Do we need a Community Hydrological Model?

We believe that there are too many models in hydrology and we should ask ourselves the question, if we are currently wasting time and effort in developing another model again instead of focusing on the development of a Community Hydrological Model. In other fields, this kind of models has been quite successful, but due to several reasons, no single community model has been developed in the field of hydrology yet. The concept, strength, and weakness of a community model were discussed at the Chapman Conference on Catchment Spatial Behaviour and Complex Organisation held in Luxembourg in September 2014. This discussion as well as our own opinions about the potential of a community models or at least the necessary discussion to establish one are debated in this commentary. 1. There are Too Many Models in Hydrology This is partly because every generation of PhD students thinks they could do better, partly because there is a wide range of different types of applications with different needs in hydrology, and partly because there is no agreement on a common set of concepts for the process representations. However, many of those models are similar in structure (for both the ‘‘bucket’’ model and the ‘‘physically based’’ model cases) so that the same, or very similar, concepts have been programmed and tested over and over again. There are advantages to this when models are simple, of course, but as model and data implementations get more and more complex this could be considered a waste of community effort and resources. In the atmospheric modeling domain, there have been initiatives to have community models, both for the atmosphere itself (e.g., CESM [Hurrell et al., 2013] and WRF [Skamarock and Klemp, 2008]) and for the land surface boundary condition, including a representation of the hydrology (e.g., NOAH-MP within WRF [Niu et al., 2011] and CLM [Lawrence et al., 2011]). These models have structured version release and development test bed programs. The Community Land Model (CLM) is widely used as a land surface parameterization in the hydrometeorological community but has not made much impact on the hydrological modeling community. It is not even mentioned in Beven [2012], because it is aimed at partitioning the surface energy balance for different surfaces within a grid, rather than getting better predictions of river discharges. Indeed, it would not pass scrutiny as a satisfactory hydrological model for many hydrologists. Some models developed within the hydrological modeling community have also been released in open source form with a view to building up a large user community (e.g., SWAT, HYPE, or different model codes in R and MATLAB for HBV, TOPMODEL, or Dynamic TOPMODEL (see R codes for both models at http://cran. r-project.org) [e.g., Str€ omqvist et al., 2012; Buytaert, 2013; Fuka et al., 2014; Metcalfe et al., 2015]), and a number of model repositories also exist (e.g., at CSDMS, http://csdms.colorado.edu). The trend in hydrology has rather been to create even more new models, or to provide metamodeling frameworks (SUPERFLEX [Fenicia et al., 2011], FUSE [Clark et al., 2008], FARM [Euser et al., 2013], and SUMMA [Clark et al., 2015a]) that allow users to combine different process representations or try many different conceptual model constructs. There has also been continuing development of physically based models built around Darcian representations of the subsurface as both open source (PARFLOW [Maxwell et al., 2014], MODFLOW [Banta et al., 2000], and commercial products (HYDRUS3D [ Simůnek et al., 2008], MIKE-SHE [Refsgaard et al., 1995], and HYDROGEOSPHERE [Therrien et al., 2010]). In both these types of catchment model, there are commonalities in approach and structure but no real convergence to a common goal. Correspondence to: M. Weiler, [email protected] Citation: Weiler, M., and K. Beven (2015), Do we need a Community Hydrological Model?, Water Resour. Res., 51, 7777– 7784, doi:10.1002/2014WR016731. Received 26 NOV 2014