Historical coastline change analysis between 1903/04 and

2017

By Roman Sorgenfrei & Stefan Groenewold

Based on historical maps from the French National Overseas Archives (1903-1953), U.S.

Army Map Services (map series L7014, 1965-1993), satellite images (Landsat, 1988-

2017), and GPS tracks a collection of in total 131 historical coastlines could be

reconstructed after careful digitalization and rectification of the data. With this, the

changes of coastlines from 1903 until 2017 can be observed and the rates of changes

in meter per year could be relative accurately calculated. The collection forms a unique

dataset that reaches further back than the mostly used data from satellite imagery only.

This rate of change for the last decades (10 and 30 years) and indicated trends over the

past century were the main criteria for classification of the 71 Coastal Protection

Segments (CPSs) of the Coastal Protection Tool (CPMD). Only about 10 % of the

coastline is still accreting by more than 20 m per year, especially at the most south-

western spits of Ca Mau. Meanwhile, more than 50 % of the 720-kilometre long coastline

are currently eroding, of which more than 70 kilometres are eroding with rates between

20 and 50 meters per year.

Fig. 1: Coastline regression from 1904 until 2014 in eastern Ca Mau Province based on

georeferenced historical maps and recent satellite images. Between 1904 and 2014 about 5,8

km of coastal land was lost (on average 52 m per year).

1

1. Introduction

The analysis of historical changes of the coastlines is one of the main tools for coastal

protection planning. The Mekong Delta is geological a relative young system and developed

around 7,500 years ago with higher rates of progradation during the last 3,000 years with

falling sea level. During the last two decades, there were several studies carried out on

changes of the sediment balances and coastal processes.

2. State of the knowledge about coastline changes in the Mekong Delta

Not every detail of the complex Delta system and its evolution is understood yet but there is

certain agreement about the following facts which have direct consequences for coastal

protection planning:

• About 80 % of the sediment discharge of the Mekong River into the East Sea is trapped

on the subaqueous (= underwater) Delta area within 20-30 km off the shore.

• The peak of sediment discharge is taking place in August - November while during the

stormy period from January - April the sediment is re-suspended and distributed mainly

thru an underwater channel system relative close to the shore (within ca. 20 km),

(Nittrouer et al., 2017).

• Roughly a third of this sediment is deposited near the northern and southern Delta

(proximal deposits) front east of Ben Tre, Tra Vinh, Soc Trang and northern Bac Lieu

provinces. Between 40-66 % of the sediment is transported south-eastwards and

deposited southeast (distal deposit) near cape Ca Mau and to some lesser extend near

the bay of Kien Giang at a water depth between 5-20 m (Unverricht, 2014; Liu et al.,

2017a, b).

• For the sediment supply shorewards along the West Sea coast re-suspension

processes and cross-current transport plays a role, too, although it cannot be fully

quantified yet (LMDCZ, 2017).

• A decrease of sediment supply to the mentioned sediment deposits will under-nourish

the sediment demand of the coast (Nittrouer et al., 2017).

• The discharge of sediment by the Mekong River system into the East Sea was

estimated at 150-160,000 tons per year before the period of dam and reservoir

construction and is now decreasing. However, there is disagreement about the amount

since a time lag of effects and inconsistent data impede analysis. Currently the

discharge of sediment is estimated to have decreased to 110,000 tons per year

(Milliman & Fainsworth, 2011) and a sharper decrease is expected after the dam

construction works in the upper Mekong region will be finished.

• The trapping of sediment particle is not only a physical process but is strongly

determined by biological and chemical aggregation processes. An important

aggregation process takes place during the so-called estuarine circulation where the

discharge of freshwater by the Mekong river is causing a coastwards counter current

that transports oceanic water and sediments towards the shore. Due to an increasing

SLR this coupling is diminished and the trapping efficiency is decreasing. The SLR and

land subsidence also lead to deeper channel beds resulting in even lower sediment

trapping efficiency (Allison et al., 2017).

• There is high agreement by different sources of the fact that the erosion trends along

the Mekong Delta coast is accelerating (Anthony et al., 2015). While in the period

between 1973-95 there was still an accretion of 7,2 m annually (on average), the

2

accretion rate fell down to about 2,8 m annually in the period 1995-2005 and even

became negative (which means erosion) between 2005-15, estimated on -1,4 m

annually (Liu et al., 2017a, b). Erosion occurs over about half of the entire coastline.

This is in agreement with the following analysis using historical coastlines going back

to the year 1904. The erosion rate is one main criteria for the coastal classification in

the CPMD.

• There is also agreement by cited studies about the main causes for the increasing

erosion. Decreasing sediment discharge by the Mekong river, sand mining in the

middle and lower Mekong river, land subsidence in coastal areas caused by

groundwater extraction, SLR and destructive land-use in the mangrove belt.

3. The use of historical maps and satellite images for detecting coastline

changes

In order to understand the morphological development of coastlines, the original hydrology of

coastal areas as well as the former distribution of mangrove forests, the analysis of historical

maps, aerial images, and satellite images is of highest value (Albers et al., 2013). The

technique was piloted in Soc Trang and extended to the entire Mekong Delta coast. The

collected maps and satellite images were used to digitise the coastlines from 1903 until 2017.

These were then used for coastline change analyses to calculate regression and transgression

rates as well as area changes. The rates were calculated using the ArcGIS 10 extension DSAS

- Digital Shoreline Analysis System (U.S. Geological Survey, 2018a).

To remove uncertainty regarding the terms coastline and shoreline, here a brief definition of

the two. Even though they differ, they are often interchanged in literature:

• Shoreline is internationally mostly defined as Mean High Water Line although in the

general use the shoreline is often moving up and down with the tides.

• The coastline is usually (depending on the context) defined as the boundary of

terrestrial vegetation and sea, or the Normal High Spring Tide Line on beaches or the

cliff foot at rocky coast. The definitions are not handled consistently among policy

makers and legal frameworks in Vietnam.

The following technical information on how to proceed from original maps to a coastline

analysis of time series are mainly directed to GIS experts. An even more detailed description

of the workflow can be found in the CPMD library (Sorgenfrei, 2016) Readers less interested

in technical GIS procedures may continue with section 10 where some selected results are

illustrated.

4. Collection of historical maps

The maps collected in the French colonial archives have a scale of at least 1/100 000. If the

scale is smaller, map features such as the coastline, channels, river mouths etc. are too

inaccurate and often too highly aggregated to be used for coastline change analysis.

The historical maps from the French colonial archives cover the time of the first half of the 20th

century and were published in different coordinate reference systems (CRS). Table 1 gives

an overview of the properties of these maps.

3

Tab. 1: Properties of historical maps selected.

Year Scale Ellipsoid / Spheroid Datum Projection

1903/04 1/100 000 Clarke ellipsoid no datum given Bonne 1923/24 1/100 000 Clarke ellipsoid no datum given Bonne 1951 1/100 000 Clarke ellipsoid no datum given Bonne 1953 1/100 000 Everest spheroid no datum given UTM 48N

In addition to these maps, topographic maps of Vietnam published by the U.S. Army Map

Service were used (University of Texas Libraries, 2018a), in particular the published “new”

map series L7014 for the whole of Vietnam. For this map series old French topographical

maps were updated supported by several aerial photography missions and increased to the

scale of 1/50 000 by the U.S. Army Map Service (Dang & Le, 2001). These maps are the most

accurate and detailed scanned maps used for this compilation. The majority of the maps from

this series displaying the area of the Mekong Delta are from 1965, but some are from later

years (see Tab. 2). Few were updated last in the year 1989 or even 1993, for which no older

versions are available online anymore. One map displaying a part of Phú Quốc Island is from

1957, while the map displaying the border between Vietnam and Cambodia is from 1993. The

latter is the only map from the U.S. National Imagery and Mapping Agency, Series L7015,

Cambodia Topographic Maps (University of Texas Libraries, 2018b).

Tab. 2: Topographic maps of Vietnam from U.S. Army Map Service, Series L7014 (University of Texas Libraries, 2018a).

Year Scale Spheroid Datum Projection EPSG code Quantity

1957 1/50 000 Everest Indian 1960 UTM 48N 3148 1 1965 1/50 000 Everest Indian 1960 UTM 48N 3148 35 1967 1/50 000 Everest Indian 1960 UTM 48N 3148 1 1968 1/50 000 Everest Indian 1960 UTM 48N 3148 1 1969 1/50 000 Everest Indian 1960 UTM 48N 3148 5 1970 1/50 000 Everest Indian 1960 UTM 48N 3148 2 1983 1/50 000 Everest Indian 1960 UTM 48N 3148 1 1984 1/50 000 Everest Indian 1960 UTM 48N 3148 2 1989 1/50 000 WGS 84 WGS 84 UTM 48N 32648 3 1993 1/50 000 WGS 84 WGS 84 UTM 48N 32648 1

In order to ensure that geodata from different sources, and in the current case especially

different times, can be analysed and compared, it is necessary to use a common coordinate

reference system (CRS). Internationally the CRS usually used to exchange geodata is the

geographic coordinate system WGS 84 (EPSG: 4326). For most of the other CRS, which are

usually applied in regional mappings, transformation algorithms into WGS 84 exist and are

usually provided with GIS software. Therefore WGS 84 is the CRS to choose to make geodata

available for as many users as possible.

4

5. Processing historical maps from the French colonial archives and map

series L7014

Unfortunately, older maps lacked information about the geodetic datum, which is necessary

in order to georeferenced the scanned maps into their original CRS and then transform them

into WGS 84. Therefore, the map series L7014 provided in GeoPDF format was processed

first and based on these result, the older maps were georeferenced directly into WGS 84. The

workflow of the georeferencing for the map series L7014 is shown in table 3. The historical

maps from the French colonial archives were georeferenced using QGIS. As reference the

maps from series L7014 were used in combination with Google Maps Satellite Images

(QuickMapServices plugin for QGIS). Between 19 and 61 GCP were used to georeference

each historical map with a focus of the points in the coastal areas.

All historical maps which were georeferenced are now available in WGS 84 in a locally stored

geodatabase. Before publishing they should be projected into UTM 48N (WGS 84 / UTM 48N

= EPSG: 32648).

Tab. 3: Workflow for georeferencing map series L7014.

Processing step Example resulting file name

1 Download GeoPDF from University of Texas Libraries, 2018a

tra_cu-6228-2.pdf

2 Save as Tiff file format using Adobe Photoshop tra_cu-6228-2.tif 3 Georeference into Indian 1960/UTM 48N or

WGS84/UTM 48N, depending on original CRS (see Tab. 2) using ESRI ArcGIS (define CRS to the Tiff files in ArcCatalog before using them in ArcGIS; transformation type ‘Adjust’; compression type ‘LZW’)

tra_cu-6228-2_.tif

4 Transform into WGS 84 using ESRI ArcGIS tool ‘Project Raster’ (transformation technique 2)

tra_cu-6228-2_WGS84.tif

5 Clip the surrounding borders (frame and legend) of the GeoTiff files (ArcToolbox -> Data Management Tools -> Raster -> Raster Processing -> Clip)

tra_cu-6228-2_WGS84_Clip.tif

Tab. 4: Workflow geo-referencing the historical maps from the French colonial archives.

Processing step Example resulting file name

1 Open .JP2 files and save them as tiff files (Adobe Photoshop)

237_100K_1904.tif

2 Crop maps to delete parts showing open sea (Photoshop)

237_100K_1904_.tif

3 Enhance imagery in Photoshop (colour, tone and saturation)

237_100K_1904__.tif

4 Georeferencing in QGIS based on map series L7014, file name indicates the transformation type (Thin Plate Spline) and the number of ground control points (GCPs) used

237_100K_1904_TPS19.tif TPS = Thin Plate Spline 19 = number of GCPs

5 Transform into WGS 84 using ESRI ArcGIS 237_100K_1904_TPS19_WGS84.tif

5

6. Limitations due to map quality

It needs to be mentioned that the map series L7014 is the most accurate in all terms. In

comparison to the older maps in particular the hydrological information, like irrigation canals,

is more accurate. In this regard especially, the maps from 1953 need to be handled carefully,

because they suggest accuracy due to their detailed content, but the georeferencing process

showed a high degree of generalisation for the hydrological map features. The coastal areas,

especially the coastline itself, between the map borders close to the border of Kien Giang

Province and Ca Mau Province, need to be considered inaccurate for all maps from the French

colonial archives. The same is true for the maps from 1953 between the map tiles Ca Mau

West and Ca Mau South.

7. Google Earth satellite images

The Google servers store a large collection of satellite images for the Mekong Delta region

ranging for some parts back to the year 2000 and are still getting updated. These images give

good, and in latest years also in high resolution, information about the development of the

coastal areas. Still they have to be treated carefully since their georeferencing is not always

accurate (offsets of around 20 m are possible). In comparison to this, the small offsets resulting

from the manual georeferencing process of exported Google Earth images (screenshots) can

be neglected.

In general, the Google Earth maps can be divided into recent satellite images, usually

shown when using the Google Earth GUI, and the historical satellite images which are

provided additionally. Because Google chooses the date of the images shown in standard

view based on their date as well as their quality (e.g. cloud cover), some of the satellite images

available in the timeline view are from a more recent date than the usually shown ones.

The recent satellite images can be downloaded directly from the Google servers using

the software ‘Google Satellite Maps Downloader’. It loads all the tiles stored for a chosen

rectangular extend from the Google servers and provides an option to combine the tiles into a

single BMP file, which is already georeferenced (CRS = WGS 84, EPSG: 4326).

The historical Google Earth images can be best exported as high-resolution

screenshots, a function available in ‘Google Earth Pro’, which is available for free (Google

Earth Pro, 2018). To make georeferencing of these images easier and faster, the view in

Google Earth Pro needs to be oriented north, not tilted and the terrain needs to be switched

off to eliminate vertical exaggeration which alters the images. By adding ‘Placemarks’ as

ground control points (GCPs) in three corners of the chosen view area before exporting the

screenshots can be georeferenced easily using the ‘Georeferencer’ tool of QGIS, a free open

source GIS software (QGIS, 2018). The coordinates can be extracted from the properties of

the Google Earth Placemarks, thus current as well as historical satellite images from Google

Earth can be georeferenced accurately. The images from Google Earth used in this analysis

are coming from two suppliers which hold the copyrights. For images until the year 2013 this

is DigitalGlobe, while the images since 2014 are from CNES/Astrium. Before publishing maps

with satellite images in the background provided by Google it is necessary to clarify licence

issues. For further information see Google Earth Help (2018) and Google Permissions (2015).

For deriving data, as in our case digitising the coastline, defined as seaward forest border, the

usage of the data is allowed.

6

8. Landsat satellite images complement the dataset

To complement the gap between the data from historical maps and the Google images from

above, 40 freely available satellite images from Landsat (Landsat 5 TM, Landsat 5 TM C1

Level 1, and Landsat 8 OLI/TIRS) were included as data source within the last two years. They

cover the complete Mekong Delta coast for the years 1988, 1995, 2000, 2005, 2010, 2015

and 2017. In future, further images from Landsat 8 OLI/TIRS or Sentinel, another free available

satellite image source, can be used to expand the dataset. These images can be downloaded,

among other satellite data, via EarthExplorer (U.S. Geological Survey, 2018c).

As for the Google Earth images, the coastline for the Landsat satellite images was

defined as seaward forest border. To make the identification easier, the multispectral image

information of the Landsat images was used to calculate the so-called red-edge, taking

advantage of the near infrared wavelengths recorded by the satellites. Thus, vegetation was

easily identified on the images and digitised.

9. Application of the programme “Coastline change analysis system (DSAS)”

The application of DSAS is described in detail in the ‘Installation Instructions and User Guide’,

available on the website of the U.S. Geological Survey (U.S. Geological Survey, 2018b). In

addition, Madore (2014) provides a very good step by step manual for the complete DSAS

workflow. A general dataset with a collection of coastlines spanning the whole available time

from 1903/04 till 2017 and covering the complete Mekong Delta was created for overview

analyses. Therefore, the coastlines of the georeferenced historical maps from 1903/04,

1951/53 and the map series L7014 were digitised. The maps from 1923/24 only differ from the

1903/04 maps in their areal information like land use and hydrology but have exactly the same

coastline as the older maps. Consequently, they were not digitised and excluded from

analyses.

Because the digitised coastlines from 1903/04 and 1951/53 are less accurate than the

more recent ones from the map series L7014 and Google Earth satellite images, the results

calculated with them should be treated with caution. They are less exact than the results from

the later coastlines and can therefore only be used as indicators for long time coastline

development.

For the period from 1988 till 2017 the shorelines, defined as the seaward vegetation border,

were digitised based on satellite images. The recent Google Earth satellite images were

digitised for the complete delta. For in depth analysis of several focus areas the additionally

available historical satellite images from Google Earth were also digitised and supplemented

with available shorelines from shoreline monitoring surveys done by GIZ projects in Bac Lieu

and Soc Trang as well as digitised historical maps from the Soc Trang geodatabase.

Additionally, Landsat satellite images were digitised to complement the dataset until 2017. All

digitisations were made in clockwise direction, which simplifies the use of several tools of

ArcGIS. The complete dataset can be downloaded via the online CPMD. A collection of 14

coastlines spanning the complete time can be displayed in the online CPMD under Surveys -

> Historical coastlines.

At least two coastlines from different time steps are necessary to calculate change

rates with the DSAS, which offers several statistical calculation methods (U.S. Geological

Survey, 2018b). The most frequent cited methods are EPR (end point rate) and LRR (linear

regression) (Thi et al., 2014). This is because they are the least expecting for the data input.

The EPR is calculated by dividing the distance of coastline movement by the time elapsed

7

between the oldest and the most recent coastline in a given data set. Its major advantages

are the ease of computation and minimal requirement of only two coastline dates. The major

disadvantage is that in cases where more data are available, the additional information is

ignored (U.S. Geological Survey, 2018b). This method was used for the overview coastline

change rates between the data from the historical maps and the most recent Google Earth

satellite images used in this analysis (2014/15).

The LRR is the result of estimating the average rate of change using a number of coastline

positions over time, with the change statistic of fitting a least-squared regression line to all

coastline points for each transect. The linear regression rate is the slope of the line (Thi et al.,

2014). This method uses all available data regardless of changes in trend or accuracy and is

easy to employ. But it is susceptible for outliers and tends to underestimate the rate of change

relative to other statistics, like EPR (U.S. Geological Survey, 2018b). Both methods were used

for coastline analysis (Sorgenfrei, 2016).

While it is not possible to allocate values of accuracy for the old historical maps, it would

be possible for the Google Earth satellite images as well as for the map series L7014. Because

of this, and because both used calculation methods don’t use accuracy information during

calculation, the accuracies of the single coastlines were not taken into account in the

conducted analyses. For further analyses of the more recent data, for which accuracy can be

allocated, other statistical methods like weighted linear regression (WLR), can be used in

DSAS (U.S. Geological Survey, 2018b).

Using the Landsat satellite images two extra coastline change calculations were

conducted covering the complete Mekong Delta for transects every 100 meters along the

coast. Only the parts of the coastline classified as special segments in the CPMD were

excluded, those are the harbour development in Song Doc, Ca Mau, land reclamation in Rach

Gia, Kien Giang and the canal and thermal power plant in Duyen Hai, Tra Vinh, which account

in total for 20 km of coastline. The first calculation is the EPR based on the coastlines 1988

and 2015 (30 years), the second the EPR based on 2005 and 2015 (10 years). The results for

all transects surrounding the Mekong Delta coast can be displayed in the online CPMD (layers

are in Surveys -> Coastline changes over 10/30 years). The change rates were classified in 6

classes.

8

10. Some selected results on coastline changes in the Mekong Delta

The results of the aforementioned analysis of the erosion rate (EPR) over 10 years for transect

every 100 meters along the complete Mekong Delta coast show that only 10% of the coastline

is currently still accreting, while more than 50% are eroding. More than 70 km are eroding with

erosion rates between 20 and 50 meters per year (see table 5.).

Tab. 5: Result overview of the coastline change analysis using EPR (end point rate) in DSAS for the period 2005-2015.

EPR 10a in m/year Count km %

> +40 304 30.4 4.3

+20 to +40 322 32.2 4.6

0 to +20 2607 260.7 37.2

-20 to 0 3040 304.0 43.4

-40 to -20 637 63.7 9.1

< -40 94 9.4 1.3

Total sum 7004 700.4* 100 * Of the in total 720 km long coastline the harbour development in Song Doc, Ca Mau, land reclamation in Rach Gia, Kien Giang

and the canal and thermal power plant in Duyen Hai, Tra Vinh, which account for in total 20 km of coastline, were excluded.

In average the coastline change rate along the complete Mekong Delta coastline in the period

from 2005 to 2015 was -0.25 m/year, the delta is therefore shrinking. Looking at the results of

the analysis for the period of 1988 to 2015 (EPR 30), the average coastline change rate was

2.08 m/year. In table 6 further results are aggregated for the latter period.

Tab. 6: Result overview of the coastline change analysis using EPR (end point rate) in DSAS for the period 1988-2015.

EPR 30a in m/year Count km %

> +40 324 32.4 4.6

+20 to +40 578 57.8 8.3

0 to +20 2744 274.4 39.2

-20 to 0 2716 271.6 38.8

-40 to -20 552 55.2 7.9

< -40 93 9.3 1.3

Total sum 7006 700.6* 100 * Of the in total 720 km long coastline the harbour development in Song Doc, Ca Mau, land reclamation in Rach Gia, Kien Giang

and the canal and thermal power plant in Duyen Hai, Tra Vinh, which account for in total 20 km of coastline, were excluded.

In the following some selected sites around the Mekong Delta are illustrated as examples,

they as well as others can also be displayed in the CPMD (Surveys -> Coastline changes over

10/30 years and Surveys -> Historical Coastlines).

9

Figure 2. This example shows the changes of coastline in the border area between Bac Lieu

and Ca Mau province (from online CPMD) indicating steady retreat. Coastlines from 1904

(white line) to 2017 (blue line) with 10-year trends of changes indicated as hatched lines as

showed in the online CPMD. Processed data for coastlines derived from historical maps and

satellite images are available as download (as GIS shapefile format) on the online CPMD.

10

Figure 3. Ca Mau coast around 1889. Source: The National Overseas Archives (Archives

nationales d'outre-mer or ANOM), originally the Centre for Overseas Archives (Centre des

archives d'outre-mer), in Aix-en-Provence, France.

Figure 4. The south tip of Ca Mau is one of the few remaining areas in the Mekong Delta that

shows continuing progradation since 1903. From a typical ‘spit’ formation at that time sediment

accreted up to 5 km northwards. This progradation can be considered as an important

indicator of sediment supply to the distal sediment deposits southwest of the tip. The western

part of the mangrove forest is the protected National Park Mui Ca Mau. On the right side (the

southeast coast of Ca Mau) a long-term erosion can be recognised.

11

Figure 5. The larger spatial scope including the area pictured in Figure 4. It is obvious that

very high land loss due to erosion since 1904 along the East Sea coast (Ca Mau and southern

Bac Lieu) contrasts with large accretion along the West Sea coast north of the spit. Despite

uncertainty left, it can be assumed that this trend will continue and indicates constant sediment

starvation along the SE coast (CPR 5). GIS analysis also provides an estimate of the land

area lost and gained: the southern coastal stretch along the East Sea lost persistently land

since 1904.

Figure 6. Coastal area in the south of the estuarine region. A typical pattern is the dominating erosion in the northern part of the coastal head pictured here (Vinh Hai, Soc Trang province) and accretion towards the southern end. This pattern is found even more clearly expressed in all estuarine river islands. Changes during between 2006 and 2016 are indicated here.

12

Figure 7. Coastline changes on Cu Lao Dung island (Soc Trang) from 1904 till 2012 in the

river mouth of the southern Mekong branch. Result of an analysis in GIS, base map provided

by ESRI, satellite images from DigitalGlobe 2008 & 2011, GeoEye 2000 & 2009 and i-cubed

1999 (Roman Sorgenfrei, published in SCHMITT & ALBERS 2014). The core of the island

was much narrower in 1904 and land progradation extended mainly southwards and

eastwards into the open East Sea. For this analysis, the coastline was defined as being the

seaward mangrove forest edge which is not exactly following official definitions of the coast-

or shoreline but is pragmatic if working with historical maps. Accretion rates are ranging from

6.2 to 68.2 m per year over a period of 108 years.

13

Figure 8. History of coastline in Kien Giang bay at the border between An Minh and An Bien

district in detail. The white line shows the supposed coastline around 1904. Generally, the

coast was progressing in north-western direction since then. However, at the western (left side

below of the picture) coast there is a tendency to retreat after the year 1988. This trend is

generally recognizable along the entire stretch south of this point up to Phu Tan (Ca Mau

province).

14

Figure 9. Changes of coastlines in the Mekong Delta from 1903/4 to 2015. Huge change rates

of more than 60 m/year erosion as well as more than 100 m/year accretion can be seen in this

analysis. Mayor accretion during the last century was along the coasts of Tra Vinh, Soc Trang

and Bac Lieu Provinces. While the east coast of Ca Mau province had the highest erosion

rates during this period, the west coast had during the same time the highest accretion rates.

In western Ca Mau close to and across the border with Kien Giang no larger changes can be

observed over the last 110 years despite the losses during the last two decades.

11. Conclusion and recommendations

Although quite complex regarding technical expertise if applied for longer stretches of

coastline, the analysis of historical coastlines is adding clearly value to coastal protection

planning. Trends can be derived from broader data base in space and time in order to

recognise patterns which are not visible from short-term erosion hotspots and one-time

surveys. Sediment coasts are highly dynamic by nature being constantly exposed to changing

hydrological forces. The entire Mekong Delta is the result of these forces and still in the

process of shaping. By including not only satellite imagery but also historical maps (at least

back to 1903/4) much deeper insight into the dynamic is provided. The trends on a larger

temporal and spatial scale are reflected in the classification of the coasts in the Mekong Delta

into 7 “Coastal Protection Regions (CPR)”. The understanding of the coastal dynamic and the

classification into respective regions helps to move from emergency-response driven coastal

protection towards strategic coastal protection. It is recommended to extend the collection of

coastlines using future satellite images which are being published freely (e.g. Sentinel satellite

imagery).

15

References

Albers, T., Dinh, C. S. & K. Schmitt (2013): Shoreline Management Guidelines - Coastal

Protection in the Lower Mekong Delta.

Allison, M.A., C.A. Nittrouer, A.S. Ogston, J.C. Mullarney and T.T. Nguyen (2017):

Sedimentation and survival of the Mekong Delta: A cases study of decreasing sediment supply

and accelerating rates of relative sea level rise. Oceanography 30 (3):98-109,

https://doi.org/10.5670/oceanog.2017.318.

Anthony, E.J., G.Brunier, M. Besset, M. Goichot, P. Dussouillez and V.L. Nguyen (2015):

Linking rapid erosion of the Mekong River Delta to human activities. Science Reports 5:14745,

Doi:10.1038/srep14745. www.Nature.com/scientificreports.

Dang, H. V. & Q. T. Le (2001): Development of Surveying and Mapping Technology in Vietnam. – <http://www.fig.net/resources/proceedings/fig_proceedings/korea/full-papers/pdf/session3/dang-le.pdf> (accessed: 02.07.2018). Google Earth Help (2018): Use of Google Earth Imagery. – <https://support.google.com/earth/search?q=Use+of+Google+Earth+Imagery > (accessed: 02.07.2018). Google Earth Pro (2018): Download Google Earth Pro for PC, Mac, or Linux. – <https://www.google.com/earth/download/gep/agree.html> (accessed: 02.07.2018). Google Permissions (2015): Using Google Maps and Google Earth. – <http://www.google.com/permissions/geoguidelines.html> (accessed: 02.07.2018). Liu, J.P., D.J. DeMaster, T.T. Nguyen, Y. Saito, V.L. Nguyen, T.K.O. Ta, and X. Li. (2017a):

Stratigraphic formation of the Mekong River Delta and its recent shoreline changes.

Oceanography 30(3):72–83, https://doi.org/10.5670/oceanog.2017.316.

Lui, J.P., D.J. DeMaster,, C. Nittrouer, A. Nittrouer, E.F. Eidam, and T.T. Nguyen (2017b): A

seismic study of the Mekong subaqueous delta - Proximal versus distal sediment

accumulation. Continental Shelf Research, Vol 147, pp 197-212.

LMDCZ (2017): Lower Mekong Delta Coastal Zones, Project website and final reports, European Union, Agence Francais Developpement (AFD) and Southern Institute for Water Resource Research (SIWRR). http://lmdcz.siwrr.org.vn. Madore, B. (2014): Shoreline Change from a Storm Event Procedure using the Digital

Shoreline Analysis System. –

<http://sandy.ccom.unh.edu/publications/library/Shoreline_change_procedural_

2014_11_18.pdf> (28.09.2015).

Milliman, J.D. and K.L. Fainsworth, (2011): Discharge to the coastal oceans: A global

synthesis. Cambridge University Press, New York, pp 1-394.

Nguyen, V.L., T.K.O. Ta, M. Tateishi, I. Kobayashi and Y. Saito (2005): Holocene evolution of the Mekong River Delta, Vietnam. Journal of the Geological Society of Japan 111(11):XXI. https://dol.org/10.5575/geosoc.111.11.XXI.

16

Nittrouer, C.A., D.J. DeMaster, E.F. Eidam, T.T. Nguyen, J.P. Liu, A.S. Ogston and P.V. Phung

(2017a): The Mekong continental shelf: The primary sink for deltaic sediment particles and

their passengers. Oceanography 30 (3):60-70, https://doi.org/10.5670/oceanog.2017.314.

Nittrouer, C.A., J.C. Mullarney, M.A. Allison, and A.S. Ogston. (2017b): Introduction to the

special issue on sedimentary processes building a tropical delta yesterday, today, and

tomorrow: The Mekong System. Oceanography 30(3):10–21,

https://doi.org/10.5670/oceanog.2017.310.

QGIS (2018): QGIS - A Free and Open Source Geographic Information System. – <https://qgis.org/> (accessed: 02.07.2018). Schmitt, K. & T. Albers (2014): Area Coastal Protection and the Use of Bamboo Breakwaters in the Mekong Delta. – In: Nguyen, D. T.; Takagi, H. & M. Esteban [eds.] (2014): Coastal Disasters and Climate Change in Vietnam. Engineering and Planning Perspectives. 107-132. Amsterdam, Boston, Heidelberg. Sorgenfrei, R. (2016): Report on consulting services for the creation of a coastal area related geodatabase including processed historical maps from French archives as well as recent GIS material and the estimation of coastline regression since 1904 in the southern Mekong Delta. – <http://coastal-protection-mekongdelta.com/download/library/8.R.Sorgenfrei2016_Coastal_geodatabase_estimation_coastline_regressionsince1904_southernMekongDelta.pdf> (accessed: 02.07.2018). Thi, V. T., Xuan, A. T. T., Nguyen, H. P., Dahdouh-Guebas, F. & N. Koedam (2014): Application of remote sensing and GIS for detection of long-term mangrove shoreline changes in Mui Ca Mau, Vietnam. – Biogeosciences 11:3781-3795. University of Texas Libraries (2018a): U.S. Army Map Service, Series L7014, Vietnam Topographic Maps 1/50 000. – <http://www.lib.utexas.edu/maps/topo/vietnam/> (accessed: 02.07.2018). University of Texas Libraries (2018b): U.S. National Imagery and Mapping Agency, Series L7015, Cambodia Topographic Maps 1/50 000. – <http://www.lib.utexas.edu/maps/topo/cambodia/> (accessed: 02.07.2018). U.S. Geological Survey (2018a): DSAS - Digital Shoreline Analysis System. Computer Software for Calculating Shoreline Change. – <http://woodshole.er.usgs.gov/project-pages/DSAS/> (accessed: 02.07.2018). U.S. Geological Survey (2018b): The Digital Shoreline Analysis System (DSAS) version 4.0 — an ArcGIS Extension for Calculating Shoreline Change. – <http://woodshole.er.usgs.gov/project-pages/DSAS/version4/index.html> (02.07.2018). U.S. Geological Survey (2018c): EarthExplorer. – <http://earthexplorer.usgs.gov/> (02.07.2018).

The National Overseas Archives (Archives nationales d'outre-mer or ANOM), originally the

Centre for Overseas Archives (Centre des archives d'outre-mer), opened in Aix-en-Provence

in 1966.

17

Unverricht, D, (2014): Hydro-, morpho- and sediment processes in the subaqueous Mekong

Delta, Southern Vietnam. Dissertation thesis, Christian-Albrechts University Kiel.