Evaluation of several calibration procedures for a portable soil moisture sensor

- General calibration equations result in higher root mean square errors and bias for some conditions.
- Calibrations based on soil texture or land cover reduce these errors and eliminate bias.
- Field specific calibrations are necessary for reduced root mean square errors and low bias.

SummaryThe calibration and validation of remotely sensed soil moisture products relies upon an accurate source of ground truth data. The primary method of providing this ground truth is to conduct intensive field campaigns with manual surface soil moisture sampling measurements, which utilize gravimetric sampling, soil moisture probes, or both, to estimate the volumetric soil water content. Soil moisture probes eliminate the need for labor-intensive gravimetric sampling. To ensure the accuracy of these probes, several studies have determined these probes need various degrees of localized calibration. This study examines six possible calibration techniques using data collected during a field campaign conducted in 2012, with soil moisture samples being collected over 55 fields in southern Manitoba, as part of the Soil Moisture Active Passive Validation Experiment 2012 (SMAPVEX12). The use of a general equation, applied to all collected data, resulted in the largest error regardless of whether a linear or third order polynomial relationship was established for the calibration of the soil moisture probes. Calibration equations based on soil texture or vegetation land cover reduced the error; however, the individual calibration equations established for each field in the study had the lowest error of all the calibration techniques. Although average bias was low for all of the calibration techniques, the use of the general equation to calibrate individual fields resulted in high biases for some fields.