It depends on the model's purpose and the parameters to be calibrated, or how "good" assumptions we can support for each parameter's behaviour.

The literature has also suggested ways to cope with such situations, here are two recent relevant works that might be of help:

Article The assessment of water resources in ungauged catchments in Rwanda

Article Method of Prediction the Stream Flows in Poorly Gauged and U...

Hi Mounirou,

I am not sure whether by "minimum hydrometric information" you mean the number model input series. I think the number of series to use is case-specifically restricted by the model structure. Data requirements by lumped conceptual and process-based models are different. Since you talked about parsimonious models I think the model with the least number of input series is better than others which make use of many input datasets.

The question on the extent to which models can be useful and effective in a situation of data-scarcity could be appropriately answered when you know what you want to use the model outputs for. For instance, you can have a regional analysis of river flow to estimate design quantiles for risk-based water resources applications. In line, I refer you to the articles appearing after the next sentence. I also think the use of a statistical metric (such as adjusted R-squared) which applies penalties to the number of inputs you use to generate outputs could be relevant in your case.

Article Empirical statistical characterization and regionalization o...

Article Uncertainties in Flow-Duration-Frequency Relationships of Hi...

In case you were using statistical models, you would make use of the novel fitness function to concurrently seek for the most accurate and compact information you require - see details via the link below:

https://doi.org/10.1080/01431160802549278

Charles

Perhaps this might be useful. I am not saying perfect for every situation. The information is based on a general guide found by Dr. David Rosgen, and requires one with experience in identifying bankfull flow indicators, such as found at top of point bars. Bankfull flow is approximately the mean annual flood with a variable return period of about 1-2 years, most commonly about 1.5 years. This is the channel forming flow, and where most sediment is moved through time due to its frequency, and the stream power to move sediment declines once the floodplain is engaged. If one takes the bankfull depth at the thalweg and multiplies it by 2, this approximately defines the cross section of the 50 year flood. Entrenched streams such as gullies may have difficulty in identifying point bars, and have no active floodplain, so there may not be a energy loss above bankfull, so sediment movement and delivery remains high. So obviously, if one installs a culvert to contain only bankfull flow, it’s going to be overtopped most years and needs heavy protection. If the culvert can pass the flow contained in the two times bankfull flow cross section, it roughly will have a 2 percent chance each year for overtopping, so in 20 years, there is at least a 40% chance of overtopping, unless there are other issues with sediment, bedload, large wood, etc. Just take this guidance as rough, but if culverts we being installed with no guidance, reading and using the channel morphology may help. If your work is primarily in urban areas, zip would suggest developing an envelop curve for the physiographic area, which is the highest measured flow for all gauged stations against the station drainage size. Some specialists in geomorphology may be able to estimate flows at sites with history of high water marks or geomorphic indicators. The importance, technical and funding support for a long term stream gauging network is undeniable. If there are some long term sites available within the physiographic area, one might be able to assess use and applicability of various models, making adjustments as appropriate. But one must note that even with long term data, the confidence limits may be plus or minus 40%, so there still is a need to identify and evaluate rainfall variances such as caused by topographic factors, differences in land use, soil types, vegetation types, etc.

I found some indicators concerning low flows such as looking for aquatic insects to help determine flow permanence (perennial, intermittent and ephemeral flow). Also in some areas, water lines on the inside of culverts may develop at normal low flows. Identifying the meaning of these water lines relative to gauged streams in your area might develop as useful indicators. Also sometimes the moss, lichens or other indicators on stable channel substrates might also be useful. Even if one does not have full equipment and expertise to gauge streams, installing channel staff gauges, surveyed cross sections, peak gauges or vented or barometric pressure adjusted transducers to record stream stage can be very useful, and hire specialist to develop stage discharge relationship, with checks for adjustment through time.

Thanks you Mr Charles Onyutha

Thanks you Mr William F. Hansen Angelos Alamanos