Temperature and logger calibration
Why do we need to calibrate loggers?
SRTMN provides a reliable record of “true” river temperatures. Unfortunately, different makes and models of datalogger, and even different individual dataloggers of the same make and model, will record temperature with different systematic differences from the “true” temperature. These systematic differences between the recorded temperature and true temperature are known as “logger bias”. If they are not adequately addressed then logger biases can accentuate or reduce long-term trends in river temperature, or introduce spatial differences where they do not occur.
How does SRTMN correct for logger bias?
A 2-stage logger calibration procedure has been established under SRTMN to correct observed temperatures to “true” river temperatures. This procedure is shown as a schematic in Figure 1.
Schematic diagram showing the steps involved in the 2-stage logger calibration procedure established under SRTMN to correct observed temperatures to “true” river temperatures. All SRTMN are calibrated against an internal reference logger, which in turn is calibrated against external reference standards. The joint uncertainty is calculated and calibration coefficients , including coefficients that characterise the uncertainty are recorded on the Marine Scotland FLEObs database. Observed values are corrected on export from the database.
Step 1: Two ‘internal reference’ dataloggers are calibrated against an external UKAS accredited standard at 5° C intervals ranging between 0 - 30° C. This calibration is repeated annually.
Step 2: Field deployed dataloggers are calibrated against the ‘internal reference’ loggers in a recirculating water bath (Figure 2), over a temperature range of 0 - 30° C. The water bath is initially filled with crushed ice to reduce the temperature to 0°C, it is then allowed to slowly come up to room temperature over a period of several days. Next the water bath is heated to 30°C, before cooling slowly to room temperature. Measurements are obtained every minute.
Water bath containing different types of temperature logger being calibrated to ‘internal reference’ logger
Step 3: Joint bias and uncertainty between external / internal (step 1) and internal / individual logger (step 2) is calculated using an in-house R script and correction coefficients are obtained that relate the recorded temperature to the “true temperature”. These coefficients are then uploaded into the Freshwater Laboratory Environmental Observations (FLEObs) database.
Step 4: Calibrated loggers are then deployed in the field. These loggers are routinely downloaded, and raw data is uploaded into FLEObs. Finally, corrected data (adjusted using the appropriate calibration coefficients) is exported out from FLEObs (Figure 3).
Time series graph showing raw and corrected water temperatures where raw values show a consistent positive bias.
Step 5: Loggers are rotated on a multi-year programme. When loggers are returned to the lab, they are re-calibrated and new-coefficients uploaded to FLEObs. When data are exported from FLEObs, they are corrected using the calibration equation appropriate to the dates of data collection.