This article was downloaded by: [Wu, Weicheng] On: 30 November 2008 Access details: Sample Issue Voucher: International Journal of Remote SensingAccess Details: [subscription number 906194077] Publisher Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK International Journal of Remote Sensing Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713722504 Coastline evolution monitoring and estimation—a case study in the region of Nouakchott, Mauritania W. Wu a a PRODIG, Ecole Pratique des Hautes Etudes, 75005 Paris, France First Published on: 18 June 2007 To cite this Article Wu, W.(2007)'Coastline evolution monitoring and estimation—a case study in the region of Nouakchott, Mauritania',International Journal of Remote Sensing,28:24,5461 — 5484 To link to this Article: DOI: 10.1080/01431160701227612 URL: http://dx.doi.org/10.1080/01431160701227612 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. International Journal of Remote Sensing Vol. 28, No. 24, 20 December 2007, 5461–5484 Coastline evolution monitoring and estimation—a case study in the region of Nouakchott, Mauritania W. WU* PRODIG, Ecole Pratique des Hautes Etudes, 191 Rue St-Jacques, 75005 Paris, France Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 (Received 21 November 2005; in final form 16 November 2006 ) Since the construction of a harbour, Port de l’Amitié, an important importation gate for Nouakchott, northwestern Africa in 1987, the previous coast dynamic equilibrium has been destroyed and thus a significant littoral geomorphologic change has occurred, which has produced a severe degradation of the littoral and urban environment. This research is focused on coastline evolution monitoring and its potential change estimation by remote sensing techniques using multitemporal Satellite pour l’Observation de la Terre (SPOT) images and Markov chain analysis. It is the objectives of the study to measure and estimate the coastal current hydrodynamics, coastline evolution rates, harbour life-span and to provide useful reference for the local authorities to make decisions for their future coastal management. The results show that the north beach of the harbour has extended by 0.92 km2 (92 ha) from 1989 to 2001 and the accretion will probably reach its maximum limit in about 13 years¡6 months (in 2014–2015) while the harbour will gradually arrive at the end of service. The south sandbar has been eroded by 1.34 km2 (134 ha) and the coastline has retreated landwards by 362 m at the maximum point. Another 0.91 km2 of land will be eroded in the next 10 years from 2001 to 2011. This erosion has caused several inundations to the suburb and urban areas, provoking a deterioration of the urban environment. 1. Introduction The coastline of the region of Nouakchott, located in the Meso-Cenozoic SenegaloMauritanian Basin (Caruba and Dars 1991) in northwestern Africa (figure 1), has experienced significant modification in recent decades. This change continues today because of poor coastal management involving the build-up of the Port de l’Amitié harbour on the coast in 1987. Beach accretion occurs on the north of the harbour and there is significant erosion in the south part (figures 1, 4, 5 and 6). Furthermore, extractions of natural resources, for example, shell and sand, which are used for urban construction, have weakened the strength of the sandbar and altered its original dynamic equilibrium. As a result, the lagoon barrier in the south burst in 1991 and a flood extended to the whole southwest part of the city; in 1995, seawater crossed over the sandbar around the wharf, drowning whole sebkha and a slum district. In 1998, in addition to the above described zones, more than six sections of the sandbar—the only natural defence of Nouakchott from ocean invasion, were breached and an unprecedented inundation into the western and southern parts of the city occurred. This coastal environmental change, together with other natural *Email: w.wu@cgiar.org. Current address: NRD, University of Sassari, Viale Italia No.57, Sassari 07100, Italy. International Journal of Remote Sensing ISSN 0143-1161 print/ISSN 1366-5901 online # 2007 Taylor & Francis http://www.tandf.co.uk/journals DOI: 10.1080/01431160701227612 Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5462 W. Wu Figure 1. Location of the study area, Nouakchott, Mauritania in northwest Africa (after Wu 2003). disadvantages, e.g. serious aridity, elevation of the salty underground water level and invasion of the desert (Courel 1993, 1998, Courel et al. 1996, 1998, Wu 2003, Wu et al. 2003), are severe challenges to the capital of Mauritania. Beach accretion on the up-drift side of a breakwater and erosion on the downdrift side are common phenomena in coastal management. In addition to Nouakchott, similar littoral geomorphologic changes have also taken places in Santa Barbara, USA (Komar 1998), Lanshan, China (Chang 1997), Hualien, Taiwan (Hsu et al. 2000), Fortaleza, Brazil (Maia et al. 1998), Northeastern Nile Delta, Egypt (Frihy et al. 1998). Thus, this kind of change monitoring, mapping and evolution estimation are of common importance for understanding the coastline evolution originated from artificial engineering and evaluating the impact on the littoral environment. The coastal change in Nouakchott came into existence as soon as the Port de l’Amitié had been constructed. But it was not apparent until 1991, when a flood breached the eroded barrier in the south of the harbour and extended to the Coastline evolution 5463 Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 southwest of the region. Since then, a research project ‘Littoral and urban environment of Nouakchott’ has been undertaken at PRODIG. The fragility of the sandbar and its behaviour under wave attack have been tentatively analysed (Courel et al. 1996, 2000), urban extension and its related problems, seawater flood risks in the region have been studied (Wu 2003, Wu et al. 2003, Abou Dagga et al. 2004). However, some other important factors, such as coastal morphological change rate and harbour life-span have still not been investigated. Based on previous studies (Wu et al. 2002b), effective approaches have been applied to monitor the coastline modification in the past 11 years and forecast the evolution in the near future using multi-temporal Satellite pour l’Observation de la Terre (SPOT) High Resolution Visible (HRV) images and Markov chain analysis. The objectives of this study are to reveal the alongshore current hydrodynamics, coastline change properties and their potential evolution in the near future and to calculate useful reference data for the authorities to control the littoral environment degradation and design future coastal engineering better. 2. 2.1 Methodology for coastline evolution monitoring Background and data selection The application of remote sensing techniques to coastal evolution monitoring has a certain history. As early as 1963, Verstappen had already begun to use aerial photographs to measure changes in the Solo Delta into Java Sea and in 1971, Stafford and Langfelder employed air photographs to survey the coastal erosion. Since then, various remote sensing data and techniques such as aircraft multispectral scanner (MSS) (Jensen et al. 1987), aerial photogrammetry (Gorman et al. 1998), Landsat Thematic Mapper (TM) (Ramsey et al. 1998), SPOT (Populus et al. 1991, Chen 1998), Radar (Jol et al. 1996, 1998, Ramsey et al. 1998) and airborne light detection and ranging (LIDAR) (Daniel et al. 1998, Stockdon et al. 2002, Robertson et al. 2004), have been utilized in different littoral research projects around the world. The recently developed measuring techniques such as airborne photogrammetry and LIDAR provide high resolution measurements with less error. They are, however, expensive and require extensive training (Moore 2000). Thus the type of remotely sensed data that can and should be selected for monitoring coastal changes depends on the cost, pixel resolution and measurement scale (Raffy 1994, Quattrochi and Goodchild 1997, Cracknell 1999, Marceau and Hay 1999, Hwang and Ku 2004, Boak and Turner 2005). As the detailed extent of land surface features discerned is different we can observe them on different scales or by ‘zooming’ (Raffy 1994). Each scale of measurement requires correspondingly its optimal pixel resolution data. This is the so-called scale effect in remote sensing (Quattrochi and Goodchild 1997, Marceau and Hay 1999, Hwang and Ku 2004). As a local case study, a prime requirement of the remote sensing data lies in their capacity to highlight as exactly as possible the coastal geomorphological change and shoreline evolution in Nouakchott. Coarse resolution images such as Advanced Very High Resolution Radiometer (AVHRR), even middle resolution ones like the Moderate Resolution Imaging Spectroradiometer (MODIS) and Multi-spectral Scanner (MSS) are not suitable for this purpose. High resolution satellite data first taken into consideration were Landsat TM and SPOT HRV. The latter with a finer pixel size (10 m for the Panchromatic, abbreviated as PAN, and Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5464 W. Wu 20 m for XS bands) allows a more accurate change rate to be discriminated (around 5–33 my21, see §3) and thus seems optimal for this research. In fact, as one of the most frequently used remote sensing data, SPOT images have been widely used for monitoring and assessing coastal changes and shoreline evolution (Populus et al. 1991, Chen 1998, Courel et al. 2000, Lin et al. 2001, Maktav et al. 2002, Wu et al. 2002b, Abou Dagga et al. 2004, Bayram et al. 2004, Elkoushy and Tolba 2004). Some national and international programmes, e.g. National Coastal Environment Programme (PNEC) in France (http://www.programme-pnec.org/), NOAA Coastal Change Analysis Program (C-CAP) in the USA (http:// www.csc.noaa.gov/), Global Monitoring for Environment and Security (GMES) in Europe (http://europa.eu.int/comm/space/gmes/whatisgmes_en.htm), all make use of SPOT images as one of the major information sources for coastal ecosystem research. Moreover, with decadal time-series of coverage since 1986, SPOT images are also cost-effective. The author was thus highly encouraged to select multitemporal HRV, particularly PAN images shown in table 1, for evaluating the coastline evolution in the region of Nouakchott. For change detection, it is usually necessary to acquire multi-date images around the same dates of the year or rather ‘anniversary dates’ (Lillesand and Kiefer 2000) of different years to avoid the seasonal difference in both spectral radiation (Sun elevation angle, Sun–Earth distance, etc.) and surface reflection. However such an acquisition relies to a great extent upon the availability of images. If there happened to be some clouds over the area of interest when satellites were passing, the images acquired can not be used for this purpose. Fortunately, the study area is located in a low-latitude region (latitude: ¡18u N). Winter images (with Sun elevation angle: 38– 44u) may also be useful for monitoring change. These are the reason that the images selected in table 1 are not on the same dates of the year. To analyse coastline change, it is necessary to clarify first which coastline we are concerned with in this research and its influencing factors. ‘Coastline’ has been numerously defined and can be summarized as (1) outline of a coast and (2) the boundary between land and sea/water. However, with ‘zooming’ this outline or boundary is amplified and becomes a concrete coastal zone, a completely dynamic Table 1. SPOT images acquired and their corresponding tide data. Tidal rangea Date Image types 3 November SPOT 1 XS 1989 converted into PAN 4 February SPOT 3 1995 PAN SPOT 1 11 November PAN 1999 22 January SPOT 4 2001 PAN a Max Min Spatial height resolution height (m) Time (m) (m) Time Tidal height when SPOT Passing time of passingb SPOT (m) 20 1.74 1:44 0.41 7h59 11 h 44 1.32 10 1.71 2:40 0.35 8h43 11 h 46 1.10 10 1.82 1:04 0.30 7h15 11 h 40 1.73 10 1.95 11:23 0.13 5h07 12 h 1.90 Tidal data were provided by Ibrahima Dia, University of Nouakchott (2002). Tidal heights were calculated according to the regularity of the semidiurnal tide variation. b Coastline evolution 5465 interface between land and water being subject to tide and wave actions, alongshore and cross-shore erosion and sedimentation, storm surge (Chang 1997, Komar 1998, Boak and Turner 2005) and anthropogenic activity. Due to its dynamic nature, the coastline considered in the research is neither the high water-line nor the low one but the Instantaneous Water-Line. This line, generally associated with tidal movement, sea-level rising, wave and storm surges and artificial engineering, is located between the breaking zone and wet-sand belt. Its position can be determined according to the brightness difference of the two related zones in this case study (see §3). Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 2.2 Change detection and estimation algorithm To extract the changes from the remotely sensed data of different dates, several processing methods, for example, post-classification comparison, change vector analysis and image differencing (Weismiller et al. 1977, Gordon 1980, Jensen and Toll 1982, Quarmby and Cushnie 1989, Singh 1989, Lambin and Strahler 1994, Lillesand and Kiefer 2000) are possible at present. Some authors (Wu et al. 2003) suggested a ‘superposition vectorization’ approach to quantify the changes when these traditional methods are not applicable. Since there is an evident difference in the reflectance of the visible channels in water versus land, the differencing technique was applied to the multi-date SPOT images to highlight the littoral geomorphologic changes and coastline evolution. A Markov chain analysis was introduced to estimate the potential coastline evolution. This analysis will be discussed in detail in §3. Figure 2 shows the methodology involved in this study. The procedures are unfurled as follows: Figure 2. Methodology and procedures adopted in this study. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5466 W. Wu (1) Pre-processing The PAN image dated 1999 was first geometrically corrected based on a topographic map (IGN 1998) in datum WGS84 and projection UTM (28) within a RMS error of 1.5 pixels. Then this corrected image was used as a reference to rectify the other SPOT images listed in table 1. The RMS error of the image-to-image rectification comes between 0.21–0.31 pixels. The three channels XS1, XS2 and XS3 of the images from 1989 were calculated according to the formula: S(XS1 + XS2 + XS3)/3 to obtain a PAN equivalence but with a spatial resolution of 20 m. Then a relative atmospheric correction (Caselles et al. 1989) was conducted among the four PAN images to reduce atmospheric effect as they were acquired on different days of the year. It is assumed that the reflectance of the deep water surface remains constant throughout the year. The digital numbers (DNs) of the pixels located in the wave troughs of deep waters (.30 m in depth) were measured 50 times in each of the four PAN images. The average DN values are 32 in the image of 1989, 44 in 1995, 46 in 1999 and 28 in 2001. The one from 1999 was regarded as a standard and used to correct the others by either adding or subtracting a value to all pixels in the images of the other three dates. The atmospheric effect was thus relatively corrected. At last, the wave breakers that bring about confusion with the beach in all PAN images were visually recognized and masked. The beachside border of this mask is the Instantaneous Water-Line, or rather, coastline. This line was defined along the visually ‘darkest’ points between the wave-Breaking zone and Dry beach (white in the image). From the brightness histogram, there is a significantly recognizable narrow ‘valley’ (1–3 pixels in the PAN image from 1995, ,1 pixel in images 1989, 1999 and 2001) in a series of west–east profiles cutting Breaking zone, Wet-sand and Dry beach. The lowest points in the valley represent the transition from Wet-sand to Water or Water to Wet-sand. The line drawn along the darkest points can be thus regarded as the Instantaneous Water-Line with little difference from the real instantaneous land to shallow-water transition. This line was considered comparable in the observed images (see §4). Then the masked areas were given a DN value similar to the adjacent coastal water (75 in this case study) to avoid leaving them blank. (2) Differencing After pre-processing, the PAN images of four dates were incorporated into an image with multiple bands. A subtraction was then applied to two of these bands to carry out image differencing. A new three-band image was thus produced where band 1 represents the subtraction result of the digital numbers between PAN 2001 and 1999, band 2 between PAN 1999 and 1995 and band 3 between PAN 1995 and PAN 1989: Band 1~D2001
1999 ~DN of PAN 2001
DN of PAN 1999 ; Band 2~D1999
1995 ~DN of PAN 1999
DN of PAN 1995 ; Band 3~D1995
1989 ~DN of PAN 1995
DN of PAN 1989 : After subtraction, the D values are between 2149 and + 165. The brightness profiles of the differencing results are shown in figure 3. A thresholding technique was followed to highlight the beach accretion (significant positive) and sandbar erosion 5467 Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Coastline evolution Figure 3. Brightness profiles of the differencing results after masking the wave-breaking zones (IWL, Instantaneous Water-Line). (significant negative). The thresholds checked and considered reliable are listed in table 2. (3) Change extraction and colour composite A colour composition for the new three-band image was carried out after differencing and thresholding. To highlight the extension of the north beach, only positive change layers were integrated and given in RGB colour, the accretion from 1989 to 1995, 1995 to 1999 and 1999 to 2001 was clearly revealed in figure 4(a). Similarly, the erosion of the south beach was underlined by three negative change layers. A colourful image showing the erosion was formed (figure 4(b)). 5468 2.3 W. Wu Change quantification and results A vectorization was followed to quantify the beach accretion and erosion of different stages around the harbour and spur. The surface change rates from 1989 to 1995, 1995 to 1999 and 1999 to 2001 were measured and demonstrated in table 2. Table 2. Thresholds and littoral changes around the harbour, Port de l’Amitié. Position Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Up-drift side of the har bour Up-drift side of the spur Downdrift side of the har bour Change type and rate 1989–1995 1995–1999 1999–2001 Total Error* Threshold for accretion (DN) Accretion (km2) Average accretion rate (km2 year21) Accretion (km2) > + 70 > + 85 > + 40 0.411 0.078 0.334 0.070 0.170 0.136 0.916 0.023 0.081 0.002 0.022 0.008 0.008 0.038 0.007 Threshold for erosion (DN) Erosion (km2) Average erosion rate (km2 year21) (235 (230 (230 0.632 0.120 0.639 0.135 0.067 0.053 1.338 0.041 0.119 0.011 *The error represents the possible difference between the detected results and true changes and estimated based on the geometric correction RMS error. (a) (b) Figure 4. Zooms showing the coastal changes around the Port de l’Amitié in Nouakchott from 1989 to 2001 resulted from image differencing and thresholding. (a) Seaward accretion of the up-drift side and (b) landwards erosion of the down-drift side from 1989 to 1995, 1995 to 1999 and 1999 to 2001.The changed areas are quantified and listed in table 2. Available in colour online. Coastline evolution 5469 The linear change rates and evolution probabilities along the normal direction of the coastline were measured and listed in tables 4 and 5 (see §3). The change discrimination reveals that the north beach has extended by 0.92¡0.02 km2 at an average rate of 0.08 km2 year21 and the south beach was eroded by 1.34¡0.04 km2 at a rate of 0.12 km2 year21 during the period 1989–2001. The eroded area is larger than the accreted one. More than 15 km of the south beach has been in erosion. The latter has caused the disappearance of the lagoon barrier in 1995 and whole lagoon in 1999. A sand accumulation of around 0.04 km2 was observed on the up-drift beach of the spur (figure 4(a)). Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 3. Potential coastline evolution analysis A coastline evolves by beach erosion and sedimentation. These phenomena can be explained by the actions of the wave and tide and their related currents. Man has, however, become voluntarily and involuntarily an essential agent of the evolution of many littoral areas (Paskoff 1998). The geomorphologic changes in Nouakchott are due to a modification of the coastal hydrodynamic conditions, such as coastal current direction and velocity, resulting from the harbour construction. Knowledge of the coastal current and morphological changes in the past are useful for understanding the future potential evolution. 3.1 Wave and alongshore current Waves appear when the speed of wind exceeds 3–4 m s21. Their characteristics depend thus on the speed of wind, the duration and the dimension of waters. Generally, the following relations are used to characterize the waves (Verstappen 1977, Paskoff 1998):  v~gT=2p or v2 ~gl=2p and l~gT 2 2p ð1Þ where l is the wavelength in metres, T is the period in seconds, n is the velocity in m s21 and g is the acceleration of gravity (9.81 m s22). If the wavelength is measured, it is possible to calculate the wave propagation velocity. As other objects, waves can be identified in remote sensing images according to their different reflectance in different directions in the visible bands (Cuq 1991, Populus et al. 1991). Additionally, wave direction measurement can reveal the wave– current interaction (Cuq 1991). These are helpful for understanding the coastal current dynamics and littoral drift features. The wave propagation and the average wavelength were recognized and measured in the SPOT images by visual enhancement in the low DN section in the histogram. It is found that the waves mainly came from the northwest. In fact, such direction occupies 48.6% of the total waves (Courel 1998). It is thus the dominant direction. The results are shown in table 3 and figure 5. It is known from the wave propagation that a coastal current flows from north to south. This is the longshore component of the Canaries’ Current, which moves from northeast to southwest in the Mauritanian waters (Lanjamet 1988). It is this component that carries littoral drift and plays a key role in the coastal morphological changes. It puts sediments into movement and transports them when its velocity is higher than a certain criterion and deposits them when it slows down to certain limit. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5470 Figure 5. W. Wu Wave propagation observed in the SPOT images dated 1989, 1995 and 2001. Table 3. Wave propagation properties. Date Direction Incidence angle (a) Wavelength (l) Period (T) Wave velocity (n) 3 November 1989 4 February 1995 22 January 2001 N 30u W N 60u W N 40u W 60u 30u 50u 880 125 42 23.7 8.9 5.2 37.0 14.0 8.1 The velocity of the alongshore current depends on several factors: wave period, incidence angle, surging height, slope and roughness of the infralittoral seafloor (Paskoff 1998). It is not easy to exactly calculate this velocity. It can, however, be estimated based on some empirical formulas. One of them, proposed by Putnam et al. (1949) is often employed as the results derived from this formula have been confirmed by field measurements (Chang 1997). It has the following form: V~  4X
2 {1 zY 0:5 X ~646 Hp cosðaÞT {1 {ð2X Þ {1 0:5 ð2Þ Y ~ð2:28gH Þ0:5 sinðaÞ where V is the velocity of the littoral current in m s21, p is the slope of the infralittoral seafloor, H is the surging wave height in metres, a is the incidence angle of the wave relative to the shoreline and T is the wave period in seconds. From the bathymetric map, p is measured as 0.18% to 0.3% with an average of 0.24%. The height H varies from 1 m to 2 m with a mean of 1.5 m (Lanjamet 1988). Coastline evolution 5471 Using the a and T obtained from the SPOT images shown in table 3, the velocities of the littoral current were calculated and shown as follows: X ~0:005, Y ~155:120, and V ~0:87 m s
1 on 3 November 1989 ; X ~0:225, Y ~2:896, and V ~0:76 m s
1 on 4 February 1995 ; and Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 X ~0:288, Y ~4:435, and V ~0:81 m s
1 on 22 January 2001 : These results are not far away from the 0.5 m s21 reported by Lanjamet (1988). According to Dou et al. (1995) and Chang (1997), such currents can put sands with granulity of 0.01–4 mm into movement. If they meet obstacles, for example, breakwater, they will slow down or change their direction and discharge their carrying substances. Thus the wave actions and littoral currents modify the coastal geomorphology by either erosion or accretion. This is the mechanism of coastline evolution. Tide and its currents (flood and ebb) also play a certain role in littoral environmental evolution. Due to a microtidal range (,2 m, Lanjamet (1988) and table 1) and the low slope of the coastal seafloor, the tidal energy has been largely dissipated before reaching the shoreline. Therefore, the tide impacts on the shoreline modification would be less important than those of waves. The build-up of the Port de l’Amitié has modified the local wave and flow direction. That is why the up-drift accretion and down-drift erosion have occurred. In 1991, a spur was constructed to protect the down-drift beach from further erosion (figures 1, 4 and 5). As a matter of fact, it has caused a more serious landwards retreat as it modifies the wave direction as the harbour does (Wu et al. 2002b). 3.2 Prediction of coastline evolution In fact, a number of authors and researchers have carried out studies on the prediction of coastline changes by applying different approaches and methods, for example, linear methods such as linear regression analysis (De Vriend 1991, Crowell et al. 1997, Douglas and Crowell 2000, Galgano and Douglas 2000, Fenster et al. 2001); empirical orthogonal functions, canonical correlation, principal oscillation pattern (Larson et al. 2003); nonlinear techniques such as time-delay embedding, singular spectrum analysis, forecasting signatures, fractal analysis and neural networks (Southgate et al. 2003); coastal-tract approach (Cowell et al. 2003a,b); Airborne Topographic LIDAR data (Stockdon et al. 2002, Robertson et al. 2004) and so on. It was reported that linear regression analysis is a reliable approach for long-term period prediction of shoreline trend (30 + years) where physical changes such as opening of inlets or shore engineering are absent (Crowell et al. 1997, Douglas et al. 1998, Galgano and Douglas 2000). This implies, however, its imperfection and even inapplicability for a coastline with anthropogenic intervention, e.g. harbour construction, where hydrodynamic features have been modified. Other linear and nonlinear approaches are useful both for natural and anthropogenically affected coastal systems but need longtime data series with both higher spatial and temporal resolution (Larson et al. 2003, Southgate et al. 2003). Due to a paucity of the abundant and long time-series field observation data, the above mentioned approaches seem to be difficult to be applied in the case of Nouakchott. Thus, a Markov chain analysis is introduced here for predicting the coastline evolution as a complement of the above mentioned methods. 5472 W. Wu A Markov chain is a mathematical model for describing a certain type of process that moves in a sequence of steps through a set of states. It was introduced to geography by Brown (1970) and Collins et al. (1974) and applied to land use change by Burnham (1973), Bell (1974), Bell et al. (1977), Robinson (1978), Jahan 1986, Baker 1989, Brown et al. (2000) and to deforestation by Miller et al. (1978), Nualchawee et al. (1981) and Lambin (1994). Wang and Yang (1992) utilized this theory for analysing the birth and death process in demography. As Lambin (1994) reviewed, the advantage of Markov chain analysis lies in its mathematical and operational simplicity. For the landscape change study, the central mechanism of the Markov chain is the probability, Pij, which refers to the likelihood of transition or movement from state i to state j in a given time interval, represented by the fraction or percentage of land cover types. As to a discrete landscape model, the Markov process can be expressed as (Lambin 1994): Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 ntz1 ~M nt ð3Þ where, nt is a column vector, n5(n1,….nm), whose elements are the fractions of land cover area in each of m states at time t, and M is a m6m matrix whose elements, Pij, are the transition probabilities during the time interval from t to t + 1. The matrix M is row-standardized such that the sum of transition probabilities from a given state is always equal to one (Jahan 1986). The limitation is that the first order model can not be used for long-term prediction due to the possible change in transition probability. Furthermore, there is a difficulty with ascribing causality within the model, i.e. the transition probabilities are often derived empirically from multi-temporal land use and land cover maps or remote sensing images with no description of the process (Baker 1989, Lambin 1994). Coastline modification is in fact a continuous transition from land to sea or from sea to land. This continuous transition can be observed by multi-temporal sampling in time, e.g. repeatedly scanning by remote sensors. Each observation can be regarded as an instantaneous point of the process chain. The Markov chain model seems thus applicable in this evolution estimation. The components to be determined in nt are the linear accretion and erosion rates in the normal directions of the coastline. Generally, at any observation point on the coastline there exist two possible changes: accretion and erosion. The matrix M can be expressed as (Jahan 1986, Brown et al. 2000): qPa 1{Pa r t1{Pe Pe s ð4Þ where Pa represents accretion probability and Pe erosion probability. If the accretion and erosion rates are symbolized as Ra and Re, they can be respectively in the light of the equation (3) formulated as: Raðtz1Þ ~Pa RaðtÞ zð1{Pa ÞReðtÞ ð5Þ Reðtz1Þ ~ð1{Pe ÞRaðtÞ zPe ReðtÞ : ð6Þ As mentioned above, in a given section, the coastal evolution controlled by the littoral hydrodynamics is known: the up-drift side has been experiencing accretion, a transition from sea (state 1) to land (state 2); and the down-drift side suffering Coastline evolution 5473 Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 erosion, a transition from land (state 1) to sea (state 2). These are direction-oriented changes and there is only one possibility for the pixels in the given section. The transitional rates (equations 5 and 6) can be further simplified as: Up
drift side : Re ~0, Raðtz1Þ ~Pa RaðtÞ ð7Þ Down
drift side : Ra ~0, Reðtz1Þ ~Pe ReðtÞ : ð8Þ Usually, the transition probability is calculated by measuring the proportion of land cover (e.g. forest or farmland) of pixels within certain unit (e.g. a 565 pixels moving window) of two different dates. In the case of poor understanding of the change dynamics and other impacts, such a calculation is appropriate. However, it is not applicable for this case study as the change direction of a point on the coastline in the given section is known and unique either from sea to land (upper-drift accretion) or from land to sea (down-drift land erosion). Thus a relative transition probability approach is introduced here. For the upside beach, the accretion probability of a point along the normal direction of the coastline was calculated by ratioing the accretion distance (Da) at the observed point with the maximum accretion distance (Dam) in the same time period. The erosion probability was obtained in the same way. Knowing the transition probability, it is possible to estimate the future evolution rates in the normal directions of the coastline at sampling points and the forthcoming position for both sides (tables 4 and 5) according to equations (7) and (8). The results are shown in figure 6(a) and (b). Here the sampling points were randomly selected on the coastline of 1989. One requirement is that at the hydrodynamic changing part, sampling points were more densely chosen than in other parts. Dh is the distance of such a sampling point to the harbour, which was measured on the SPOT image. 3.3 Maximum accretion and harbour life-span The harbour life-span depends mainly on the depth of the harbour pool. After the maximum accretion, the alongshore drift will largely deposit into the harbour due to current diffraction and relenting. This will gradually lead to the end of harbour service if no dredging engineering is conducted. The question is how to estimate this maximum accretion limit. According to the past accumulation features under the actions of the coastal flow, a probable maximum accretion limit that completely covers the harbour breakwater was projected (figure 6(a)). The time for the accretion to reach this limit will be that of the harbour still in service, or rather, the life-span. This time was obtained according to the probable accretion rates—Ra(post-01) (table 4) and ranges from 12.10 to 14.38 years or 13 years¡6 months. That is to say, the Port de l’Amitié harbour is likely to arrive at the end of service after 2014–2015 if measures are not taken. 3.4 Erosion tendency It is inferred from the Markov chain analysis that around 0.94 km2 of land will be eroded again by marine abrasion from 2001 to 2011. The maximum land retreat will amount to 322 m (at sampling point No. 8) and the coastline behind the spur will Da (m) No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Dh (km) 89–95 95–99 99–01 Da(89–01) (m) Ra(89–01) (m year21) Pa(89–01) Ra(post 01) (m year21) 0.04 0.13 0.23 0.34 0.44 0.55 0.67 0.80 0.93 1.06 1.20 1.30 1.41 1.53 1.63 1.73 1.84 1.94 2.04 2.13 2.22 2.32 2.42 2.5 2.61 2.71 2.83 185.20 182.43 175.83 185.01 188.08 188.99 188.79 180.07 168.90 162.77 155.87 145.03 139.82 133.33 121.61 113.20 100.50 93.88 88.85 76.68 72.10 70.20 75.02 77.85 75.12 79.20 71.39 121.90 123.65 119.16 110.51 104.38 98.99 90.24 93.33 101.26 100.49 101.57 99.77 96.43 94.57 94.79 99.29 100.08 111.13 102.23 100.64 100.51 94.15 95.85 95.31 87.33 83.85 79.72 33.79 35.90 40.35 44.71 52.29 59.48 63.36 67.18 68.19 60.57 61.02 62.36 60.59 62.71 66.09 59.20 54.87 49.16 52.94 51.22 52.01 49.94 45.63 43.15 44.37 46.12 39.22 340.89 341.98 335.34 340.23 344.75 347.46 342.39 340.58 338.35 323.83 318.46 307.16 296.84 290.61 282.49 271.69 255.45 254.17 244.02 228.54 224.62 214.29 216.50 216.31 206.82 209.17 190.33 30.44 30.53 29.94 30.38 30.78 31.02 30.57 30.41 30.21 28.91 28.43 27.43 26.50 25.95 25.22 24.26 22.81 22.69 21.79 20.40 20.06 19.13 19.33 19.31 18.47 18.68 16.99 0.98 0.98 0.97 0.98 0.99 1.00 0.99 0.98 0.97 0.93 0.92 0.88 0.85 0.84 0.81 0.78 0.74 0.73 0.70 0.66 0.65 0.62 0.62 0.62 0.60 0.60 0.55 29.86 30.05 28.90 29.75 30.54 31.02 30.12 29.81 29.42 26.94 26.06 24.25 22.64 21.70 20.50 18.97 16.77 16.60 15.30 13.42 12.97 11.80 12.04 12.02 10.99 11.25 9.31 Dma(post (m) 01) 401.00 408.00 406.00 415.00 403.00 387.00 389.00 370.00 352.00 334.00 315.00 310.00 297.00 306.00 286.00 265.00 233.00 232.00 219.00 189.00 185.00 167.00 162.00 167.00 154.00 135.00 130.00 Tma (year) 13.43 13.58 14.05 13.95 13.20 12.48 12.91 12.41 11.97 12.40 12.09 12.78 13.12 14.10 13.95 13.97 13.89 13.98 14.31 14.09 14.27 14.15 13.45 13.89 14.01 12.01 13.97 W. Wu Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5474 Table 4. Accretion on the up-drift side of the harbour. Da (m) No. 28 29 30 Dh (km) 89–95 95–99 99–01 Da(89–01) (m) Ra(89–01) (m year21) Pa(89–01) 2.93 3.05 3.16 60.74 49.81 29.57 79.38 75.03 73.37 38.38 39.35 34.92 178.50 164.19 137.86 15.94 14.66 12.31 0.51 0.47 0.40 Ra(post 01) (m year21) 8.19 6.93 4.88 Dma(post (m) 01) 118.00 97.00 70.00 Tma (year) 14.41 14.00 14.33 Dh, distance from the sampling point to the harbour; Da(89–01), accreted distance from 1989 to 2001; Ra(89–01), accretion rate from 1989 to 2001; Pa(89–01), relative accretion probability from 1989–2001; Ra(post 01), accretion rate after 2001 estimated from equation (7); Dma(post 01), maximum accretion distance after 2001 projected from the alongshore hydrodynamics and past evolution trend; Tma, time needed to reach the maximum accretion result from Dma(post 01)/ Ra(post 01). Coastline evolution Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Table 4. (Continued.) 5475 De (m) No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Dh (km) 89–95 95–99 99–01 De(89–01) (m) Pe(89–01) Re(89–01) (m year21) Re(post 01) (m year21) De(01–11) (m) 0.78 0.83 0.88 0.94 1.00 1.15 1.36 1.58 1.80 2.08 2.30 2.55 2.72 2.86 3.00 3.18 3.38 3.51 3.72 3.96 4.03 4.19 4.26 4.33 4.45 4.59 4.72 4.82 4.90 5.01 97.65 134.10 161.66 184.86 194.99 206.36 216.85 199.23 186.34 182.81 173.88 165.80 147.55 135.14 128.87 116.24 105.06 103.78 86.00 71.49 59.89 53.59 50.14 42.30 36.67 36.92 32.12 31.71 29.13 25.82 45.31 45.09 47.55 50.16 61.09 98.45 125.46 150.39 159.02 146.84 145.12 132.52 127.73 129.31 126.94 130.24 122.78 121.34 115.45 111.09 110.96 104.34 105.43 97.57 87.43 85.76 84.90 82.83 83.05 82.76 0 21.90 27.58 33.20 31.32 19.58 11.16 12.06 12.64 10.88 12.97 16.34 21.54 16.91 9.02 10.55 8.56 7.16 13.85 8.91 7.02 6.92 0 10.35 14.32 7.16 4.71 0 0 0 142.96 201.09 236.79 268.22 287.40 324.39 353.47 361.68 358.00 340.53 331.97 314.66 296.82 281.36 264.83 257.03 236.40 232.28 215.30 191.49 177.87 164.85 155.57 150.22 138.42 129.84 121.73 114.54 112.18 108.58 0.40 0.56 0.66 0.74 0.80 0.90 0.98 1.00 0.99 0.94 0.92 0.87 0.82 0.78 0.73 0.71 0.65 0.64 0.60 0.53 0.49 0.46 0.43 0.42 0.38 0.36 0.34 0.32 0.31 0.30 12.76 17.95 21.14 23.95 25.66 28.96 31.56 32.29 31.96 30.40 29.64 28.09 26.50 25.12 23.65 22.95 21.11 20.74 19.22 17.10 15.88 14.72 13.89 13.41 12.36 11.59 10.87 10.23 10.02 9.69 5.05 9.98 13.84 17.76 20.39 25.98 30.84 32.29 31.64 28.63 27.21 24.44 21.75 19.54 17.31 16.31 13.80 13.32 11.44 9.05 7.81 6.71 5.97 5.57 4.73 4.16 3.66 3.24 3.11 2.91 50.45 99.82 138.42 177.60 203.91 259.77 308.43 322.93 316.39 286.26 272.05 244.42 217.49 195.43 173.14 163.09 137.96 133.19 114.43 90.52 78.10 67.09 59.75 55.71 47.30 41.62 36.58 32.39 31.07 29.10 W. Wu Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5476 Table 5. Erosion at the different observation points on the down-drift side of the harbour. De (m) No. 31 32 33 34 35 36 37 38 39 40 41 42 Dh (km) 89–95 95–99 99–01 De(89–01) (m) Pe(89–01) 5.13 5.23 5.33 5.43 5.55 5.66 5.74 5.85 5.92 6.02 6.09 6.18 23.88 25.20 27.88 30.78 29.43 32.11 36.64 34.59 34.94 30.69 29.05 27.96 78.48 73.99 72.62 65.13 59.69 52.63 51.71 47.79 39.31 40.61 41.57 40.33 0 0 0 3.32 5.31 8.29 6.69 6.18 7.32 3.55 0 0 102.36 99.19 100.50 99.23 94.43 93.03 95.04 88.56 81.57 74.85 70.62 68.29 0.28 0.27 0.28 0.27 0.26 0.26 0.26 0.25 0.23 0.21 0.19 0.19 Re(89–01) (m year21) 9.14 8.86 8.97 8.86 8.43 8.31 8.49 7.91 7.28 6.68 6.31 6.10 Re(post 01) (m year21) 2.59 2.43 2.49 2.43 2.20 2.14 2.23 1.94 1.64 1.38 1.23 1.15 De(01–11) (m) 25.87 24.29 24.93 24.31 22.01 21.37 22.30 19.36 16.43 13.83 12.31 11.51 Coastline evolution Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Table 5. (Continued.) De(89–01), eroded distance from 1989–2001; Re(89–01), erosion rate in the period 1989–2001; Pe(89–01), relative erosion probability from 1989–2001; Re(post 01), potential erosion rate after 2001 acquired from equation (8); De(01–11), probable landwards erosion distance in the period 2001–2011 derived from Re(post 01)610 years. 5477 5478 W. Wu have a significant north and northeast withdrawal (sampling points 1, 2, 3 and 4 in figure 6(b)) due to wave diffraction. 4. Discussion and conclusions Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Remote sensing techniques, together with Markov chain analysis have been applied to alongshore current hydrodynamic characterization, coastline evolution monitoring and estimation. This study shows that the results acquired seem coherent to the reality and the methodology is adaptive to such kind of research. Some points worthy of attention and results obtained are depicted as follows: (1) In the course of change detection, a number of influencing factors should be taken into account: (i) Image-to-image rectification whose RMS error should be controlled within a small value, for example, 0.5 pixels. Those in our research are 0.21 to 0.33 pixels. (ii) Tide effect. Since the Instantaneous Water-Line was considered as the coastline for monitoring change, it is important to analyse the comparability of this line in different years. As mentioned above, the region of Nouakchott is a microtidal area (,2 m). The beach slope is between 10u and 20u (Philippon 1999). The maximum intertidal zone between high and low tides is a belt with a horizontal width of 5.5–11.5 m along the coastline. These Instantaneous Water-Lines of different years were all situated in this belt. So the tidal effect on these Lines would be less than 1 pixel size of SPOT PAN (a) (b) Figure 6. Potential coastline evolution around the harbour, Port de l’Amitié, in the near future: (a) the accretion on the north beach of the harbour will probably reach its maximum limit in 13 years¡6 months (2014–2015) and (b) the possible coastline in the down-drift side in 2011, forecasted based on the evolution analysis using a Markov chain model. Available in colour online. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Coastline evolution 5479 images. Furthermore, as seen in table 1, a significant difference in tidal level is distinguished between the images from 1999 and 1995. Thus the error of subtraction may be produced between these two dates. However, a difference of 0.63 m on the beach may lead to a horizontal error of 1.7–3.6 m with a surface area of 0.008–0.015 km2 for the north beach and 0.015–0.030 km2 for the south coast. These fall into the error permissions (table 2). (iii) Sea-level rising due to climate warming. According to UNDP (1999) and IPCC (2001), the global sea-level has risen by 10 cm to 25 cm at an average rate of 0.1–0.2 cm year21 over the past 100 years. If we project this rate to our observed period 1989–2001, the average sea-level rising was only 1.1–2.2 cm. It is far less important than that of tidal influence and thus negligible. In general, coastal seasonal variation, storms and flooding events will also take a certain part in modifying coastal morphology, position of coastline and sand budget accumulation. However, whether these mentioned catastrophes have enhanced erosion and decreased accretion magnitudes or coastline movement in the study region has not been particularly investigated. It is suggested that this should be done in future research. (iv) Wave-breaking zone with a width of 5 m to 60 m in the different SPOT images. This zone is easily confused with the sand beach. It was perceived that the influence of this zone on the detection results is much more important than any other factor. It is thus necessary to mask it before image differencing. (2) A first-order Markov model was applied to evaluate the coastline evolution tendency in this study. Since this is a simple chain with a known change direction between two states (from land to sea or from sea towards land), imperfections and differences in the estimation results from the real evolution in the future are inevitable, especially with intervening human activity, which leads to change in the hydrodynamic condition and transition probability. The question arising here is which order of the model would be more suitable for such a case study and whether we should consider the transitional probability changing with time P(t). If this is the case, the calculation is more complex. Any comments or criticisms from experts in various fields are thus warmly welcome to improve the remote sensing based modelling and prediction procedures. Additionally, to produce a more reasonable estimation, it is better to apply multiple models to see whether they obtain the same or similar results. This work is to be completed in the future. (3) Artificial management has produced littoral environmental changes and coastline evolution in the region of Nouakchott. The north beach has extended by 0.92 km2 at a rate of 0.08 km2 year21 and will have another accretion of about 1.32 km2 when it reaches the maximum limit in 13 years¡6 months (2014–2015) while the harbour will gradually reach the end of service if no measure is taken. The south beach has been eroded by 1.34 km2 at a rate of 0.12 km2 year21. The maximum retreat from 1989 to 2001 is measured as 362 m and a potential of 322 m is estimated for the period 2001–2011. So another 0.91 km2 of land will be eroded in the next 10 years. (4) The coastal currents, especially the littoral drift, generated by the waves from the northwest and flowing mainly north–south at a velocity of 0.76–0.87 m s21, play an important role in coastal geomorphologic evolution. Such currents can put the sands with a grain size of 0.01–4 mm into movement and deposit them when they slow down or change their direction before the obstacle. It is this hydrodynamic factor that controls the beach erosion and accretion in Nouakchott. 5480 W. Wu These coastal changes have provoked serious modification of the littoral and urban environment. Any future management engineering should take warning from this lesson. Acknowledgment The author would like to thank Professor D. Pumain of the University of Paris I for her useful discussion on the usage of the Markov chain analysis and Professor G. B. Bénié of the University of Sherbrooke for his suggestion for the composition of this paper. Sincere thank to Dr B. Lacaze for reading the paper and giving me precious ideas for improvement. The author is also grateful to Mr I. Dia for providing the tidal measurement data, to Dr M. El Bah, Professor M.-F. Courel and Dr J. Le Rhun for their cooperation at the first stage of the research. Finally, the author would like to thank the reviewers for their pertinent comments and suggestions that allowed me to improve the paper. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 References ABOU DAGGA, N., COUREL, M.F., BADRAN, F. and ANSELM, B., 2004, The evolution study of the natural hazard zones by the Kohonen self-organising map classification. International Archives of Photogrammetry and Remote Sensing, 35, pp. 503–508. BAKER, W.L., 1989, A review of models of landscape change. Landscape Ecology, 2, pp. 111–133. BAYRAM, B., BAYRAKTAR, H., HELVACI, C. and ACAR, U., 2004, Coastline change detection using CORONA, SPOT and IRS 1D images. International Archives of Photogrammetry and Remote Sensing, 35, pp. 437–441. BELL, E.J., 1974, Markov analysis of land use change—an application of stochastic processes to remotely sensed data. Socio-Economic Planning Sciences, 8, pp. 311–316. BELL, E.J. and HINOJOSA, R.C., 1977, Markov analysis of land use change: continuous time and stationary processes. Socio-Economic Planning Sciences, 11, pp. 13–17. BOAK, E.H. and TURNER, I.L., 2005, Shoreline definition and detection: a review. Journal of Coastal Research, 21, pp. 688–703. BROWN, D.G., PIJANOWSKI, B.C. and DUH, J.D., 2000, Modeling the relationship between land use and land cover on private lands in the Upper Midwest, USA. Journal of Environmental Management, 59, pp. 247–263. BROWN, L.A., 1970, On the use of Markov chains in movement research. Economic Geography, 46, pp. 393–403. BURNHAM, B.O., 1973, Markov intertemporal land use simulation model. Southern Journal of Agricultural Economics, July, pp. 253–258. CARUBA, R. and DARS, R., 1991, Géologie de la Mauritanie (Nice Centre Régional de Documentation Pédagogique). CASELLES, V. and GARCIA, M.J.L., 1989, An alternative simple approach to estimate atmospheric correction in multitemporal studies. International Journal of Remote Sensing, 10, pp. 1127–1134. CHANG, R.F., 1997, Coastal Engineering Environment (in Chinese) (China: Qingdao Ocean University Press). CHEN, L.C., 1998, Detection of shoreline changes for tideland areas using multi-temporal satellite images. International Journal of Remote Sensing, 19, pp. 3383–3397. COLLINS, L., DREWETT, R. and FERGUSON, R., 1974, Markov models in geography. The Statistician, 23, pp. 179–209. COOPER, J.A.G. and MCLAUGHLIN S., 1998, Contemporary multidisciplinary approaches to coastal classification and environmental risk analysis. Journal of Coastal Research, 14, pp. 512–524. COUREL, M.-F., 1993, Les écosystèmes littoraux de l’Aftout es Sahéli en Mauritanie. Journées du Programme Environnement, CNRS/ORSTOM, Lyon, 13–15 January. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Coastline evolution 5481 COUREL, M.-F., 1998, Nouakchott, Site à risques multiples. Espace et Société en Mauritanie/ URBAMA, Fascicule de Recherche No.33, Tours, France. COUREL, M.-F., LETERRIER, E., LE RHUN, J., JAOUEN, X., ROLANDO, C. and BARRY, J.-P., 1996, Evolution récente d’un milieu lagunaire mauritanien: les écosystèmes littoraux de l’Aftout-es-sahéli. Sécheresse, 7, pp. 33–39. COUREL, M.-F., LETERRIER, E. and LE RHUN, J., 1998, L’environnement de la ville de Nouakchott: un site à risques multiples. Géochronique, 6, pp. 32–36. COUREL, M.-F., RUDANT, J.P. and CHAMARD, PH. C., 2000, SPOT et Radar, deux outils complémentaires d’analyse en milieux arides: l’exemple de Nouakchott en Mauritanie. Publications de l’Association Internationale de Climatologie, 12, pp. 228–236. COWELL, P.J., STIVE, M.J.F., NIEDORODA, A.W., DE VRIEND, H.J., SWIFT, D.J.P., KAMINSKY, G.M. and CAPOBIANCO, M., 2003a, The Coastal-Tract (Part 1): a conceptual approach to aggregated modeling of low-order coastal change. Journal of Coastal Research, 19, pp. 812–827. COWELL, P.J., STIVE, M.J.F., NIEDORODA, A.W., DE VRIEND, H.J., SWIFT, D.J.P., KAMINSKY, G.M. and CAPOBIANCO, M., 2003b, The Coastal-Tract (Part 2): a conceptual approach to aggregated modeling of low-order coastal change. Journal of Coastal Research, 19, pp. 828–848. CRACKNELL, A.P., 1999, Remote sensing techniques in estuaries and coastal zones—an update. International Journal of Remote Sensing, 20, pp. 485–496. CROWELL, M., DOUGLAS, B. and LEATHERMAN, S., 1997, On forecasting futture U. S. shoreline positions: A test of algorithms. Journal of Coastal Research, 13, pp. 1245–1255. CUQ, F., 1991, Etude des houles à la côte à partir d’images SAR et SPOT. Rapport de Recherche du Programme National de Télédétection Spatiale (PNTS) (Paris: CNRS). DANIEL, R.C., SHUXFORD, R.H. and MCCANDLESS D., 1998, Coastline mapping and identification of erosion hazard areas in Pacific County. Proceedings of the ESRI 1998 User Conference, San Diego, CA. Available online at: http://gis.esri.com/library/ userconf/proc98/PROCEED/TO2SO/PAP214/P214.HTM DE VRIEND, H.J., 1991, Mathematical modelling and large-scale coastal behaviour, Part 1: physical processes. Journal of Hydraulic Research, 29, pp. 727–740. DE VRIEND, H.J., 1991, Mathematical modelling and large-scale coastal behaviour, Part 2: physical processes. Journal of Hydraulic Research, 29, pp. 741–753. DOU, G.R., DONG, F.W. and DOU, X.P., 1995, Research on the mathematical models of the coastal embouchure sands (in Chinese with English abstract). Sciences in China (A), 25, pp. 995–1001. DOUGLAS, B.C., CROWELL, M. and LEATHERMAN, S.P., 1998, Considerations for shoreline position prediction. Journal of Coastal Research, 14, pp. 1025–1033. DOUGLAS, B.C. and CROWELL, M., 2000, Long-term shoreline position prediction and error propagation. Journal of Coastal Research, 16, pp. 145–152. ELKOUSHY, A.A. and TOLBA, E.R.A., 2004, Prediction of shoreline change by using satellite aerial imagery. International Archives of Photogrammetry and Remote Sensing, 35, pp. 209–214. FENSTER, M.S., DOLAN, R. and MORTON, R.A., 2001, Coastal storms and shoreline change: signal or noise? Journal of Coastal Research, 17, pp. 714–720. FRIHY, O.E., DEWINDAR, K.M. and BANNA, L.M., 1998, Natural and human impact on the northeastern Nile delta coast of Egypt. Journal of Coastal Research, 14, pp. 1109–1118. GALGANO, F.A. and DOUGLAS, B.C., 2000, Shoreline position prediction: methods and errors. Environmental Geosciences, 7, pp. 23–31. GORDON, S., 1980, Utilizing Landsat imagery to monitor land use change: a case study in Ohio. Remote Sensing of Environment, 9, pp. 189–196. GORMAN, L., MORANG, A. and LARSON, R., 1998, Monitoring the coastal environment, Part IV: mapping, shoreline changes and bathymetric analysis. Journal of Coastal Research, 14, pp. 61–92. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5482 W. Wu HANSON, H., AARNINKHOF, S., CAPOBIANCO, M., JIMÉNEZ, J.A., LARSON, M., NICHOLLS, R.J., PLANT, N.G., SOUTHGATE, H.N., STEETZEL, H.J., STIVE, M.J.F. and DE VRIEND, H.J., 2003, Modelling of coastal evolution on yearly to decadal time scales. Journal of Coastal Research, 19, pp. 790–811. HSU, T., OU, S. and TZANG, S., 2000, Evaluations on coastal topographical changes at Hualien Coast, Taiwan. Journal of Coastal Research, 16, pp. 790–799. HWANG, C.S. and KU, C.Y., 2004, Exploring the optimum spatial resolution for satellite imagery. In GIS for Coastal Zone Management, D. Bartlett and J. Smith (Eds), pp. 77–94 (London: Taylor & Francis). IGN, 1998, Carte Topographique de Nouakchott (Paris: IGN). IPCC (Working Group II), 2001, Coastal zones and marine ecosystems. In Climate Change 2001: Impacts, Adaptation and Vulnerability, Available online at: http://www.grida.no/ climate/ipcc_tar/wg2/286.htm. JAHAN, S., 1986, The determination of stability and similarity of Markovian land use change processes: a theoretical and empirical analysis. Socio-Economic Planning Science, 20, pp. 243–251. JENSEN, J.R. and TOLL, D.R., 1982, Detecting residential land use development at the urban fringe. Photogrammetric Engineering and Remote Sensing, 48, pp. 629–643. JENSEN, J.R., RAMSEY III, E., CHRISTENSEN, E., MACKERY, H.E. and SHARITZ, R., 1987, Inland wetland change detection using aircraft MSS data. Photogrammetric Engineering and Remote Sensing, 53, pp. 521–529. JOL, H.M., PETERSON, C.D., VANDERBURGH, S. and PHIPPS, J., 1998, GPR as a regional geomorphic tool: shoreline accretion/erosion along the Columbia River littoral cell. Proceedings of the Seventh International Conference on Ground Penetrating Radar (GPR’98), May 27–30, University of Kansas, Lawrence, KS, USA, 1, p. 257–262. JOL, H.M., SMITH, D.G. and MEYER, R.A., 1996, Digital ground penetrating radar (GPR): a new geophysical tool for coastal barrier research (examples from the Atlantic, Gulf, and Pacific coasts, USA). Journal of Coastal Research, 12, pp. 960–968. KOMAR, P.D., 1998, Beach Processes and Sedimentation (Upper Saddle River, NJ: PrenticeHall, Inc.). LANJAMET, I., 1988, La Grande Plage mauritanienne (géographie, écologie et faune), Centre culturel français de Nouakchott. Collection ‘Connaissance de la Mauritanie’ (Paris: Centre National de Recherche Scientifique). LAMBIN, E.F., 1994, Modelling Deforestation Process (A Review). Tropical ecosystem environment observations by satellites, TREES Series B: Research Report no. 1, EUR15744EN. LAMBIN, E.F. and STRAHLER, A.H., 1994, Change-vector analysis in multitemporal space: a tool to detect and categorize land cover change processes using high temporal resolution satellite data. Remote Sensing of Environment, 48, pp. 231–244. LARSON, M., CAPOBIANCO, M., JANSEN, H., ROZYNSKI, G., SOUTHGATE, H.N., STIVE, M., WIJNBERG, K.M. and HULSCHER, S., 2003, Analysis and modeling of field data on coastal morphological evolution over yearly and decadal time scales. Part 1: Background and linear techniques. Journal of Coastal Research, 19, pp. 760–775. LILLESAND, T.M. and KIEFER, R.W., 2000, Remote Sensing and Image Interpretation, 4th edn (New York: John Wiley & Sons, Inc.). LIN, T.H., LIU, G.R., CHEN, A.J. and KUO, T.H., 2001, Applying satellite data for shoreline determination in tideland area. Proceedings of the 22nd Asian Conference on Remote Sensing, 5–9 November, Singapore. Available online at: http://www.crisp.nus.edu.sg/ acrs2001/pdf/064LIN.PDF MAIA, L.P., JIMENEZ, J.A., SERRA, J., MORAIS, J. and SÁNCHEZ-ARCILLA, A., 1998, The Fortaleza (NE Brazil) Waterfront: port versus coastal management. Journal of Coastal Research, 14, pp. 1284–1292. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 Coastline evolution 5483 MAKTAV, D., SUNAR ERBEK, F. and KABDASLI, S., 2002, Monitoring coastal erosion at the Black Sea coasts in Turkey using satellite data: a case study at the Lake Terkos, northwest Istanbul. International Journal of Remote Sensing, 23, pp. 4115–4124. MARCEAU, D.J. and HAY, G.J., 1999, Remote sensing contributions to the scale issue. Canadian Journal of Remote Sensing, 25, pp. 357–366. MILLER, L.D., NUALCHAWEE, K. and TOM, C., 1978, Analysis of the dynamics of shifting cultivation in the tropical forest of northern Thailand using landscape modelling and classification of Landsat imagery. Proceedings of the 12th International Symposium on Remote Sensing of Environment, pp. 1167–1185 (Ann Arbor: Environment Research Institute). MOORE, L.J., 2000, Shoreline mapping techniques. Journal of Coastal Research, 16, pp. 111–124. NUALCHAWEE, K., MILLER, L., TOM, C., CHRISTENSON, J. and WILLIAMS, D., 1981, Spatial inventory and modelling of shifting cultivation and forest land cover of northern Thailand, with inputs from maps, airphotos and Landsat. Remote Sensing Center Technical Report, No.4177, Texas, Texas A&M University, College Station. PASKOFF, R., 1998, Les littoraux, impacts des aménagements sur leur évolution, Troisième édition (Paris: Armand Colin/Masson). PHILIPPON, S., 1999, La côte de Nouakchott (Mauritanie): une évolution naturelle et anthropique, Mémoire de Maı̂trise, sous la direction de Prof. Courel, Université de Paris 1, France. POPULUS, J., ARISTAGHES, C., JONSSON, L., AUGUSTIN, J.M. and POULIQUEN, E., 1991, The use of SPOT data for wave analysis. Remote Sensing of Environment, 36, pp. 55–65. PUTNAM, J.A., MUNK, W.H. and TRAYLOR, M.A., 1949, The prediction of longshore currents. Transactions of the American Geophysical Union, 30, pp. 337–345. QUARMBY, N.A. and CUSHNIE, J.L., 1989, Monitoring urban land cover changes at the urban fringe from SPOT HRV imagery in south-east England. International Journal of Remote Sensing, 10, pp. 951–963. QUATTROCHI, D.A., and GOODCHILD, M.F. (Eds),1997, Scale in Remote Sensing and GIS (Boca Raton, FL: Lewis Publishers). RAFFY, M., 1994, Change of scale theory: a capital challenge for space observation of Earth. International Journal of Remote Sensing, 15, pp. 2353–2357. RAMSEY III E.W. and NELSON, G.A., et al. 1998, Generation of coastal marsh topography with radar and ground-based measurements. Journal of Coastal Research, 14, pp. 1158–1164. ROBERTSON, W.V., WHITMAN, D., ZHANG, K. and LEATHEKMAN, S.P., 2004, Mapping shoreline position using curborne laser altimetry. Journal of Coastal Research, 20, pp. 884–892. ROBINSON, V.B., 1978, Information theory and sequences of land use: an application. Professional Geographer, 30, pp. 174–179. SINGH, A., 1989, Digital change detection techniques using remotely sensed data. International Journal of Remote Sensing, 10, pp. 989–1003. SOUTHGATE, H.N., WIJNBERG, K.M., LARSON, M., CAPOBIANCO, M. and JANSON, H., 2003, Analysis of field data of coastal morphological evolution over yearly and decadal timescales. Part 2: non–linear techniques. Journal of Coastal Research, 19, pp. 776–789. STAFFORD, D.B. and LANGFIELDER, J., 1971, Air photo survey of coastal erosion. Photogrammetric Engineering, 37, pp. 565–575. STOCKDON, H.E., SALLENGER (JR)., A.H., LIST, J.H. and HOLMAN, R.A., 2002, Estimation of shoreline position and change using airborne topographic LIDAR data. Journal of Coastal Research, 18, pp. 502–513. UNDP, 1999, Climate Change Information Kit, Available online at: http://www.undp.org/gef/ new/ccinfo.htm. Downloaded By: [Wu, Weicheng] At: 08:30 30 November 2008 5484 Coastline evolution VERSTAPPEN, H. Th., 1963, The application of aerial photograph interpretation in geomorphological research. Geogr aphical. Studies of the Polish Institute of Scrence, 46, pp. 122–126. VERSTAPPEN, H. Th., 1977, Remote Sensing in Geomorphology (Amsterdam: Elsevier). WANG, Z. and YANG, X., 1992, Birth and Death Processes and Markov Chains (New York: Springer-Verlag and Science Press of China). WEISMILLER, R.A., KRISTOF, S.J., SCHOLZ, D.K., ANUTA, P.E. and MOMEN, S.A., 1977, Change detection in coastal zone environments. Photogrammetric Engineering and Remote Sensing, 43, pp. 1533–1539. WU, W., 2003, Application de la géomatique au suivi de la dynamique environnementale en zones arides. PhD thesis, Université de Paris 1, France. WU, W., LAMBIN, E.F. and COUREL, M.-F., 2002a, Land use and cover change detection and modelling for North Ningxia, China. Proceedings of the Map Asia 2002, Bangkok, Thailand, 7–9 August 2002. Available online at: http://www.gisdevelopment.net/ application/environment/land/index.htm. WU, W., COUREL, M.-F. and LE RHUN, J., 2002b, Coastal geomorphological change monitoring by remote sensing techniques in Nouakchott, Mauritania. Proceedings SPIE, 4886, pp. 667–679. WU, W., COUREL, M.-F. and LE RHUN, J., 2003, Application of remote sensing to the urban expansion analysis for Nouakchott, Mauritania. Geocarto International, 18, pp. 17–24.