappa index of agreement produced the best means of assessing accuracy with which a given threshold could distinguish between the ground sites and the choropleth map. Kappa is simply a measurement of the proportion of the test sites that have correctly been classified.

Observations were impossible and insufficient to develop an error matr!x for the other datasets due to land use conditions changing rapidly since the time the aerial photographs were taken, the visual comparison however did appear to generally confirm the land use maps validity.

Dokumen terkait