How do I interpret my accuracy assessment?

Clearly, given the complex issues surrounding accuracy, it is not a simple matter of equating accuracy with usefulness. The example shown below shows two versions of the same map: the data in the left hand were originally interpreted to 12 life-form habitat classes. A number of these classes can be considered to be part of a more general 'reef habitat' (with outcrops of bedrock and boulders) with varied life forms, with the remainder being sediment habitats. If the classes are amalgamated into these two broad groups, as depicted in the right-hand map, map accuracy increases considerably. This might seem obvious. The detailed map has lower accuracy because there is greater similarity between the reef habitats when modelling their distribution based on physical and biological characteristics: for example, kelp forest and kelp park differ mainly in their depth distribution. In contrast, there is a considerable difference in the physical and biological characteristics of the the rocky habitats and the sediment habitats. It might well be the case that managers simply wish to know where reefs occur. However, the map on the left shows far more information about the distribution pattern of the component reef habitats and a degree of overlap might be acceptable especially if this is between similar habitats.
Two version of the same map showing different levels of accuracy:
12 life-form classes (left) and two habitat types (right).
The map on the right is more accurate, but is it more useful? Accuracy is usually traded off against information content in the interpretation and analysis of a dataset. At one extreme, the interpretation may attempt to show subtle variations in habitat content that are simply not supported by the data. At the other extreme, the habitats are so general that the information is not useful for most purposes. An alternative to measuring the accuracy of a map is to determine the confidence someone using the map for a particular purpose has in the map. For more information see 'How much confidence should I put in my map?' and 'The MESH approach to confidence assessment'.
Links to other sections:

All material variously copyrighted by MESH project partners 2004-2010

| List Access Keys | Mapping European Seabed Habitats (MESH) - Home | Overview | News | Work areas in the MESH Project | Partners | Contact Us | Search | Site Map | Partner Extranet |