The dilemma- the heated debate

As I briefly mentioned in the previous post, there is an ongoing historical debate going on about lumped models, which are usually conceptual, and physically-based distributed models. I personally enjoyed reading the literature on this issue. It is quite interesting to see how different authors criticize others’ work!

 

So I decided to write a brief post summarising some of the papers I read and hopefully you will get a sense of the situation. 

You can trace the debates back to 1989, when Beven responded to some the claims of physically-based model promoters. 

Some argue that due to the fact that lumped models are not physically- based and do not consider the physical processes that happen in catchments are not a good choice of modelling approach and that they are over-simplified version of the real world system. Furthermore, there is this criticism towards lumped models that they require long hydrological and meteorological data for calibration which may not always be available. (Abbott et al., 1986)

At this point the other party is clenching their teeth! Saying that distributed models themselves are “lumped” at grid scale and they are far from the real world system as they do not consider the spatial heterogeneity of the system. (Beven, 1989)

Some also argue that since these models consider the spatial variability of the catchment response, they are assumed to be a better approach especially when modelling the environmental changes as these changes only occur in some parts of the catchment (Bormann et al., 2009)

And lumped modelling approach supporters say what can be seen in many papers, that spatially explicit models are complex, and difficult to apply and are demanding parametrically and computationally. (Beven, 2012)

My personal view on this issue is that there is no generally right or wrong decision, as models (whatever their type is) in hydrology are simplification of what really happens at catchments. So I don’t think one can generalize a single best approach as the performance of models vary depending on catchment specification and the aim of the modelling itself. As Seiller, Anctil, and Perrin (2012) concluded from their study, the accuracy and predictive ability of models cannot be anticipated for a specific hydrological and catchment condition, thus, intercomparison approach can be used as an appropriate tool to handle this issue.  

Hope you got a brief perspective of the discussions made in the literature. Quite interesting I have to say, scientific soap opera!

So long,