All along the chain of hydrometric data processing, from the raw measurements in the field to the validation of discharge, there are a number of recurrent errors which occur and which we should concentrate on reducing. Methodological rigor will allow us to obtain consistently reproducible and satisfactory results.

In fact, a quality approach doesn't mean concentrating all of our attention one particular factor but rather eliminating the main causes of error, then using the recommendations of the good practices guide which are set out in the following chapters.

At the risk of being obvious, we will first present some major error factors. Our experience has lead us to observe that every hydrometrist has at one time or another made a major mistake. Distance is sometimes necessary for us to evaluate our practices. We hope that this guide, which is intended for everyone, will be useful for developing high quality hydrometry.

1-For Gauging, measure the entire discharge

A backwater discharge forgotten in high water conditions, a partially obstructed (and thus unmeasured) mill's runway in an urban zone, a secondary branch tangled in plant growth or other similar oversights can affect measurement results anywhere from 10 to 100% no matter what the discharge.

This inaccuracy could be even greater then errors made when determining discharges using floaters to measure water surface speed.

Measurement depends, then, on careful, meticulous field observations.

Also daily practices should include care attention to equipment maintenance and to the number and positioning of vertical gauges.

2- Often at a given site, only part of the recorded levels can be converted into discharge

On some sites, because of downstream conditions such as a blockage or gates, the levels might only be valid when greater than a certain value.

At other sites, the opposite could be true: information might only be valid at lower levels, for example, when there is a backwash of a flooding tributary.

Still other sites could be affected by plant growth and require a greater number of gauges.

These site constraints should result in:

It is not sufficient to continue using a site whose information does not meet quality criteria solely because that site has been providing discharge data for a long time.

3-The main uncertainty about rating curves- extrapolating discharges beyond measured values

The use of simple arithmetic or logarithmic extrapolation from the office is a practice used too often. Easy to use, these methods are dangerous because the results are often far too weak or too strong. Moreover, extrapolation based on an extreme measurement is fragile; the absence of measurement after overflow brings about estimations which could be highly erroneous. It would be more appropriate to insist on the fundamental role of real measurement.

A hydrologist's good knowledge of the area allows him to take a suitable approach to discharge measures and thus derive more reliable extrapolations.

Methods of fine tuning estimations at high discharges include:

1) making a cross profile which allows extrapolations from average speed and wetted section;

2) taking into account the geometric characteristics of measuring sites and their control sections (high flow channel). These can also be useful in setting up a small model. This fundamental work, when it is done, has a significant impact on the outlying values of high water measurements. This approach is always more reliable than "office extrapolation".

4- General conception

Hydrometry is not a laboratory science; it should be done in the field. Dividing up the work into specialized tasks is not recommended. On the contrary, project leaders should be out in the field with their researchers, to understand the real political, motivational and material constraints present.

Hydrometry should not be just a routine but rather a profession open to technology, intelligence and re-evaluation.