GIS as a Tool to Identify a Coastal Sediment Management Option using...

GIS as a Tool to Identify a Coastal Sediment Management Option using Remote Sensing – A Study of Songkla Bay in Southern Thailand


Wijesekera N.T.S.
Tel: +66 2 524 6184
Fax: +66 2 524 6147
Email: [email protected]

Picton Phillipps G.P.
Tel: +66 2 524 6148
Fax: +66 2 524 6147
Email: [email protected]
GIS Application Center, Asian Institute of Technology
Bangkok, Thailand

1. Introduction
Throughout the world coastal sediment management is an issue that has attracted the attention of many governments. Sediments in the coastal zone interact with the sea waves and get transported in either a long-shore or cross-shore direction. This phenomenon causes either coastal erosion or accretion and has been an ongoing challenge for coastal managers, engineers, planners and the like. Since most countries coastal zones stretch for hundreds or thousands of kilometers, it is not an easy task to identify the coastal reaches that require early attention. Remote sensing images that have the coverage of significant spatial extents are excellent tools to provide information with respect to beach sections that need attention. Multi-temporal remote sensing images over a particular region provide straightforward information about the beach profile changes. However efforts to manage beach profile changes using RS have to be supported with associated sediment quantities so that useful decisions on coast management and financing options can be made.

Fig. 1 Landsat Image of Songkla Coastline (1999). Inset – location of Songkla in Southern Thailand
Geographic information systems incorporating temporal changes in the bathymetry of a region can be effectively used to identify the volumetric change of coastal sediments over time together with the associated beach profile change. In the present work, the bathymetry of a coastal reach near Songkla bay of Thailand for the years 1955, 1977 and 1990 were used to identify the sediment change and the beach profile change. The present work is based on a dataset that has been previously used for training activities (GACR 2001). Report of a related previous work raises many concerns (Lacoul, Samarakoon & Honda 2001). A critical analysis, based on previous experience, was performed on the same base data to identify the nature of any relationship between the volumetric changes of sediment in a beach section over time and the linear shift in the coastal extent over the same period of time (GACT 2002). Sound theoretical reasoning and achieving rational delineation of shoreline cells with uniform thickness, establishing a logical long shore direction suitable for spatial analysis, method of delineation and spatial demarcation using true GIS potential while avoiding visual ambiguities, a critical evaluation of results etc., are some of the differences of the present study. The present study also critically looks at results over different spatial and temporal resolutions.

2 Methodology
Songkla bay in southern Thailand is located approximately between 100° 32’E – 100° 37’E and 7° 10’N – 7° 17’N. Digital Bathymetric information for the study area in 1950, 1977 and 1990 were available for the study. The coastlines for each year were extracted from the same nautical charts that were used to extract bathymetric data. Computations were done by considering the Kinematic Element Model (One-Line Model) on coastal sediment transport. The One-Line Model that ensures the conservation of volume of sediment in a shoreline cell indicates the possible existence of a direct relationship when coastal sediment transport is considered in the long term.

The availability of bathymetric information for Songkla beach enables the computation of sediment volume change using GIS. The same information can also be used to identify the coastline shift. In the present study, the volume of sediment moved in pre-determined beach cells were compared with the coastline shifts during the periods of 1950 – 1990 and 1977 – 1990. In Huq (1990), it has been identified that in the study area, the ocean depth that demarcates the surf boundary could be taken as 3.2 meters. A depth of 4 meters was taken to compute the limits of the dynamic zone for each year under consideration. This depth was used to identify the surf boundary limits of each year under consideration. Taken together, the surf boundaries for the three years indicated a common direction, which was taken as the long-shore direction. The seaward extent of the beach cells was defined by the outer limits of the surf boundaries from all 3 years. The beach cells were selected perpendicular to the identified general direction of sediment transport and a cell width of 100 meters was taken for initial computations. The landward boundary of the cells was taken as the boundary defined by the outer limits (inland) of all 3 coastlines from 1950, 1977 and 1990. The GIS used both vector and raster models for computations and employed standard functions to draw perpendiculars, perform densification of lines etc, to make the shoreline cell delineation accurate. The bathymetric surfaces were constructed using a grid cell resolution of 20 meters. Using these surfaces, the volume of water in each cell was computed for each year under consideration. The difference of the volumes of water between each set of years was taken as the volume of sediment change during that period. Computation of beach profile shift was done by taking the average shift of the beach for two parallel sections of each shoreline cell. In an attempt to ensure whether the entire dynamic zone in the cross-shore direction was captured while computing volumes for each cell, the surf boundary was extended seawards by another 500 meters and calculations were repeated. This selection was made, taking an island formation near the bay into consideration to ensure that the assumptions made in the one line model were maintained.

3 Results
Surf zones for each year with bathymetry data, the respective coast lines, the common direction of long-shore from surf zone computations and the beach cells at 100m apart are shown in Fig. 2. The surf boundaries for the years 1977 and 1990 are in close proximity to each other, where as the 1950 surf boundary appeared further seaward, thereby contributing significantly to the overall surf boundary. The volumetric change within each beach cell for the three concerned periods were computed and plotted against the respective linear shifts in coastal extent (Figs. 3 & 4). It was observed that changes in sediment volume varied between the profile lines taken to the 4m surf boundary and the profile lines that extended a further 500 seaward. This suggests that the delineation at 4m is insufficient to characterise the dynamic zone for along shore sediment transport. The present work limited the investigation to this boundary, 500m beyond the 4m depth line, and the study of sediment movement was performed on the data set for the profile lines extended to this limit.

Fig2. Illustration of the GIS model used to compute shifts in extent of the coastline and corresponding changes in sediment volume in each of the 100m wide “shoreline cells”. Orange shading shows change in bathymetry

The correlation between change in coastal extent and sediment volume for the period 1977 – 1990 indicates that a weak relationship exists (r2 0.377), and that no linear correlation through the origin exists between the data from 1950 – 1977(figs3 & 4). The behaviour of two sets of data appears insufficient to use a single model to enable estimation of sediment change by linear shift in the coastline. However, a look at the two data sets indicates that the 1950 – 1977 data has shown a decrease in shoreline sediments, irrespective of the direction of shoreline movement. The behaviour of the sediments from 1977 – 1990, although with significant scatter, shows a relationship closer to linear through the origin. The scatter in the results could be that, at such a fine spatial resolution, coastal sediment dynamics play a major role in the relationship between shoreline length and the associated sediment volume change. In addition, the One Line model concept is said to be valid only at coarser spatial resolutions. Therefore, to investigate the effects of spatial scale on the study, the data was re-sampled for beach cells of 500m width and 1km width, and then fresh relationships were extracted (figs.5 -8). The correlations illustrated for the resampled data show that for the period 1977 – 1990 the correlation improves as the spatial resolution becomes coarser (@500m R2~0.45). Although at the 1 km level the correlation is sufficient to allow estimation of sediment volume from the shift in the linear extent of the coastline, it may be necessary to extend the study area for a conclusive regression. The 1950 – 1977 results showed a clearer relationship when aggregated to 500m and 1km shoreline cells. This relationship indicated that the shoreline shift and variation of sediment volume follows a linear trend that does not fall through the origin (@500m, R2~0.6).

Figure 3 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1950 – 1977 Figure 4 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1977 – 1990
Figure 5 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1950 – 1977 for beach cells resampled to 500m width Figure 6 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1977 – 1990 for beach cells resampled to 500m width


Figure 7 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1950 – 1977 for beach cells resampled to 1 km width Figure 8 Scatter plot showing the relationship between coastline shift and change in sediment volume between 1977 – 1990 for beach cells resampled to 1 km width

4 Discussion
1. The determination of long-shore direction is very important for the delineation of coastal cells. The one-line model is based on the long shore transport of sediments and the long-term cross-shore equilibrium. This necessitates the identification of long-shore direction and the long shore direction changes. Depending on the spatial resolution, the irregularities in the land and sea interface such as bays and projections makes it very difficult to identify uniformly spaced beach cells that would have the surf zone as one boundary of the cell. In this study the identification of general long shore direction using bathymetric information enabled the smooth delineation of shoreline cells. This is an important identification of a physical feature in the Songkla bay area to study long shore sediment transport. The general direction method also avoided errors that may arise from the beach segments not being equally spaced. The approach of the present study, using a general direction showed a clear rationale.

2. It is of extreme importance that the shoreline cell delineation uses the strengths of today’s GIS software to automatically draw perpendicular lines to the general direction of long-shore movement. Eye estimated lines using visual image of shore line projected on screen would raise serious concerns with respect to the sediment estimations done using such work. In the present work all computations and delineations were carried out with the use of in-built GIS functions and therefore is a good example for similar work.

3. The analysis of sediment movement relationships with beach profile change identified the possibility that the surf zone boundary may be beyond a 4 meter depth. Wave data, adjacent coastal structures and other features of the coastline need to be carefully studied to improve the identified relationships.

4. The water level for each year was assumed constant and the tidal influence was considered minimal. The digitizing from bathymetry maps, registration of bathymetric data, remote sensing data play an important part in the quality of results. The present study used the previously used data set for computations. The survey data and methods also significantly influence the quality of outputs. These need to be subject to careful investigation to improve results.

5. The present work identified the presence of a linear relationship between linear coastline shift to associated sediment volumes. The 1950 – 1977 dataset indicated a linear trend away from the origin. The 1977 – 1990 relationship was linear and through the origin, indicating a stable coastline sediment behaviour that could be used for future and coastal zone sediment management.

6. The study revealed that the coastal change near Songkla bay from 1950 – 1977 has had a significant sediment movement in the long shore direction. This has caused the linear trend to move away from the origin

7. Long term sediment transport was assumed to be taking place only in the long shore direction within each beach cell. The sediment contribution to a beach is not always from long shore and cross shore sources. They can get transported by inland waters or could be influenced by dredging and disposal operations. Therefore for application of this methodology to any beach section it is necessary to carefully consider beach sediment balance and ensure that each component is well accounted for to suit the model assumptions.

8. This study raises the issue of both temporal and spatial scale. The general trend that we have observed is that the relationship between linear shift in coastline and sediment volumes can only be determined at the coarser spatial resolutions (for example 500m). As for temporal scale, the results suggest that this relationship is not constant over a large period of time. The temporal study shows the effect of coastal construction activities on the coastal environment (notably the construction of a pier in 1968 and an oil terminal). The results of the relationships between sediment volumes and coastline changes should be studied along with coastal geomorphology to identify the realities of human interventions


  • The construction of a model to extract sediment movement and the shift of coastlines in beach cells could be effectively achieved using Geographic Information Systems.
  • This study indicated the existence of a linear relationship between the volumetric change and coastline shift at spatial resolutions in the order of 500m – 1km. The trend identified using the 1977 – 1990 can be used for future planning and management purposes. The use of remote sensing imagery was shown viable when used alongside the regression formulas identified in the study.
  • The considerations and assumptions such as the accuracy of data registration of data, delineation of the active zone of sediment transport, temporal and spatial resolution, action of cross shore sediment transport, computation of bathymetric information need to be given careful consideration for more reliable results.
  • A general “best-fit” line, established from analysis of the surf boundary line, can be used to indicate the overall direction of sediment transport through an area, and can assist in the construction of beach cells. This can be used in future studies to avoid the issues arising out of undulations in the coastline.
  • The details of bathymetry well beyond the surf zone need consideration to minimise uncertainties in the dynamic zone limits.
  • Long-term sediment movement of Songkla indicates that between 1950 – 1977 there had been a significant sediment movement along the entire coastal stretch studied. The study of 1977 – 1990 shows that the beach has reached stable conditions. As such Long-term sediment transport studies are important to identify relationships for coastal zone management

This study was performed as part of a training module developed at the GIS Application Center (GAC) of Asian Institute of Technology for the Coastal Zone Management using RS and GIS held in Bangkok 24 June – 5th July 2002. Support extended by GAC, Asian Center for Research on Remote Sensing (ACRoRS) and associated staff is gratefully acknowledged. Valuable comments by Dr. Sutat Weeraskul and Dr. Patrick Hesp were very helpful in performing this analysis. We wish to send our sincere thanks to the ESRI user group and especially to Douglas R. Guess, Lancaster County Assessor’s Office and James M. Gerst for the assistance provided in times of need.


  • GACR (2001) GIS Application Center Training Report, Training Program on Remote Sensing & GIS for Coastal Zone Management – Report to National Space Development Agency, Japan, GIS Application Centre, Asian Institute of Technology, Thailand, July.
  • GACT (2002) GIS Application Center Training Material, Training Program on Remote Sensing and Geographic Information Systems for Coastal Zone Management – Training Material. GIS Application Centre, Asian Institute of Technology, Thailand, June.
  • Huq, A. (1991) Coastal Erosion on the East Coast of the Southern Penninsular, Thailand. M.Eng Thesis, Asian Institute of Technology, Thailand. Pp79.
  • Lacoul, M., Samarakoon, L., & Honda, K. (2001) A study on Coastal Erosion/Accretion in Southern Thailand (Songkla Inlet), Proceedings of the 1st Regional Seminar on Geo-Informatics for Asian Eco-System Management, Kathmandu, Nepal, December.