|
|
|
Land Use
|
Land Use-Cover Change Detection Using Knowledge based approaches: Remote Sensing and GIS
Maximum Likelihood Classification
Training samples wee first selected from various spectral class for images 1988 and 1995. After selecting the training samples, classification was urn on the data using maximum likelihood algorithm. Grouping of spectral classes were done on the basis of land cover types which are: forest, rubberforest, oil palm, rubber, mixed horticulture, recreational areas, grass land, urban, construction, cleared land and water body.
Results and Discussion
The performance of the knowledge-based classification and multi-source information classification was tested by comparing with conventional visual interpretation and maximum likelihood classification for change detection. A confusion matrix was used to assess the accuracy of each resulting land use by comparing or cross-tabulating the classified land use to the actual land use observed in the field and existing land use map 1988.
The overall accuracy before and after grouping ranges from 31 per cent to 78 per cent. The knowledge-based classification 1988 compared by land
Use 1988 yielded the lowest accuracy. Knowledge-based classification 1995 versus land use-cover 1995 shows the highest accuracy 78 per cent followed by maximum likelihood classification 76 per cent after grouping. Other overall accuracy after grouping does not exceed 70 per cent.
Conclusions
The knowledge- based classifications and maximum likelihood classifications of images 1988 and 1995 gave comparble means overall accuracy of 44 per cent. However, knowledge-based classification has advantage of quick classification and less field work than maximum likelihood classification.
By using remote sensing and GIS to analyses the nature, growth rate and location of land use changes, the central and local governments can identify cities and towns. They can formulate plans and policies to deal with such uncontrolled growth. The integration of remote sensing and GIS is a powerful tool and decision support system for urban growth management.
Table2 Accuracy Assessment before and after grouping
| Result from |
Average Before grouping |
accuracy after grouping |
average Before grouping |
reliability after grouping |
overall Before grouping |
accuracy after grouping |
% improved |
| KBC95vsLU95 |
0.59 |
0.77 |
0.59 |
0.58 |
0.56 |
0.78 |
22 |
| MLH95vsLU95 |
0.38 |
0.54 |
0.35 |
0.56 |
0.46 |
0.66 |
26 |
| VI95vsLU95 |
0.53 |
0.52 |
0.41 |
0.54 |
0.51 |
0.62 |
11 |
| KBC88vsLU88 |
0.24 |
0.46 |
0.28 |
0.64 |
0.31 |
0.62 |
31 |
| MLH88vsLU88 |
0.34 |
0.77 |
0.29 |
0.75 |
0.54 |
0.76 |
22 |
KBC95=knowledge-based for image 95
MLH95 =maximum likelihood for image 95
KBC88=knowledge-based for image88
MLH88=maximum likelihood for image88
LU95=land use 95,VI95=visual interpretation
LU88= LAND USE 88 |
Reference
-
Bronsveld, K., Chutirattanapan, S., Pattanakakok, B., Suwanwerakamtom, R., and Trakooldit, P., 1994, The use of local knowledge in land use/land cover mapping from satellite images. ITC Journal, 4,349-358.
- De Bie, C.A., K. Beek, J, Driessen, P.M, Zinck, J.A, (1995):" Towards Operationalization of soil Information
for Sustainable Land Management". Position Paper, Brazil soil Conference.
-
Hutchinson, C.F (1982): " Techniques for Combining Landsat and Ancillary Data for Digital Classification improvement". Photogrammetric Engineeing and Remote Sensing, 48, pp 123-130.
- Jenssen L.L. F and H. Middelkoop (1992): "Knowledge-base crop classification of Landsat Thermatic Mapper image". International Journal Of Remote Sensing, 1992,
Vol. 13.No.15, pp 2827-2837.
- Maniere, R., Poisson, M., and Lecharpentier, M., 1984, Land use mapping by remote sensing in theMediterranean region: Development of a software pakage for processing Landsat data. Proceeding of the Eighteenth Internation Symposium on Remote Sensing on Environment, Paris, 1-5 October (Ann Arbor, MI: ERIM), pp. 1937-1944.
- Peddle, D.R, and S.E. Franklin (1992): "Multisource Evidential Classificationo f Surface cover and Frozen Ground". International Journal of Remote Sensing 13(17), pp 3375-3380.
- Quarmby, N.A. (1989): " Monitoring urban land cover changes at the urban fringe from SPOT HRV imagery in south-east England". International Jouranl Remote Sensing, 1989, vol. 10.6, pp 953-963.
-
Shafer, G., and Logan, R., 1987, Implementing Dempster's Rule of Hierarchical Evidance. Artifical Intelligence, 33 271-298.
-
Srinivasan, A., and Richards, J.A., 1990, Knowledge-based techniques for multi-source classification. International Journal of Remote Sensing, 11, 505-525.
|
|
|
|
|
|
|
