|
Titel |
Are maps different? Why we do not effectively communicate uncertainty in geographical space. |
VerfasserIn |
Ronald Corstanje, Rachel Creamer, Jack Hannam, Rogier Schulte, Bob Jones, Thomas Mayr |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250094440
|
Publikation (Nr.) |
EGU/EGU2014-9846.pdf |
|
|
|
Zusammenfassung |
There can be considerable amount of uncertainty in geographical (mapped) data. This
may be related to the quality and quantity of the underlying data, spatial models
or manipulations (e.g. scaling), which have been used to derive the mapped data
from the observed data. Whereas significant effort has been made in quantifying
the spatial uncertainty, it is still not clear how we communicate this uncertainty
both to the scientific, practitioner and general public. Moreover, as maps are often
representations of spatial data, and produced with a specific purpose or intent (e.g. soil
or geological maps) and there are different consequences to spatial uncertainty
under different circumstances, depending on the use of the spatial product. The
most effective way to represent this uncertainty therefore depends as much on the
type of uncertainty as it does on the manner in which the uncertainty is quantified.
Here we explore various ways of determining and representing this uncertainty in
spatial data based on a large-scale soil survey project for the Republic of Ireland, in
which only 44 % of the country was ever mapped in detail. A field and predictive
mapping process was designed for the remaining 66 % of the country at a scale of
1:250,000, in order to fulfil national and European policy requirements. The project
combined traditional soil surveying with spatial modelling techniques (digital soil
mapping). Empirical data was collected in to the form of 232 soil pits and over 10,000
auger points. Spatial data has therefore been generated in form of modelled soil
characteristics, and uncertainty estimated from the surveyed data. We determined both the
overall map quality by various model performance and behaviour statistics, and local
uncertainty through various forms of ‘hot spot’ analysis. We also represented the
uncertainty in a set of causal models, in which we described uncertainty in terms of
how well the mathematical models corresponded to the underlying soil landscape
relationships. We found that no singular method effectively describes or communicates
uncertainty in these maps and present a method that combines the various forms of
uncertainty as possibly the best representation of uncertainty for the map users. It is as
yet unclear though whether this method, or the more basic underlying measures,
or any uncertainty consideration at all, will be considered by the soil map users. |
|
|
|
|
|