Rationalising uncertainty when using total pressure loggers for groundwater level measurements

Total pressure loggers are less expensive and for many professionals more convenient to stock and to deploy in the field rather than using vented systems which require connection to expensive and bulky cables.

I know of some NGOs working on development projects in Africa, and elsewhere, who faced with limited budgets and the need to simplify processes in the field, use unvented total pressure loggers for water level measurements without post-correcting the data for variation in atmospheric (barometric) pressure. They simply take the view that the accuracy will be good enough for the type of work they are doing and the need to post-process data is eliminated thereby saving time and requiring less training and the need for fewer troubleshooting skills if data is not collected properly.

So how reasonable is this approach? Can this choice be justified scientifically? Here’s some barometric data to help put this issue in context.

Barometric Variation in the UK

atmospheric pressure variationIn the UK, atmospheric pressure at sea level can vary over a range of 83 mbar or a water head equivalent of approx. 85 cm[i]. Figure 1 shoes the barometric variation over a 10 year period from the National Physical Laboratory at Teddington to the west of London. In this record, measurements are taken every 5 minutes. What is notable is that this variation could occur over a relatively short period of a few hours or a few days.  Consequently, for most applications in the UK barometric pressure correction is essential to avoid a significant margin of uncertainty in the water level record.

The only reasonable exception to this would be when loggers are used to record water levels for short periods of time – e.g. for slug tests in boreholes lasting a few minutes (possibly up to 1 hour).  It may even be possible to justify using uncompensated data for carrying out short term pumping tests or water level monitoring over say 1 to 2 day periods, though these would need to demonstrably coincide with a period of stable pressure, and I doubt this latter application would be perceived favourably by any of our Environment Agencies!

Barometric Variation in Sierra Leone

baro logger use in sierra leoneIn Sierra Leone, and many parts of Sub-Saharan Africa, barometric variations are less variable and atmospheric pressure systems more stable. However, it is difficult to obtain reliable barometric data from on-line sources for anywhere in Africa. It is only through direct measurement that we were able to confidently build up a record of natural barometric variations for our own needs in Sierra Leone (Figure 2). 

Figure 2 illustrates barometric records from 3 of the 11 In-Situ Rugged BaroTroll loggers used in country. These were located at different elevations across the Rokel Seli river catchment with pressure recorded every 15 minutes. This allowed us to capture both annual variability and the change in pressure with barometric pressure variationaltitude. The overall annual barometric range for any one of these records is in the order of 15 mbar – i.e. less than 20% of the range observed in the UK.

What is particularly interesting is that the annual variation is underpinned by a semi-diurnal variation in pressure with a range of up to 10 mbar per day (Figure 3), typical of tropical climates.

The average annual groundwater level variation in the basement complex of Sierra Leone (see "Monitoring Groundwater Levels in Sierra Leone") in response to a total rainfall of 3000 mm per year is in the order of 6 m (Figure 4), and can exceed 10 m.[ii]  Given this huge range, if your objective is to record seasonal variations in groundwater levelswater level, then a potential uncertainty of 15 cm is probably acceptable and, in terms of simplifying monitoring procedures locally, may be a compromise well worth considering.

If however, you are carrying out pumping tests in water wells over shorter periods, the drawdown needs to be significantly greater than 10 cm to accommodate the potential semi-diurnal variation. In most cases where monitoring level change sin a pumping well, this order of uncertainty should be tolerable. It may not be acceptable if observation boreholes  are being monitored.

Concluding remarks

I should be clear that I am not advocating the routine use of water level loggers without carrying out barometric compensation. This is simply bad practice and goes against all my technical instincts! Don’t do this without evidence and justification. 

But, where climate allows and where skills and resources are in short supply and where you can justify the degree of uncertainty, then it may be scientifically defensible to simplify.

If you do choose this approach, then make sure you really do know the pressure variation locally and are not guessing. If there are no published data records, then use an inexpensive barometer and manually record pressure at 3 hourly intervals or less (to pick up semi-diurnal changes – or use a spare total pressure logger if you have one to hand).  You should also be recording water levels using a manual dip tape periodically – or at least at the beginning and end of a pumping test or monitoring cycle - and validating against the logger record (e.g. Figure 4).

And this is certainly a situation in which the measurement accuracy should be clearly stated on the data record and reports – see earlier blog (Groundwater Level Measurements - Being Professional).

Find out more about In-Situ Europe's water level loggers and water quality instruments today.

©Peter Dumble 2017



[i] 1 mbar is equivalent to a head of 1.02 cm of water at a density of 1 kg/m3 and at 4 deg C.

[ii] For data and reports see: http://www.salonewatersecurity.com/