jprenaud.info

In 2006, Jones et al. wrote a paper entitled “An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow” in Water Resources Research concluding that the interpretation of tracer based studies of pre-event contribution to streamflow might need re-interpreting. After discussing the methodology used in this paper at the 2006 European Geosciences Union Spring Assembly with Hannah Cloke and Markus Weiler, we submitted a comment to Water Resources Research. Our comment was published in September together with the reply from the authors of the original paper.

However, Hannah, Markus and I are disappointed by the answer from Sudicky et al. to our initial comment because it does not really answer our criticism. As the comment system in Water Resources Research does not allow comments on a reply (to a comment…), we wrote this short comment in October and sent it to Sudicky et al. who were invited to reply. They have not yet provided us with their answer. We will update this page and make their answer available here when we have received it.

Read our full comment below or download it (PDF version, 36KB) from the website of the University of Bristol.

Comments on Sudicky et al. [2007] in Water Resources Research

JP Renaud^(a,*), Hannah Cloke^(b), Markus Weiler^(c)

^(a) University of Bristol, ^(b) King’s College London, ^(c) University of British Columbia

In 2006, Jones et al. wrote a paper entitled “An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow” (Jones et al. [2006]) in Water Resources Research concluding that the interpretation of tracer based studies of pre-event contribution to streamflow might need re-interpreting. After discussing the methodology used in this paper at the 2006 European Geosciences Union Spring Assembly, we submitted a comment to Water Resources Research. Our comment has been now been published (Renaud et al. [2007]) and the reply from the authors of the original paper is also now available (Sudicky et al. [2007]).

We are disappointed by the answer from Sudicky et al. to our comment. Sudicky et al. do not seem to answer our criticism. As the comment system in Water Resources Research does not allow for a comment on a reply (to a comment…), we have written this short answer to Sudicky et al.

Sudicky et al. start their reply with a set of two mass-balance equations for the flow (1a) and a tracer in solution in the water (1b), where

“[...] the subscripts t, o and gw refer to the total flow as measured in the stream, the surface runoff component arising directly from the precipitation event, and the pre-event portion from the groundwater system, respectively.” - Sudicky et al. [2007], paragraph 2

It is very important to note how these terms are defined. Indeed, research on hydrograph separation focuses mainly on the separation between pre-event water and event water from the hydrograph rather than groundwater and runoff components. This was the bulk of our disagreement with the Jones et al. [2006] paper. It seems that, at the very start of their reply, Sudicky et al. are still confusing two different things: water sources (event and pre-event) and water pathways (surface and subsurface).

They then write equation (1b) in discrete form and note that:

“The subscripts o and gw here denote overland (i.e., event) and groundwater (i.e., pre-event) contributions, respectively, [...]” - Sudicky et al. [2007], paragraph 4

We are very disappointed to read that Sudicky et al. still associate pre-event water with groundwater flow and event water with runoff with no further justification in their reply to our comment. We went to great length in our comment (paragraphs 2, 3 and 4) to explain that looking at overland flow and groundwater flow is not sufficient to distinguish between event and pre-event water. In particular, we said:

“[...] we feel that the quantities used by Jones et al. mix two very distinct attributes, namely the water origin or temporal water sources (new/old or event/pre-event) and the water flow pathways (overland or subsurface, saturated or unsaturated).” - Renaud et al. [2007], paragraph 2

Furthermore, we made clear that when the term “groundwater” was used in lieu of “pre-event water” in previous studies, it was a terminology error. Pre-event water, the water present in the catchment before it starts to rain, can reach the stream via underground or surface flowpaths. We said:

“[...] pre-event water can follow different pathways. Buttle [1994] provides an overview about these different pathways of pre-event water and he points out that groundwater flow is only one out of six processes that can deliver pre-event water to the stream.” - Renaud et al. [2007], paragraph 3

It is very difficult to follow, let alone understand, a paper such as Jones et al. [2006] where the terms are used interchangeably. Using the notation above, the analysis done in Jones et al. [2006] calculates the ratios of groundwater/surface water in the hydrograph and not pre-event/event water ratios. This makes any comparison of their model runs with data obtained by field work pretty much irrelevant; this “apples and pears” comparison was the impetus for our comment.

One of the main points of the reply by Sudicky et al. seems to be to convince us that dispersion is an important process that needs to be taken into when modelling tracing experiments. Indeed it is … and this has been known for a while. This is why models based on the advection-dispersion equation have been developed instead of using the streamlines from a Darcy-Richards model; we fail to see the novelty here. In particular, Sudicky et al. explain that

“[...] in a given time step in InHM, q is calculated from the solution of the mixed form of Richards’ equation before being used in the tracer transport computations. Therefore any alteration of the dispersive transport parameters in the model will not affect the value of q.” - Sudicky et al. [2007], paragraph 4

Of course, this is the reason why flow models cannot be used to trace water on their own. We actually explained why in our comment too:

“The reason this process is not apparent in the Darcy-Richards’ equation is that if a molecule of water is replaced by another molecule of water it does not change anything for the sake of pore water pressure calculations. However, when different types of water are considered, i.e. pre-event and event water, and if interested in the relative contributions of each type to the flow, kinematic dispersion does matter.” - Renaud et al. [2007], paragraph 8

Sudicky et al. provided a separate paragraph about isotopic tracers and again emphasized that dispersion should be taken into account when using this type of tracer too. Of course it should! And we never said that it should not… Our argument is that it is possible in the general sense that a geochemical tracer moves from the pre-event water to the event water flow by diffusion, therefore introducing noise as some event water could be identified as pre-event and vice versa because of a diffusive flux. Our point was that this diffusive flux is less of an issue when using isotopic tracers because the tracer is then a molecule of water and not an ion in solution, therefore effectively tracking water.

In their final paragrah, Sudicky et al. explain that

“It needs to be strongly emphasized here that we are not questioning the veracity of tracer data itself. Instead, we simply questioned the manner in which the data are commonly interpreted.” - Sudicky et al. [2007], paragraph 7

The original text in Jones et al. [2006] said:

“[...] the paradox of the rapid mobilization of old water is perhaps not paradoxical and that interpretations based on tracer-based hydrograph separation techniques should be reexamined with respect to their applicability for estimating hydraulically driven groundwater flow contributions to streamflow.” - Jones et al. [2006], paragraph 34

Again, it is important to note the term “hydraulically driven groundwater flow contributions to streamflow” in the statement above. Tracer based methods such as those used in most of the papers referenced by Jones et al. [2006] do not calculate this contribution but rather the pre-event water contribution to streamflow ratio which is different from the “groundwater flow contribution”. Our comment even provided a short reminder of how tracer based hydrograph separation is done in the field to clarify this confusion:

“When using conservative tracers in the field, both the subsurface water and rainwater are sampled once to determine their respective tracer signature (geochemical or isotopic). Then multiple sampling is then done in the stream at regular time intervals during the storm event and usually for a while after it has ended. The proportion of event and pre-event water is eventually calculated from the ratios of the different concentrations in the stream water and a unit hydrograph derived from these. Hydrograph separation done this way can only separate between sources of water (pre-event/event) and cannot separate between pathways.” - Renaud et al. [2007], paragraph 6

So the ratios obtained from field work are not the same as these calculated by Jones et al. [2006] and then defended by Sudicky et al.

To conclude, we want to repeat that Jones et al. [2006] were comparing apples and pears, that there is no need (emerging from their work) to re-interpret tracer results, and that the paradox is still paradoxical.

References:

Jones, J. P., E. A. Sudicky, A. E. Brookfield, and Y.-J. Park (2006), An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow, Water Resources Research, 42, W02407, doi:10.1029/2005WR004130.

Renaud, J.-P., H. L. Cloke, and M. Weiler (2007), Comment on ‘‘An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow’’ by J. P. Jones et al., Water Resources Research, 43, W09601, doi:10.1029/2006WR005157.

Sudicky, E. A., J. P. Jones, A. E. Brookfield, and Y.-J. Park (2007), Reply to comment by J.-P. Renaud et al. on ‘‘An assessment of the tracer-based approach to quantifying groundwater contributions to streamflow,’’ Water Resources Research, 43, W09602, doi:10.1029/2006WR005416.

Note about paragraph 5:

Paragraph 5 in Sudicky et al. is about the inclusion of an unsaturated component in their analysis; i.e. they separated between saturated and unsaturated components. This is something that we mentioned in the very first draft of our comment but it was then dropped from the argument (following a suggestion by an anonymous reviewer) as it was a minor point not bringing much to the discussion. We are not sure why this paragraph was not removed from their reply.