Climate transposability- case of Canada and Germany

As some components of the water cycle are already affected by climate change, such as intensity and frequency of precipitation, duration of snow cover, surface runoff, evapotranspiration, etc. Hence, there is a need to quantify the impacts of climate change on water bodies.

Seiller,Anctill, and Perrin (2012) recently carried out a study that investigated the robustness of twenty conceptual lumped models when climate conditions are significantly different between calibration and validation. The study investigated the models both individually and collectively for temporal transposability, the capacity of a model to perform with same accuracy when the conditions differ from those used in calibration process.

They studied two catchment areas, Au Saumon in Canada and Schlehdorf in Germany. The method of Differential Split-Sample Test (DSST) was used which is based on calibrating the model on pre-change conditions and validate it on post-change data. So, past climate conditions were used for calibration while present and future climate projections were used for validation. This puts the model in demanding conditions where calibration climatic conditions differ from validation data Seiller,Anctill, and Perrin (2012) . 

The four configurations are shown below, where the boxes on the left show the climatic conditions used for calibration and the boxes on the right illustrate the climate condition of validation.

 

Individual assessments of the model concluded that it is difficult to identify a single best model that performs well under all the contrasting conditions. For instance, the MARTINE model performed better when it was calibrated for dry conditions and validated for humid years, while GARDENIA model was the opposite. The collective analysis, however, gave better results for all DSSTs for Au Saumon, while this held true only for one DSST on Schlehdorf catchment Seiller,Anctill, and Perrin (2012).They finally concluded that in general lumped models performed poorly in terms of climate transposability, however, the models combined resulted in a better climate transposability, “as if the many model structures compensate for one another’s weaknesses”.

This has been suggested by other studies as well. For example, Bormannet al.(2009) study investigated the performance of models with different spatial resolutions on two catchments with different environments. This study concluded that while each mode offers its own specific advantages depending on data availability, boundary and scale conditions, combining the results from different models and applying a multi-model approach reduces the uncertainty in prediction.

These conclusions to me personally were really interesting as it, to some degree, might calm the storm of which modelling approach is better (an ongoing discussion about lumped and distributed models). 

Hope that this post showed an interesting example of how the impacts of climate change on water bodies are carried out.

So long,