A –Streams of Thought– contribution by  Wouter Knoben, Shaun Harrigan, David Wright, Wouter Berghuijs

Scaling (i.e. the transfer of knowledge across scales) and scale issues (i.e. the associated problems) are at the heart of most hydrologic puzzles. In the most recent “Meet the Expert in Hydrology” session, organized at the EGU General Assembly 2017 in Vienna, YHS invited three speakers to identify to what degree their research is connected, influenced by, and influencing research at other spatial scales. By evaluating the current state of research and discussing future directions we tried to shed some new light on the question “Is hydrological research at different spatial scales connected?”. This is what we learned…

Hydrological research takes place across extremely different spatial scales (from pore-scale to global). Experimentalists have an understanding of detailed hydrological processes of hillslopes and headwater catchments, model developers generate larger-scale hydrological descriptions to simulate streamflow, while global hydrologists use such models to simulate the global water cycle. In theory, research at these different spatial scales could be strongly connected. For example, field experiments lead to new understanding and models, which lead to better hydrologic predictions at larger scales. In practice, hydrological research across different spatial scales can, at times, be disconnected. Barriers, such as limited data availability and consistency and the enormous heterogeneity across diverse landscapes, may limit the interaction between scales.

In the “Meet the Expert in Hydrology: Is research at different spatial scales connected?” we invited three experts who in their work focus on different spatial scales: Prof. Jeffrey McDonnell (University of Saskatchewan) focuses on novel experimental work at the hillslope and catchment scale, Dr. Markus Hrachowitz (Delft University of Technology) focuses on catchment scale models, and Prof. Reed Maxwell (Colorado School of Mines) develops models for hydrologic simulations across continental and global scales. The three experts were invited to share their experiences on bridging the connections between different research scales, the difficulties, and opportunities of transferring knowledge between these scales and their views on future developments in hydrological knowledge from the field to the global scale.

Figure 1: Three experts were invited to talk about their respective research scale, and the connection to other spatial scales: Prof. Jeffrey McDonnell (University of Saskatchewan) focuses on novel experimental work at the hillslope and catchment scale, Dr. Markus Hrachowitz (Delft University of Technology) focuses on catchment scale models, and Prof. Reed Maxwell (Colorado School of Mines) develops models for hydrologic simulations across continental and global scales.

The role of fieldwork at small spatial scales

Jeffrey McDonnell first talked about the tremendous amount of heterogeneity that is present in small hillslopes. An example he highlighted was the study that led to the “fill-and-spill” hypothesis (Tromp‐van Meerveld & McDonnell, 2006). This analysis of a single hillslope over various storm events shows that subsurface depressions first have to fill up before the hillslope starts to produce runoff. This is potentially very similar to the processes that generate runoff on the land surface.

Jeffrey McDonnell argued similar threshold processes occur in many different parts of the hydrological cycle and that the heterogeneities are essentially unknowable. Emergent behavior might be the best option to deduce hydrological functioning for scales larger than experimental plots. He then proceeded with an example of comparing two relatively close catchments that show wildly different dominant features. In such cases, tracer hydrology can be valuable to confirm (or to challenge) hypotheses of catchment functioning; tracer experiments can be used to better understand how catchments work. He showed a further example of how experimental-scale research can be expanded to a global scale using a meta-analysis of experimental studies across the planet (Evaristo & McDonnell, 2017).

The role of catchment-scale model development

Markus Hrachowitz argued that hydrology has an observation problem, not a science problem. He gave examples of how during his Ph.D. research he found that it was extremely challenging to find hillslopes that were representative for other parts of the (very small) catchment he studied. He gave another example of measurement challenges by emphasizing that snow measurements may not be representative for snowfall only just a few meters further in the catchment.

He subsequently argued that we can still make effective predictions at the catchment scale by looking at emergent relationships. While we may not be able to measure and model the full heterogeneity present at the catchment scale simple emergent relationships at larger scales can help to develop useful models that we can parameterize and that can represent the catchment’s overall functioning. He discussed the example of root-zone storage, which while highly heterogeneous at small scales, is predictable at the catchment scale (Gao et al., 2014).

The role of large-scale simulations

Reed Maxwell argued that computational models are a key tool in bridging the gap between research scales. In his diverse research group (working both on field studies and large-scale modeling), they use models as a place to integrate the knowledge and observations across scales and test whether that matches with our understanding described in hydrological models.

He gave examples where modeling and field observations have been integrated; a recent publication tested the role of increased radiation versus the role of rain to snow transition for total runoff generation (Foster et al., 2016). A second example he discussed is how including lateral flow seems to strongly improve the match between modeled transpiration and observed transpiration (Maxwell & Condon, 2016).

However, he carefully stated that, while models are useful we need to realize that all models are predicting the water cycle in a comparable manner, and they may all be wrong.

Discussion on increasing the synergy between research scales
During the final 40 minutes of the session, there was a panel discussion on how we can increase the synergy between research scales.

The speakers agreed that focussing at emergent behavior on larger scales can inform hypotheses, that then should be tested by modelling and experimental work.

Hydrologists have focused too long on reproducing the hydrograph. There are plenty of models that can accurately reproduce the hydrograph. While this is very useful for operational purposes, it does not allow much progress in better understanding the water cycle.  

Hydrologists have been “farmers”  for too long (i.e. collecting data and running models because we can) and we should become “hunters” instead (i.e. actively pursuing the relevant and interesting hypothesis that are testable). In this way, we can become much more efficient in making progress.

As the resolution of models is improved, the hope is that the heterogeneity will disappear and the governing equation that we use can represent the processes that we will model. In reality, heterogeneity will not disappear, at any scale. We need to critically think how we better deal with sub-grid heterogeneity as currently we often apply the equations that are representative for other scales and hope that the heterogeneity can be adequately described by parameters that we tune. This premise may be wrong for particular processes. For example, is Darcy-Richards equation really the best equation for gridded models covering large areas?

The large watershed is not a linear superposition of soils. We should go after the characteristic forms of nonlinearity that can be used in models. For example, it is currently unclear if large scale models can represent the filling and spilling of the landscape.

Observations indicate that water in catchments can be thousands and sometimes even billions of years old. This is a part of reality that is not often represented by current-day hydrological models. However, Reed Maxwell emphasized that large-scale models can generate very old water ages.

References

• Evaristo, J. and McDonnell, J. J. (2017). Prevalence and magnitude of groundwater use by vegetation: a global stable isotope meta-analysis. Sci. Rep. 7, 44110;
• Foster, L. M., Bearup, L. A., Molotch, N. P., Brooks, P. D., & Maxwell, R. M. (2016). Energy budget increases reduce mean streamflow more than snow–rain transitions: using integrated modeling to isolate climate change impacts on Rocky Mountain hydrology. Environmental Research Letters, 11(4), 044015.
• Gao, H., Hrachowitz, M., Schymanski, S. J., Fenicia, F., Sriwongsitanon, N., & Savenije, H. H. G. (2014). Climate controls how ecosystems size the root zone storage capacity at catchment scale. Geophysical Research Letters, 41(22), 7916-7923.
• Maxwell, R. M., & Condon, L. E. (2016). Connections between groundwater flow and transpiration partitioning. Science, 353(6297), 377-380.
• Tromp‐van Meerveld, H. J., & McDonnell, J. J. (2006). Threshold relations in subsurface stormflow: 1. A 147‐storm analysis of the Panola hillslope. Water Resources Research, 42(2).

Citation: Knoben, W. J. M., Wright, D., Harrigan, S, & Berghuijs, W. R. (2017), Is research at different spatial scales connected, Streams of Thought (Young Hydrologic Society), Published May 2017.

About the authors Wouter Knoben is a PhD student at the University of Bristol. Shaun Harrigan (@shaunharrigan) is a research associate at the Centre for Ecology & Hydrology (CEH) in the UK and Early Career Scientist representative for the EGU Hydrological Sciences Division. Wouter Berghuijs (@wberghuijs) is a postdoctoral researcher at ETH Zurich. David Wright is a PhD Candidate at the University of Adelaide

Correspondence to: Wouter Knoben (w.j.m.knoben@bristol.ac.uk)

Acknowledgments We thank Jeffrey McDonnell (University of Saskatchewan), Markus Hrachowitz (Delft University of Technology), and Reed Maxwell (Colorado School of Mines) for their enthusiastic participation in the ‘Meet the expert in hydrology’ session at EGU 2017. The material in this article is strongly based on their presentations and ideas generated during the fruitful open discussion with the audience.