Preliminary evaluation of photogrammetric-Remote Sensing approach in monitoring shoreline erosion

Preliminary evaluation of photogrammetric-Remote Sensing approach in monitoring shoreline erosion

Mazian Hashim, Dr. Aziz Ibrahim and Adeli Abdullah
university of technology malaysia
locked bag 791, 80990 johor bahru Malyasia
Tel (07)-576160, Telex:MA 60205, Fax:572-555

Abstract
Photogrammetric-Remote Sensing approach for monitoring shoreline erosion is devised and evaluated. Shoreline analysed from aerial photograph were digitized and later merged to the recent satellite data of the corresponding area. The output of this study indicates that this techniques provide an agreeable result to the conventional field measurements within tolerable extents.

Introduction
For best planning, in controlling or retarding the process of shoreline erosion, for the coastal evolution process need to be fully understood .Hence, for the purpose of monitoring the coastal evolution process at one effected beach site require the data to be recorded for a reasonably long period of time, sufficient enough for the movement of the shoreline trend to be tracked .in practice, however, due to certain planning circumstances such as financial, tine constraints, etc; the consulting coastal engineers who were tendered to a coastal protection work often nave to implement their project in a given period of time. This in turn would definitely be an impossible task to monitor the evolution trend .As a result, monitoring beach phemenonal carried out in the early stage of the project within a short time frame, although it will not represent the true evolution of the shoreline erosion that were taking place.

Consequently, this study will concentrate on the preliminary evaluation of the photogrammetric-remote sensing analysis for providing alternative on spatial information of the movement trend of shoreline. In this approach, the shoreline from previous record such as aerial photographs of the effected area were analysed and later merged them to a recent satellite remote sensing data. The merging of the data captured by photogrammetric means to the satellite remote sensing data requires the vectorised shorelines be rasterized, registered to a common geodetic base. The validation and verification of the photogrammetric-remote sensing approach were then made by comparing the amount of shorelline movements detected from ancillary information gained. From the certified plan of the cadastral survey and hydrographic charts.

Methods

Study area
The study area, located to the north of Terengganu river, comprises approximately 6 km of sandy beaches, extending northerly from Tanjong Takir to sultan Mahmud Airport. Within the study reach, there are several fishing villages which are now being seriously threatened by beach erosion (did, 1985). See figure 1.

This study area is part of littoral drift cell along the coast of peninsular Malaysia extending from Terengganu River mouth at 10 km northwards tip to Kuala Merabang Telipok. Changes in this littoral cell will certainly have an impact on the shoreline of the study area (DID, 1982).

Fig. 1. Location of the study area and its vicinity
Sources of information
There are essentially two main sources of information: Those information obtained from field investigations conducted as part of this study and those from previous related studies and past records. Aerial photographs acquired in 1966, 1968, 1975 and 1980 were used to complement the 1986 and 1988 satellite imagery for analysing the changes particularly the shoreline. The 1966, 1968 1975 aerial photographs were taken at scale of 1:40,000 and 1:25,000 for 1980. The Thematic Mapper data of Landsat-5 acquired July 1988, and multispectral SPOT-1 data of 1986were used to provide the latest thematic information of the study site. Information obtained from field investigation were used as control to evaluate the movement of shoreline analysed by the proposed photogrammetric remote sensing approach.

Ancillary data that were obtained from current field surveys or past records including hydrographic charts. And resurvey of cadastral lots aligning the shoreline of the test site. Early small scale hydrographic surveys completed in 1937 and 1972 for the study area were used to quantify shoreline and seabed changes. The latest hydrographic survey of the site was carried out in 1988 as part of this study.

Geomertic Correction and Data merging
The aerial photographs and the satellite imagery of the area of interest were initially registered to a common geodetic base prior extraction of shoreline information. The Rectified Skew Orthomorphic (RSO) coordinate system adopted by Directorate of Surveying and Mapping, Malaysia (DSMM) were used for this purpose.

For the aerial photographs the geometric correction is performed by creating a corresponding model of the terrain from the overlapping photographs by means of stereoplotter. A WILD-AG1 stereoplotter linked to a microcomputer is used in and exterior orientation of the aerial photograph is performed emprically at six model points configured as in conventional photogrammetric approach; two at upper left and right corner, two at left and right of the principal line, and two being at lower left and right corner of the overlapped area.

The absolute orientation of the model is then performed by properly scaling and levelling it to three known ground control points priorly plottted geodetic base at scale of 1:10,000 . the three ground control points were among the nine control points used later for registering the satellite imagery to this geometric corrected model. A difference of 0.5 metre planimetry and 1.0 metre for heights is tolerable at all control points in the absolute orientation. The shoreline is then extracted and plotted. The procedure involved in the interpretation and capturing of the shoreline from this model is explained in the following section.

On the other hand, the subscene of the satellite digital data were processed in the Diplx ARIES -III , digital image analysis system. The satellite subscene data is registered to the same 1: 10,000 geodetic base by image-to-image basis. Second degree polynomial transformation function given in equation (1) were best found .to suit the transformation of the systematically corrected satellite data to Rectified othomorphic projection of Malaysia. Hashim et. al. (1988) also addressed that, second degree polynomial function show a more stable residuals pattern in the transforming level l B SOPT-1 data to RSO projection. The cubic convolution resampling scheme to a pixel size of 5 metres is then then followed ,The summary of the image-to-image registration in merging the two data set is tabulated in table l.

X= A ₁+A₂ x² +A₃ x +A₄ Y² +A₅ Y+A₆ xy
Y= B ₁+B₂ x² +B₃ x +B₄ Y² +B₅ Y+B₆ xy
Where As' , Bs are the transformation coefficients ,
X,Y are the ground control points ,and
X,Y are the image coordinates .
In merging both data sets, the shoreline extracted from the analysis by interpreting the appropriate clues of the desired shoreline is firstly converted to a raster file .the rasterization of the photogrammetrically captured shorelines is carried out by a simple vector-raster conversion as illustrared in the next section.

Table.1. Summary of the image-image registration involved in merging the data captured by photogrammetric means to the satellite subscene.

GCP	SLAVE POINTS		MASTER POINTS		RESIDUALS		Rem.
GCP	Line	Pixel	North	East	N	E	Rem.
1	149.0	114.0	592773.0	569506.0	0.2	-0.3
2	336.0	507.0	592149.0	570684.0	-0.6	-0.6
3	324.0	810.0	591106.0	571606.0	-0.1	0.6
4	619.0	199.0	591383.0	569783.0	1.4	1.1
5	786.0	370.0	590897.0	570299.0	-0.1	-0.9
6	624.0	59.0	591358.0	569366.0	-0.9	-0.2
7	76.0	281 .0	573000.0	570000.0	0.7	0.2	C
8	82.0	617.0	593000.0	571000.0	0.1	.0 .4
9	41.20	26 .0	592000.0	570000.0	0.1	0.1
10	418.0	612.0	592000.0	571000.0	-1.2	0.4	C
11	425.0	942.0	592000.0	572000.0	-0.2	0.4
12	761.0	941.0	591000.0	572000.0	4.5	-2.3	C
13	755.0	606.0	591000.0	571000.0	2.6	-0.1	C
14	748.0	270.0	591000.0	570000.0	-3.7	-0.4	C
Standard errors of pixel estimate = 0.53 m Standard errors of line estimate = 0.52 m Standard errors of pixel estimate of check points = 1.03 m Standard errors of pixel estimate of check points = 2.07 m;

Shoreline Extraction
Extraction of shoreline form the geometrically corrected model is carried out by a "stream mode" following the anticipated line by fusion of cursor marks of the right and left of the viewing system of a stereoplotter. Before extraction of the shoreline in this particular area, the water level at time of photography is firstly determined and later used as a guide in providing useful clues during the interpretation of the present shoreline. Once, the water level at time of photography is identified in the model (viewed as land -water interface), the normal high tide mark along the beach can be located. This, however, is not the shoreline wanted, but it will assist in locating the nearest location of the highest high tide mark which were presently adopted by the DSMM as the state shoreline in topographical survey practice.

The exercise of locating the shoreline in this approach mark rely rather heavily on the available clues that exists on the beach. These clues include of those like the extent of occupation, cropped lines, etc. the association of the reasonable clues to the water levels at a particular site formed an important element in the shoreline interpretation. This association is illustrated in figure 2. The shorelines delineated from aerial photographs acquired in different years were then automatically digitized by the online microcomputer, and stored in a digital vector format.

For the satellite thematic MLA data, the infrared band (band 7), and band 5 were used in the shoreline extraction. In the infrared band, the land-water interface are clearly defined as the water absorbed the radiation energy and, thereby, contribute nearly no energy returns. As for the land features, the radiation reflects according to the nature of properties of materials it hits. The water-land interface, however, does not represented the present shoreline but it is the extent of water level at time of data acquisition. Enhancing this land -water interface with a high pass filter orthogonally depict this line and its aligning features more significantly. The linear features which have been cropped by pioneering small marine shrubs, near to the end of occupance limit can be reasonable interpreted as an approximate present shoreline. Similarly for the SPOT -1 MLA data, the infrared band 3 and band 2 were processed accordingly.

Misinterpretation by this technique could be very minimum because the normal high tide mark is very pronounced, and the shoreline required to be located in this approach is normally within a for metres inland, hence it is justifiable within the spatial resolution of 20 meters and 30 meters of the SPOT-1 and TM data used.

The digital vector format of the shorelines which have been extracted by photogrammetric means from aerial photographs were then integrated to the geo-coded remotely sensed TM and SPOT-1me rasterizing the shoreline raster data. This Is done by a simple vector-toraster conversion as illustrated in table2. The raster on the display unit of the digital image analysis system.

Fig. 2. Prominent water levels and clues association in the shoreline interpretation.
Results and discussion

Shoreline Changes
The extent of shoreline changes can be quantified from the photogrammetric-remote sensing analysis carried out. Based the analysis carried out using the information obtained from the photogrammetric-remote sensing approach, the findings can be summerised as follows:

The northern spit at the Terengganu river mouth has been growing in a southeast direction between the observation periods of 1966, 1968, 1975, 1980 and 1986 at an approximate rate of 10 metres per year.
At the same time, the shoreline of the spit retreated at about the same magnitude and the centre of the spit has become narrower.
From kg. Seberang Takir to Kg Bahru Seberang Takir, the shoreline retreats at 2 to 4 metres per year.
From Kg. Bahru Seberang Takir to Kg. Telaga Daing, the shoreline retreats at 1.4 to 2.5 metres per year.
From Kg. Telaga Daing to Kg. Ketapang, the shoreline retreats at 1 to 2 metres per year.
From Kg. Ketapang to kg. Telaga Batin, the shoreline advances at 1 to 3 metri\es per year.

Quantitatively, the movement of the shoreline identified in the whole 6 kilometres study reach is presented in figure 3. Direct comparison with the shoreline determined from hydrographic records of 1972-1988 have shown that the shoreline obtained by photogrammetric remote sensing significabntly show the same trend and magnitude of shoreline movement for 1968-1988 aerial photographs Figure 4 (a) shows the variation of shoreline of 1937 and 1968 obtained from hydrographic records, while the variation of the snoreline movements outline in photogrammetric remote sensing approach for the period of 22 years is shown in figure. 4(b).

Fig. 3. Shoreline movements analysed in Photogrammetric-Remote
Sensing approach within the study area.

Fig. 4(a). Comparison of 1937 and 1972 shoreline obtained
from Hydrographic surveys.

Fig. 4(b). Shoreline variations 1966-1988 extracted from
Photogrammetric-Remote Sensing approach.
Assessments to justify the above findings were later carried out by visiting to affected study area. The test site were of 2.5 km southern beach of the study area. For the purpose of analytical evaluation on the quantity of shoreline that have taken place, two stable points which were recognisable in the image as well as in aerial photographs were used. The two points were navigational beacon at southern end of Kg. Seberang Takir, and end-road at central point of Kg. Seberang Takir. Both the test points were located at 1.5 km apart, Table 3 summarizes the comparison of the erosion measured from field surveys and the photogrammetric remote sensing approach.

Field checks on the affected sites also involved the distance measurement from the present shoreline to those marks identified by local inhabitants as shoreline used to be in the previous 10 or more years, which strongly show that the findings gained from the analysis does comply with the real situation. The certified plan of the cadastral survey of the effected sites where also consulted in the field check. A comprehensive quantitative analysis on different test sites of other areas is yet to be implemented for further evaluation of the shoreline movement gained by this technique.

Conclusion
Shoreline movements can be monitored using the photogrammetric remote sensing approach as demonstrated. Aerial photographs alone with the established photogrammetric mapping is the best way of mapping the shoreline movements provided that it is acquired at timely basis as required in a particular project. Satellite remote sensing data which is affordable to be in timely and its synoptic view can be an optimal tool to complement the quantification of the shoreline movements.

Table 2 : Pseudo coding for Vector-to-Raster Conversion of the shoreline captured from Aerial photographs.

1. Define data, open input file, open output file

2. Begin toop Read input vector coordinates (shoreline)
Find miniuwr x, y and maximum x, y vector coordinates Close loop

3. Built a raster overlay based oj min x.v and max. x.y input file.

4. Begin loop Verify whether the input string coorodinates lie within a pixel of a raster file.
If yes, assign a value I to this pixel
If NOT. Go to next pixel and verify the same thingClose loop

Vector data overlaid on the raster file created based on the range of x and y coordinates of captured vector data.
Zoomed -in resterization criterion assignment

minimum distance from centroid to the vector data is the
indicator of inclusion of this cell to a raster file

Table 3 : Comparison Of Shoreline Erosion Analysed From Hydrographic Survey
And Photogrammetric-Remote Sensing Approach.

Test site	* H.S		** P.RS		*** F.SUR
Test site	1972-1988	mean	1968-1988	mean	1952-89	mean
1	65m	4m/yr	81m	4.1m/yr	120.0	3.3
2	30m	2m/yr	45m	2.3m/yr	110.0	3.0

1. Southern end of Kg. Seberang Takir Navigational beacon,
2. Central end -road of Kg. Seberang Takir.
* H.S. for hydrographic survey.
** P.RS for photogrammetric remote sensing approach.
*** Resurvey of the cadastral lots of the existing certified plan.

Acknowledgements
This paper described the work carried out as part of feasibility study and detailed design of coastal protection work at Kuala terengganu-seberang Takir, Malaysia. The authors gratefully acknowledge the Drainage and Irrigation Department and the Terengganu Survey Department for providing the necessary information required in this study.

References

Drainage and Irrigation Dept. (DID). (Dec. 1985) Final Report on Kuala Terengganu Sector, Natiional Coastal Erosion Study. (Unpublished).
Drainage and Irrigation Dept. (DID). (Sept. 1982) Final Report on Tanjong Berhala harbour Hydraulic Investigations. (Unpublished).
Hashim, M; S. Ahmad and S.K. Ho. (1988). Evaluation of Image-to-Grid Registration Accuracy of SPOT-1 MLA Data with Mathematical Transformation. Proceedings of Earth Resources: Data, Systems and Applications 88'. July 12-14 1988 Kuala Lumpur.