data

Links between models and data

Each modeling project uses a set of assumptions that are based on empirical data or observations in the field, i.e. the model input. The model then processes these into an output information, which can be quantitative numeric or spatial data or qualitative information.

On the modeling workshop in August 2014 we listed all input and output data:

The models are connected to real world data of different kind, e.g.:

• expert knowledge: structural information about how natural mechanisms work, or an estimation on which mechanisms are most relevant.
• aerial images: photographs (rgb or spectral) that provide sufficient resolution and topographic correction to assess pattern and type of vegetation cover on the landscape scale. Spatial distribution of plant stress can be assessed from spectral images by calculating NDVI.
• cover data: percentages of plant cover of the entire vegetation or of certain plant species or functional groups (e.g. seeders and resprouters)
• soil data: characteristics of soil
• management data: what kind of management is applied to a site? How do management practices affect the mechanisms of the model (quantitatively or qualitatively)?
• species traits: particular features of functional groups, e.g. nursing, resprouters, seeders.
• climate data
• comparative data: originating from an experiment, these data answer a particular question, e.g. how does seedling survival differ below plant canopy and in the open, soil characteristics in the open or below a patch.
• anecdotal data: data that answer a particular question for model parameterization, e.g. average plant lifespans, data on plant cover. These might be taken out of any research publication or be unpublished data from related experiments within or outside CASCADE or single measurements.
• complete datasets: data that are providing a parameter value under a range of conditional values, e.g. vegetation cover or plant composition on all CASCADE sites. These data naturally would come from other Workpackages in CASCADE and can be used for model specification and validation.