An error analysis of marine habitat mapping methods and prioritised work packages required to reduce errors and improve consistency

Strong, James Asa ORCID: https://orcid.org/0000-0001-8603-097X. 2020 An error analysis of marine habitat mapping methods and prioritised work packages required to reduce errors and improve consistency. Estuarine, Coastal and Shelf Science, 240, 106684. https://doi.org/10.1016/j.ecss.2020.106684

Full text not available from this repository.

Official URL: http://dx.doi.org/10.1016/j.ecss.2020.106684

Abstract/Summary

Many technical issues influence the accuracy and repeatability of marine habitat mapping studies – identifying, quantifying and suggesting further work to reduce these issues was the objective of this investigation. Issues were identified by firstly defining the methods that are common used for marine habitat mapping and then by identifying the ‘methodological variables’ (i.e. decision points within a map production methodology) within each method that entrained error, and ultimately contribute to inaccuries in end products. Following the identification of issues, potential solutions for addressing the methodological variables that were subsequently ranked by efficacy and cost-effectiveness. A total of 56 marine habitat mapping methods for were constructed using combinations of 18 common mapping techniques and platforms (e.g. ship, UAV). A total of 39 significant methodological variables were identified for all of the methods with individual methods having between 6 and 18 methodological variables each. The error analysis requires that the potential of each methodological variable to influence the overall accuracy of a map be estimated. These estimates of influence were taken from published studies that have compared different approaches within a methodological variable, and subsequently compared the accuracy of the resulting maps (typically via cross-validation). Sometimes it was necessary to use expert judgement to estimate the contribution of a methodological variables to changes in map accuracy when other forms of information were not available. Recommended work packages to reduce the influence of each of the 39 methodological variables (termed error reduction solutions here) are listed and prioritised. Error reduction solutions relevant to: (i) classification analysis (also referred to as segmentation); (ii) matching sampling resolution to the habitat resolution (via a habitat resolution catalogue); (iii) improving the positional error of ground-truthing sampling; (iv) increasing the replication and improving distribution of ground-truthing; and (v) reducing reader error during the processing of benthic footage were particularly important for reducing error within the final mapped outputs across multiple methods.

Item Type:	Publication - Article
Digital Object Identifier (DOI):	https://doi.org/10.1016/j.ecss.2020.106684
ISSN:	02727714
Date made live:	02 Jul 2020 14:17 +0 (UTC)
URI:	https://nora.nerc.ac.uk/id/eprint/528057

Item Type:

Publication - Article

Digital Object Identifier (DOI):

https://doi.org/10.1016/j.ecss.2020.106684

ISSN:

02727714

Date made live:

02 Jul 2020 14:17 +0 (UTC)

URI:

https://nora.nerc.ac.uk/id/eprint/528057