As the largest low-lying river delta in the world, located at the confluence of the mighty Ganges-Brahmaputra-Meghna rivers, and as one of the most densely populated countries with more than 163 million people, Bangladesh already faces tremendous vulnerability. Accelerated sea-level rise, along with tectonic, sediment load and groundwater extraction induced land uplift/subsidence, have exacerbated Bangladesh’s coastal vulnerability. Climate change has further intensified these risks with increasing temperatures, greater rainfall volatility, and increased incidence of intensified cyclones and cyclone-induced storm surges, in addition to its seasonal transboundary monsoonal flooding, tides, large seasonal river discharges along with the associated sediment transport causing load/compaction of the coastal regions. As a result, Bangladesh coastal region has become the most dynamic region with the highest erosion and accretion rate in the world. For decades, the shape of the shoreline has changed greatly affecting millions of people living in the region. Our objective is to quantify the long-term or multi-decadal, seasonal shoreline changes for coastal Bangladesh to assess the impacts of the complex geophysical and climatic processes.
In this study, the shoreline from 1970’s to the year 2017 are extracted from a four-decade time-series of Landsat imagery. An automated shoreline extraction method based on Google Earth Engine (GEE) Application Programming Interface (API) is developed and applied to quantify Bangladesh coastal shoreline changes. This method involves Normalized Difference Water Index/Modified Normalized Water Index (NDWI/MNDWI) and the Otsu Threshold Method to enhance the accuracy of the digital imagery processing. The extracted Landsat imagery shorelines in three example regions are validated by comparing with independent DigitalGlobe and with CNES/Airbus higher resolution imagery at several m using Google Earth (GE). We concluded that the extracted Landsat shoreline distance in this study are in general <30 m accuracy, which is the Landsat pixel size, and thus validating our data processing method.
The multi-decadal shoreline change rate is computed using Digital Shoreline Analysis Software (DSAS). End Point Rate (EPR) and Weighted Linear Regression (WLR) rate are computed in a comparison, to quantify the shoreline change rate in eleven study regions depicting the entire Bangladesh coastal region. In addition, we provide an estimate of uncertainties for each averaged shoreline study location, within a given time span, for the whole 40-year time series, for each of the study regions for both erosion and accretion processes.
Seasonal shoreline variations are also investigated by comparing the quarterly least cloud covered Landsat image during the same year. We find a significant correlation causing shoreline changes at seasonal scales with the monthly precipitation data and seasonal river discharges which presumably are correlated with sediment load. For the study regions to assess seasonal shoreline variations, we find that several hundred meters of shoreline changes are primarily caused by seasonal river discharge.
The impacts on the shoreline change due to storm surges is investigated by an exhaustive search for the before and after storm Landsat imagery. Two severe cyclone storms with several meters’ surge height which occurred during 1991 and during 1995 are investigated. The observed water extent change is used to evaluate the impacts of these two example cyclones. We find significant correlations between these cyclones at high wind speeds and the shoreline erosions, resulting from these cyclone-induced storm surges. Selected coastal regions with Landsat observed long-term shoreline changes, namely due to erosion and sediment accretion, are compared with a published study, which has a much shorter data span using Landsat imagery than the four-decade long data span used in this study. However, they indicated qualitative agreement, partially confirming the validity of the shoreline retrieval technique.
Finally, the results have revealed the comprehensive dynamic change of Bangladesh coastal region at multi-decadal time span, for the first time, providing the most thorough and accurate shoreline change analysis to-date. Results of this study have shown the distinct features in different coastal regions in coastal Bangladesh. The east coast is accretion dominated. All the three regions of Chittagong, Kutubdia Island, and Cox’s Bazar have shown more sediment accretion than erosion. The Chittagong north region has an average accretion rate of 12.93±9.25 m/yr, while the average erosion rate is 1.20±3.61 m/yr. The Chittagong region shoreline has an average accretion rate of 3.46±8.18 m/yr, while the average erosion rate is 6.85±1.50 m/yr. The Kutubdia Island shoreline has an average accretion rate of 0.93±1.07 m/yr, while the average erosion rate is 7.37±1.17 m/yr. The Cox’s Bazar region shoreline has an average small accretion rate of 0.12±1.97 m/yr, while the average erosion rate is 1.61±1.37 m/yr. The central estuary is the most dynamic region with the highest erosion and accretion rate in the whole Bangladesh coastal region. The maximum erosion rate of 303.50 m/yr occurred on the Hatiya Island. The maximum accretion rate of 249.60 m/yr has been observed on the east bank of the central estuary. The Hatiya Island shoreline has an average accretion rate of 22.26±17.69 m/yr, while the average erosion rate is 49.48±3.19 m/yr. The Sandwip Island shoreline has an average accretion rate of 42.35±11.76 m/yr, while the average erosion rate is 38.41±3.17 m/yr. The west coastal region is erosion dominated with more than 60% of the shoreline is eroding. However, the erosion rate of is much smaller as compared with the central estuary. The Barguna region shoreline has an average accretion rate of 6.15±1.02 m/yr, while the average erosion rate is 8.59±0.62 m/yr. The Sundarban region has an average accretion rate of 23.40±8.10 m/yr, while the average erosion rate is 6.24±1.24 m/yr. Shorelines in the form of land area gain or loss for the central estuary study region in Bangladesh are computed for the last 40 years to illustrate the significant evolution of the coastal areas.
In summary, we provided a comprehensive quantification of Bangladesh shoreline change in terms of erosion and accretion over the last four decades. Our findings reveal that there is large variability in different parts of the Bangladesh coastal region with accretions due to excessive sedimentation transport, and erosions due to river discharge from inland, and storm surge from the ocean. These results are anticipated to improve future or other studies such as relative sea level rise at fine spatial scales, and improvement of cyclone-induced storm surge models.