Masante, D.; Thomas, A.R.C. (2019). Simulated landscape metrics of configuration and Ecosystem Service provision across virtual landscapes, based within River Conwy . NERC Environmental Information Data Centre. https://doi.org/10.5285/67f9fe33-14dd-4676-9a6d-65fdbafe2a46
These data are metrics of landscape configuration and modelled provision of Ecosystem services for a large number of virtual landscapes (c. 7500) superimposed on real topography. The landscapes are made up of patches of woodland interspersed across a grassland, and were generated with the landscapeR package in R. The topography used is from the Conwy catchment, split into 10 sections to enable comparison between topographies. Metrics were generated for each virtual landscape to quantify landscape configuration. An Ecosystem Services model (LUCI) was run to calculate a metric of “area mitigated” as a proxy for the provision of runoff mitigation Ecosystem Services. Simulated landscapes were established to answer two questions: firstly to identify the relative controls of patch area and fragmentation on service provision and secondly to identify catchment feature controls on these relationships.
The work was done by Dario Masante and Amy Thomas, with input from Laurence Jones, as part of work under the NERC Biodiversity and Ecosystem Services (BESS) project NERC Grant Ref: NE/K015508/1.
Relatively simple landscapes were produced consisting of two components in varying landscape configurations. To do this LandscapeR, was artificially set with the whole area as improved grassland. Patches of woodland were then added within this landscape, in controlled combinations of patch area and number of patches. LandscapeR was set to expand patches in circular patterns to minimise the influence of patch shape on our analysis.
In the first stage of the simulation, 100 simulated landscape configurations of woodland patches and improved grassland were created for the Hiraethlyn sub-catchment, allocating patches randomly up to the user input number of patches and area of coverage.
In the second phase of analysis, the landscape generation approach was repeated for 10 test areas. For this phase, landscape proportion and patch number were not pre-set, so the landscapes generated have random numbers of patches and fall along a continuum from 0-100% woodland, instead of being clustered around set landscape proportions.
The Land Utilisation Capability Indicator (LUCI) model was run for each landscape to provide a value of “area mitigated”. For both datasets, landscape metrics were calculated using SDMTools and SAGA terrain modules in R.
QA: Note that as patches start accumulating in the landscape, they merge and so the complexity of patch shapes inevitably increases. For dataset “Output_SummaryHiraethlyn”, since patch placement was random, some user set combinations were not achieved in the output, due to merging of patches, reducing the actual number of patches. To minimise merging, when allocation resulted in a lesser number of patches than seeded, the process was repeated until the seed number was reached, or the highest in one hundred trials. User input landscape proportion was given preference over achieving the number of patches. Merging may therefore result in a negative correlation between landscape proportion and number of patches across the dataset, and we must take this into account when analysing the outputs.
Data were output to Excel spreadsheets. Data was then exported to comma separated files for ingestion into the EIDC.