Slides for the "Wrap your model in an R package" talk @ useR2016
Environmental models are:
- (Usually) not implemented in R
- Generic: i.e. the model parameterisations need to be defined for each case-study (at least one model run), possibly followed by a sensitivity analysis (multiple model runs) to identify the model parameters with the highest impact on the model predictions. Finally a model calibration step (multiple model runs) for minimising the error between measured and predicted model output is performed by varying the most sensitive input parameters (which cannot be determined precisely enough through measurements) within a realistic range.
Rule-of-thumb: "If you’re going to do something three times or more, you should think about writing a small package"(Roger D. Peng, 2016)
- R package: https://github.com/KWB-R/kwb.wtaq
- Tutorial: https://kwb-r.github.io/kwb.wtaq
In addition we also "wrapped" the following models (but these are not yet available on Github):
Check out our conference paper "Wrap your model!" (Sonnenberg et al. 2014)