Isotopic Composition of Precipitation in a Southeastern Region of Brazil during the Action of the South Atlantic Convergence Zone

The use of stable isotopes of hydrogen and oxygen is a tool widely used to trace water paths along the hydrological cycle, providing support for understanding climatic conditions in different spatial scales. One of the main synoptic scale events acting in southeastern Brazil is the South Atlantic Convergence Zone (SACZ), which causes a large amount of precipitation from southern Amazonia to southeastern Brazil during the southern summer. In order to determine the isotopic composition of precipitation during the action of SACZ in São Francisco Xavier in southeastern Brazil, information from the Weather Forecasting and Climate Studies Center of the National Institute for Space Research (CPTEC) was used regarding SACZ performance days, the retrograde trajectories of the HYSPLIT model, and images from the GOES-16 satellite, in addition to the non-parametric statistical tests by Spearman and Kruskal–Wallis. A high frequency of air mass trajectories from the Amazon to southeastern Brazil was observed when the SACZ was operating. During the SACZ events, the average isotopic composition of precipitation was more depleted, with a δ18O of −9.9‰ (±2.1‰), a δ2H of −69.3‰ (±17.9‰), and d-excess of 10.1‰ (±4.0‰). When disregarding the SACZ performance, the annual isotopic composition can present an enrichment of 1.0‰ for δ18O and 8.8‰ for the δ2H. The long-term monitoring of trends in the isotopic composition of precipitation during the SACZ events can assist in indicating the evapotranspiration contribution of the Amazon rainforest to the water supply of southeastern Brazil.


Introduction
Stable isotopes of hydrogen and oxygen are one of the main ecohydrological techniques for tracking water during its cycle, acting as a kind of "water fingerprint" [1]. This tracking is only possible due to the fractionation process between light and heavy isotopes of hydrogen and oxygen [2]. Different associations between the stable isotopes of hydrogen and oxygen result in the most varied combinations of water molecules (isotopologues), enabling tracking through the atomic mass difference of molecules [3].
The growing number of scientific studies focused on the use of isotopes shows the importance in different contexts, such as in studies related to climate change, evapotranspiration, surface runoff, and others [4][5][6]. One of the applications of the use of isotopes is to determine their composition during precipitation, more specifically with the performance of synoptic scale events, enabling the tracking of moisture that contributes to their formation [7][8][9]. In addition, this application can serve as a useful tool for validating atmospheric models [10] and for a better understanding of isotopic values in the tropical region [11].

Study Region
This study was performed in the Ecohydrologic Experimental Station of São Francisco Xavier (EES-SFX) located in São Francisco Xavier (SFX), a district of the city of São José dos Campos (SP), approximately 1100 m above average sea level ( Figure 1). The EES-SFX comprises the experimental branch of the Laboratory of Isotopic Ecohydrology (LabEcoh) of the National Institute for Space Research (INPE). The region is characterized by the presence of remnants of the Atlantic forest [23] and is one of the slopes of the Paraíba do Sul River Basin.
According to the Köppen climate classification, the climate in the region is defined as Cwa, humid subtropical with dry winters and hot summers [24], with the hottest month in February (average 23.9 • C) and the coldest month in July (average 16.4 • C) [25]. The annual rainfall is~1885.0 mm , weather station code 02245050), when the rainy period occurs between October and March, presenting 78.3% (1475.9 mm) of precipitation during the year, and the dry period occurring between April and September, with 21.7% (409.1 mm) of annual precipitation [26].
his study was performed in the Ecohydrologic Experimental Station of São Francisco Xavier (EES-SFX) located n São Francisco Xavier (SFX), a district of the city of São José dos Campos (SP), approximately 1100 m above verage sea level (Figure 1). The EES-SFX comprises the experimental branch of the Laboratory of Isotopic cohydrology (LabEcoh) of the National Institute for Space Research (INPE). The region is characterized by the resence of remnants of the Atlantic forest [23] and is one of the slopes of the Paraíba do Sul River Basin. ccording to the Köppen climate classification, the climate in the region is defined as Cwa, humid subtropical wi ry winters and hot summers [24], with the hottest month in February (average 23.9 °C) and the coldest month in uly (average 16.4 °C) [25]. The annual rainfall is ~1885.0 mm (1950-2000, weather station code 02245050), wh he rainy period occurs between October and March, presenting 78.3% (1475.9 mm) of precipitation during the In the study region, there are several meteorological systems that promote precipitation events. For example, during the austral summer, mesoscale convective systems occur, as well as isolated storms, beyond the performance of the SACZ, capable of generating deep convection and large amount of precipitation [27]. On the other side, during the austral winter, the amount of precipitation is low due to the performance of the South Atlantic Subtropical High [28], a high pressure system that inhibits cloud formation. During this period, precipitation occurs mainly due to the passage of cold fronts that can overcome this block with approximately 3.3 events each month [29].

Samples Collection
Water samples from precipitation events (≥1 mm) that occurred in SFX were collected from February 2019 to February 2020, totaling 178 samples in 124 days with precipitation. The rain collector used was a "funnel ball" type [30] and followed the technical standards indicated by Global Network of Isotopes in Precipitation [31]. This collector had the funnel attached to a glass jar, properly covered with a metallic adhesive tape, and a highdensity polyethylene plastic bag attached to the bottom of the funnel to reduce any type of evaporation in addition to facilitating collection and storage. After the end of each precipitation event, the collected samples were immediately stored in a refrigerator at a temperature of~4 • C, subsequently transferred to 1.5 mL glass vials for isotopic analysis. The process of transferring the water to the flask occurred through a syringe with a needle, coupled to a hydrophilic PTFE filter, with pore diameter of 0.22 µm, able to eliminate possible impurities and microorganisms.

Isotopic Analysis
All precipitation samples were analyzed at the LabEcoh/INPE in São José dos Campos, Brazil. The measurements were performed on a Picarro Cavity Ring-Down Spectrometer, L2130-I Model. The standards used were those made available by the manufacturer, considering Zero (δ 18 O 0.3 ± 0.2‰; δ 2 H 1.8 ± 0.9‰), Mid (δ 18 O −20.6 ± 0.2‰; δ 2 H −159.0 ± 1.3‰) and Depl (δ 18 O −29.6 ± 0.2‰; δ 2 H −235.0 ± 1.8‰), organized for analysis every 10 samples, enabling the correction of any deviations [32]. Each sample, as well as the standards, was analyzed six times, but the first three analyzes were discarded to avoid the memory effect, and the last three samples were used to calculate the arithmetic mean, as a final result. This routine ensures greater precision to the analyzed data [33]. The analytical precision was ±0.08‰ for oxygen and ±0.9‰ for hydrogen.
Isotopic ratios were defined in terms of ‰ (per mile) using the notation δ, relative to Vienna Standard Mean Ocean Water (V-SMOW) (Equation (1)), where R sample and R standard are ratios between heavy and light isotopes ( 2 H/ 1 H or 18 O/ 16 O) in the collected water and in the standard, respectively [34]. Deuterium excess (d-excess) was calculated as shown in Equation (2).
To calculate isotopes monthly averages and also the SACZ events, the weighted average by the amount of precipitation was used, as shown in Equation (3): where δ w informs the weighted average value of the isotopic ratio over the amount of precipitation, P i represents the amount of precipitation in the event (mm), δ i is the isotopic ratio of an individual sample (‰), and n represents the number of collections.

SACZ Events Determination and Meteorological Data
To determine the performance and positioning of the SACZ during precipitation collections, images from Geostationary Operational Environmental Satellite (GOES- 16) in the infrared field [35], with a 10-min time interval, were used together with daily bulletins provided by the Weather Forecasting and Climate Studies Center of the National Institute for Space Research (CPTEC), with a 6-h time interval [36]. Surface meteorological data (precipitation uncertainty of ±4.0%, ±1 rainfall between 0.2 and 50.0 mm per hour and ±5.0%, ±1 rainfall between 50.0 and 100.0 mm per hour; temperature-uncertainty of ±0.21 • C; relative humidity-uncertainty of ±2.5%) were obtained through a weather station HOBO U30 with standard ONSET sensors, installed exactly next to the precipitation collection site, with a data logging frequency of every 5 min.

Air Mass Sources
To identify the retrograde trajectory of main air masses that contribute to precipitation events, the Hybrid Single-Particle Lagrangian Integrated Trajectory Model (HYS-PLIT) was used [37]. This model is widely used in studies involving precipitation stable isotopes [38][39][40]. For each time the precipitation started, retrograde trajectories of 10 days (240 h) were calculated, at an altitude of 2,000 m, using the meteorological data set from Global Data Assimilation System (GDAS), with a 1 • × 1 • (1 •~1 11 km) spatial resolution [41]. Due to the data resolution and uncertainties inherent to the model, the retrograde trajectory is just a general approximation of the air mass origin because interactions with local humidity can occur and cannot be considered by the trajectories [38].

Statistical Analysis
As the Shapiro-Wilk normality test [42] revealed the data were not extracted from a normal distribution (p-value < 0.001), nonparametric statistical tests were adopted. To correlate the meteorological variables with the isotopic signal of precipitation, the Spearman test was used [43]. The requirements and interpretation for the Spearman Test are given in the Table 1 [44]. The Kruskal-Wallis test was adopted to verify the existence of statistically significant differences between the isotopic values of δ 18 O, δ 2 H, and d-excess [9]. The significance level adopted for the tests was 0.05.

Meteorological and Isotopic Characterization
The temporal variability of meteorological data (precipitation, temperature, and relative humidity of the air) in SFX are shown in Figure 2a. During the study period (February 2019-February 2020), the total precipitate was 2373.8 mm, when the lower amount of precipitation occurred between the months of May and October (497.4 mm, 21% of the total), representing the least rainy period. On the other hand, the rest of the months showed a higher amount of precipitation (1876.4 mm, 79% of the total), representing the rainiest period. The average air temperature for the period was 18.5 • C, the coldest month occurred in July (average of 14.7 • C), and the hottest month occurred in January 2020 (average of 20.7 • C). Regarding the relative humidity of the air, June was considered the driest month (average of 76%), and February 2020 was the wettest month (average of 93%) (Figure 2a).
The Spearman correlation between meteorological variables and the isotopic signal of precipitation ranged from moderate, strong, and very strong ( Table 2). The precipitation presented a very strong negative correlation for δ 18 O, δ 2 H, and d-excess (more than −0.80; p-value ≤ 0.003). The relative humidity of the air also showed a very strong negative correlation for δ 18 O and δ 2 H (superior than −0.91; p-value < 0.001) and a strong correlation for d-excess (more than −0.78; p-value 0.004). The air temperature showed a moderate negative correlation but without statistical significance with the δ 18 O (more than −0.58; p-value 0.052) and showed a strong correlation with the δ 2 H (more than −0.61; p-value 0.040) and d-excess (more than −0.76; p-value 0.007).    Figure S1).

Isotopic Composition during SACZ Events
According to the synoptic analyzes generated by the CPTEC, nine SACZ events occurred in SFX during the period from February 2019 to February 2020. During the nine SACZ events, 29 rainwater samples were collected in 17 days of precipitation. The average isotopic composition of precipitation (precipitation-weighted values) during SACZ events (Table 3)   The averages per precipitation event (arithmetic means) reveal isotopic differences (δ 18 O, δ 2 H and d-excess) that occur during the SACZ performance and without its influence ( Figure 3). As previously reported, the isotopic value of precipitation was extremely depleted during the SACZ events. Applying the Kruskal-Wallis test, it was possible to observe significant differences between the mean values of the entire collection period (ALL) events and SACZ (δ 18 O, p-value < 0.001; δ 2 H, p-value < 0.001; d-excess, p-value < 0.001) and between SACZ and NO SACZ (δ 18 O, p-value < 0.001; δ 2 H, p-value < 0.001; d-excess, p-value < 0.001). In addition, there was a significant difference between the mean values of ALL and NO SACZ for δ 18 O (p-value 0.03) and δ 2 H (p-value 0.03). These results show that the average annual value of the isotopic composition is more enriched without the SACZ performance, surpassing the annual average by 1.0‰ for δ 18 O and 8.8‰ for δ 2 H. On the other hand, there was no significant difference for the values of d-excess (p-value 0.17).
(δ 18 O, p-value < 0.001; δ 2 H, p-value < 0.001; d-excess, p-value < 0.001). In addition, there was a significant difference between the mean values of ALL and NO SACZ for δ 18 O (p-value 0.03) and δ 2 H (p-value 0.03). These results show that the average annual value of the isotopic composition is more enriched without the SACZ performance, surpassing the annual average by 1.0‰ for δ 18 O and 8.8‰ for δ 2 H. On the other hand, there was no significant difference for the values of d-excess (p-value 0.17).  (Table  S2).

Retrograde Trajectory of The Air Parcel
Using the 178 retrograde trajectories generated by the HYSPLIT model, three air mass transport patterns have been identified in SFX (Figure 4). The first pattern of trajectories refers to the days with SACZ performance (29 trajectories), presenting the existence of a potential moisture transport from the Amazon Region to SFX due to the large percentage of trajectories from this region (Figure 4a). The second pattern of trajectories for all collections  (Table S2).

Retrograde Trajectory of the Air Parcel
Using the 178 retrograde trajectories generated by the HYSPLIT model, three air mass transport patterns have been identified in SFX ( Figure 4). The first pattern of trajectories refers to the days with SACZ performance (29 trajectories), presenting the existence of a potential moisture transport from the Amazon Region to SFX due to the large percentage of trajectories from this region (Figure 4a). The second pattern of trajectories for all collections (178 trajectories) (Figure 4b) shows that, in addition to the humidity from the Amazon, moisture shifts from the Atlantic Ocean to SFX also occur frequently. In the third pattern of trajectories, when SACZ events are excluded (149 trajectories) (Figure 4c), there is a reduction in the density of humidity trajectories from the Amazon region, compared to previous patterns. (178 trajectories) (Figure 4b) shows that, in addition to the humidity from the Amazon, moisture shifts from the Atlantic Ocean to SFX also occur frequently. In the third pattern of trajectories, when SACZ events are excluded (149 trajectories) (Figure 4c), there is a reduction in the density of humidity trajectories from the Amazon region, compared to previous patterns.

Isotopic Seasonal Variability
The seasonality of the hydrogen and oxygen isotopic signal (Figure 2b)

Isotopic Seasonal Variability
The seasonality of the hydrogen and oxygen isotopic signal (Figure 2b) was in agreement with the quantity effect reported by [1], when there is a negative correlation with the amount of precipitation. The d-excess values significantly higher than the global average (~10‰) [45], registered between the months of July and September, may be associated with lower relative humidity [46], characteristic of the period for the region [47]. It is also important to note that both the isotopic signal and the seasonality of found results in this study are in accordance with the results obtained by Gastmans et al. [21] and Santos et al. [22] for the Southeastern region of Brazil. The slope and intercept (Figure 2c) higher than the global average may reflect the high amount of precipitation during the year [48], characteristic of the study region, in addition to local evaporation being an important factor in increasing the intercept, in relation to the global average [49]. These values are typical for humid tropical and humid temperate regions [50].
The very strong negative correlation of precipitation with δ 18 O and δ 2 H (Table 2) indicates the quantity effect: the greater (less) the amount of precipitation, the more depleted (enriched) the isotopic signal will be [51]. The negative correlation of relative air humidity with δ 18 O and δ 2 H was also an expected effect, because the drier atmosphere favors evaporation, causing isotopic enrichment of precipitation [52]. In contrast, despite a moderate correlation, the effect of air temperature showed a different result to that described in the classic literature [1,38]. According to the literature [51], a positive correlation between isotopic enrichment and air temperature was expected. However, this effect was not observed, which may be associated with low thermal amplitude, characteristic of tropical regions, and with the amount effect. This negative correlation was also found by Santos et al. [22] in a study on southeastern Brazil and by Baker et al. [53] in a study on the southwestern Amazon and some regions of Asia [54,55]. This inverse relation suggests the temperature effect may not be a good indicator to control the isotopic signal of precipitation for the region. In addition, this effect may have been masked by another one (relative humidity, for example), or the source of moisture that precipitates on the spot may be a determining factor for its seasonal variability [56]. In relation to d-excess, its negative correlation with the relative humidity of the air and precipitation was already expected [51,57], although the last one may not cause significant variations in the d-excess in some particular cases [58]. The inverse correlation of d-excess with the air temperature was not expected, since the d-excess usually presents correlation directly proportional to temperature [47,59]. This adverse result suggests the air temperature is not the best predictor for d-excess, but relative humidity, as presented by Li et al. [60].

Air Parcel Trajectory and Isotopic Signal of Precipitation during SACZ
As shown in Figure 4, there are three air-mass transport patterns for SFX. The trajectories of oceanic origin are mainly caused by the frequent passage of cold fronts and by the action of South Atlantic Subtropical Anticyclone (SASA), that inject water vapor into the continent, mainly during austral summer [61][62][63][64]. Several studies corroborate with the plot's trajectories during the SACZ performance (Figure 4a), highlighting the expressive contribution of the Amazon region to provide moisture for SACZ formation [36,[65][66][67][68]. This moisture transport shows the importance of the Amazon in the provision of ecosystem services to different locations, similar to other rainforests, which are also of great importance in moisture transport for water maintenance in other regions [69,70]. This service strengthens the constant search for the maintenance and preservation of tropical forests [71].
With the aid of GOES-16 satellite images in the infrared channel, it is possible to observe the range of cloudiness associated with positioning during SACZ events, as well as its magnitude ( Figure 5). The SACZ positioning was mostly well defined, mainly referring to moisture transport from the Amazon region to southeastern Brazil, as well as identified with the retrograde trajectories of the HYSPLIT (Figure 4a). During the nine episodes that SACZ was active, individual isotopic values varied from a δ 18 O of −16.1‰ and a δ 2 H of −123.1‰ (during the fifth collection of the second event) to a δ 18 O of −8.1‰ and a δ 2 H of −52.1‰ (during the collection in the fourth event), as shown in Figure 5. In addition, d-excess varied individually from 1.7‰ (during the first collection in the second event) to 14.7‰ (during the second collection in the eighth event). Furthermore, in the first two SACZ events, the d-excess presented values below the average (Table 3). A joint analysis of the data suggests the d-excess increase may occur gradually, due to the total coupling of systems in the region for the SACZ formation [17]. For example, in the events that enabled the higher number of collections (events 5, 8, and 9), d-excess increased gradually (values shown individually in the Supplementary Material- Table S1). Atmosphere 2021, 12, x FOR PEER REVIEW 22 of 32 Figure 5. Infrared images from the GOES-16 satellite on the days when the SACZ was active in SFX (magenta star). Images selected on days and times when the precipitation collection presented more depleted isotopic signals for each SACZ event.
During the SACZ events, the occurrence of the most depleted isotopic signal may be associated with the continentality effect and latitude [72,73], interpreted as Rayleigh distillation [1,74]. Rain depletion during the SACZ event, compared to the average isotopic signals of the other precipitating events that act in the region, may occur due to the displacement of the large cloudiness complex, along with humidity from the Amazon to southeastern Brazil. Because this system extends over a distance for a distance greater than 2000 km, from an average latitude of −7 • S to −23 • S and can last ap-proximately 10 days [14,17].
As shown by Salati et al. [75], the displacement of moisture from the Atlantic Ocean to the Amazon basin would not be enough to cause an expressive isotopic gradient to its easternmost region, as Rayleigh distillation proposes. Thus, it was suggested that the moisture recycling process, produced by evapotranspiration from the Amazon basin, would be a major cause of the low isotopic gradient of precipitation (δ 18 O −0.75‰/1000 km) in relation to that expected by the continentality effect [75][76][77][78]. Njitchoua et al. [79] also found a low isotopic gradient in the Cameroon rainforest (δ 18 O −0.9‰/1000 km) in Central Africa, attributing this value to the moisture recycling driven by the transpiration of local forests having a more isotopically enriched water. Studies developed by Salati et al. [75] and Gat and Matsui [76] reveal that the isotopic composition of δ 18 O of precipitation in the Amazon basin was approximately −4.5‰ during the months that the SACZ is formed. The fact that the average δ 18 O of precipitation presents the value of −6.4‰ (±1.0‰) during the months of SACZ operation in SFX demonstrates the occurrence of an isotopic gradient between the Amazon and southeastern Brazil, of approximately −1‰/1000 km for δ 18 O, driven by the Rayleigh distillation. This gradient could possibly be higher if the data available by Salati et al. [75] and Gat and Matsui [76] occurred in the SACZ event scale, enabling a more refined comparison.
It is important to mention that in regions where the rain recycling process is reduced or absent, the Rayleigh distillation has been more clearly evidenced. Kern et al. [80] used data from an isotopic monitoring network in Austria, Slovenia, and Hungary and found an extremely high gradient of δ 18 O, of −2.4‰/100 km, during the boreal winter period, attributed to the reduction of evapotranspiration for the formation of rain. Rozanski et al. [81] found the annual mean of the isotopic gradient of δ 18 O, in the order of −2‰/1000 km for the European continent, a region where an expressive contribution of atmospheric vapor from densely forested areas is not expected. The occurrence of the Rayleigh distillation along the air mass trajectory from the Amazon to southeastern Brazil can be explained by the fact that along this path, there are regions with a high degree of deforestation, such as the arc of deforestation region [82] and areas of the Cerrado biome in in the Midwest of Brazil, with a predominance of pastures or monocultures [20] drastically reducing the moisture recycling process through evapotranspiration [83,84]. The joint analysis of the results, based on similar works developed in the Amazon and in highly anthropized regions of the European continent, support the hypothesis that the greatest deforestation in the Amazon may be an important factor in the greater depletion of the rainwater isotopic signal that occurs in SFX in the southeastern region of Brazil during events where the SACZ is active.
In a future scenario, with the evapotranspiration reduction in the Amazon basin [71,85] mainly due to the loosening of Brazilian public policies related to forest management and the reduction of deforestation [86,87], the isotopic composition of precipitation may present more depletion in the southeastern region of Brazil during SACZ events, and consequently in its annual average, because the recycling process is one of the main processes responsible for isotopic enrichment [78,88,89].

Conclusions
The results of this study revealed that during the periods in which the SACZ was active in the southeastern region of Brazil, particularly in São Francisco Xavier, the isotopic signal of precipitation was more depleted (δ 18 O of −9.9‰ (±2.1‰) and δ 2 H of −69.3‰ (±17.9‰)) compared to the average isotopic signal (−4.9‰ (±2.2‰) to δ 18 O and −24.4‰ (±21.3‰) to δ 2 H). By disregarding the isotopic signal of precipitation during the presence of the SACZ, the annual averages of δ 18 O and δ 2 H can increase by 1.0‰ and 8.8‰, respectively. The isotopic depletion of precipitation during SACZ events is explained by the Rayleigh distillation process that occurs during humidity transportation from Amazon to southeastern Brazil.
The results found in this study support other works that show the strong contribution of the Amazon to the water supply in southeastern Brazil. In addition, the results obtained also suggest that the degradation of the Amazon forest may be an impact factor for an isotopic signal of more depleted precipitation, observed during SACZ events. This depletion suggests less participation of evapotranspiration from forest areas during SACZ events. Thus, the proposed study suggests that long-term monitoring of the variability of the isotopic signal of precipitation during the SACZ, associated with the detection of the most depleted isotopic signal, can serve as an indicator of decreased of the recycling process, impacting humidity formation and contributing to SACZ occurrence.

Supplementary Materials:
The following are available online at https://www.mdpi.com/2073-44 33/12/4/418/s1, Table S1: Set of isotopic and meteorological data used in this study. The samples collected during the performance of SACZ are in red. Table S2: Detailed Boxplot values used in Figure 3. Figure S1: Graphical details of the Spearman test between the monthly meteorological variables and the isotopic signal, shown in Table 2.
Author Contributions: C.S. collected samples as part of his PhD, processed and analyzed the data, and prepared the first draft of this paper and the write-up of the manuscript. C.S., R.C., C.B., D.G. and L.B. analyzed the data, provided guidelines for the write-up of the manuscript, and contributed to its editing and finalization. All authors have read and agreed to the published version of the manuscript.