Figure 2. Evaluation of the concentration effect for the (a) STRASSE, (b) PIRATA FR24, (c) RARA, (d) ACTIV, and (e) BERMUDA cruise.

Multiple evaluations have been carried out before, during, and after the cruises to validate the applied correction (see Table 6). The correction applied is shown by the black dotted regression curves (we chose 20 kppmv humidity level as a reference). For STRASSE, PIRATA and RARA, the tests have been done with a Picarro autosampler. The reproducibility of the method is 0.05‰ for δ¹⁸O and 0.40‰ for δD. For BERMUDA, each measurement corresponds to 4 min average at a given humidity. We estimate the precision for the 4 min averages to be better than 0.08 ‰ and 0.25 ‰ for δ¹⁸O and δD. For ACTIV, each measurement was carried out manually and corresponds to averaging over 10–15 min and the precision on the δ¹⁸O and δD measurements are about 0.04 ‰ and 0.20 ‰.