Sensitivity analysis for two ground heat flux calculation approaches

The ground heat flux (soil heat flux at the surface) can be calculated from in situ soil measurements in several different ways; all these approaches produce results that are affected by errors in the measured input data set. In this study, we analyse the influence of measurement errors on the results of two methods to determine the ground heat flux: a combination of the heat flux plate approach and calorimetry and a combination of the gradient approach and calorimetry. For that purpose, a sensitivity analysis is performed on a soil data set from the LITFASS-2003 experiment: the ground heat flux is calculated from the measured input data set and recalculated after modifying it (10,000 repetitions). Subsequently, the original and the recalculated results are compared and assessed by computing a quality flag. From this analysis, we conclude that the reference depth (the splitting depth between plate and calorimetry or between gradient and calorimetry, respectively) is the most important variable influencing the quality of the results data set and that it should be as deep as possible. Furthermore, amongst the variables modified, temperature measurements have the greatest influence on the quality of the results data set. Finally, the combination of plate and calorimetry gave slightly better results than the combination of gradient and calorimetry; nevertheless, using the latter method is recommended because it is based on measurements that are more reliable and less destructive.