Recommendations on data use are difficult to provide, given each application may contain bespoke conditions which will guide dataset selection. However, we provide the following general recommendations on data use:

  • Extent data that best matches the mapped time period should be used

  • For contemporary mapping, the GMW maps provide the most up-to-date high-quality maps

  • Simard et al. should be used for AGB estimation as it relies upon physical measurements of structure and not climatic variables. Furthermore Hutchinson et al. do not recommend that their data is used for specific study areas

  • Rovai et al. and Sanderman et al. both provide high-quality maps of soil c, yet were derived using separate methods. Rovai et al. uses climatic and typology to predict soil c while Sanderman et al. use machine learning approaches that use both climatic data and spectral reflection of land cover. These datasets are produced at very different scales and provide very different estimates. We recommend the Sanderman et al. dataset where spatial resolution is most important, but Rovai et al. for regional scale analyses.

We note the following Pro's and Cons of each dataset below:

Mangrove Extent



  • The GMW provide the most consistently detailed map with a spatial resolution of 25 m, thus is useful for mapping small stands

  • The GMW maps are the most recent global mangrove extent product ensuring the mapped area is the most up-to-date

  • GMW updates are produced anually, so the baseline is continuously revised and gain/loss layers are available


  • Version 1 of the GMW map does not include many small island states so is not appropriate for mapping these regions

  • The GMW baeline is made from optical remote sensing data and thus the quality of the dataset is decreased in areas of persistent cloud cover



  • The MFW map is detailed map with a spatial resolution of 30 m, enabling it to map all but the smallest mangrove stands

  • It is spatially continuous so there are no gaps or missing regions


  • Th MFW map is relevant for time periods aroundthe year 2000 and more recent data is unavailable

  • In persistently cloudy areasmangrove areas are omitted and not mapped due to the lack of availability of data



  • The CGMFC-21 map has a spatial resolution of 30 m and is comparabletot he GMW and MFW maps

  • The CGMFC-21 map provides an estimate of mangrove cover as well as extent, providing information on mangrove density and structure

  • Maps are available for the period 2000-2012


  • CGFMC-21 are not continually updated as with the GMW dataset

  • The CGMFC-21 is limited to regions of already mapped mangrove and so omits scrub or juvenile mangroves and some regions of new growth



  • Data for other wetland types (bogs, marshes) available from same source allowing wider wetland maps to be constructed


  • The CIFO map has a comparitively low spatial resolution of approximately 236 m, which is much coarser than othe ravailbale products

  • Small islands are omitted from the map due to the large spatial resolution

  • The CIFOR dataset maps mangrove extent at higher latitudes than those agreed upon by other sources and is thus less reliablein these locations

Mangrove atlas


  • Earliest source of mangrove extent that provide a historic estimate of mangrove extent


  • The dataset is made up of contributions from countries and thus the map is composed of many resolutions and data from a range of time periods

  • The extent map for a given country cannot be relibaly validated and thus the mapp error is difficult to quantify

  • Th eworld Mangrove Atlas is not applicable for contemporary extent mapping

Aboveground Biomass



  • The hutchinson map is a climate-driven map so is able to capture modeled biomass at regional scales and is able to estimate biomass in adjacent wetlands also


  • The model is not updated and is valid only for the early-2010s

  • The resolution of the model prevents fine scale variation from being captured. As climateic variables do not vary at the scale mangrove forest structure does, this is reflectaed in the map as a lack of detail

  • For the same reason, anthropogenic influcences are also not detected, so areas of cleared of mangrove could be predicted to have anomalously high biomass, as determined from the climatic variables



  • The Simard biomass map has a high resolution of 30 m providing increased detail over other biomass products

  • The high spatial resolution enables variation in mangrove forest structure to be captured, thus estimates are more accurate than climatic driven data

  • Field data was used to calibrate the biomass model

  • Height as well as biomass is available


  • The uncertainty around height and biomass estimates is increased for low stature forests and detailed variations elow heights of 1 m are not well characterized

  • The mangrove height and biomass data are derived from data captured in 2000 so are thus outdated

Soil Carbon



  • The Sanderman map is the highest resolution of all the soil carbon maps with 30 m resolution, capturing local scale variations in C

  • The model data is recent giving an up-to-date estimate of soil C

  • The map conatins modeled soil C to multiple soil depths


  • The model is based on Landsat spectral data which may not be indicative of soil C, particularly in low stature forests



  • Extensive use of 900 field locations to build a global model of soil C

  • Machine learning algorithms (Random Forest) which are able to model soil C more spatially accuractely were used


  • Input data may be affected by imperfect information on plot location

  • The spatial resolution is approximately 1 km, thus the map lacks some spatial detail in comparison to others



  • Country-specific approach allows identification of where mangrove losses have the highest contribution to GHG emissions

  • Based on field data collected per coutry


  • Recent data unavailable as field data is not routinely collected in all countries

  • Data quality among countries is highly variable with less reliable data available in some countries

  • Where no data is available a global avergae is applied which may not be representative of the true values

  • No map is provided as only country estimates are given

  • Total C values are based on CGMFC-21 which estimates cover and thus underestimate extent. Therefore, soil C estimates are expected to be underestimated



  • Contains information for 57 nations that previously lacked soil C information

  • Uses mangrove typlolgy and enviornmental settings to understand controls and soil C


  • Coarse resolution (25 km) in comparisson with other mangrove soil C maps

  • Total C estimates are based on CGMFC-21 which may underestimate soil C