The Potential of Spatial Tools for Tsunami Hazard Assessment

The Potential of Spatial Tools for Tsunami Hazard Assessment

SHARE

Wan Nor Zanariah Bt. Zainol @ Abdullah
Tutor
University Putra Malaysia
Malaysia
[email protected]

Loh Kok Fook,
Fellow Researcher
Institutes of Advanced Technology
Malaysia
[email protected]

Mohd. Ibrahim Seeni Mohd.
Professor Dr.
University Technology Malaysia
Malaysia
[email protected]

Abdul Rashid Mohammed Shariff
Associate Professor Dr.
University Putra Malaysia
Malaysia
[email protected]

Abstract
The Indonesian tsunami, which was triggered by a massive undersea earthquake registering 9.0 on the Richter scale on December 26, 2004 had caused extensive damages to many Asian countries. Kuala Muda, Kedah, Malaysia was one of the tsunami affected areas. This study has adopted spatial tools to assess the tsunami impacts in the coastal areas of Kuala Muda. It has adopted the post-classification change detection technique using SPOT-5 data sets acquired over two dates, one before and the other after December 26, 2004. Contextual information – Digital Elevation Model (DEM) and texture were incorporated in the digital classification of the SPOT-5 datasets. The change map was generated showing the tsunami affected coastal features. A digitized cadastral lot map was superimposed on this change map and the damage valuation analysis, based on a model developed in this study, was done on a lot by lot basis. The valaution result would be useful in estimating compensation to be paid by the affected land owners.

1.0 Introduction
A tsunami, with its epicenter off the west coast of Sumatra, Indonesia, was triggered by a massive undersea earthquake, registering 9.0 on the Richter scale on December 26, 2004. It had claimed human lives in excess of 100000 in several affected countries – Sri Lanka, India, Indonesia, Myanmar, Malaysia, Maldives, Somalia and Thailand. The worst hit areas were Tamil Nadu (India), Sri Lanka and Aceh (Indonesia) which had official death tolls surpassing 6000, 22000 and 45000 respectively (Shattri and Loh, 2005).

In Malaysia, the worst affected area is Kuala Muda, Kedah, where most agricultural areas were badly hit. A study has been conducted on the valuation of tsunami damaged areas using SPOT-5 data sets over this area. The overall objective of the study was to detect and evaluate the damage using an appropriate model. This paper highlights the methodology adopted and the results attained.

2.0 Study Area
The study area is located between UL = 660000, 240005 and LR = 600010, 299995 as shown in Figure 1. It has mainly flat terrain with elevation below 20m amsl. The mean monthly temperature is 26.053 °C and the mean monthly rainfall is 234.6 mm. Wetland rice is the main land use, while others include mangrove, inland forest, rubber and horticulture are also present.

3.0 Materials
SPOT-5 data, before the tsunami (December 14, 2004) and after the tsunami (January 15, 2005). For the purpose of this research both panchromatic and multi-spectral data sets were processed and analyzed. Ancillary data were also acquired to serve specific purposes in the research. The softwares used for specific image processing were ERDAS Imagine 8.7 and ENVI 4.0, while Arc View 3.2 was used for spatial analysis functions.

4.0 Methodology
Radiometric and geometric corrections were carried out before image fusion. Contextual information – DEM and textural layers were incorporated with the green, red and near infra red bands in the supervised classification analysis (Gong and Howarth, 1990; Robert, 1997; and Stuckens et al., 1999). The digital classifiaction was done using the Maximum Likelihood Classifier for both dates – pre and post tsunami. Accuracy assessment using stratified randomized reference points were then carried out for both maps (Congaton et al., 1998). The digital subtraction of the post tsunami classification layer from that of the pre tsunami classification layer has given the change layer Field survey was conducted twice before and after image classification for familiarization with the area and accuracy assessment respectively.

A cadastral lot map was superimposed on the classification change map in the GIS. The GIS then computed the percentage damage on a lot by lot basis. Three classes of damage – slight, moderate and severe were adopted based on the percentage damage per lot as shown in the Table 1. Each class was then assigned weights for computation of damage valuation (Alpar et al., 1999; Margaret et al., 1999; and Graciela et al., 2005).

Table 2 gives the market land value in the study area determined by Department of Valuation and Property Services (Ismail Omar, 1997). Valuation of the individual affected lots was done using Equation 4 as below.

Damaged value = damaged area (ha) x land use valuation (RM) x Weight Equation 4 It is apparent that the area (ha) damaged, the land cover or land use type and the percent area (%) damaged are important criteria considered in this research to estimate valuation for the tsunami damaged areas. In case of mixed land cover or land use, the weighted average of the land use/cover valuation was taken.

5. Results and Discussion

5.1 Classification results
Figures 2 and 3 depict respectively the contextual classification maps before and after the tsunami. There are nine classes of land use or land cover for each map. It is observed that most changes were from mangrove, urban, horticulture and paddy to mud deposited areas. This is because the tsunami has flooded these areas with mud brought from the erosion of the beach. The wave energy was so strong that some of the muds were deposited far inland.

This map is considered reliable as the respective overall accuracies of the pre and post tsunami classification maps were 85.60% and 82.86%. In addition the kappa statistics for the pre and post tsunami classification maps were calculated respectively as 0.8330 and 0.8105. Tables 3 and 4 present the contingency error statistics for computation of both the overall accuracy and kappa statistics. The tsunami affected change map then produced by subtracting the post-tsunami classification map with the pre-tsunami classification map. The cadastral lot was supeimposed on chnage map in the GIS. Then the percentage area of damages were computed on a lot by lot basis.


5.3 Evaluation of Damage Area
There were 48 cadastral lots affected by tsunami of which 11 were slightly damaged, 15 moderately damaged and 22 severely damaged as presented in Table 5. Severely affected areas were paddy, mangrove and horticulture, while urban areas were slight to moderately affected.

The slightly damaged areas were evaluated between RM 297.60 and RM 2062.62 representing respectiuvely valuation for LOT 37 with 0.0372 ha damaged and LOT 1010 with 0.1637 ha damaged. As mentioned earlier the valuation has depended not only on the area (ha) damaged but also the type of land use or land cover.

The moderately damaged areas were valuated between RM 1636.74 (LOT 853, 0.0433 ha) and RM 15879.78 (LOT 844, 0.4201 ha). This lower limit of the valuation was less than the lower limit for the slightly damaged area. This was due to the significantly less area (ha) damaged in the former.

The severly damaged areas were valuated between RM 2320.00 (LOT 17, 0.0580 ha) and RM 124014.21 (LOT 286, 0.1759 ha) The total damage valuation of the affected cadastral lots was RM 709216.00. Table 7 shows that LOT 1044 experienced the least impact from tsunami as the damaged area covered only 7% of the lot area. On the contrary LOT 850, with 99.93 % damaged, was considered the worst case scenario.

6.0 Conclusion
The damage valuation, using a simple model developed by the researchers, has shown that 48 cadastral lots were affected by the tsunami, where 14 lots were found to be severely affected. The total estimated loss for all the lots was RM 709216.00.

This research has proven that SPOT-5, a moderate resolution sensor, was reliable in assessing tsunami damage valuation of the affected study area. The result of the study gave a generalised valuation of damage for the purpose of compensation to be paid by the Government of Malaysia to the affected land owners. The research can be refined using higher resolution satellite data sets like the Quickbird and Ikonos as well as making available valuation data on properties and crop loss.

List of References

  • Alpar, B., Gazioglu, C., Altinok, Y., Yucel, Z.Y., and Dengiz S., (1999) Tsunami Hazard Assessment In Istanbul Using By High Resolution Satellite Data (IKONOS) and DTM. Institute of Marine Sciences and Management, Istanbul University, Istanbul.
  • Congaton, Rusell, G., and Green K., (1998). Basic Analysis Techniques, Assessing the Accuracy of Remotely Sensed Data: Principles and Practices, (2nd ed.). (pp. 43-49) New Jersey: Prentice Hall.
  • Gong, P., and Howarth, P. J., (1990), The Use of Structural Information for Improving Land-Cover Classification Accuracies at the Rural-urban Fringe. Photogrammatic Engineering and Remote Sensing, 56 (1), 67-73.
  • Graciela, M., Lorenz, H., and Radu, G., (2005), Remote Sensing of Landslides: An Analysis of The Potential Contribution to Geo-spatial systems for Hazard Assessment in Mountainous Environments. Remote Sensing of Environment, 98 (2-3), 284-303.
  • Ismail Omar, (1997). Penilaian Harta Tanah. Kuala Lumpur : Dewan Bahasa dan Pustaka.
  • Jampoler, S. M., and Haack, B. N., (1990). The use of SPOT Data to Map Urbanization and Analyze Change: Kathmandu, Nepal. In Technical Papers: 1990 ACSM-ASPRS Annual Convention, 4, 208-17.
  • Jensen, J.R., (1996). Introductory Digital Image Processing: A Remote Sensing Perspective, 2nd edition, (pp 108-117, 245-251, and 257-279). New Jersey: Prentice Hall.
  • Jensen, L. M., (1990). Knowledge-Based Classification of An Urban Area Using Texture And Context Information in Landsat-TM Imagery. Photogrammetry Engineering Remote Sensing, 56, 899-904.
  • Kartikeyan, B., Gopalakrishna, B., Kalubarme, M. H., and Majumder, K. L., (1994). Contextual Techniques For Classification of High and Low Resolution Remote Sensing Data. International Journal Remote Sening, 15, 1037-1051.
  • Margaret. D., Richard, K., Williams, S. J., and Charles, C., (1999). Managing the Impacts of Coastal Hazards, Sea Technology, ProQuest Science Journals, 40 (1), 65.
  • Robert, A. S., (1997), Remote Sensing: Model and Methods for Image Processing, (2nd ed.). (Department of Electrical and Computer Engineering, University of Arizona, Tucson Arizona).
  • Shattri, M., and Loh, K. F., (2005), Tsunami Disaster Preparedness in Malaysia. Journal of Malaysian Surveyors.
  • Sohl, T. L., (1999). Change Analysis in the United Emirates: An Investigation of Techniques. Photogrammatic Engineering and Remote Sensing, 65, GRS, 31 (1), 136-45.
  • Sunar, F., (1998). An Analysis of Change Detection in a Multi-date Data Set: A Case Study in the Ikitelli Area, Istanbul, Turkey. International Journal of Remote Sensing, 19, 225-235.
  • Stuckens, J., Coppin P. R., and Bauer M. E., (1999), Integrating Contextual Information with Per-Pixel Classification for Improved Land Cover Classification. IEEE Transaction of Geoscience and Remote Sensing, 35, 1136-1145.
  • Wang, F., (1993). A Knowledge-Based Vision System for Detecting Land Changes at Urban Fringes. IEEE Transaction of Geoscience and Remote Sensing, 31 (1), 136-45.