The wide range of riverbed hydraulic conductivity values suggests that one should try estimate the hydraulic conductivity of the specific river bed/underlying aquifer or system one is interested in; in order to decrease error. Also, depending on the location, the conductivity value may decrease seasonally (such as the dry season for intermittent streams) where the bed can get choked with fine sediment and algae, together with lower head.

Thanks a lot for your comments. You are right but it seems also reasonable to think that processes and mechanism that create or transform river beds are similar at least at regional level depending on factors climatology, water resources, seasonality, flood occurrence, channel slopes and many others

Yes Javier, i think what you say is true. In the end it depends upon what level of accuracy is desired, and what level of error one may be comfortable with. That would depend on the application -- for instance if its a water balance for a catchment, one can neglect seasonal decreases in hyd cond (i guess!), whereas if its a study of percolation processes or surface water-groundwater interactions, perhaps the average hydraulic conductivity of the specific shallow aquifer may be desired.

Dear Javier,

You may try the following framework, in order to find reasonable values of leakage coefficients (lc):

If you intend to predict in ungauged river reaches, you should try finding similarities regarding to hydroclimatic controls, hydrogeological controls and river morphology between your area of study and those of other studies. Then, you are able to select some lc values. However, note that this process may compromise large uncertainties, which should be evaluated in your study.

If you have data and can calibrate your model, you may also use the aforementioned process to estimate a first guess and/or the ranges of the possible lc values.

kind regards

That is the point Alexandre. What similarities may i use and how do you use them, for example by means of a gis in order to predict these values to large territories

Thanks a lot for your contribution

Dear Javier,

I would begin with a review of all works that estimated leakage coefficients in the climate zone of interest (e.g. Dryland). Then, you have to determine the best group of predictors (e.g. bed sediments, river morphology characteristics) for the leakage coefficient, using a statistical model. Finally, validate your statistical model.

I am looking forward to seeing your results.

Thanks. Of course, that's the procedure. But what we've found are many experiences based on calibration and few results of a conceptual scheme. Do you have any idea or recommendation about a work or article referring a way of organizing climatic, geological and hydrological characteristics?

Usually River-Aquifer Interaction studies are site-specific ones and I did not know any regionalization study on leakage coefficients.

However, I think that is very interesting research topic and you can count with contributions of my side.

best regards

Thanks a lot for your comments

The best way (and to my concern the only reliable one) is to determine this coefficient by discharge measurements up- and downstream the stretch under concern.

however, when there are no discharge and head difference (between the stream and the aquifer) measurements and you still must estimate a leakage coefficient, what to do?

I think it would be interesting to evaluate whether there are predictors that enable us to estimate the order of the magnitude of this parameter.

Dear Ronny, folowing Alexandre answer, do you know any experimentation across different basins, from upper to lower territories and articles referring main conclusions about this leakeage coefficient?

Hello Alexandre, one would need to compute a water balance for the stretch of river of the form

inflows - outflows = +/- net groundwater-surfaceater interaction +/- errors

inflows = inflow at upstream end + rainfall over stretch;

outflows = outflow at discharge end - ET

The estimate of net GW-SW interaction thus obtained could then be used to estimate lc based upon river bed area.

As regards experimentation across different basins, one can do a review of the literature like you (Javier) have. The sheer variability of lc though will lead to high uncertainity in results from extrapolating lc values from other basins; and finally the application would determine the acceptable range of uncertainity, as far as i understand. Certainly a review can help make different models based upon river bed /aquifer geology and soil types (alluvial, mud, loam, sandy etc).

Haha,

Ronny Verhoeven is right. Measurements, actual field measurements are the only way for reliable River/Wadi Losses.

Is your river far away, inaccessible even for the most coarse, rudimentary field estimates?

The range of k-values (10-9 - 10-2) is so huge, that it seems to me entirely pointless to use them in any model, blindly. So, rather leave this whole thing out of your model and insert any arbitrary value at will (or a range, if you can manage in the model).

There is a huge set of criteria that drive the river leakage, both in he river (bed) itself as well as in the connected aquifer beneath. I took measurements up- and downstream of occasional storm runoff events in my Wadi but also looked at the underlying geology, the gravel fill and the silt layer between the Wadi fill and the bedrock to get an approximate idea of the local k-values.

I must say, each river is different - maybe you have to share some more info on your specific case first.

Dear Clemens,

Initially the question is referred to information about predictive physiographic, climatic or hydrological parameters in order to estimate a reasonable lc. There is no specific river in which calculate lc, but a conceptual idea or concept reasonable enough is needed. For example, may we use a potential equation depending on the area of a basin in which the exponent over the area varies depending on geology or climatic factor?

Best regards

Ola, Javi

well, in my opinion, this is really problematic as your range of k-values indicates

and because there so many (!) factors that can influence the gw/sw exchange between river and aquifers/unsaturated zone. To weigh these factors is almost impossible as a general world formula.

So, depending on your overall model, you might consider leaving this factor out entirely since a variation and hence uncertainty by the factor 1:1000 (let alone 10million!) really cannot be addressed by general considerations. In my little research area alone, leakage coefficients range from close to zero (no transmission losses) to almost infinite (total loss of flow at a certain point of the Wadi).

Thanks a lot

But it seems to be reasonable that erosion works in upper rivers making permeabilities higher and narrowing river bed. On the contrary, sedimentation of finer materials tend to occur in lower basins widening river beds and reducing permeabilities. Best regrads and best wishes for Christmas

As has been said above, measurements are the best option. The only problem with using the difference between two flow measurements is the uncertainty in the flow measurements. Basically, the loss has to be significantly greater than the uncertainty in the difference of the two flows. So it comes down to a question of the quality of the streamflow measurements. If the gauges are well constructed and maintained, then this method will work (at least where the infiltration loss is high enough, and there are not other more significant losses over the reach). For many gauges in Australia, this approach is not viable as the uncertainty in the flow measurements is too high, and other losses (e.g. extraction) are also too high.

There are other methods that can be used. If there is a sufficient difference between surface and groundwater, then tracers can be used (e.g. temperature). Again, it comes down to the uncertainty in the measurements, and the difference in the measurements of the tracer between surface and subsurface water.

A recent study on infiltration in ephemeral streams (reference below) used the approach of blocking off a section of a stream, and adding water to this to measure the infiltration loss. Does require a considerable amount of water, but gets a far better measure than using an infiltrometer.

Batlle-Aguilar, J., and P. G. Cook (2012), Transient infiltration from ephemeral streams: A field experiment at the reach scale, Water Resources Research, 48(11), W11518.

Dear Barry, is there any free access to reference cited?. Best regards