Uncertainty of modeling results is a major issue in hydrological and environmental modeling. Uncertainties depend on a number of factors including simplifications in the conceptual model, processes occurring in the watershed but not included in the model, process that are included in the model, but their occurrences in the watershed are unknown to the modeller or unaccountable, processes that are not known to the modeller and not included in the model and errors in the input variables such as rainfall and temperature.
Article SWAT Model Application and Prediction Uncertainty Analysis i...
It is a complicated issue, however it could be solve by using proper optimization procedure. We try to verify ice roughness coefficient on 40 km reservoir using genetic algorithms and our experiment could produce up to 6 parameters. If we try more parameters results did not converged well.
Don't know if my answer would be confusing. But, people on climate modelling usually use the ensemble framework to evaluate uncertainties in modelling. This is also used in hydrological modelling to give a measure of uncertainty in results. Don't know if your question is really focussed on the analysis of variable and parameter influence as well as sensititvity. Best regards
Good point Thomas. Any model can only be conditionally validated. Application of the model outside the conditions under which it has been tested results in increased uncertainty in the accuracy of the predictions - but also difficulty in assessing what those uncertainties are. However, it should be noted that the period available for model testing depends on both the available length of record as well as the length of the calibration period. Thus it is best to keep the calibration period as short as possible, while still obtaining good predictive accuracy.
How well a model can be tested depends on the variability in the conditions. This includes climate and land use (as pointed out by Thomas). So you need to look into the historical record to see what has happened over the period where you have available data.
I suggest you "ERROR ANALYSIS". I am uploading a manual atached to the message which is a manual for engineering student in UCLA.
- Badges
- Science topic
- Similar topics
- Agricultural Science
- Agronomy