1. GEOG-4045 Environmental Remote Sensing Larry Kiage & Naresh K. Sonti
Evaluation of land loss and or gain in the Indus River delta using Landsat Enhanced Thematic Mapper
GEOG-4045 Environmental Remote Sensing
Project Presentation By
Larry Kiage & Naresh K. Sonti
Spring 2003

2. Contents Introduction Problem Statement Methodology Results Conclusion
Research Question
Methodology
Classification
Change detection
Principal Component Analysis
Results
Conclusion

3. Introduction East coast of Pakistan
200 kms out of 900 kms of coastline.
100 million people.
Human intervention endangers future of the delta. (Up to 78% of Indus water is diverted for irrigation and damming).

4. Statement of the Problem
This study is geared towards conducting land cover/use change detection in the Indus River delta, coastal Pakistan, using the Enhanced Thematic Mapper images of the delta acquired on a temporal scale of 8 years (1992 and 2000) with a view of establishing whether there has any land gain or loss in the delta and to infer into the possible causes of the changes in land cover.

5. Research Questions
What changes have occurred in the delta between 1992 and 2000 in terms of land loss/gain?
What changes have occurred in vegetation?
Which areas have experienced most change?
Which spectral bands are most appropriate for the change detection?
What are the likely causes of these changes?

6. Methodology Sub setting Classify Change detection PCA Unsupervised
Matrix Analysis
PCA
Plate 1(a): Indus River Delta (Path: 152, Row:043, Scene ID: LT , acquisition date: April 27, 1992)
Plate 1(a): Indus River Delta (Path: 152, Row:043, Scene ID: LE , acquisition date: March 8, 2000)

7. Methodology
Subsets
Plate 2 (a): Subset of Indus River Delta (1992) used for change detection
Plate 2 (b): Subset of Indus River Delta (2000) used for change detection

8. Methodology … Choice of ETM Bands.
Band 4 & 5 were selected: determining vegetation types and delineating land and water surfaces.

9. Methodology … Major considerations in choice of image
Cloud free (less than 10% cloud cover).
Same season
To limit sun angle differences and azimuth.
Images acquired for the March/April period water level in the river is optimum.
June – September : Flood season
December – February: Dry season

10. Methodology … Approaches adopted for Change detection First Approach:
Classifying band 5 (1992 & 2000) into 2 classes (Land and water).
Unsupervised classification (20 classes)
Supervised classification (2 classes: Land and Water
GIS Analysis (Matrix Method).

11. Approach 1
1992
2000

12. Results Dominant changes Land/water interface Water (unchanged)
Land gain
Land loss
Land (unchanged)

13. Results …
Results of Matrix Analysis showing Land loss/gain between 1992 and 2000.
No. of Pixels
1992
2000
Sq Miles
Hectares
Change
238649
water
74.84
Water (no change)
29204
land
9.15
Land gain
72601
22.76
Land loss
708122
222.07
Land (no change)

14. Methodology … Second Approach:
Classifying band 4 (1992 & 2000) into general vegetation, land and water classes.
Lack of Reference data.
Change detection.

15. Results …
10% change
20% change

16. Principal Component Analysis
Multispectral image data analysis is the presence of extensive interband correlation, i.e., images generated by digital data from various wavelength bands often appear similar and convey essentially the same information.
Principal component analysis (transformation) reduces such redundancy in multispectral data. When applied as a preprocessing procedure to Unsupervised classification, this transformation increases the computational efficiency of the classification process.

17. PCA Component 2 Component 3 Component 4 Component 5 Original
PCA Image
Component 6
Component 7
PCA
Component #
Value 1 2 3 4 5 6 7

18. Principal Component Analysis
Value
Percentage 1
88.24 2
7.47 3
2.11 4
1.31 5
0.48 6
0.26 7
0.10

19. Conclusions
Significant changes in land cover occurred during the 8-year period from 1992 to 2000.
The most affected land areas are those that border the ocean and along tidal channels which were lost during the period under review.

20. Conclusions…
The areas under water cover gained significantly and this has been interpreted as to be a result of reduction in water and subsequent sediment discharge of the Indus River.
Because of human intervention in the natural delta building processes of the Indus, its future looks bleak.

21. Conclusions…
Continued reduction in mangrove vegetation will not only deprive land protection from river and tide erosion but will be catastrophic to the biodiversity the many marine fauna that use the mangrove habitats as spawning and nursery grounds.

22. Thank you!