OVERVIEW
This dataset contains the CoSMoS-COAST model code, modeled shoreline position 
timeseries and related data for multiple globally distributed sea turtle nesting
beaches, underlying the analysis presented in Chapter 4 of the dissertation 
"Coastal Science for Sea Turtle Conservation" by Jakob C. Christiaanse (2025).

For each study site, the data includes model input data (CoSMoS-COAST) and 
model results for the hind- and forecast periods, under various SLR scenarios. 
The dataset is organized to facilitate reproducible coastal change analysis 
and supports research on coastal dynamics and sea turtle habitat. Data are 
provided in CSV files, compatible with common data analysis tools.

The Chapter is currently in preparation to be submitted as a journal article.
Until it is published, please cite the dissertation when using this data (see 
references).

CONTENTS
The model code is provided in two matlab (.m) files:
- CoSMoS_COAST_CONV.m
    This is the main model code used to run the hindcast period, optimizing the
    model parameters to the satellite-derived shoreline data. The model is based
    on the version in Vitousek et al. (2023), but written as a convolution instead
    of traditional numerical timestepping, as in Vitousek et al. (2025). Further
    instructions to run the model are given in the code.
- CoSMoS_COAST_CONV_fixed.m
    This is a simplified version of the model code, which runs with a fixed parameter
    set. As there is no optimization, it does not require shorelines as input, making 
    it very fast. This code was used to run the forecast scenarios with the parameter
    values obtained from the hindcast. Further instructions given in the code.

The remaining data is divided in one folder per study site:
- AI_LB: Long Beach, Ascension Island
- AU_DH: Dirk Hartog Island, Australia
- BR_BV: Busca Vida, Brazil
- CR_TO: Tortuguero, Costa Rica
- CV_JB: Joao Barrosa, Cabo Verde
- CY_AL: Alagadi, Cyprus
- MX_LE: La Escobilla, Mexico
- MX_RN: Rancho Nuevo, Mexico
- OM_MI: Masirah Island, Oman

Each study site folder is divided into three subfolders:
- 0_input: All input data used to run the model.
- 1_hindcast: All model output from the hindcast.
- 2_forecast: All model output from the forecasts.

DATA STRUCTURE
The three subfolders all contain CSV files with the following data:

0_input
- **_**_sealevel_1979_2024.csv
    Yearly sea level from 1997 to 2024 extracted from the nearest GTSM node.
    --Columns--
        - wl: yearly sea level [m] relative to MSL

- **_**_sealevel_2025_2100.csv
    Yearly sea level from 2025 to 2100 extracted from the AR6 regional sea level
    projections for three emmision scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) and
    three percentiles per scneario (5th, 50th or median, and 95th).
    --Columns--
    One column for each scenario-percentile combination, denoted as:
        - wl_*scen*_*q*: yearly sea level [m] relative to MSL, where scen is the 
        emmision scenario (e.g., slr126 for SSP1-2.6) and q is the percentile (e.g.,
        q50 for the 50th percentile).

- **_**_shorelines.csv
    Tide-corrected shoreline position time series extracted from CoastSat, normalized
    by subtracting the median between 2021-2024 (reference position), and reindexed to 
    a daily timeseries (therefore contains a lot of gaps/NaNs)
    --Columns--
    One column per transect (T1, T2, T3) giving the shoreline position [m].

- **_**_slopes.csv
    Slopes used for the regression slope in the SLR term of the model. These have either
    been estimated from CoastSat or use a default value of 0.1 (see Chapter).
    --Columns--
    One value per transect (T1, T2, T3) giving the slope.

- **_**_waves_1979_2024.csv
    Historical wave time series extracted from the nearest ERA5 node.
    --Columns--
        - Hs: Daily maximum significant wave height [m].
        - Tp: Corresponding peak wave period [s] (not used in model).
        - Dir: Corresponding mean wave direction [degrees].

- **_**_waves_2025_2100.csv
    Future wave time series created by shuffling the historical time series (see Chapter).
    --Columns--
        - Hs: Daily maximum significant wave height [m].
        - Tp: Corresponding peak wave period [s] (not used in model).
        - Dir: Corresponding mean wave direction [degrees].

1_hindcast
All remaining CSV files have one column per transect (T1, T2, T3) and all shoreline
position timeseries are in meters relative to the reference position (median 2021-2024).
- **_**_hind_comp_crossshore.csv
    Cross-shore component of the modeled shoreline position.

- **_**_hind_comp_longshore.csv
    Longshore component of the modeled shoreline position.

- **_**_hind_comp_slr.csv
    SLR/Bruun component of the modeled shoreline position.

- **_**_hind_comp_trend.csv
    Residual trend (vlt) component of the modeled shoreline position.

- **_**_hind_parameters.csv
    Optimized model parameter values for each transect.

- **_**_hind_shoreline_prediction.csv
    Modeled shoreline position (sum of the four components).

- **_**_hind_shoreline_smoothed.csv
    Smoothed input shoreline positions, used to calibrate and evaluate the model.

- **_**_hind_skill_metrics.csv
    Skill metrics for the hindcast, computed from the modeled shoreline position and 
    the smoothed input shoreline positions.

2_forecast
The CSV files in the forecast folder are equivalent to those in the hindcast folder, 
but then for the forecast period (2025-2100). The parameters, shoreline_smoothed, 
and skill_metrics files are not given as they only exist for the hindcast. The the 
other five files (modeled shoreline and components) are provided for each emmission 
scenario and SLR percentile combination (see also input data).

DATA FORMAT
Model code provided as matlab files (*.m) and tabular data as CSV files.

USAGE
Any software or programming language that can read CSV files can be used to analyse
the data (Python, R, C++, Julia). We have used Python (Pandas) for all processing and 
operations on the data.

REFERENCES
Christiaanse, J. C. (2025). Coastal Science for Sea Turtle Conservation [Doctoral 
dissertation, Delft University of Technology].

Vitousek, S., Vos, K., Splinter, K. D., Erikson, L., & Barnard, P. L. (2023). A Model 
Integrating Satellite-Derived Shoreline Observations for Predicting Fine-Scale Shoreline 
Response to Waves and Sea-Level Rise Across Large Coastal Regions. Journal of Geophysical 
Research: Earth Surface, 128(7). https://doi.org/10.1029/2022JF006936

Vitousek, S., Buscombe, D., Gomez‐de la Peña, E., Calcraft, K., Lundine, M., Splinter, 
K. D., Coco, G., & Barnard, P. L. (2025). Are Equilibrium Shoreline Models Just 
Convolutions? Journal of Geophysical Research: Earth Surface, 130(6), e2025JF008452. 
https://doi.org/10.1029/2025JF008452

LICENSE
CC BY 4.0