The large number of empirical/conceptual procedures included into SWAT,there is much doubt about the really ‘physically-based’ character of SWAT model.
Yes Hamzeh, SWAT model has a conceptual character. I found that most of the hydrological models that need a calibration of some parameters should be called semi-physical. Because, parameters derived through calibration are just conceptual. But, if all parameters are empirical (no need of calibration), the model is then physical.
Yes Dear Bcz it takes lots of physical inputs and parameters like soil, dem, lulc etc.
Having used SWAT (and published) I would agree with the carefully argued points made by Dr Lampert. Not that this means SWAT is necessarily inferior to a physically-based model (if one existed) - you may never have enough data to parameterise such a model.
SWAT model is a data driven software. There are many completed types of data such as topography, land use, soil, vegetable cover, climate, human activities,.... In practical saying, SWAT model users may not collect all necessary data for the model.
SWAT has grown over the years by adding several different modules that were partly taken from other models (that were developed for different scale applications) and simplified. Hence, some parts of SWAT are more "process-based" than others. Using a model and interpreting the results requires always a careful look at its structure and how processes are described (regardless whether "physically-based" or not).
I am working since years in the SWAT developers group and we found that some users are still not aware of this fact.
However, despite the use of default values for every "physical" parameter in the equations, conceptual and physically-based models should also undergo a calibration-validation process because of the scale effects. These arise due to the differences between point and cell /surface calculations and the numerical effects. The calibration produces effective values for the original parameters, which must, however, retain their significance and be representative of the original range of conditions.
you may be right,but hydrology it self very complicated and if we are using some model then we are simplifying it this way or other.
as per my view and reading/observing swat, it require much data and on the basis of that it can predict about runoff,sediment,erosion under different management constraints.
there are so many data and much noise created ,we have to listen which we want to listen .the simple example i can quot "every time I am telling my civil engineering students that ,if you want to become good civil engineer, then just try to see every where civil engineering!"
anyway we have to grab what we want!
Not fully because it uses the USLE method for estimating soil erosion.
Interesting discussion. Why are you asking this question? Will the answers influence your choice regarding which model to use? In my view the choice of a model should be a balance between the modelling aim (what do you want to simulate at which spatial and temporal scale and with which accuracy), complexity and characteristics of hydrological processes in the specific study area and data availability. The question whether a model is physically-based is only partly relevant, in particular to ensure that the model is providing the right answers for the right reasons.
In principle, most models are in some sense physically-based (also relatively simple conceptual ones), but less are process-based (see also the comment by Fulvio Rivano). As Martin Volk mentions, all models and their results require a critical look, and also, uncertainties should be considered, quantified and taken into account when interpreting results. This also applies to the SWAT model, in particular because the (spatial) detail of the model often suggest more physcial realism than can be supported by the nature of the process equations and the data used in case studies (e.g. number of sub-catchments vs. discharge stations). Looking at the different process formulations, SWAT is very similar to many (often called) conceptual hydrological models. The main difference regarding hydrological modelling is the spatial scale (resolution) where conceptual models are usually applied at lumped or semi-distributed scales.
Hi Martijn,
nothing to add ;o), and SWAT simulates not only hydrological but also several other biophysical processes. By the way, we also developed a grid-based SWAT landscape version - where we are aware that this even complicates the discussed issues (how useful is the gridbased version on which scale? ;o) see (or find the article at my RG site):
http://onlinelibrary.wiley.com/doi/10.1002/hyp.10197/abstract
Best,
Martin
I just found this question and answers interesting! I believe it depends on how you define the term, "physically-based." If that term means that the values of all parameters should be able to be determined by direct physical measurements so that parameter calibration should not be required, only field or laboratory scale mathematical models would be called "physically-based" ones. If the term means, on the other hand, that equations and simulation mechanisms employed in a model should have a physical basis, I believe all hydrological models (except for purely statistical models such as ARIMA) can be called "physically-based" models including a simple storage model (e.g. TANK and ABCD models). The model classification schemes can help us understand the features of models, but the classification may not be used to judge the soundness of a model, which is totally dependent on the simulation strategy and conceptual framework used in the model. For instance, I found that a hydrologic model that has been regarded accurate and physically-based and thus very popularly used to describe overland flow processes assumes constant overland flow velocity over time, which can be applicable to a certain case, but I may not use that model for my overland routing research. I believe we may need better and more precise model classification schemes so as to avoid unnecessary confusions and prejudgement against any model.