Global distribution and dynamics of muddy coasts

Muddy coasts provide ecological habitats, supply food and form a natural coastal defence. Relative sea level rise, changing wave energy and human interventions will increase the pressure on muddy coastal zones. For sustainable coastal management it is key to obtain information on the geomorphology of and historical changes along muddy areas. So far, little is known about the distribution and behaviour of muddy coasts at a global scale. In this study we present a global scale assessment of the occurrence of muddy coasts and rates of coastline change therein. We combine publicly available satellite imagery and coastal geospatial datasets, to train an automated classification method to identify muddy coasts. We find that 14% of the world’s ice-free coastline is muddy, of which 60% is located in the tropics. Furthermore, the majority of the world’s muddy coasts are eroding at rates exceeding 1 m/yr over the last three decades.


Validation of pixel classification
The pixel-based image classification is subsequently validated against in situ sediment observations of sediment grain size distribution in the Dutch Wadden Sea (Rijkswaterstaat, 1998).The exposed coastlines of the barrier islands of the Wadden Sea are sandy, whereas the interior is muddy (in line with observations -see also Alonso et al. 1 ).The sheltered coastlines, on the mainland coast as well as the barrier island coasts, are sometimes vegetated and not shown here.Our analysis reveals that pixels classified as sand generally have a low mud content, with a peak at approximately 2%.Pixels are classified as muddy for a much wider range in the observed mud content.As a result, 99% of the pixels classified as sandy have a mud content below 25%, while only 64% of the pixels classified as muddy have a mud content exceeding 25%.The reason for low mud content in some pixels classified as muddy is that some sandy beaches are separated from clear seawater by a strip of turbid water which is incorrectly classified as mud.3/34

Validation of hybrid transect classification
For the transect-based performance, 75% of the 1,868 manually labelled training transects are used for training, and 25% for validation.Using the validation transects, we evaluate the accuracy of the multispectral classification method, the geospatial classification method, and the hybrid classification method.The f1-scores reveal that transects are most accurately classified for muddy coasts, and that adding the geospatial data especially improves the classification of rocky and vegetated coasts (compare the top plots in Supplementary Figure 3).The way the classifier is influenced by similar coastal types is visualised in a confusion matrix (see bottom plot in Supplementary Figure 3).
From the values of a confusion matrix the overall model accuracy (OMA) has been calculated using the following equation; OMA=(TP+TN)/(TP+TN+FP+FN); the number of correctly classified validation points (the sum of the diagonal) divided by the total number of validation points.
Supplementary Figure 3. Upper left: F1-score for four coastal types (sandy, muddy, rocky, vegetated), when using the multispectral transect classifier.Upper right: F1-score for four coastal types (sandy, muddy, rocky, vegetated), when using the hybrid transect classifier.The number of used training transects is on the x-axis and the overall accuracy (f1-score) on the y-axis.The solid line represents the mean accuracy (f1-score), the shaded area represents the 95th percentile of the accuracy (f1-score).Bottom: Confusion matrix of 25% of the transect training data used for validation.The predicted labels are on the x-axis, the actual labels on the y-axis.
The typical physical geospatial features are characteristic of different coastal environments.These features can therefore be used to optimise the classification.Therefore the density distribution of the muddy, sandy, rocky and vegetated coasts are visualised against the geospatial features using the 1,868 labelled training transects, see Supplementary Figure 4.  5a).In order to verify that this estimate is not heavily influenced by the sparsity of the training set, the same process was also applied in a large subset (N = 100,000 samples) from the world's predicted transects, resulting in a range of 2300 m (see Supplementary Figure 5b).was applied to avoid the influence of the spatial autocorrelation of muddy areas during the model performance test.To that effect we selected a range of R=3000 m (on the order of magnitude of the spatial decorrelation scale), and iteratively train-tested the model by removing all training points within distance R from the testing transect.This rendered a testing accuracy of 0.88 for muddy zones (and an average 0.73 for all classes), which is consistent with the baseline testing accuracy metrics.
Additionally, we tested the sensitivity of the range selection by evaluating [300,3000,30000,300000] m decorrelation ranges, resulting in a test accuracy value for muddy beaches of [0.87, 0.88, 0.86, 0.87] respectively.This insensitivity to the range selected may be explained by spatial consistency of world's muddy beaches features (which the stratified sampling scheme captures well). 6/34

Qualitative validation of muddy coast distribution using literature survey
To compare the model results an extensive benchmarking with literature has been conducted.This describes the detailed literature survey on the occurrence of muddy coasts along the worlds' coastline.A key source of information is the book 'Muddy Coasts of the World: Processes, Deposits and Function'; specifically, Chapter 6: Geographic distribution of muddy coasts 3 .In this chapter a detailed description is presented (of more than 100 pages) on the occurrence of muddy coasts and muddy systems following a stepwise approach along the entire world's coastline, based on an extensive literature survey.
Below we have copied relevant statements from Flemming (2002) 3 and associated >100 literature references and manually drawn polygons around the described areas.This resulted in an exclusive digital map of muddy coasts described by Flemming (2002) 3 , which has been used for validation of our model-predicted map of the global distribution of muddy coasts.
Supplementary Figure 6.Map showing the global distribution of outlined regions where muddy coasts have been reported according to Flemming (2002;top panel) and predicted muddy transects (lower panel).
Based on the descriptions by Flemming (2002) 3 , which vary in the level of detail depending on the availability of references, a total of 143 polygons of varying dimensions have been drawn by the authors.Forty-six of the 143 are associated to areas smaller than 50 km or which are not covered by our global coastal transect system; which can be explained by inland lakes, lagoons, etc..This yields 97 polygons in which significant presence of muddy coasts have been reported by Flemming (2002)  3 ; see top panel in Supplementary Figure 7.
In 95 of the 97 polygons we also predicted muddy transects.In fact, in 89 of the 97 polygons we found more than 5% of muddy transects.This high agreement (score of 92%) indicates that the model can well predict muddy transects at all muddy regions in the world, which can be explained by the global spreading of training locations.Supplementary Figure 7 presents the values for other thresholds for mud only, showing e.g. that in 80 out of 97 (i.e.82%) regions, up to 10% muddy transects have been predicted.The lower table presents the values for mud + vegetation, showing e.g. that in 96 out of 97 (i.e.99%) regions, up to 25% muddy transects have been predicted.
Furthermore, our analyses found 39,700 km of muddy coasts within all 89 polygons outlining the areas where Flemming (2002) 3 has reported the occurrence of muddy coasts.In addition, based on the model predictions, we have found a total of 51,700 km of muddy coasts that are located outside the Flemming polygons and hence can be seen as newly mapped muddy coasts.
Supplementary Figure 7. Relation between percentage mud (upper panel) and mud+vegetation (lower panel) found in Flemming regions and the number of Flemming regions identified.

Highly muddy regions
For each polygon the percentage of muddy transects is calculated.Twenty polygons reveal high percentage of muddy transects (all > 80%), representing a total length of about 12,500 km.When exploring the model results for these highly muddy regions, we regularly observe patches of transects labelled as 'vegetated' in a region governed by muddy transects.This can be explained by the fact that at some locations the vegetation directly faces the water when e.g. the beach has been eroded.Following this reasoning, we selected transects labelled as >80% mud and vegetation (assuming mud>60%), and found 38 polygons, adding up to about 42,000 km of highly muddy regions. 8/34

Validation using Eurosion Typology
To benchmark the performance of our classifier on muddy coasts, we utilized the Eurosion coastal typology 4 , which can be found in the EMODnet repository (https://emodnet.ec.europa.eu/en/geology).The Eurosion initiative 4     involved the aggregation of geological field data collected by local experts and national geological survey agencies.To ensure a robust comparison, we concentrated on specific geographical areas, namely the UK, German Wadden Sea, and French Atlantic coast (see Supplementary Figure 8).These areas were chosen because they only contain Eurosion field observationsother European regions contain additional observations that have not been validated yet.While the Eurosion coastal typology comprises 26 classes, our classification task focuses specifically on five coastal types.In this validation, we focused on the Eurosion coastal type "muddy coastline, including tidal flat, salt marsh."This decision was made to avoid arbitrary mapping between the 26 Eurosion coastal types and our five coastal types.Our validation process encompassed 4,317 coastal transects, which are classified as "muddy coastline including tidal flat, salt marsh" according to Eurosion (Salman et al., 2004)  To recap, our model has convincingly identified abundant muddy coasts globally in muddy coastal areas previously reported in literature.On top of that, the model has exposed a significant amount and length of new muddy coasts.Also, on a more local scale we see that the model is able to discriminate between sandy and muddy beach sections (see examples in Supplementary Figure 11).
The global transect dataset is shared through Zenodo (for details see Section: Data availability) and will be presented in an interactive viewer when publishing the manuscript. 12/34

Uncertainty statistics of shoreline changes of muddy coasts
We assumed that the adaption of the ShorelineMonitor for muddy coasts is justified, because the uncertainties are of the same order of magnitude as for the coastline dynamics of sandy coasts presented in Luijendijk et al.The bottom plot in Supplementary Figure 12 reveals that muddy coasts in general have higher change rates (erosion and accretion) than sandy coasts, thus we can conclude that muddy coasts are more dynamic than sandy coasts.
Supplementary Figure 12.Cumulative distribution functions of coastline change rate dynamics for muddy (blue) and sandy (orange) coasts (upper four plots).Change rate density distribution for muddy (blue) and sandy (orange) coasts (bottom). 13/34

Demonstration of annual muddy shorelines
Here, we demonstrate the dynamics using an expanding and eroding coastline as an example: an eroding muddy coast at the river mouth of the Bidyadhari River in West Bengal, India; an accreting muddy coast at the river mouth of the Pulau River, Papua, Indonesia.The local coastline changes evaluated above reveal realistic trends in erosion and sedimentation.
The river mouth of the Bidyadhari River in West Bengal, India, reveals high erosion rates in the period 1988 and 2016.
Erosion rates range between 30-40 m/yr over these three decades.An example of an accreting muddy coast is located at the river mouth of the Pulau River, Papua, Indonesia.This coast is facing the Arafura sea, which is a marginal sea of only 50-80 m depth 6 .The dynamics of this coastline reveals a high rate of accretion in the period 1988 to 2016.Accretion rates range between 20-30 m/yr over these three decades.

Examples of shoreline change time series per class
The following Supplementary Figures (14 -  In Ecuador the coastal character changes considerably, muddy shorelines being essentially restricted to three major localities and a number of small estuarinelagoonal sites (Ayon and Jara 1985).The second major mudflat/saltmarsh/mangrove system is developed in the shelter of Cojimies Peninsula.
Bordering the Golfo de Guayaquil is the third major and, with a total area exceeding 900 km2, also the largest muddy coastal sector of Ecuador.

Ecuador 7a
Brazil Between the Maranhense Gulf and the Orinoco delta, situated along the north-east coast of Venezuela, one of the longest continuous stretches of swampy and muddy mangrove-fringed coasts in the world is found Eisma et al. 1991;Milliman et al. 1982;Meade et al. 1985;Allison et al. 1996 Amazone -Brazil 7b French Guiana the coastal plain of French Guiana lies at an average altitude of 4 m and is composed of Holocene marine muds (20% fine silts and 60% clay, the latter predominantly comprising kaolinite) that have been accumulating over the past 8,000 years.Mud bank migration along the coast of French Guiana has been discussed by Froidefond et al. (1988).
Amazone    Perkins and Williams, 1966McManus, 1976Melusky, 1987 Solway Firth Forth Tay 13 Ireland There are many estuaries with mud flats and salt marshes along the Irish coast, but in general there is little information available on them in the international literature (e.g.Guilcher and King 1961).
Guilcher and King, 13 England Some of the better known muddy intertidal environments are The Wash, a lower macrotidal embayment along the east coast of England (Evans 1965(Evans , 1975;;Evans and Collins 1975), the muddy salt marshes along the coast of North Norfolk (Steers 1960;Pethick 1981), the mudflats along the Thames estuary, Evans 1965Evans , 1975;;Evans andCollins 1975 Steers 1960;Pethick 1981 North Folk Thames 14 Portugal Major estuaries are the Lagoon of Aveiro (e.g.Moreira et al. 1993;Cunha and Moreira 1995), the lagoon at the mouth of the Montego River (Morais 1985), the lower Tagus River or Mar dal Palha which is essentially a river mouth lagoon lined by large mud flats and salt marshes along the south-eastern shore and around the delta of the Tagus River in the north (e.g.Brotas and Catarina 1995) Moreira et al., 1993Cunha and Moreira, 1995Brotas and Catarina, 1995 Tego estuary Aveiro lagoon 14

Mediterranean
The most notable lagoons and coastal lakes are Mar Menor and Torrevieja just south of the Rio Segura mouth, and Albufera de Valencia south of the City of Valencia.
Mar Menor 14 Mediterranean Larger muddy swamps are only found in the Rhone delta, especially in the Carmargue (Massoud and Piboubes 1994).

Massoud and Piboubes Carmargue 15 Albania
The east coast of the Adriatic is mostly steep and rocky, coastal lowlands with a few muddy deltaic deposits only appearing along the coast of Albania, e.g.Karavastas and Narta Lagoons (Shuisky 1985a).

Shuisky 1985a
Narta Lagoon Karavastas the adjacent intertidal flats being today mainly composed of muddy sands with mud contents rarely reaching 50%.This essentially applies to the whole Wadden Sea shoreline between Skallingen in Denmark and Den Helder in The Netherlands.

16
Black Sea Coastal mud deposits are only found on the Cukurova deltaic plain associated with the delta swamps and marshes of the Seyhan and Ceyhan rivers in the southeastern corner of the country (e.g.Goney 1976).Goney 1976 17 Caspian sea between the Volga and the Ural, and in the large embayment to the east of the Ural River.Due to the aridity of this region, the exposed muddy coastal plain has given rise to the formation of clay dunes Volga Ural 18 West Africa As indicated above, the marshy and uniformly muddy shores of the Southern Rivers also occupy the whole coast of Guinea Bissau, the Republic of Guinea, and most of Sierra Leone (Anthony 1997).
Anthony, 1997 Gambia Senegal S Bissau Guinea Sierra Leone 18 Nigeria the Niger Delta at Mahin where a mud beach is found along the open coast (Ibe et al. 1989).In other places the swales between the ridges form muddy, elongated mangrove swamps which also reach far up into the lower valleys of the Yewa, Ogun, Oshun, and Shasha Rivers.The Niger Delta, which has the same shape as the Volta Delta, except on a very much larger scale, dominates the Nigerian coast by occupying over 60% of the roughly 800 km long coastline.The entire lower delta plain is occupied by muddy intertidal mangrove swamps (Allen 1965a).South Africa An important coastal wetland along the south-west coast, characterized by extensive mud flats and salt marshes, is Langebaan Lagoon (Day 1959;Flemming 1977).Day, 1959Flemming, 1977 Langebaan 19 Eastern Africa The coastal plain conditions observed along the north-east coast of South Africa continue into southern Mozambique up to l6°S, just north of Antonio Enes.This coastal sector is dominated by a number of large rivers such as the Maputo, Incomati, Limpopo, Save, Gorongose, Buzi, Pungue and Zambezi, and landlocked lagoons or lakes in the rear of coastal dune barriers, e.g.Lagoa Pelela (Hill et al. 1975).The rivers carry heavy loads of suspended sediments and are thus associated with extensive muddy mangrove swamps along their lower, estuarine reaches (e.g.Day 1974).Almost the entire west coast is occupied by muddy mangrove swamps, whereas along the east coast the mangrove environment is restricted to sheltered estuaries and the distributaries of the Kelantan River delta.Ong et al. 1980Ong et al. , 1991;;Koopmans 1964;Coleman et al. 1970 Malaysia -West; Malaysia -Kelantan 22f Malaysia Mud flats and mangroves also occur on both the west and the east coast of Penang Island which is located along the north-west coast.Irion 1983;Thom and Wright 1983;Pickup 1984;Wolanski et al. 1992;Alongi 1991;Alongi et al. 1992;Harris et al. 1993;Baker et al. 1995;Wolanski and Gibbs 1995 Papua New Guinea -South 23a Australia Beginning on the east coast of Victoria and proceeding in a clockwise direction, the first muddy coastal environments are associated with a number of microtidal estuaries such as those of the Cann, the Bemm and the Snowy River.

Bird 1985f
Australia -Cann; Australia -Bemm; Australia -Snowy 23b Australia Next in line are the Gippsland Lakes located in the rear of Ninety Mile Beach.Since 1889, when an artificial inlet was constructed, these lakes were converted into brakish water lagoons, with former reed swamps having been replaced by salt marsh vegetation.Bird 1965Bird , 1967bBird , 1983;;  The west of Venus Bay is Westernport Bay, the second large tidal basin found along the coast of Victoria.The inner parts of this upper mesotidal environment are lined by broad mud flats, salt marshes and mangroves.
Australia -Westernport 23f Australia Finally, Port Phillip Bay, at the head of which the city of Melbourne is situated, displays some mud flats and salt marshes near the entrance and in particular on socalled Mud Island which is located near the bay mouth.
Australia -Melbourne 23g Australia Just across the border into South Australia muddy salt marsh swamps of limited extend occur along the shores of a string of small coastal lagoons (Lake George, Lake St. Clair, and Lake Eliza) located between two calcarenite ridges between Rivoli Bay and Guichen Bay.
Australia -Rivoli 23h Australia In the Coorong barrier system narrow salt marsh belts are associated with carbonate muds in an evaporitic environment.New Caledonia On New Caledonia, for example, muddy mangrove swamps and salt marshes are found in the Bay of St. Vincent, at Mara and in association with the delta formations of the Dumbea, Néra, Poya, Iouanga, and Koumac rivers along the west coast and the Diahot River on the north coast.Baltzer 1965Baltzer , 1969;;Dugas 1974;Bird and Iltis 1985 New Caledonia -Vincent; New Caledonia -Dumbea; New Caledonia -Néra; New Caledonia -Poya; New Caledonia -Koumac; New Caledonia -Diahot 24c Fiji On the Fiji Islands, numerous small muddy mangrove pockets line sheltered embayments and river mouths (e.g. the Sambeto) on Viti Levu, the deltas of the Navua, Singatoka, Nandi, Mba, and Rewa Rivers being particularly good examples.

Bird 1985g
Fiji -Navua; Fiji -Singatoka; Viji -Nandi; Fiji -Rewa 25a China From the Yangtze delta to Haizhou Bay, the former mouth region of the Hwanghe River (Yellow River), the coast remains muddy.Kwon 1974;Wells and Huh 1979;Eisma and Park 1985;Park 1987;Alexander et al. 1990 South Korea 26 Russia n.a.; coastline is not considered in paper; no ice-free coastline

2 .
(a) Distribution of mud content, among sandy and muddy classified pixels.(b) Confusion matrix of pixel classification and mud content.(c) Map of the pixel classification and mud content of the Dutch Wadden Sea.

Supplementary Figure 4 .
Distribution plots for four coastal geospatial variables, based on 1,868 labelled transects.The vertical shaded bars show the (true) densities, the solid lines show the kernel density estimations for the four coastal types: sandy coasts, muddy coasts, rocky coasts and vegetated coasts.(a) Maximum elevation; (b) transition zone width; (c) tidal range; (d) latitude.Additionally, we tested the influence of the size of the training set in the performance of the predictions by drawing random class-balanced subsamples of different size, along with the dispersion induced by randomly subsampling the training-testing partition (Supplementary Figure 5a).This resulted in an average accuracy of 86.5% for muddy beaches during testing.The spatial dependence of mapped features may result in a biased validation of spatial classifications.This is comprehensively reviewed and discussed in Ploton et al. (2020) 2 .To avoid the effect of spatial autocorrelation in the fitting of our model, we first described the spatial structure of muddy labels.This was done by computing the indicator experimental semivariogram for the muddy class and fitting an exponential semivariogram model.This results in a decorrelation distance (range of the fitted exponential model) of 2900 m for the Training set (see Supplementary Figure

5 .
Spatial autocorrelation described by an empirical semivariogram and an exponential parametric model for (a) muddy transects from the training dataset (N=1868) and (b) from a subsample of the world's classified transects in our predicted model (N=100,000).The validation strategy buffered leave-one-out cross-validation (B-LOO CV method, as described by Ploton et al. (2020) 2 along coasts in the UK, France and Germany, with a total length of 6,936 km.Given the discrepancy in classes, we evaluated the performance of the classifier in terms of true positives (TP / (TP + FN)).When computing the recall over this validation set (N = 4,376), we find that the classifier has 82% sensitivity, indicating the model's accurate detection of muddy coasts.Supplementary Figure 8. Regions used (marked in orange) for validation with Eurosion coastal typology.Supplementary Figure 11.Examples of the model predicted transects (yellow = sandy, orange = muddy, green = vegetated).Panel A shows an area at the Amazone coast, where muddy transects alternate with vegetated transects.Panel B shows the clear transition from a muddy river section into a sandy beach environment in the north of Australia, while panel C shows the capability of the model to detect local differences in sediment type at a small river delta and adjacent sandy beaches.
Figure 12 shows the cumulative distribution function (CDF) plotted against the uncertainty bandwidth for the muddy and sandy coastlines.The change rate uncertainty represents the bootstrapped uncertainty of the linear fit through the time series data of the coastline positions.The uncertainty bandwidth is calculated as the change rate uncertainty divided by the change rate.For example, if a coastline has changed with a rate of 9 m/yr and the uncertainty was 3 m/yr, then the uncertainty bandwidth is 30%.The less uncertainty bandwidth the more accurate the change rate.
17) present examples of shoreline change time series per dynamic class as presented in Table 1 of the manuscript.Supplementary Figure 14.Examples of shoreline change time series for the class: very extreme dynamics.In the map the green transects represent accretion, red transects erosion.The time series plot presents the shoreline position over time for the transect within the yellow circle.10.Validation muddy shoreline change Literature search resulted in a few papers covering different locations, with varying environmental settings, where either 1) in-situ observations have been used or supported the determination of the reported shoreline changes or 2) some validation has been carried involving ground truthing with field work (e.g.individual shoreline positions).The following 3 Mao et al. (2021) presents a dataset for Australia on large-scale decadal shoreline change.A validation of shoreline change at a muddy beach over Roebuck Bay in Western Australia is presented in Fig. 7 in Mao et al. (2021) and shown here in Supplementary Figure 18.Mao et al. (2021) states that both digitization (blue line) and our method (red line) show a similar accretion trend within the AOI (Supplementary Figure 18b).The decadal change rates reported by Mao et al. (2021) for digitization of their shorelines agree reasonably well with our dataset: 0.05 vs. 0.036 km2/yr, contributing to our confidence in the derived long-term shoreline changes in muddy systems.Supplementary Figure 18.(a) Area of interest from Mao et al. (2021) for Roebuck Bay, (b) change in land area (km2) and rates reported by Mao et al. (2021), (c) the area selected to compare the change trends from ShorelineMonitor and (d) represents the corresponding change in land area from the ShorelineMonitor.19/34 2. Mana, French Guyana Brunier et al. (2019) reports on the exceptional rates and mechanisms of muddy shoreline retreat following mangrove removal in French Guyana.The paper presents the annual shoreline change rates over a stretch of 50km for three distinct periods between 1987 and 2014.They conducted field measurements at a survey site within this coastal stretch to verify their shoreline detection.Supplementary Figure 19 compares the change rates for the three periods spatially for both the reported values in Brunier et al. (2019) and the ShorelineMonitor dataset.Generally, the spatial patterns of erosion and accretion rates agree well for the first and last period.For period 2 some deviations are found around km30 where up to 200 m/yr erosion is reported (hence 1400 m in total) while the ShorelineMonitor shows erosion rates reaching values close to 100 m/yr.Overall, despite some deviations, the long-term spatial and temporal behaviour shows comparable results between the ShorelineMonitor and Brunier et al. (2019).Supplementary Figure 19.Annual change rates reported Brunier et al. (2019) (left column) and ShorelineMonitor dataset (right column) for the period 1987-1999 (upper row), period 1999-2006 (center row), and period 2006-2014 (lower row).Colors of the bars presented in the right column are made consistent with Brunier et al. (2019).

3 .
Fig on map country Description from Flemming References Polygon ID / name 2a Canada (British Columbia) is essentially a fjord coast and the only muddy sections of any significance are those formed by the Fraser and Skeena river deltas Kellerhals and Murray 1969; Luternauer 1980; Swinbanks and Murray 1981; Owens and Harper 1985 Fraser Panama, Costa Rica, Nicaragua and Belize has received little attention.Indications are that muddy shores are rare and widely the shores of Cuba, the largest of the Caribbean Islands (cf Bird 1985a), extensive mangrove swamps occupy the entire Peninsula de Zapata along the eastern margin of the Golfo de Bataban6, situated on the south-west coast.go unmentioned that extensive non-marine muddy wetlands occur along parts of the Great Lakes, in particular the north-eastern part of Lake Michigan, the east-central shores of Lake Huron, and the eastern shores of Lake Erie.muddy wetlands and tidal flats appears to be insignificant.Bodere, 1985 11 Iceland Arctic salt marsh develops along the muddy intertidal shores of cold regions in arctic Europe, and is known from northern Iceland EU 12 Norway Clayey beach deposits are occasionally found in wave-protected localities where muds are supplied by local streams or where a nearshore mud blanket emerges along a sheltered beach section.12 Sweden Most muddy deposits occur below sea-level and in icedammed lakes.12 Finland Muddy shorelines are restricted to sheltered pocket beaches, where muds winnowed from glacial till have accumulated 12 Finland The head of the Gulf of Finland is formed by the muddy marshes and wetlands of the Neva and Luga River deltas, much of the former having been reclaimed in former centuries for the construction of St. Petersburg.Muddy and marshy lowland shores are again found locally along lowland shores are again found locally along the Estonian Archipelago, including the eastern and southern shores of the Islands of Saaremaa and muddy wetland and marshy sections along the eastern shore of the Baltic Sea are associated with the sheltered lagoonal shores of the Kursiu Marios and Vistula Bay.
coast between Cap Saint Andre and the Mangoky delta is smoother, almost the entire coastline being lined by mangroves thriving in the shelter of sandy barrier beaches, especially along the muddy delta lobes of the rivers Manambolo, Morondava, Tsiribihina, and Mangoky.Madagascar 19Tanzania Muddy mangrove swamps occur in the sheltered bays and estuaries of these rivers, but also along the open coast north of the Pagani River.Similarly, most of the islands along the coast (e.g.Mafia, Zanzibar, and Pemba) have flat, swampy valleys along their western shores(Alexander 1985).coast of Iran initially remains flat, being characterized by the muddy swamps and marshes of the Zuhr eh Rud delta.continues into Kuwait and is subsequently displaced by the large deltaic swamps and muddy marshes of the combined Euphrates, Tigris, and Karun Rivers.Kuwait 21a Pakistan Near the Iranian border in the west the mouth region of the Dasht River is formed by a muddy coastal-plain swamp.Pakistan -Dasht 21b Pakistan Further east along the Makran Coastal Region a number of muddy coastal lagoons are encountered, the largest one being Kalmat Khor some 50 km west of Ormara.Pakistan -Kalmat Khor 21c Pakistan Large muddy tidal flats and mangrove swamps, comprising trees of Rhizophora conjugata and Avicennia alba, are found along the shores of Miani Lagoon, situated between Ras Malan and Ras Muari., presently inactive delta section tidal channels and mud flats are situated in the lee of sandy barrier bars, whereas open-coast mud flats characterize the active delta further south.Towards the border with India, the delta merges with the Rann of Kutch which forms a vast muddy salt-marsh swamp (e.g.Cori Creek) and which harbours only a few small pockets of mangrove shrubs. of the Rann of Kutch is situated on the Indian side of the border, the marshy coastline extending into the inner parts of the Gulf of Kutch which forms the northern margin of the Kathiawar Peninsula.The tidal range continues to increase, reaching upper macrotidal amplitudes in the inner Gulf of Khambhat (or Cambay) with 10.7 m at spring tide.As a consequence, the coast is lined by extensive mud flats that are backed by salt marshes and mangrove swamps.abound along the east coast of India.In contrast to the west coast major rivers discharge into the Bay of Bengal, several of them, e.g. the Cauvery, the Krishna, the Godavari, the Mahanadi, the Brahmani, and the Ganges, forming large river deltas.Sambasiva Rao and Vaidyanadhan 1979India -East 21g Bangladesh With the exception of the southern portion of the Chittagong coast, the shores of Bangladesh consist of muddy mangrove swamps or salt marshes.Bangladesh 22aBurma Extensive mangrove swamps occur around river mouths, especially in the coastal sector between the Kaladan River near Sittwe (Akyab) and the Sandoway River mouth 300 km further south.coast of Thailand muddy shores and mangrove swamps are less continuous, being concentrated in the Delta of the Pattani River in the south, on the coastal plain between Songkhla and Nakhon Si Thammarat which is characterized by a number of brakish-water lagoons, along the Ao Ban Don embayments north of Surat Thani, along the coast between Lang Suan and Chumpon, intermittently between Prachuao Khiri Khan and Hua Hin, along the entire north coast of the Gulf of Thailand south of Bangkok, and along the north-east coast between Klaeng and Trat south coast of Irian Jaya is occupied by lowland swamps associated with muddy estuaries, e.g.along the lower courses of the Digul, Mapi, and Pulau rivers.Indonesia -South Papua 22nPapua New Guinea Of the larger rivers only the Sepik, located along the north coast, has constructed a large protruding delta.Papua New Guinea -Sepik 22o Papua New Guinea Along the Gulf of Papua on the south coast the main rivers (e.g. the Fly, Wawoi, Turama, Kikori, Purari) open into estuarine channels between which extensive mangrove swamps, backed by nipa palms, are found.
Bay the muddy shoreline once more becomes embayed due to the irregularity of the coast.This coastal type continues past Qingdao and along the entire Shandong Peninsula up to Laizhou Bay, where the coastal physiography becomes controlled by the modern Hwanghe delta.China -Haizhou 25cChina West of the Hwanghe delta, along the shore of Bohai Bay up to the Luan River, the mud flats attain a width of up to 10 km.mud flats up to 2 km in width again appear at the head of Liaodong Bay, the source of the mud being the Liao River.25eTaiwan Muddy shores are found mainly along the west coast of Taiwan, where large amounts of sediment are brought down by local rivers to form wide intertidal sand and Yalu River delta eastwards into North Korea the coast is initially dominated by wide intertidal mud flats.The tidal range is about 4 m in this region.The mud flats continue up to the mouth of the Taedong Gang.appear west of Ongjin (North Korea).From here southwards the entire coast is fringed by mud flats of varying width.The west and south coasts of South Korea are extremel irregular, being characterized by numerous embayments formed by intricate headlands of outcropping igneous rocks.Most of these embayments are entirely mud-filled, the tidal flats growing upward and outward from the coast.
Venezuela to Santa Marta, just south of Cabo San Juan de Guia, the coast is rocky or sandy, being lined by some barrier beaches in its central reaches.The remainder of the coast, up to the border with Panama at Acandi, the coast is extensively lined by mud flats, the shoreward margins being fringed by mangrove swamps.
-French Guiana 7c SurinameThe coast of Surinam is 350 km long and, with a mean tidal range of 1.8 m, is lower mesotidal in character(Psuty 1985).Here the fine-grained coastal plain deposits are 20 km wide in the south-east, widening to a maximum of 140 km in north-west(Zonnefeld 1954).The general topography is dominated by groups of chenier ridges, composed of sand and shell material, separated by wide mud flats (e.g. ).
of coastal Cameroun is lined by muddy mangrove swamps, the only exception being the roughly 50 km section west of the capital Douala which consists of rocky cliffs eroded into the marginal volcanic cone of Mt.Cameroun(Schwartz  1985).The muddy mangrove-lined coast continues past Equatorial Guinea (Rio Muni) into the northern part of Gabon up to the mouth of the Ogooue River flats reappear in Exmouth Gulf and continue almost uninterrupted up to the delta of the De Grey River and reappear beyond Eighty Mile Beach to occupy almost the entire lower to upper macrotidal coast of the Kimberly district up to the boundary of the Northern Territory.extensive mangrove swamps and mud flats in their estuarine reaches are -from west to east -the Victoria, Daly, Mary, South Alligator, East Alligator, Mann, Goyder, Roper, Limmen Bight, McArthur, Robinson, and Calvert.coastal environments associated with muddy swamps and marshes are the estuary of the Clarence River, Lake Illawarra, Lake Coil, Turron estuary, the lagoons, bays and lakes between Port Macquarie and Sydney (e.g.Wallis Lake, Myall Lake, Nelson Bay, Lake Macquarie, Tuggerah Lake, Broken Bay, Port Jackson, and Botany Bay), the Shoalhaven River delta and Wapengo Lagoon located on the south coast.coast the Marion Bay barrier shelters well-developed salt marshes in Blackman Bay Lagoon, while similar deposits occur in Moulting Lagoon located in the rear of the Nine Mile Beach barrier.or salt marsh deposits associated with headland-bay coasts are found in eastern Northland (e.g.Rangaunu Bay, Doubtless Bay, and Bay of Islands), Auckland (e.g.Firth of Thames), Coromandel (e.g.Mercury Bay), and the Gisborne and Nelson districts (e.g.Poverty Bay).