Krieged BroadScale Substrate Map of Fulmar rMCZ
Description
A geostatistical analysis of the data is reported leading to the selection of a linear model of corregionalization for the composition of the sediment, based on the additive log-ratio transformation of data on mud, sand and gravel content. This model is then used for spatial prediction on a 250-m grid. At each grid node a prediction distribution is obtained, conditional on neighbouring data and the selected model. By sampling from this distribution, and backtransforming onto the original compositional simplex of the data, we obtain a conditional expectation for the proportions of sand, gravel and mud at each location, a 95% confidence interval for the value at each node, and the probability that each of the four sediment texture classes that underlie the EUNIS habitat classification is found at the node.
Resources
| Name |
Format |
Description |
Link |