2.1 Precipitation data

The precipitation data were obtained from the Servicio Meteorológico Nacional (SMN), an agency within Mexico’s National Water Commission (CONAGUA). The data consists of daily precipitation records for the period 1990-2020, recorded by several stations along the sub-basins (Angostura, Chicoasén, Malpaso, Peñitas) of the Grijalva system based on the basin delineation for the Hydrological Region 30. All sub-basins belonging to the upper and middle Grijalva hydrological subregions are considered, as well as the three sub-basins of the lower Grijalva contributing flow to the Peñitas dam (^{SEMARNAT, 2010}) (Fig. 1). These stations constitute the largest network of climatological stations along the country, with a total of 5400 units (not all stations cover the same time period), providing daily climatological data of maximum and minimum temperature, 24 h accumulated precipitation, and evaporation. Data from these stations are publicly available online on the CONAGUA website (^{CONAGUA, 2025}). For this study, 30 stations were used, some of which are located at the boundaries of the sub-basins. Using the stations of each sub-basin, the arithmetic mean of daily accumulated precipitation was calculated based solely on available daily measurements. All statistics and results obtained in this study were derived from this information. To verify the consistency of the data, the annual accumulated precipitation for each sub-basin was compared with a product from the SMN, yielding very similar results. The number of stations varies over the period of study, with more in Angostura and Chicoasén and fewer in Malpaso and Peñitas, although in the last 20 years, the number of stations increased (Fig. 2).

Fig. 1.Elevation map indicating the limits of each sub-basin (light green), the bodies of water and dams (dark blue), and the SMN climatological stations used for this study (with solid dots and letters for the dams): (a) Angostura (stations in red), (b) Chicoasén (stations in green), (c) Malpaso (stations in yellow), and (d) Peñitas (stations in cyan). Shared stations located at the boundaries of the sub-basins are represented by stars. 

Fig. 2 Yearly average number of stations for each sub-basin. 

2.2 ENSO classification and statistical differences

To assess the influence of the ENSO phenomenon on precipitation in the sub-basins of the Grijalva system during its neutral, El Niño, and La Niña phases, a monthly classification was conducted from 1990 to 2020 (Table I). The classification was based on the Oceanic Niño Index (ONI) of the Climate Prediction Center (^{NOAA, 2024}). To discern statistical differences between phase means, a set of non-parametric statistical tests was applied in all cases, including the Kruskal-Wallis test. To identify which group pairs differed significantly, a Dunn’s post hoc test was conducted. All tests were conducted with a two-sided 95% confidence interval, so a resulting p-value less than 0.05 indicated a significant difference between the means. Subsequently, a comparison was made with the results that would have been obtained using parametric tests, and no statistical differences were identified (data not shown).

2.3 Tropical cyclones

An analysis of variability in the trajectories of tropical cyclones (TCs) during ENSO phases in the Grijalva was conducted. For this purpose, a potential region of influence over the Grijalva sub-basins was established. A rectangle was delimited whose limits have an extension of 7.5º (~ 800 km) from the most remote location of each cardinal point of the Grijalva basin. This broader area allows for the inclusion of a greater number of TCs in the statistical analysis and is not intended to estimate the amount of precipitation associated with TCs in the study basin. The trajectories of hurricanes for the North Atlantic and Northeast Pacific oceans were obtained from the HURDAT2 database of the National Hurricane Center (^{Landsea and Franklin, 2013}).

As the trajectories were classified by month for each ENSO phase (Table I), it is possible to group and count the number of years associated with each phase, such that the total number of years for any given calendar month is always 31 (since the period 1990-2020 spans 31 years). For each of these groups, the number of TCs that occurred in each ocean was recorded. To compare TC occurrence as a function of ENSO phase, it is useful to normalize the count. Since the number of years under each ENSO phase is not the same, the total number of TCs in each group was divided by the number of years corresponding to each phase (i.e., it was considered a normalized value). When a TC occurred across two months, it was divided into fractions, assigning to each month the corresponding fraction of its trajectory, such that the sum of the fractions equaled 1.

3.1 Precipitation climatology

The precipitation climatology in the Grijalva sub-basins is characterized by a dry season from November to April and a rainy season from May to October. During the rainy season, a relative minimum known as mid-summer drought (MSD) occurs (^{Perdigón-Morales et al., 2017}; ^{Corrales-Suastegui et al., 2019}). Results show that long-term monthly means of precipitation for each sub-basin are consistent with previously reported precipitation climatologies for the Grijalva basin, with a mid-summer minimum in July and two relative maximums in June and September (Fig. 3, Table II), although those studies were not conducted at the sub-basin scale nor with monthly resolution (^{Arreguín-Cortés et al., 2014}; ^{Andrade-Velázquez and Medrano-Pérez, 2021}). Figure 4 shows the annual means and their trend, while Table II presents the average precipitation values for each month and for the entire year of the four sub-basins. In Angostura, Chicoasén, and Malpaso, the precipitation patterns and amounts are similar, with a dry season followed by a rainy season that includes a clearly defined MSD. These three sub-basins exhibit a first peak in June, a second one in September, and an MSD mainly in July, according to ^{Perdigón-Morales et al. (2017)} and ^{Straffon et al. (2020)}. The mean yearly precipitation is very similar between these sub-basins. The Peñitas sub-basin, as the other three, has peaks in June and September, but it receives considerably more precipitation from September to February compared to the other sub-basins. The location of the Peñitas sub-basin is a key factor in its intense precipitation regimen during autumn and winter, with precipitation caused by cold fronts that interact with other systems that transport moisture from the Gulf of Mexico, which, favored by topographic convection, develops precipitation (^{López-Méndez, 2009}; ^{Andrade-Velázquez and Medrano-Pérez, 2020}). Additionally, the Peñitas sub-basin is the only one that shows a statistically significant positive trend in precipitation during the study period, although the time series exhibits considerable interannual variability. In contrast, the trend for Chicoasén is negative. In the case of Angostura and Malpaso, although they exhibit positive and negative trends, respectively, these are not statistically significant when considering the uncertainty range (Fig. 4). Based on the previous results, it is not possible to conclude that there is a single precipitation trend within the Grijalva basin; therefore, further studies with a climate change perspective are needed to identify the implications for the amount of hydroelectric energy that may be generated in the future, as well as the considerations that dam operators should take into account for planning and management.

Fig. 3 Long-term monthly mean precipitation of the sub-basins in the Grijalva basin: (a) Angostura (red line), (b) Chicoasén (green line), (c) Malpaso (orange line), and (d) Peñitas (blue line). 

Fig. 4 Annual precipitation and trends for the period 1990-2020. 

In the Peñitas sub-basin, annual accumulated rainfall is at least twice as high as in the other sub-basins. However, the total volume in Peñitas is smaller due to its size (see Table II). To quantify the relative magnitude of the MSD for each sub-basin, the proportions between monthly means that have relative maxima or minima were calculated as shown in Table III. For Angostura and Malpaso, the June-July and August-July ratios are similar. For Peñitas, the September-July ratio is considerably higher, whereas for Chicoasén the average rainfall in September is only slightly greater than in July. For the September-June relationship, that is, between the two precipitation peaks, the ratio is close to 1, except for Peñitas, where it is considerably higher. These results show that the Peñitas sub-basin has a different climatology than the other in terms of the amount of precipitation and its distribution throughout the annual cycle.

The 25th, 50th, 75th, 95th, and 99th percentiles were computed and plotted, as well as the maximum and minimum average daily precipitation of each month of the Grijalva sub-basins (Fig. 5). The 25th, 50th and 75th percentiles show a similar behavior as the monthly mean, with peaks in June and September, and relative minimums between these peaks. This means that in at least 75% of the years, precipitation over the sub-basins exhibits a strong MSD, according to climatology. The behavior of the 95th, 99th percentiles, and the maximum, differs from the other statistics and varies for each sub-basin. Several patterns are observed. In Angostura, the 99th percentile and the maximum have a year-around peak in August, with most of the maximum precipitation occurring between July and October. Chicoasén shows a bimodal behavior, having peaks in June and August for the 95th, 99th percentiles, and the maximum, and a clear MSD in July. In Malpaso, the maximum and 99th percentile also peak in June and August, with highest values of the maximum precipitation occurring from June to October, and a clear midsummer minimum in July. The Peñitas sub-basin exhibits a completely different climatology compared to the other sub-basins. For the percentiles, absolute maximum, and minimum, the first peak occurs in June, and a second peak occurs in September. However, for the 95th, 99th percentiles, and the absolute maximum, the peaks occur in October-November. This suggests that under extreme conditions, such as the interaction of a cold front with a tropical system, extreme precipitation can occur, as it did in November of 2020 when Cold Front 11 interacted with Hurricane Eta (^{CONAGUA, 2020}), causing a severe flooding in Tabasco.

Fig. 5 25th, 50th, 75th, 95th, and 99th percentiles of the monthly mean of the daily precipitation (green lines), maximum (red line), and minimum (yellow-gold line) values for the period from 1990 to 2020 in the Grijalva sub-basins. (a) Angostura, (b) Chicoasén, (c) Malpaso, and (d) Peñitas. 

The importance of considering the interquartile range and percentiles, such as the 95th and 99th, lies in enabling dam system operators to respond appropriately to extreme events. For example, November in the Peñitas sub-basin exhibits the largest range of variability among all sub-basins (Fig. 5d). It has an average precipitation of just over 10 mm day^-1, which nearly doubles (triples) at the 95th (99th) percentile. This highlights the relevance of incorporating broader ranges of variability in the planification of dam operation.

3.2 ENSO influence

Precipitation over the sub-basins of the Grijalva behaves differently according to the ENSO phase. Figure 6 shows a comparison of the monthly mean for each ENSO phase and sub-basin. The neutral episodes follow the seasonal mean in both behavior and precipitation amount across all sub-basins. The patterns for El Niño and La Niña differ across sub-basins. El Niño’s mean in Angostura follows the main mean, but from June to September precipitation decreases to a level similar to the 25th percentile. The opposite occurs for La Niña’s mean, which has similar precipitation to that of the 75th percentile. In Chicoasén, El Niño’s mean follows the precipitation of the 25th percentile line from July to September, while La Niña’s mean exceeds the 75th percentile in August. For Malpaso, El Niño’s mean behavior is similar to that in Chicoasén sub-basin, but for La Niña’s mean, the precipitation exceeds the 75th percentile during several months. In Peñitas, the differences between phases are clearer because of the amount of precipitation in the sub-basin: El Niño’s mean precipitation is similar to that of the 25th percentile in April-May and July-October, while La Niña’s mean exceeds the 75th percentile in July, August, and October. It’s worth noting that in December, the mean neutral episodes for Malpaso and Peñitas are similar to or exceed the 75th percentile.

Fig. 6 Monthly mean daily precipitation for each phase: neutral (green line), El Niño (red line), and La Niña (blue line) for the period 1990-2020 in the Grijalva sub-basins. (a) Angostura, (b) Chicoasén, (c) Malpaso, and (d) Peñitas. Additionally, the mean that includes all phases (black dashed line) and the 25th and 75th percentiles (green dashed line) are shown. 

To assess significant differences between the phase means across months and sub-basins, nonparametric tests were performed. In Table IV, the p-values of the tests are shown. A statistical difference was considered only when the p-value was lower than 0.05. The results in Table IV show the same trend as in Fig. 4. For the rainy months July, August, and September, the tests show a difference between phases, particularly in August. In all sub-basins, the mean values for El Niño years are substantially lower than those for the other phases. In particular, during La Niña years, the tests also show a statistical difference across all sub-basins, with higher precipitation values than in El Niño years in August, and in some cases also in July and November (see right-hand columns of Table IV). Compared to years with a neutral ENSO phase, La Niña episodes show higher precipitation values-even if statistical significance is not achieved-in August across all sub-basins, in July in Angostura, Malpaso, and Peñitas, and in October in Peñitas. During the dry season, the mean values are very low due to the lack of precipitation, although the tests do not show significant differences across all sub-basins. For Angostura and Malpaso, a significant difference is observed in November between the El Niño and La Niña phases; for Peñitas, there is no difference apart from summer. These results provide clear evidence that during months with higher precipitation in the rainy season, precipitation is consistently lower in El Niño years and significantly higher in La Niña years. The dry season shows a more complex pattern across sub-basins, with relatively important higher values during extreme years in the Peñitas sub-basin.

3.3 Tropical cyclones

A count of the trajectories of TCs that occurred between 1990 and 2020 in a region with potential precipitation influence over the Grijalva sub-basins was performed. Table V shows the total and normalized number of TC by month, ENSO phase, and ocean. The comparisons between phases are clearer with the normalized count, because it divides the total number of TCs by the number of years that occurred in each phase, as described in section 2.3. The normalized count was graphed in Figure 7. In the Atlantic, during the summer and autumn of El Niño years, the number of TCs is very low compared to the other phases, while La Niña years have a higher number of TCs (Fig. 5a). During August, September, and October, La Niña years have at least 7.5, 3.6, and 2.5 more TCs than El Niño, respectively. For the Pacific, during May and August in El Niño years, there are considerably fewer TCs than in the other ENSO phases (Fig. 5b). In September and October of La Niña years, there are 2.8 and 2.1 fewer TCs than during El Niño years. Figure 8 displays the trajectory of TCs that occurred in the region over the period of interest, showing the spatial pattern of all the trajectories. For the Atlantic (Gulf of Mexico and Caribbean Sea), there is a clear pattern of very few TCs during El Niño years (Fig. 6b). The opposite occurs for the neutral and La Niña phases (Fig. 6a, c). In the Pacific, the number of trajectories looks similar for El Niño and La Niña years, which is explained by the fact that in some months each ENSO phase has more or fewer TC than the other. In neutral years, there is a clear abundance of TCs in the Pacific Ocean (Fig. 6a). These results are consistent with the differences shown in Table IV and agree with the reduction in precipitation during El Niño years, with a decrease in the number of TCs from the Atlantic. On the other hand, it is inferred that the lower number of TCs in the Pacific during La Niña years does not have a great impact on the precipitation over the Grijalva sub-basins. Table IV shows that in comparison to the other phases, La Niña’s mean precipitation is lower in November for all sub-basins. However, in Peñitas, the second month with maximum precipitation can occur until November (Fig. 5d). This is explained by the higher number of TCs in the Atlantic during November, which is three and two times higher than in the neutral and El Niño phases, respectively.

Fig. 7 Normalized number of tropical cyclones by month in (a) Atlantic and (b) Pacific, grouped by ENSO phase: neutral (green line), El Niño (red line), and La Niña (blue line), in the potential influence zone of the Grijalva sub-basins from 1990 to 2020. 

Fig. 8 Tropical cyclones from 1990 to 2020 from the Atlantic and Pacific within the potential influence zone (yellow-gold) over the Grijalva sub-basins (purple) by ENSO phase: (a) neutral (green), (b) El Niño (red), and (c) La Niña (blue).