2.2. Regional moisture contributions from FLEXPART
We set the regional sources of atmospheric moisture (Fig. 2), based on Hoyos et al. (2018), as follows: Tropical North Pacific (TNP), Tropical South Pacific (TSP), Subtropical North Atlantic (STNA), Tropical North Atlantic (TNA), Tropical Atlantic (TA), Tropical South Atlantic (TSA), Caribbean Sea (CARS), Northern South America (NOSA), Orinoco Basin (ORIC), Northern Amazon Basin (NAMZ), and Southern Amazon Basin (SAMZ). In Hoyos et al. (2018), these zones were determined by the best agreement among the Dynamical Recycling Model (DRM), the Quasi-Isentropic Model (QIBT), and the Flexible Particle Dispersion Model (FLEXPART).
[Insert Figure 2]
The FLEXPART is a 3D Lagrangian dispersion model that accounts for the net loss (or gain) of specific humidity, q, along with a large number of backward/forward trajectories between source and target areas (Stohl & James, 2005, 2004; Gimeno et al., 2012; Sodeman et al., 2015). The moisture contribution from a source area is estimated by accounting for the vertically-integrated long-term balance of precipitation P and evaporation E along the trajectories for a large number of computational particles (2 million). Net E − P < 0 represents a loss of moisture (precipitation exceeds evaporation), and E − P > 0 represents a gain of moisture (evaporation exceeds precipitation) (Stohl & James, 2005, 2004). The diagnostic precipitation (P) corresponds to the climatological sum of |E - P < 0| over the entire period.
Here, we estimate the composition of moisture contributions using the experiment developed by Hoyos et al., (2018). However, we calculate the seasonal values considering the optimal transport day (that is, the day when the moisture transference is maximum) instead of the 10-day global mean lifetime of water vapor in the atmosphere (Numaguti, 1999; Gimeno, Nieto, Drumond, Castillo, & Trigo, 2013; Van der Ent & Tuinenburg, 2017; Gimeno et al., 2021). This experimental design accounts for the fact that the time scale at which the regional atmospheric moisture is exchanged depends on the dynamic relationship between source and target regions (including not only distance but also the intensity of the advective processes and mechanisms that cause precipitation). We expect that, as the integration time of the moisture trajectories increases, the contribution of each source region correspondingly increases until reaching the maximum contribution value, after which moisture contribution shows a decreasing or asymptotic behavior (Hoyos et al., 2017; Nieto & Gimeno, 2019). Setting integration times at the optimum allows us to avoid over (or under) estimations in the relative contributions of moisture sources. The period simulated 1980-2012, includes the availability period of rainwater isotope records.