This repository containts the datasets from the article "Mapping saltmarsh communities in South Portugal using high spatiotemporal resolution satellite imagery" (Submitted). The dataset was used in a workflow used to create saltmarsh maps for the Algarve region (South Portugal), focused on the 4 main costal systems of the region: Alvor, Arade, Ria Formosa and Guadiana.

For a description of the methodology see the article [link] and Github repo [link].

Repository content

1. system-masks.zip

Contains 4 geojsonfiles with a polygon which delimits the areas included in the study. The files are named after the respective systems that they delimit. Any region outside of these polygons were not used in the analysis.

CRS - EPSG:4326

2. manual-clean-up-masks.gpkg

Polygons which were manually created to mask out (exclude) pixels which were classified as saltmarsh, but are clearly not.

File contains a single layer with 52 polygons and one variable.

Variables:

• system [string] - Which system the polygon delimits

3. saltmarsh-training-data.gpkg

Data used for supervised model training. Each row represents one quadrat, and each column contains either quadrat identifiers, target classes, or predictor classes.

File contains a single layer with 2448 points and 18 variables.

Variables:

• water_system [string] - Study system in which the quadrat was sampled
• transect [string] - Name of transect in which the quadrat was sampled
• quad_id [integer] - Unique identifier per quadrat
• cluster [integer] - Vegetation cluster identified via hierarchical clustering. They are nested within water_system, and the same number within different systems will not correspond to the same vegetation type.
• marsh_type [string] - Functional groupings of saltmarsh vegetation (low, middle or high), created by grouping cluster based on niche of the defined clusters.
• train [boolean] - Was quadrat used in the train (TRUE) or test (FALSE) stage of model training?
• ndvi [numerical] - Normalized Difference Vegetation Index, calculated from the satellite image mosaic as (nir – red) / (nir + red).
• ndwi_high [numerical] - Normalized Difference Water Index estimated from images at high tide, calculated as (green – nir) / (green + nir)
• ndwi_low [numerical] - Normalized Difference Water Index estimated from images at low tide, calculated as (green – nir) / (green + nir)
• subtime [numerical] - Fraction of time that a cell is estimated to be submerged in water over one year.
• coastal_blue [numerical] - Surface reflectance values at 443 nm.
• blue [numerical] - Surface reflectance values at 490 nm.
• green_i [numerical] - Surface reflectance values at 531 nm.
• green [numerical] - Surface reflectance values at 565 nm.
• yellow [numerical] - Surface reflectance values at 610 nm.
• red [numerical] - Surface reflectance values at 665 nm.
• rededge [numerical] - Surface reflectance values at 705 nm.
• nir [numerical] - Surface reflectance values at 865 nm.

4. saltmarsh-transect-metadata.csv

Comma-delimited file with information about vegetation sampling transects. Each row represents one transect.

File contains 6 variables.

Variables:

• water_system [string] - Study system in which the transect was sampled
• transect_set [string] - Which set of transects was this transect sampled in? Set A was performed in 2019, set B in 2023.
• site [string] - Name of the site within the study system. This was used exclusively to plan transects.
• transect [string] - Name of transect in which the quadrat was sampled
• date [date yyyy-mm-dd] - Date of transect sampling.
• notes [string] - Notes taken during transect sampling and which might be relevant to understand data issues.

5. saltmarsh-vegetation-quadrats.gpkg

Data used for to create vegetation clusters (cluster) and saltmarsh community types (marsh_type). The later was used as the target class in the supervised model training. Each row represents one quadrat, and each column contains either quadrat identifiers, or presence/absence of species.

File contains a single layer with 2448 points and 32 variables.

Variables:

• water_system [string] - Study system in which the transect was sampled
• transect [string] - Name of transect in which the quadrat was sampled
• transect_set [string] - Which set of transects was this transect sampled in? Set A was performed in 2019, set B in 2023.
• quad_id [integer] - Unique identifier per quadrat
• distance_from_water [integer] - Distance from start of quadrat, which was the point closes to the water where saltmarsh was found for that transect.
• cluster [integer] - Vegetation cluster identified via hierarchical clustering. They are nested within water_system, and the same number within different systems will not correspond to the same vegetation type.
• marsh_type [string] - Functional groupings of saltmarsh vegetation (low, middle or high), created by grouping cluster based on niche of the defined clusters.
• Arthrocaulon.macrostachyum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Tripolium.pannonicum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Atriplex.halimus [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Cistanche.phelypaea [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Atriplex.portulacoides [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Limbarda.crithmoides [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Juncus.effusus [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Limoniastrum.monopetalum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Myriolimon.ferulaceum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Limonium.vulgare [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Phragmites.australis [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Polygonum.maritimum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Puccinellia.maritima [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Salicornia.procumbens [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Salicornia.europaea [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Caroxylon.vermiculatum [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Salicornia.fruticosa [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Salicornia.perennis [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Bolboschoenus.maritimus [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Sporobolus.maritimus [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Spergularia.bocconei [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Suaeda.vera [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Triglochin.maritima [boolean] - Presence (1) or absence (0) of the species with the variable name.
• Sporobolus.montevidensis [boolean] - Presence (1) or absence (0) of the species with the variable name.

6. predicted-map.tif

Geotiff file with a single layer for predicted saltmarsh community. Values are:
  - no data - Not saltmarsh
  - 1 - Low saltmarsh
  - 2 - Middle saltmarsh
  - 3 - High saltmarsh

CRS - EPSG:32629