3.2. Local Meteoric Water Line and hydroclimatic mechanisms 
The resulting LMWLs (Fig. 7) follow the expected trends based on depletion or enrichment of isotopic composition of δ18O and δ2H from theoretical considerations such as i) the orographic distillation that generates depletion of δ18O and δ2H values due to fractionation generated in the advance of moisture flow into the continent  (Dansgaard, 1964; Rozanski et al., 1993; Clark & Fritz, 1999;  Mook, 2002; Aggarwal et al., 2005) ii) the proximity to atmospheric moisture sources (Guan, Zhang, Skrzypek, Sun, & Xu, 2013); iii) the thermodynamic conditions of sources; iv) the mixing ratio of sources (Rindsberger, Magaritz, Carmi, & Gilad, 1983; Rindsberger, Jaffe, Rahamim, & Gat, 1990), and v) the amount effect that generates depletion in hydrogen and oxygen isotopes with the increase in monthly and annual precipitation of different places and with the intensity of the storms (Dansgaard, 1964).
[Insert Figure 7]
As in our Flexpart experiment, isotopic signals from the LMWLs also highlight the seasonal behavior of moisture sources for the two regions. More specifically, from December through February (Fig. 7a, 7b, 7c), the Andean region exhibits a variation range of δ2H (δ18O) that oscillates between -20 and -60 ( -5 and -9) ‰ VSMOW, and the Caribbean also varies with the highest values of δ2H oscillating between 0 and -20 ‰ VSMOW, and for δ18O between 0 and -5 ‰ VSMOW. In the Andean region these values represent the depletion of the heavy isotopes from coastal regions towards the continental interior (orographic distillation), while in the Caribbean, more depleted values represent the first condensate from marine moisture, indicative of meteorological conditions (such as relative humidity (72%-79%) and sea surface temperature (25-28°С)) from warm sources (Yurtsevert,1981; Rozanski et al. 1993), indicating the prevalence of oceanic sources (from the Atlantic ocean as indicated in our modeling results in Fig. 5). More specifically, the values of Oxygen and Hydrogen isotopes for both regions are similar to those proposed by Bowen (2003) for the TNA zone, and the FLEXPART results indicate that TNA was the most active source in this month for the study area (>27%, Fig. 5).
For the MAM season (Fig. 7d, 7e, 7f) in the Andean region, March represents the highest enrichment for this season, and April- May also exhibit an oscillation range for δ2H between 0 and -120 ‰ VSMOW, and for δ18O between 0 and -20 ‰ VSMOW. In the Caribbean region, March exhibits similar variations to DJF, while April-May observations vary along the LMWL with oscillations for δ2H between 0 and -120 ‰ VSMOW, and for δ18O between 0 and -20 ‰ VSMOW. The main regional terrestrial contribution for MAM corresponds to ORIC and NOSA sources (Fig. 5). For both regions, the footprint of these terrestrial contributions can be seen in the local LMWL located above the GMWL (Interpretation of Fig. 7, following guidelines on Fig. 3). Moisture from ORIC is located in the lower part of the figure, indicating its warm origin, and moisture from NOSA along the LMWL indicating sources with different temperatures. The large variation for both regions in April-May (Fig. 7d, Fig. 7e) responds to the mixing ratio effect. This spread in the isotopic composition of rainwater is produced by the mixing of different air masses that precipitate over Colombia. During this season, the study area is characterized by a complex combination of terrestrial and oceanic sources of moisture that contribute in different relative amounts to regional precipitation, associated with the dynamics of the ITCZ (Tables 2 and 3).
[Insert Table 2]
[Insert Table 3]
In JJA (Fig. 7g, 7h, 7i) the range of variations for δ2H in both regions oscillates between -20 and -80 ‰ VSMOW, and δ18O between -3 and -12 ‰ VSMOW. Especially for the months of June-July (Fig. 7g, 7h), observations are located in the middle of the variation range, coinciding with a dry period, and the contributions are a combination of terrestrial and oceanic moisture sources. For the Caribbean region, although not predominant, Atlantic sources are important (Fig. 6b), as indicated by enrichment observations of the LMWL. Similarly, terrestrial sources make their contributions from ORIC and NOSA (Fig. 5f), explaining the combination of terrestrial and oceanic sources. This is similar for the Andean region, which receives more contributions from terrestrial sources, followed by contributions from the Pacific Ocean (Fig. 6a). Particularly, for the Andean region the major contributions of terrestrial sources from NAMZ occurred during June (>17%), and coincide with observations localized above of the GMWL and the lower part of LMWL due to the minor temperature of the NAMZ source.
In the SON season (Fig. 7j, 7k, 7l), the isotopic composition of precipitation shows more depleted values compared to the rest of the year, with a variation range for δ2H (δ18O) between 0 and -120 (-2‰ and -15) ‰ VSMOW. In particular, October and November exhibit more depleted values, coinciding with an increment of the contribution from the cold Pacific Ocean (Fig. 6a). Also, this season corresponds to one of the two rainy seasons of the year, generated by the seasonal migration of the ITCZ, producing an increase of monthly precipitation and intensity of the storms, causing depletion in hydrogen and oxygen (amount effect). Likewise, the Caribbean zone receives predominant moisture sources from the Pacific Ocean, intensifying these contributions in October – November due to the ITCZ staying in the northern hemisphere over the Atlantic and eastern Pacific.