Use of any aspect of this study requires full attribution (see licence). Please cite the published paper:
Murray, N. J., T. A. Worthington, P. Bunting, S. Duce, V. Hagger, C. E. Lovelock, R. Lucas, M. I. Saunders, M. Sheaves, M. Spalding, N. J. Waltham & M. B. Lyons (2022). "High-resolution mapping of losses and gains of Earth's tidal wetlands." Science. 376, 744-749. doi:10.1126/science.abm9583
A free access version of the paper is available at this link (link to free access PDF of paper).
These data products are licensed under a Creative Commons Attribution 4.0 International License. [CC- BY-4.0](https://creativecommons.org/licenses/by/4.0/)
A full description of the methods, validation, and limitations of the global intertidal change data is available in the published paper.
The principal data products from the global intertidal change analysis of the Landsat archive is a set of change maps, comprising:
"loss", 0-1 - integer representing loss, where 1 (loss) and 0 (no loss).
"lossYear", 01:19, 3 - integer representing the end year of the time-step of analysis (e.g., 19 = 2017-2019).
"lossType", 2,3,5 - integer representing intertidal ecosystem type: Tidal flat (2), Mangrove (3), Tidal marsh (5)).
"gain", 0-1 - integer representing gain, where 1 (gain) and 0 (no gain).
"gainYear", 01:19, 3 - integer representing the end year of the time-step of analysis (e.g., 19 = 2017-2019).
"gainType", 2,3,5 - integer representing intertidal ecosystem type: Tidal flat (2), Mangrove (3), Tidal marsh (5)).
In addition, we provide two tidal wetland extent data layers:
"twprobability_start", 0-100 - integer which represents the agreement of random forest decision trees for the tidal wetland class in t1 of the analysis, 1999-2001
"twprobability_end", 0-100 - integer which represents the agreement of random forest decision trees for the tidal wetland class in t7 of the analysis, 2017-2019
For analysis, the dataset is made directly available in Google Earth Engine via the Earth Engine data catalogue (dataset link):
The dataset is also available as shards on GCP Cloud Storage and can be downloaded using the `gsutil` command `gsutil -m cp "gs://gic-archive/gic-2019-v1-0/v1-0-1/*.tif" "PATH-TO-LOCAL-FOLDER"`
The datasets generated for this study are available for viewing at the Global Intertidal Change website.
See the Data Viewer pages for:
Refer to our archived data descriptor for the latest updates.
The high-resolution maps of losses and gains of Earth's tidal wetlands have several known limitations (including commission and omission errors) that are characterised and discussed in detail in the supplementary material of Murray et al. (2022).
We note three issues here:
Any downstream use of these maps should account for data uncertainty and propagate them throughout analyses. Responsible use of spatial data requires propagating known uncertainties, and thus area estimations should not be made directly from the map datasets without reporting associated uncertainty estimates. Many methods are available for this, including probability-sample based (Olofsson et al, 2014) and resample-based (Lyons et al 2018) quantitative approaches. Please refer to supplementary material of Murray et al. (2022) for further information on the source of known uncertainties and the impact they have on the area estimates derived from this product.
The global intertidal change maps were developed with training data that met strict definitions of intertidal ecosystem type and change (loss, gain and no change) for use at a specific spatial scale. These definitions may not suit the specific needs of other coastal ecosystem monitoring or conservation studies, and we therefore recommend all users to familiarise themselves with the definitions used to develop these maps, as well as the definitions of loss, gain and no change.
A minimum mapping unit is applied to the data, please refer to the published paper.
Analyses of these map data should therefore only be conducted with a detailed understanding of appropriate use of the data and careful consideration of uncertainty. For further information please contact Dr Nicholas Murray (James Cook University).