 |
| Titel |
Spatial prediction of potential wetlands at the French national scale based on hydroecoregions stratification and inference modelling. |
| VerfasserIn |
Lionel Berthier, Lubomira Guzmova, Bertrand Laroche, Sébastien Lehmann, Hervé Squivident, Manuel Martin, Jean-Philippe Chenu, Emmanuel Thiry, Blandine Lemercier, Marion Bardy, Philippe Mérot, Christian Walter |
| Konferenz |
EGU General Assembly 2014
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 16 (2014) |
| Datensatznummer |
250097225
|
| Publikation (Nr.) |
 EGU/EGU2014-12780.pdf |
|
|
|
|
|
| Zusammenfassung |
| The prediction of wetlands at the national scale is crucial to design and implement consistent
national policies aiming at preserving these fragile but essential ecosystems involved in the
production of many fundamental ecosystem services. Depending on the topographic
situation, the occurrence of potential wetlands is mostly driven by geomorphology and
rainfall and the vertical distance to the channel network. This study aims to test for France the
nationwide implementation of a method based on a topo-climatic index in combination with
the vertical distance to the channel network to predict the spatial distribution of potential
wetlands.
Following the PEEW approach used in this study (based on potential, existing, and
efficient wetland functional assessment), potential wetlands include all geographical
situations where geomorphological and/or climatic criteria may entail high probability of
wetland occurence. Potential wetlands include therefore former wetlands areas destructed for
instance by artificial drainage, urbanization or river straightening. Topo-climatic index and
the vertical distance to the channel network were calculated nationwide. To determine the
threshold of these indices, France was stratified into hydroecoregions, based on geology,
climate, topography, vegetation and channel network density. Within each hydroecoregion,
cumulative frequencies of the topo-climatic index were calculated and a threshold was
determined according to pedological data (percentage of hydromorphic soils) available in the
area and using a probabilistic approach, whereas the threshold of the vertical distance to
the channel network was computed alongside the potential flood zone. Both were
combined according to a decision tree based on permeability of the soil surface and
topographic situations in order to get a national prediction of the location of potential
wetlands. Results were validated against point-specific soil data and detailed soil
maps.
Combination of climato-topographic index and the vertical distance to the channel
network was better at predicting the nationwide spatial distribution of wetlands with a highly
improved accuracy by comparison with monofactorial modelling. The first validation test
using point-specific soil data were satisfactory, the overall accuracy of validation equaled
67.8%, the accuracy for potential wetlands equaled 54.6% and the Kappa index amounted to
0.17. The accuracy seems to be poor for areas where few soil data were available to
determine the threshold, thus stressing the need for further acquisition of soil data. The
potential of this strategy to predict wetlands over large areas was therefore confirmed. |
|
|
|
|
|
|