The main island of Puerto Rico is between the Atlantic Ocean and the Caribbean Sea, approximately between latitudes 18°31′ and 17°55′ N and longitudes 65°37′ and 67°17′ W. The main island is nearly rectangular, with an east-to-west distance of 178 kilometers (km) and a north-to-south distance of 62 km. The island coastline is approximately 1,285 km. East of the main island are two small island municipalities, Vieques and Culebra, with total coastline lengths of approximately 110 km and 48 km, respectively.

Caribbean islands are frequented by hurricanes, winter storms, and North Atlantic swells that may modify geomorphological features along the coastline in both short- and long-term periods. They are also affected by other natural hazards, such as sea-level rise and earthquakes, which have a potential to alter the location where the sea meets the land. In Puerto Rico, shoreline changes were identified after the passage of two major hurricanes (Irma and Maria) approximately 2 weeks apart in September 2017. Hurricane Maria produced major losses in beach width and elevation on beaches in the west-northwestern (from Mayagüez to Camuy), north-central (from Arecibo to west San Juan), and southeastern (from Ceiba to Santa Isabel) municipalities of Puerto Rico (Barreto-Orta and others, 2019). The impacts of these types of storms and flooding can be carried further inland as sea levels rise, affecting a greater area. Changes in relative sea levels at San Juan have been documented at an average of 1.8 centimeters (cm) per decade since 1962 (Runkle and others, 2022), corresponding to an increase of approximately 10.7 cm in the last 60 years. Moreover, since December 2019, the southern part of Puerto Rico has experienced an active seismic period, including a 6.4-magnitude earthquake on January 7, 2020. This event caused remarkable geomorphic changes, the destruction of coastal infrastructure, and coastal subsidence resulting in flooding (Pérez-Valentín and others, 2021). The effects of subsidence from recent earthquake activity make this region more vulnerable to storms and waves, influencing shoreline changes in the future.

The coast of the main island of Puerto Rico was divided into five regions for the shoreline change analysis: North, East, South, West, and Northwest. These were determined based on two criteria: general geographic location and wave energy regimes (Barreto-Orta, 1997). In addition, the islands of Vieques and Culebra are summarized as separate regions (fig. 1). Regarding wave energy regimes, the coast of the North region is characterized by swells and winter storm waves. From Aguadilla to Fajardo, coastal sites are exposed to North Atlantic winter swells (Canals Silander and García Moreno, 2019). In contrast, the coasts of the South and West regions are sheltered from these large waves and, in some areas, are surrounded by a series of keys (Cayo Enrique and Cayo Laurel in Lajas) and mangrove shorelines that offer some protection from wave energy. Similarly, the coast of the East region is sheltered from larger waves by Vieques and Culebra. It is also important to distinguish between the West and Northwest regions because of differences in the insular platform and exposure to swell (Barreto-Orta, 1997).

The insular shelf surrounding Puerto Rico ranges in width from less than 0.5 km on the northern shelf to more than 25 km on the western shelf. The greater width of the shelf reduces the occurrence of high amplitude waves. As presented by Canals Silander and García Moreno (2019), the calculation of wave power surrounding Puerto Rico includes terms for wave height and the impact they have in the near shore. The variation of wave energy environments for Puerto Rico, Vieques, and Culebra are identified in figure 2. The white lines indicate wave power contours, and the black dashed line represents the 30-m depth contour.

Four main shoreline types were identified in Puerto Rico, Vieques, and Culebra: beach (unconsolidated sediment deposit), vegetated (wetlands, mangrove, forests, and pasture, among other vegetation), rocky shorelines (including bedrock bluffs or cliffs, beach rock, and aeolianites), and human engineered (coastlines engineered with hard structures or other anthropogenic modifications). The most commonly occurring shoreline type throughout Puerto Rico is beach (30 percent), followed by vegetated (28 percent), rocky shorelines (15 percent), and human engineered (18 percent). The island has approximately 1,225 beaches in 44 coastal municipalities, including Vieques and Culebra (Barreto-Orta and others, 2019). These beaches vary greatly in width and length, slope, sediment grain size, and composition because of the complexity of both geologic and oceanographic processes identified at island systems. Beach rock formations and aeolianites exposed around the tidal platform are found in many locations and may offer protection from erosion by providing a natural wave break. Other areas of the coast, particularly the South coast, are protected by dense mangrove forests. The areas that experience the highest rates of erosion are generally of the shoreline type classified as beach. The geologic setting, coastal classification, and the connection to erosion following Hurricane Maria are detailed by Barreto-Orta and others (2019).

The shoreline change assessment was primarily focused in areas classified as sandy shorelines; however, there is some variation of shoreline types included in the assessment because change was observed in areas classified as developed, along vegetated shorelines, or at the base of a rocky headland. In most cases transects classified as mangrove, rocky cliff, developed or human engineered were removed from the analysis. Figure 3 provides an overview of the shoreline types by region, synthesized from the 2010 shoreline classification by Bracero-Marrero and others (2021).

The shoreline data from 1970 and 2010 (Bracero-Marrero and others, 2021), was digitized from georeferenced photographs and used various shoreline proxies: high-water line (HWL), groundwater exit point, wet-dry line (WDL), or the instantaneous water line as specified in the shoreline data attribute table and described in the metadata of the datasets used. Common shoreline proxies are reviewed in detail by Boak and Turner (2005). Shoreline data ranging from 1901 to 1987 (Heslin and others, 2021b) rely on user interpretation of shoreline location from the data source (historical aerial photography and topographic surveys) and may vary according to the proxy used. The variations are accounted for in the calculation of shoreline positional uncertainty as described in the “Uncertainty and Error” section of this report.

The shoreline data derived from lidar are defined by the datum-based mean high-water (MHW) elevation using either a contour or profile method to extract shorelines (Farris and others, 2018). Both methods use the MHW elevation to represent the shoreline. The average MHW for Puerto Rico was calculated by Heslin and others (2021a) using the National Oceanic and Atmospheric Administration (2020) VDatum tool (version 4.1.2) to model the regional surface using local MHW values based on the Puerto Rico Vertical Datum of 2002. Following the method used by Weber and others (2005), a single, average MHW value for a continuous section of coast was used (as opposed to using a continuously varying value). The MHW value from all tide observation stations on the main island of Puerto Rico, Vieques, and Culebra ranged from 0.09 to 0.19 meter (m). Taking the mean from those MHW observations resulted in an elevation of 0.13 m. This one contour value was used to delineate the shoreline and is recorded as the contour attribute in the MHW shoreline data shapefile for 2015, 2016, and 2018 (Heslin and others, 2021a). For additional details of the contour and profile method of shoreline extraction for Puerto Rico, see the complete data release by Heslin and others (2021a). For a discussion on the impact of choosing one MHW over multiple regional MHW values, see the “Considerations for Future Work” section of this report.

Each of the published data releases accounted for sources of uncertainty. In general, each shoreline positional accuracy measurement attempts to account for (1) errors inherent to the data source, (2) errors and uncertainty related to the interpretation of the shoreline proxy, and (3) errors related to the extraction of the shoreline data from the data source. Various terms are defined for each shoreline data product, and they are combined in quadrature to calculate the shoreline positional uncertainty for each shoreline date. For additional information and equations for calculating shoreline positional uncertainty, see Ruggiero and others (2013, p. 7–17).

For the lidar shorelines, Heslin and others (2021a) used terms related to the vertical and horizontal uncertainty of the lidar data, the cell size of the lidar data, uncertainty of the MHW elevation value, and uncertainty caused by extrapolation. For the shorelines derived from aerial photos and topographic surveys, Heslin and others (2021b) and Bracero-Marrero and others (2021) used terms that include uncertainty related to georeferencing of the source material, an error accounting for digitization error, the pixel size of the digitized data, and an estimated horizontal uncertainty in the shoreline position since the HWL varies with tide. For complete details on the calculation of individual shoreline positional uncertainty, see the process steps in the metadata of each data release (Heslin and others 2021a, b; Bracero-Marrero and others, 2021).

The range of horizontal positional uncertainty for all shorelines varies from less than a meter for the lidar data, up to 13 m for some of the oldest datasets. The positional uncertainty for each shoreline is included as an attribute with the compiled shoreline data in Henderson and others (2021). Summarized in table 1 are the year, source, shoreline proxy, source data agency, and range of published uncertainty for the given year.

Shoreline change for Puerto Rico, Vieques, and Culebra was computed using the DSAS version 5.1 software (Himmelstoss and others, 2018). The software was used to generate cross-shore transects at 50 m interval spacing along the coast. A visual assessment of the transects with shorelines, shoreline classification data (Barreto and others, 2021), and underlying base-map imagery was performed to make the final determination of transect suitability. Once transects were established and verified, DSAS was used to calculate long-term (approximately 120 years) and short-term (approximately 40 years) linear regression rates of shoreline change. Figure 4 illustrates an example of a transect with eight shoreline intersection points and how the distance from each point to a baseline is measured. The figure also demonstrates the different range of shoreline dates used to calculate short-term and long-term change.

The shoreline change rates and rate uncertainties at individual transect locations in Puerto Rico are available in the data release associated with this report (Henderson and others, 2021). This report also provides summary information for long-term and short-term shoreline change for each region, as well as maximum values of erosion and accretion at individual locations.

This section is modified from the discussion of the methods for calculating and interpreting shoreline change results produced by a study of the Georgia and Florida coasts by Kratzmann (2022); where needed, parts of the report were replicated verbatim. Following Ruggiero and others (2013), each transect rate uncertainty was reasoned to be partially independent of the others. Given that some cancellation of the individual uncertainty components is likely in a regional analysis and transect uncertainties are not likely to be independent of all the others, a partial independence approach reduces both overestimation and underestimation of the uncertainty (Ruggiero and others, 2013). To estimate the regionally averaged uncertainty of partially independent transect rates, the effective number of independent uncertainty values, n* (reduced n) was evaluated. Following Garrett and Toulany (1981), n* was found based on the spatially lagged autocorrelation of measurement uncertainty. For our measurement uncertainty, we used each measure of shoreline change rate uncertainty. This method resulted in a large reduction of the original sample size, n. Assuming that the uncertainty of an area can be represented by , we found the uncertainty of a regionally averaged change rate as follows:

The reduced effective sample size (n*) and regional uncertainty () was determined for the seven study regions (North, East, South, West, Northwest, Vieques, and Culebra). Average uncertainty values found using equation 1 are generally much smaller than the arithmetic mean confidence interval (CI) but larger than the quadrature-averaged CI (Ruggiero and others, 2013). See appendix 1 for the regionally averaged uncertainty and averaged CI values for each region.

The most commonly occurring shoreline type in the North region (fig. 3B) is beach (57 percent), followed by rocky shorelines (17 percent; includes bluffs, beach rock, and aeolianites), human engineered structures and modifications (16 percent), and finally vegetated shoreline type (9 percent; mainly mangroves).

The percentage coverage (length of each shoreline year compared to the total shoreline for the region) of available shorelines over time for the North region is shown in figure 6. The long-term rate includes up to 10 shorelines from 1901 to 2018. While the coverage of some of these older shorelines was less than 30 percent of the study area extent. Up to six shoreline years from 1977 to 2018 were used to calculate short-term rates.

For the North region of Puerto Rico, long-term shoreline change was measured for approximately 70 percent of the region, and short-term rates were measured for approximately 60 percent. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For the North region, areas not included in the shoreline change analysis were inland bays (such as Bahía de San Juan), select areas developed with armored structures, and locations with rocky cliffs.

Average and maximum values of long-term and short-term shoreline change for the North region are shown in figure 7. These graphics are provided to assist users in a visual understanding of the statistical data provided in tables 3 and 4. The locations of maximum rates of change and other areas of interest are shown in figure 8, alongside plots of the short-term and long-term rates of change along the coast.

The percentage of erosional transects decreased from 45 percent (long term) to 31 percent (short term), suggesting decreased erosion (or increased accretion) over time. However, this shift might also be explained by the differences between proxy-based HWL shorelines and datum-based MHW shorelines. Because MHW shorelines tend to be unidirectionally biased in a seaward direction compared with the HWL shorelines, the greater percentage of MHW shorelines used in the short-term analysis may result in an increased accretional signal. This offset is known as the proxy-datum bias (PDB; Ruggiero and List, 2009). In other regional assessments, the USGS calculated a PDB that was applied to older shorelines to account for the difference between the datum-based MHW position and proxy-based HWL shoreline. For Puerto Rico, the PDB for this research effort could not be generated. The “Proxy-Datum Bias” section of this report illustrates an example and includes additional details about the PDB.

The average short-term rate of change for the North region was positive, at 0.46±0.15 meter per year (m/yr). The average short-term erosion rate was −0.31±0.02 m/yr, and the average short-term rate of accretion was 0.81±0.04 m/yr. The largest short-term erosion rate, −1.83±8.73 m/yr, was observed at Punta Sardina, Isabela, whereas the maximum short-term accretion rate, 6.04±8.21 m/yr, was observed at Punta Salinas, Dorada.

The average long-term rate of change for the North region was positive, at 0.04±0.03 m/yr. The average long-term erosion rate was −0.25±0.01 m/yr, and the average long-term rate of accretion was 0.29±0.01 m/yr. The largest long-term erosion rate, −1.89±2.0 m/yr, was observed east of Punta Sardina, Isabela, whereas the maximum long-term accretion rate, 2.43±1.67 m/yr, was observed to the east of Punta Manatí, Manatí.

The most commonly occurring shoreline type in the East region (fig. 3B) is vegetated (34 percent; mainly mangroves), followed closely by beach (33 percent), human engineered structures and modifications (22 percent), and rocky shorelines (10 percent; includes bluffs, beach rock, and aeolianites).

The percentage coverage of available shorelines over time for the East region is shown in figure 9. There were up to six shorelines available for short-term analysis, and up to nine shorelines available for the long-term calculation from the 1930s to 2018. The coverage of shorelines from the 1930s to the 1960s is limited, so in many locations the short-term and long-term analysis was based on a similar range of dates.

For the East region of Puerto Rico, long-term shoreline change was measured for 60 percent of the coastline, and short-term analysis was conducted for approximately 55 percent. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For the East region, areas not included in the analysis are developed shorelines with armored structures, rocky cliffs, mangrove shorelines, and areas with insufficient shoreline data to generate a linear regression (less than three shorelines).

Average and maximum values of long-term and short-term shoreline change are compiled for the East region in figure 10. The locations of maximum rates of change and other areas of interest are shown in figure 11, alongside plots of the short-term and long-term rates of change along the coast. The percentage of erosional transects decreased from 51 percent (long term) to 34 percent (short term), suggesting either decreased erosion (or increased accretion) over time or the effect of PDB.

The average short-term rate of change for the East region was positive, at 0.31±0.11 m/yr. The average rate of short-term erosion was −0.23±0.03 m/yr, and the average rate of accretion was 0.58±0.04 m/yr. The largest short-term erosion rate, −1.95±0.36 m/yr, was observed at Yabucoa Harbor, Yabucoa, and the maximum short-term accretion rate, 4.37±5.89 m/yr, was observed at Las Croabas, Fajardo.

The average long-term rate of change for the East region was positive, 0.06±0.04 m/yr. The average rate of erosion was −0.18±0.02 m/yr, and the average rate of accretion was 0.32±0.02 m/yr. The largest erosion rate, −1.66±0.45 m/yr, was observed at Yabucoa Harbor, Yabucoa. The maximum long-term accretion rate, 1.32±0.39 m/yr, was observed at Punta Fajardo, Fajardo. Note the similarity in shoreline trend between short term and long term at the 60- to 70-km distance mark alongshore (fig. 11). Similar to those observed in the North region, rates tend to be less erosional and more accretional from long term to short term.

The most commonly occurring shoreline type in the South region (fig. 3B) is the vegetated (52 percent; mainly mangroves), followed by beach (33 percent), human engineered structures and modifications (7 percent), and rocky shorelines (7 percent; includes bluffs, beach rock, and aeolianites).

The percentage coverage of available shorelines over time for the South region is shown in figure 12. There were up to five shorelines available for short-term analysis, and a total of eight shorelines available for the long-term calculation, ranging from the 1930s to 2018. Because the number of historical shorelines before the 1970s are limited in some cases, the long-term and short-term rates are based on a similar span of shoreline dates.

For the South region, long-term shoreline change was measured for only 45 percent of the shoreline, whereas short-term change was measured for approximately 40 percent of the region. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For the South region, areas not included in analysis were primarily mangrove shorelines and areas with fewer than the three required shorelines needed to complete the linear regression.

Average and maximum values of long-term and short-term shoreline change for the South region are shown in figure 13. The locations of maximum rates of change and other areas of interest are shown in figure 14, alongside plots of the short-term and long-term rates of change along the coast. The percentage of erosional transects decreased slightly from 63 percent (long term) to 51 percent (short term), suggesting decreased erosion, increased accretion, or the presence of the PDB.

The average short-term rate of change for the South region was positive, at 0.07±0.06 m/yr. The average rate of erosion was −0.36±0.02 m/yr, and the average short-term rate of accretion was 0.54±0.04 m/yr. The largest erosion rate, −3.17±0.58 m/yr, was observed at Guánica, and the maximum short-term accretion rate, 6.25±9.79 m/yr, was observed at Punta Figuras, Arroyo.

The average long-term rate of change for the South region was negative, at −0.05±0.03 m/yr. The average rate of erosion was −0.31±0.02 m/yr, and the average rate of accretion was 0.41±0.04 m/yr. The maximum erosion rate, −2.49±0.08 m/yr, was observed north of Punta Cayito, Santa Isabel. The maximum long-term accretion rate, 5.2±4.92 m/yr, was observed at Punta Figuras, Arroyo.

The most commonly occurring shoreline type in the West region (fig. 3B) is beach (47 percent), followed by vegetated (39 percent), human engineered structures and modifications (10 percent), and rocky shorelines (3 percent; includes bluffs, beach rock, and aeolianites).

The percentage coverage of available shorelines over time for the West region is shown in figure 15. For the West region, there were up to 12 shorelines available for short-term analysis, and as many as 17 shorelines available for the long-term calculation ranging from the 1930s to 2018. The data density along the West region is the result of existing shoreline change data from a USGS report that focused on the municipality of Rincón (Thieler and others, 2007). These data were published as part of the USGS historical shoreline data release (Heslin and others, 2021b) and included in this analysis.

For the West region, long-term shoreline change was measured for 70 percent of the coastline, and short-term change was measured for approximately 60 percent of the region. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For the West region, areas not included in analysis were primarily mangrove shorelines and locations with fewer than three shoreline positions available for analysis.

Average and maximum values of long-term and short-term shoreline change for the West region are shown in figure 16. The locations of maximum rates of change and other areas of interest are shown in figure 17, alongside plots of the short-term and long-term rates of change along the coast. The percentage of erosional transects decreased slightly from long term (52 percent) to short term (48 percent), suggesting a shift to accretional or stable and less of an effect from the PDB offset.

The average short-term rate of change for the West region is positive, at 0.25±0.25 m/yr. The average erosion rate was −0.32±0.02 m/yr, and the average rate of accretion was 0.8±0.08 m/yr. The largest erosion rate, −1.97±0.81 m/yr, was observed Balneario de Rincón, Rincón, and the maximum short-term accretion rate, 5.69±6.86 m/yr, was observed at Caño La Puente, Añasco.

The average long-term rate of change for the West region is positive, at 0.10±0.05 m/yr. The average erosion was −0.23±0.02 m/yr, and the average rate of accretion was 0.48±0.04 m/yr. The maximum accretion rate, 2.56±3.01 m/yr, was observed at Punta Gorda, Rincón. The maximum long-term erosion rate, −1.39±0.41 m/yr, was observed at Balneario de Rincón, Rincón.

Maximum erosion rates for the short and long term for the West region are found in the same location, Balneario de Rincón, Rincón. Previous shoreline change studies (Thieler and others, 2007) for the West region concluded that Rincón was on average experiencing consistent erosion from 1936 to 2006. When compared with the results of this analysis, reaches of the coast identified by Thieler and others (2007) were found to have consistent, if slightly larger rates of erosion than this study. For example, the reach identified from Balneario de Rincón to just south of the mouth of Quebrada Los Ramos was found to have an average erosion rate of −1.1±0.3 m/yr from 1936 to 2006 (Thieler and others, 2007). For the same location, shoreline change rates from 1936 to 2018 were found to be slightly less erosional at −0.91±0.21 m/yr. This shift to a less erosional period might be explained by the differences between proxy-based HWL shorelines and datum-based MHW shorelines. The greater percentage of MHW shorelines used in the more recent analysis may result in a seaward shift of the average shoreline position because of PDB. Another conclusion supported by this study is the suggestion that long-term and short-term rates of shoreline change are statistically similar. This is seen in the similar pattern of change observed in the plots on figure 17, which show the rates of change along the West region.

Although the Northwest region is relatively small compared with the other regions, it was important to separate it from the West and North because of the significant differences in the insular platform, resulting in higher exposure to swell and high amplitude waves (Barreto-Orta, 1997).

The most commonly occurring shoreline type in the Northwest region (fig. 3B) is beach (48 percent), followed by rocky shorelines (31 percent; includes bluffs, beach rock, and aeolianites), human engineered structures and modifications (11 percent), and vegetated shoreline type (10 percent).

The percentage coverage of available shorelines over time for the Northwest region is shown in figure 18. There were up to six shorelines available for short-term analysis and as many as eight shorelines available for the long-term calculation from the 1930s to 2018.

For the Northwest region, shoreline change for both short-term and long-term analyses were measured for approximately 80 percent of the shoreline. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For the Northwest, areas not included in the analysis were primarily classified as rocky shoreline.

Average and maximum values of long-term and short-term shoreline change are compiled for the Northwest region in figure 19. The locations of maximum rates and other areas of interest are specified in figure 20, alongside plots of the short-term and long-term rates of change along the coast. The percentage of erosional transects decreased from 42 percent (long-term) to 30 percent (short-term), suggesting significant decreased erosion (or increased accretion) over time or the influence of the PDB offset.

The average short-term rate of change for the Northwest region was positive, at 0.17±0.12 m/yr. The average erosion rate was −0.25±0.04 m/yr, and the average rate of accretion was 0.36±0.05 m/yr. The largest erosion rate, −1.4±13.42 m/yr, was observed Playa Crash Boat, Aguadilla, and the maximum short-term accretion rate, 1.23±1.8 m/yr, was observed at Playa Punta Borinquen, Aguadilla.

The average long-term rate of change for the Northwest region was positive, at 0.10±0.09 m/yr. The average rate of erosion was −0.18±0.02 m/yr, and the average rate of accretion was 0.31±0.05 m/yr. The maximum long-term erosion rate, −0.84±0.93 m/yr, was observed at Punta Borinquen, Aguadilla. The maximum accretion rate for both long term and short term, 1.23±1.8 m/yr, was observed at Playa Punta Borinquen, Aguadilla.

The most commonly occurring shoreline type on the coast of Vieques (fig. 3B) is beach (41 percent), followed by vegetated (30 percent), rocky shorelines (29 percent includes bluffs, beach rock, and aeolianites), and human engineered structures and modifications (less than 1 percent).

The percentage coverage of available shorelines over time for the island of Vieques is shown in figure 21. For Vieques, there were a total of three shorelines available for the short-term analysis and only up to five available in the long-term analysis, with shoreline measurements ranging from the 1960s to 2018. Large areas of data gaps and locations with less than three shorelines resulted in fewer locations where short-term rates were possible and an increased likelihood that the long-term and short-term rates were calculated using a similar range and number of dates.

For the coast of Vieques, long-term shoreline change was measured for 60 percent of the island shoreline, and short-term analysis was measured for approximately 45 percent of the shoreline. For Vieques, areas not included in analysis were primarily areas of rocky cliffs (north shore), mangrove shorelines (south coast), and a large area (western portion of the island) with fewer than three shorelines needed for short-term analysis.

Average and maximum values of long-term and short-term shoreline change are shown in figure 22. The distribution of erosion and accretion are similar for short- and long-term change. Locations of interest, as well as plots of long-term rates of change along the coast of Vieques are illustrated in figure 23. For this location, short-term rates are not plotted because they appear similar to the long-term rates and would be difficult to distinguish from the plot of long-term change. Unlike previous regions, this plot displays two graphs for long-term rates of change for each side of the island: north and south. The percentage of erosional transects decreased slightly from 49 percent (long term) to 44 percent (short term). This could be due to the PDB offset or to the decrease in sample size (number of transects) from the long-term (1,289) to short-term (968) shoreline analysis.

The average short-term rate of change for Vieques was positive, at 0.03±0.07 m/yr. The average erosion rate was −0.16±0.01 m/yr, and the average rate of accretion was 0.18±0.01 m/yr. The largest erosion rate, −1.01±1.46 m/yr was observed at Black sand beach, Vieques, and the maximum short-term accretion rate, 0.9±0.4 m/yr, was observed at Playa La Chiva, Vieques.

The average long-term rate of change for the Vieques was positive, at 0.02±0.06 m/yr. The average erosion rate was −0.17±0.01 m/yr, and the average accretion rate 0.21±0.02 m/yr. The maximum erosion rate, −1.59±2.60 m/yr, was observed at Punta Arenas south, Vieques. The maximum accretion rate, 2.66±15.73 m/yr, was observed on the northern side of Punta Arenas where the spit appeared to erode from the south and accrete on the northern side.

The most commonly occurring shoreline type on the coast of Culebra (fig. 3B) is rocky shorelines (36 percent; includes bluffs, beach rock, and aeolianites), followed closely by vegetated (31 percent), beach (30 percent), and human engineered structures and modifications (2 percent).

The percentage coverage of available shorelines over time for the island of Culebra is shown in figure 24. There were only three shorelines available for short-term analysis and only up to four available for the long-term calculation from the 1960s to 2018. Large areas of data gaps and locations with less than three shorelines resulted in fewer locations where short-term rates were possible and an increased likelihood that the long-term and short-term rates were calculated using a similar range and number of dates.

For Culebra, long-term shoreline change was measured for 60 percent of the island coastline. Short-term analysis was measured for approximately 50 percent of the island. The percentage of coastline suitable for shoreline analysis is influenced by the regional distribution of shoreline type and the number of available shoreline dates for the period. For Culebra, areas not included in analysis were primarily areas of rocky shorelines, developed and mangrove shorelines, and the small rocky islands surrounding Culebra.

Average and maximum values of long-term and short-term shoreline change are shown in figure 25. The location of maximum rates and other areas of interest are specified in figure 26, alongside plots of the short-term and long-term rates of change along the coast. For this plot, long-term and short-term rates of change are plotted together for each side of the island: north and south. The percentage of erosional transects decreased significantly over time from 37 percent (long term) to 9 percent (short term). This could be because of the PDB offset, the limited sample size, or the range of dates used for the short-term analysis. Culebra is the only location where the short-term rate used data from 2006 to 2018.

The average short-term rate of change for Culebra was positive, at 0.90±0.77 m/yr. The average erosion rate was −0.20±0.05 m/yr, and the average rate of accretion was 1.01±0.10 m/yr. The largest erosion rate, −0.40±1.91 m/yr, was observed at Punta Vaca, Culebra, and the maximum short-term accretion rate, 6.92±11.99 m/yr, was observed at Punta Manchita, Culebra.

The average long-term rate of change for Culebra was positive at 0.08±0.08 m/yr. The average rate of erosion was −0.11±0.01 m/yr, and the average accretion rate was 0.19±0.02 m/yr. The maximum erosion rate, −0.59±0.30 m/yr, was observed northwest of Flamenco Beach, Culebra. The maximum long-term accretion rate, 0.94±1.36 m/yr, was observed at Playa Resaca, Culebra.

Previous USGS national shoreline products have accounted for the observed unidirectional horizontal offset between the proxy-based HWL shorelines and datum-based MHW shorelines (Ruggiero and List, 2009). This is known as the PDB and, if needed, is generally applied as a correction during the shoreline rate calculation. The PDB calculation requires historical wave data (wave height and wavelength) and localized average slope data from available elevation surveys. At the time of this study, these data products were not available for Puerto Rico. In addition, scoping work done to determine the suitability of Puerto Rico for applying the PDB found that certain factors will complicate the MHW to HWL relationship, including offshore platforms, beach rock of varying elevation, bluffs and cliffs, variable coastline with small pocket beaches, and the location of MHW in the beach profile (generally low and near the waterline or in the water). These factors influence the horizontal distance between the MHW and HWL shorelines, making it difficult to apply a consistent bias value.

Although we were unable to calculate the PDB for Puerto Rico, in many locations, there appears to be a prominent visual offset that would most easily be explained as the PDB. The observed offset between the two different shoreline types is highlighted in figure 27. The 2016 and 2018 shorelines extracted from lidar are datum-based MHW shorelines and they appear seaward of the 1970 and 2010 proxy-based shorelines derived from aerial photograph interpretation of the HWL. We consider that this is due to an offset between the proxy-based and datum-based shoreline types and not a true pattern of erosion and accretion. The HWL is typically observed landward of the last high tide swash, and the MHW occurs on the beach face seaward of the high tide line. As the slope of the beach changes from gradual to steep (as seen in fig. 27 along the eastern and western sides of the Bahía Flamenco), the shorelines get closer together. In this example, the offset is most prominent in the middle of the lowest-sloping portion of Playa Flamenco.

To account for the potential unidirectional offset (bias) in MHW and HWL shorelines that could not be corrected using a standard PDB, additional terms were included in the calculation of shoreline positional uncertainty. Any future updates to the Puerto Rico shoreline datasets would benefit from the creation and incorporation of the PDB feature to the shoreline rates calculation or extraction of all consistent, proxy-based shorelines that would not need to account for any horizontal offset between proxy-based shorelines.

The MHW contour value was extracted using an averaged MHW value from 18 tide gages from Puerto Rico’s main island as well as Vieques and Culebra. Although this provides a good, generalized estimate of MHW for the islands, some of the precision in the MHW value is lost for specific areas. For instance, the North region has a higher-than-average MHW value, while the South region has a lower-than-average MHW value. Localized MHW values based on regions could be more appropriate for site-specific analyses and could be considered for future updates to Puerto Rico shoreline analyses. In addition, the location of the MHW was often towards the waterline or in the water, which makes it difficult to justify the comparison with the HWL shoreline, which was often much higher up on the beach. Consideration of a different datum-based elevation such as the Mean Higher High Water or an offset based on existing elevation could be explored.

Given some data limitations and cross-agency coordination of shoreline data from various timeframes, there were some challenges in synthesizing shoreline data for analysis. For example, historical shorelines from 1901 to 1987 from Thieler and Danforth (1993) were incorporated into a geospatial project for analysis with more recent shorelines. In some cases, not all source images (for example, aerial photographs) were available for quality control of the original historical shorelines. Because of this, some shorelines could only be referenced using current imagery and other verified shoreline positions.

Another challenge was aggregating datasets from different agencies. The data release that accompanies this report was created with the shorelines produced by the USGS and the University of Puerto Rico. The methods used to compile the data utilized subjective decisions made by each agency when extracting shoreline data (for example, aerial photograph interpretation of shoreline classification type, georeferencing of source imagery) and could yield slightly different positions of resulting shorelines. Because of this, all shoreline data were given a final quality control and peer review within the USGS to ensure the shorelines were as accurate as possible.

It is also important to note the coordinate system that the Puerto Rico shoreline data are projected in. Previous national shoreline assessments used the local Universal Transverse Mercator zone projection for analysis. In this analysis, the data are projected using the Puerto Rico State Plane Coordinate System. This is because Puerto Rico spans two Universal Transverse Mercator zones. Although the state plane coordinate system does distort distance the further from the projection’s standard parallels, the coordinate system covers a small area and the distortion is generally minimal.

As noted in the “Nonstandard Nature of Shoreline Analysis” section of this report, the source imagery that served as the base maps for historical shoreline data extraction was georeferenced by different agencies at different times. Potential inaccuracies from the georeferenced source imagery, while accounted for in the uncertainty equation, could vary at different locations depending on its quality and spatial location. For instance, an average georeferencing error was estimated for the 1970s and 2010 shorelines, which in some cases can underestimate uncertainty of the shoreline position. The source imagery inaccuracies can also vary and different locations within the image. Locations in the source image where there are fewer and not as well-distributed control points could result in higher inaccuracy of the shoreline position.

An assessment of the horizontal offset between proxy-based and datum-based shorelines was not possible at the time of this analysis because of a lack of information or data used to calculate the PDB. The low-sloping sandy beaches of Puerto Rico are no exception. For this reason, developing the PDB would greatly benefit any future assessment.

Additional shoreline data would be useful for the short- and long-term analyses. If additional lidar shorelines can be added, the short-term rates of change could be expanded to include all locations regardless of the PDB because all locations would be the same shoreline proxy. For the long-term analysis, any additional historical source imagery and elevation data not covered in the current data releases would greatly improve the quality of the long-term rates.

Using shoreline classification data in Barreto and others (2021) as a baseline, a standardized classification field could be added to all shoreline change data, especially for all future hand-digitized shorelines. The ability to relate shoreline change to shoreline classification could allow for the exploration of other methods for a comprehensive analysis of shoreline change by different shoreline types.

Because the MHW value is averaged for all of Puerto Rico, it potentially overgeneralizes the MHW position in some areas, especially toward far northern and southern parts of the main island. Calculating localized MHW values and updating wave regime data may help capture more precise regional assessments of the island. Updated wave regime data could also provide update boundaries when deriving regional shoreline change statistics.