SHORELINE CHANGE ANALYSIS USING DIGITAL SHORELINE ANALYSIS SYSTEM METHOD IN SOUTHEAST BALI ISLAND

Based on data from the Bali Public Works Office, in 1987 the abrasion reached 51.50 km, in 2003 it reached 86.5 km, and in 2006 it increased to 140 km. Coastline change research is needed for coastal environmental protection, mitigation, and sustainable development. The objectives of this research are: 1) To predict wind speed and direction for the last 30 years; 2) To measure changes in coastlines over the last 30 years (1989-2020); and 3) Comparison of changes in coastline in 4 periods 1989-2000; 2000-2010; 2010-2020 and 2016-2020. Digital Shoreline Analysis System (DSAS) is a method that works on ArcGIS software which is used to calculate shoreline changes based on time statistics and a geospatial basis. The results of the average EPR in 19892000 (Landsat imagery), the average abrasion value was -10.43 m/y and the average accretion value was 2.35 m/y; 2000-2010 the average value of EPR abrasion was -2.61 m/y and the average accretion value of 2.65 m/y; in 2010-2020 the average EPR abrasion value was -2.72 m/y and the average accretion value was 1.60 m/y while in 2016-2020 (Sentinel Image) the average abrasion value was -4.32 m/ y and the average value of its accretion is 4.50 m/y. The conclusion of this study 1) The average wind speed ranges from 0.2 to 6.4 m/s. Wind direction shows the dominance of the Australian continent (southeast). This shows that the east monsoon is more dominant than the west monsoon; 2) In the last 30 years (1989-2020) shoreline changes can be seen from the EPR graph with an average abrasion rate of -6.39 m/ y and an accretion rate of 3.15 m/y; and 3) Identification results from 1989-2000 the villages of Padangbai and Ketewel experienced extreme accretion and high abrasion; 2000-2010 Padangbai and Jumpai villages experienced high accretion and abrasion; In 2010-2020, Jumpai and Gunaksa Villages experienced high abrasion and moderate accretion, while 2016-2020 experienced high abrasion and accretion in Tangkas and Gunaksa Villages. For further research, it can include additional variables such as tide and wave data to get better results.


INTRODUCTION
The shoreline is a dynamic line between land and ocean in the coastal that is affected by tides. The shoreline consists of the lowest receding coastline, the highest tide, and the mean sea level (Cui and Li, 2011). The shoreline changes continuously; changes occur due to the land's erosion (abrasion) and the addition of land (accretion). The abrasion and accretion process is caused by sediment transport, tides, waves, currents, anthropogenic, and land use (Arief et al., 2011). Based on the Bali Public Works Office, in 2006 the length of the beach reached 427 km, in 1987 the abrasion reached 51.50 km, in 2003 the abrasion was 86.5 km, and in 2006 it rose to 140 km. The Padanggalak Beach area in Kesiman Village, East Denpasar Regency, Denpasar City, is also still experiencing abrasion. Monitoring of shoreline changes is needed for the study of coastal dynamics, protection of the coastal environment, mitigation map, sea transportation and sustainable development of coastal areas (Kasim, 2012;Putra et al., 2015).
The Digital Shoreline Analysis System (DSAS) is a software add-in to Esri ArcGIS desktop 10.4-10.6 that enables a user to calculate rate-of-change statistics from multiple historical shoreline positions. It provides an automated method for establishing measurement locations, performs rate calculations, provides the statistical data necessary to assess the robustness of the rates, and includes a beta model of shoreline forecasting with the option to generate 10 and/or 20-year shoreline horizons and uncertainty bands (USGS, 2021).
The DSAS method has advantages over other analysis methods, including it can be used for a massive area with a large number of transects, can compare changes in coastlines from different years, and can be used to predict changes in the future coastline (Sugiarta, 2018). Nugraha et al., (2017), The results of the analysis of shoreline changes from 1995 to 2015 in the southeastern island of Bali (Klungkung and Gianyar) are the dominant shoreline changes occurring on the coast of Gianyar Regency. The technology used in this research is remote sensing. This technology can monitor shorelines for several years and be analyzed using DSAS (Digital Shoreline Analysis System) software developed by the USGS (United States Geological Survey). This technology uses Landsat satellite imagery with an accuracy of 30 square meters using the band ratio method. Landsat satellite imagery was used for 30 years, from 1989-2020. The aims of this research will be formulated as to predict wind speed and direction for the last 30 years; measure shoreline changes over the last 30 years , and compare changes in the coastline in 4 periods 1989-2000; 2000-2010; 2010-2020 and 2016-2020.

METHODOLOGY
Research locations in the coastal sub-districts of Denpasar City, Gianyar, Klungkung, and Karangasem District. In this study, primary and secondary data were used. Primary data were measured as the present coastline as existing conditions in the research area. Secondary data was interpreted satellite images and wind data by agencies. Data processing was conducted in the last thirty (30) years from 1989-2020. The year of data processing was carried out in 1989;2000, 2010, and 2020. The dynamical shoreline changes in 1989-2000, 2000-2010, and 2010-2020 are compared to determine the changes. The Satellite data were processed in the last thirty (30) years from 1989-2020 with the specific acquisition date in Table  1. The year of data processing in 1989, 2000; 2010; and 2020. The results of the dynamical shoreline changes in 1989-2000, 2000-2010, and 2010-2020 were compared to determine the changes. The Landsat imagery in each year undergoes data processing with the following stages. The first stage of image cutting (cropping Image) to ease the work process of the computer.

Image Processing
The second stage of geometric correction is to improve objects' position in the Image according to the actual situation. The geometric correction with ENVI 5.3 software help regarding the ground checkpoint from the field survey results and Bali's Province's RBI Map. The third stage of radiometric correction is to correct the Image caused by satellite damage (Istiqomah et al., 2016). Radiometric correction using ENVI 5.3 software to sharpen images, while atmospheric correction to eliminate atmospheric disturbances. The fourth stage is the delineation process to separate land and waters in the form of a coastline, which will be analyzed using a composite band technique to display the observed object's boundary (Kasim, et al., 2015).

MNDWI For Landsat
To change the MNDWI raster data to vector data using the threshold method.

MNDWI For Sentinel
According to McFeeters (1996), the NDWI method cannot efficiently reduce signal interference originating from land cover features in built-up areas. Xu (2006) noted that water bodies have a stronger absorption in the SWIR band than in the NIR band and that the built-in class has more significant radiation in the SWIR band than in the NIR band. Based on these findings, MNDWI is proposed, which is defined as: Where SWIR is a reflection of TOA from the SWIR band. In general, compared to NDWI, water bodies have a more excellent positive value in MNDWI, as water bodies generally absorb more SWIR light than NIR light; soil, vegetation, and built-up classes have smaller negative values because they reflect more SWIR light than green light (Sun et al., 2012). Xu (2006), For Sentinel-2, the green band has a spatial resolution of 10 m, while the SWIR band (Band 11) has a spatial resolution of 20 m. Thus, MNDWI needs to be calculated at a spatial resolution of 10 m or 20 m. Furthermore, MNDWI 20 m is calculated as: Where 11 is the TOA reflection of Band 11 (SWIR) of Sentinel-2 and is the TOA reflection of Band 3, which is enhanced from Sentinel-2 with a spatial resolution of 20 m. The value of is calculated as the average value of the corresponding 2 x 2 3 values. Conversely, if the spatial resolution of Band 11 is increased from 20 m to 10 m, then MNDWI becomes 10 m spatial resolution, MNDWI 10 m so that it can be calculated as: where The TOA reflectance of Band 11 at 10m is produced by downscaling the original 20-m of Band 11. This is achieved by using the PCA pansharpening method. PCA Pan-Sharpening is an approach based on the component substitution for the original data's spectral transformation (Shah et al., 2008).

DSAS Process
DSAS analysis methods used are Net Shoreline Movement (NSM), End Point Rate (EPR), Linear Regression Rate (LRR). NSM is used to calculate the distance of shoreline changes. EPR is used to calculate shoreline change rate (Sutikno, 2014). LRR is used for predicting changes in the coastline. The prediction is only made in areas that have not yet experienced breakwater development, seawall etc,. The coastline prediction also considers the LRR analysis results if the determination coefficient(R 2 ) has a high value, namely R 2 > 0.7. The next step is to determine the most appropriate regression model by conducting a regression using the variable X as a year and the Y variable as the distance of shoreline change each year with the baseline. The shoreline changes with DSAS show a positive value (+) when experiencing accretion and a negative value (-) if experiencing abrasion.

Wind Data (WR Plot)
WRPlot View is a windrose program for meteorological data. Windrose describes the wind power frequency for each specific wind sector and wind speed classes for each place in a certain period (Lakes Environmental, 2013). Based on research conducted by Fadholi (2013), it was obtained that data analysis of the direction and speed of runway winds in large quantities can be done quickly and quickly using the WR Plot (Wind Rose Plot) application. In addition to fast calculations and the resulting wind rose Image, this application also allows users to interpret the results of the analysis of wind direction and speed by providing a means of overlaying the wind rose into a Google earth map.
The problem to be studied is related to the cycle of wind direction movement and speed in 39 years (464 months), starting from January 1981 to August 2019 in Bali's southeastern waters. The further analysis uses WR Plots in annual averages with a 24 hours specification within 39 years with an average annual data.

Coastal Morphology
Identification of morphological forms and beaches is made by sampling coastal locations with changes in the highest and lowest shoreline in the study location from the DSAS analysis results. In this Figure1 below describe steps to collecting data in the field.

Result
The results of processing the average wind speed from ECMWF data for 1991 to 2000 in Figure 2 below. The graph shows the wind speed for a 10 year period from 1991 to 2000. The wind speed is between 0.2 to 6.2 m/s. Changes in wind speed tend to decrease in the final season of Transition Season 2 (September-November) to West Season (December-February) and increase in Transitional Season 1 (March-May) to East Season in June-August.

C A B
The DSAS Result in periode I shown in Figure6. In Figure 6a have 3 areas that experience significant shoreline changes. There is Padangbai Village has an average NSM/EPR abrasion value: -116.12 m/-10.19 m/y and accretion of 93.80 m / 8.24 m/y, Bugbug Village has an average NSM/EPR abrasion value: -103.06 m/ -9.05 m/y and accretion of 7.14 m/ 0.63 m/y and East Seraya Village has an average NSM/EPR abrasion of -126.78 m/ -11.13 m/y and accretion of 28.58 m / 2.51 m/y. In Figure 6b    In Figure 10a two areas experience significant shoreline changes. Padangbai has an average NSM/ EPR abrasion value: -18.81 m/ -1.99 m/y and an accretion of 29.13 m/ 3.08 m/y, and Bugbug has an average NSM/ EPR abrasion of -28.38 m/ -3.00 m/y and accretion of 13.14 m/ 1.39 m/y. In

Discussion
Changes Port. The direction of the abrasion movement is the same as the direction of the coming dominant wind from the southeast according to the prediction of the dominant wind direction and speed as described by windrose. The amount of accretion in this period is greater than in periods II, III and IV. It is possible that the beginning of the construction of the Padangbai Port was possible. (Putra et al, 2015) Pier I Padangbai Port was built in 1994, operated starting in 1997 with a maximum strength of 1,000 gross tonnage (GT). This accretion causes abrasion in other areas due to the dynamic nature of abrasion and accretion and is influenced by wind, waves, rivers, human activities and others.
Based on calculations with DSAS, from the 2000-2010 period, it can be seen that the highest level of abrasion occurred in Jumpai, -76.87 m with an abrasion rate of -7.50 m/y. Meanwhile, Padangbai was the area with the highest accretion of 27.86 m/ 2.72 m/y. This is due to sedimentation in the port due to the harbor building and coast protection. In this period, the accretion and abrasion values were smaller than the previous period. This is possible because there is no offshore development at the port.
The result form 2010-2020, Gunaksa and Jumpai have the highest accretion and abrasion. The highest accretion occurred in Gunaksa, with an average distance of 79.48 m and an average growth rate of 8.39 m/y. This is probably due to the mouth of the Telaga Waja River and Gunaksa Port. Meanwhile, we encountered abrasion of -48.53 m/ -5.12 m/y which was possible due to changes in waves and the result of sedimentation at the mouth of the Telaga Waja river.
And the result from 2016 until 2020, Tangkas and Gunaksa have the highest accretion and abrasion. Gunaksa Village has an average NSM/EPR abrasion value: -88.14 m/ -21.13 m/y and an accretion of 40.74 m/ 9.76 m/y. The highest level of abrasion occurred in Tangkas, namely -92.82 m with an abrasion rate of -22.24 m/y. The highest accretion occurred in Gunaksa, with an average distance of 40.74 m and an average accretion rate of 9.76 m/y. This is probably due to the mouth of the Telaga Waja River and Gunaksa Port.

Conclusion
Identification of shoreline changes over the past 30 years can be made using the DSAS method using Landsat and Sentinel imagery. This method can also be supported by the prediction of wind speed and wind direction forming seawater waves. In the DSAS method, NSM and EPR values are employed to assess the location/ transect where abrasion occurs (negative) or accretions (positive). The average wind speeds from ECMWF data from three periods 1991-2000; 2001-2010 and 2011-2019 illustrate the dominant wind direction from the Australian continent (southeast). This shows that the east monsoon is more dominant than the west monsoon. In the last 30 years (1989-2020) the dynamics of shoreline changes can be seen from the EPR graph for each transect which has an average abrasion rate of -6.39 m/y and an accretion rate of 3.15 m/y. The identification results of period I abrasion and accretion areas namely Padangbai and Ketewel have an average abrasion/accretion value of -21.91 m/y (extremely abrasion) and 8.24 m/y (very high); In the period II, namely Jumpai and Padangbai had an average abrasion/accretion value of -7.50 m/y (very high) and 2.72 m/y (high); In period III, namely Jumpai and Gunaksa had an average abrasion/accretion value of -5.12 m/y (very high) and 8.39 m/y (very high).
And in period IV are Tangkas and Gunaksa with an average abrasion/accretion value of -22.24 m/y (Extremely) and 9.76 m/y (very high).

Suggestion
For further research, the use of other statistical analyzes can be tried by including additional variables such as tide and wave data in order to obtain better results; and Further analysis is needed regarding sediment transport and vulnerability in the Research area