Compare input precipitation agains output runoff

[jochenklar]

This issue references the following user stories:

• As a modeller, I want to see the relation of model output variables to climate input variables in order to explore how much control an input variable has on an output variable.
• As a modeller, I want to compare my model's relation of output variable to climate input variable to those relation of other models in order to find out inter-model differences in those functional relationships.
• As a modeller, I want to assess differences (space/attribute/time) in those functional relationships in order to learn more about how those relationships change by such characteristics.

[jochenklar]

Adding @hmschmie

[thiasB]

To my understanding: As a creator of the tool, input and reference data shouldn't differ much in the demand and spectrum of possible comparisons. That's why I'm im favor of treating them the same more or less generic way. Distinguishing both only adds unnecessary complexity an code duplication. In the end the tool has to be used wisely by also acknowledging that not every test need to make sense on every (reference) data/variable relation.

[hmschmie]

Thanks. For this issue, reference data (observations) are not relevant as here, the relation is solely on already available data at the isimip repository - either input variables and model output; model output against model output or combinations of those for other models. The three user stories just show different aspects of the comparison - 1) internal for model x e.g. with a simple regression, 2) comparing such simple regressions with other model's regression and 3) to add spatial characteristics such as the xy-plots where the points are colored according to different spatial characteristics (as we have seen in the presentation from Sebastian and Thorsten).

For reference data (observations) I totally agree that there needs to be a common understanding of a data format / structure to avoid additional conversion.