2.1 Study area and data

The state of Jalisco is located in western Mexico (Fig. 2). Approximately 68% of the state’s surface area has warm sub-humid climate along the coast and in the central zone; 18% is temperate sub-humid in the higher parts of the sierras, and 14% is dry and semi-dry in the northern and northeastern parts of the state. The state’s average annual precipitation is around 850 mm; however, precipitation is more than 1000 mm in the coastal zones (^{INEGI, 2024}). The summer rain propagation from southern Mexico in May, with an important increase in June, is a major rainfall source maintained until September, when a marked reduction of the summer rainy season is observed (^{Douglas et al., 1993}; ^{Brito-Castillo et al., 2022}). On the other hand, winter rains are produced by extratropical storms that rarely reach the latitudes of the state of Jalisco (^{Jáuregui, 1995}; ^{Brito-Castillo et al., 2003}), which is the reason to focus on identifying the occurrence of extreme events during the summer season (July-September).

Fig. 2 Location of 28 weather stations in Jalisco, Mexico (dots). The solid line displays the state’s polygon, and the gray lines are the basin boundaries. COL: Colima, MICH: Michoacán, ZAC: Zacatecas, NAY: Nayarit, AGS: Aguascalientes, GTO: Guanajuato. 

Daily observations of the database by the National Meteorological Service of Mexico (SMN; https://smn.conagua.gob.mx/es/climatologia/informacion-climatologica/informacion-estadistica-climatologica) are compiled from 28 weather stations for the base period of 1980-2019 (Fig. 2). A quality analysis was performed by selecting series with more than 40 years of records and more than 90% of daily observations. For this research, one representative station was selected for each basin, according to the station with the most daily data (Table I). Total rainfall for the summer was calculated from daily records of the representative stations.

2.2 Climate indices and drought definition

Various indices monitor climate oscillations. In this study, we use the Oceanic Niño Index (ONI), which records the three-month mean SST anomalies in the tropical Pacific for the Niño 3.4 region (5º N to -5º S, 120º to 170º W). The present study defines ONI as ONI+ (ONI-) when its value exceeds 0.5 (is less than -0.5). Index values were obtained from the National Oceanic and Atmospheric Administration and Earth System/ Physical Sciences Laboratory (NOAA/PSL; https://psl. noaa.gov/data/climateindices/list/index.html).

The SPI (^{Mckee et al., 1993}), recommended by the World Meteorological Organization, was applied to representative series to characterized meteorological droughts (^{WMO, 2012}; ^{Carrão et al., 2014}; ^{Vega-Camarena et al., 2018}; ^{Nwayor and Robeson, 2023}). The SPI at monthly, semi-annual, and annual time scales involves adjusting the historical accumulated rainfall records in n months to the function of incomplete gamma probabilistic distributions (^{Guenang and Kamga, 2014}). The resulting values were transformed to standardized values (z variable values) of the normal standard distribution function (^{Steinemann, 2003}; ^{Wu et al., 2005}).

^{Edwards and McKee (1997)} stated that a gamma probability density function for a given empirical precipitation distribution from a region or an individual station is fitted by:

where α is a shape parameter (α > 0), β is a scale parameter (β > 0), and x is the amount of precipitation (x > 0); and

where Γ(α) is the gamma function.

Parameters α and β are estimated for each time scale of interest and each week or month of the year, depending on whether SPI is calculated weekly or monthly:

where

Here, n is the series length and x̅ is the mean precipitation during the time scale of interest. The cumulative probability of each observed precipitation event for the given time scale of the station of interest is then computed using the estimated shape and scale parameters. An equiprobability transformation is made from the cumulative probability to a standard normal random variable Z, with mean 0 and variance equal to 1, where the SPI takes the value of Z. Thus, positive values indicate high precipitation and negative values indicate low precipitation, with respect to the historical average. This procedure allows a determination of short-, medium-, and long-term frequencies, intensities, and drought duration (^{Guttman, 1999}; ^{Hayes et al., 1999}).

In the present research, SPI was calculated for each climatological station series for the period 1980-2019, in two time-scales, for three months (July-September), called SPI-3, and for 12 months (January-December), called SPI-12. SPI-3 is suitable for monitoring drought conditions in the surface layer over a short term, which is essential for agricultural management (^{Mishra and Singh, 2011}). SPI-12 is required for monitoring levels of groundwater and surface water supplies (^{Mishra and Singh, 2011}; ^{de la Barrera et al., 2020}). The onset of a drought event occurs when SPI takes a value ≤ -1 and lasts until the index becomes positive for at least three or 12 months, depending on the time scale (^{Mckee et al., 1993}).

2.3 Extreme daily precipitation events

The present study followed the methodology used by ^{Vega-Camarena et al. (2023)}. An EDP event was defined by initially considering a threshold of 30 mm of accumulated precipitation in 24 h for the summer series (July-September) in 1980-2019. Summer precipitation series (≥ 30 mm) were ordered chronologically from the lowest to the highest to determine the 99th percentile (P99) of each series. Local hydrological studies and civil protection reports indicate that rainfall below this value (30 mm) typically does not result in significant flooding impacts (^{Vega-Camarena et al., 2023).} While this threshold differs from global standards, it aligns with the localized conditions of the present study area, including soil saturation levels and urban drainage capacity (^{Prokešová et al., 2022}). Using the 99th percentile (P99) during the humid months (July-September) ensures the analysis focuses on the most impactful extreme precipitation events during the region’s wettest period, where flooding risks are heightened.

The P99 value for each series was taken as the threshold for determining an EDP event. Extreme events with values > P99 were identified from July 1, 1980, to September 30, 2019, for all selected climatological stations in Jalisco.

3.1 Drought events with large coverage

Drought events considered to have large area coverage were identified in at least half of the weather stations in Jalisco. Table II lists such droughts identified for SPI-3 and SPI-12 with six and four events, respectively. Note that for SPI-3, the drought that affected the state most occurred in 2005-2006, when more than 70% of the stations showed drought, including the coastal region (Fig. 3). This event was followed in importance by the 1989-1990 drought.

Fig. 3 Standardized Precipitation Index (SPI-3) series. Red (blue) dots are the groups of stations where drought events with large surface coverage were recorded (not recorded) for the periods: (a) 1982-1983, (b) 1989-1990, (c) 1994-1995, (d) 1997-1998, (e) 2005-2006, and (f) 2011-2012. Dashed lines are contours with the highest SPI-3 values at 0.2 intervals. 

For SPI-12, the drought event with the highest affected area occurred in 1989-1990, with more than 70% coverage, followed by the event in 2011-2012. Figure 3 shows the recorded drought events with large surface coverage on a seasonal scale for SPI-3. A total of six events were identified for the period 1980-2019 (Table II). The least affected region is located near the coast in the southwestern part of the state. The lowest percentage of stations recording droughts at the SPI-3 scale was registered in the periods 1994-1995 and 1997-1998 (Fig. 3c, d), while the central-northern region had a more uniform distribution of stations with recorded droughts (Fig. 3a, b, e, f). Four drought events were identified for SPI-12, corresponding to impacts on water reserve levels. The largest coverage corresponds to the periods 1989-1990 and 2011-2012, with stations distributed in the central-northern regions of the state (Fig. 4b, d). Stations on a more uniform surface recorded the 1982-1983 and 2005-2006 droughts (Fig. 4a, c). It is important to note that in Figures 3 and 4, the stations that did not maintain negative SPI values for the required period (at least three months for SPI-3 and 12 months for SPI-12) were those considered to have no record of drought.

Fig. 4 Standardized Precipitation Index (SPI-12) series; dots are the group of stations recorded (red) and not recorded (blue) drought events with large surface coverage for the periods: (a) 1982-1983, (b) 1989-1990, (c) 2005-2006, (d) 2011- 2012. Dashed lines are contours with the highest SPI-12 value at 0.2 intervals. 

3.2 Recent drought events

In addition to the drought events with large area coverage, an additional analysis was performed where droughts that have occurred in the last years (2015-2019) were identified, mainly at a seasonal scale (SPI-3). In total, nine stations were identified that recorded droughts in that period (Fig. 5), most of them in the central region of the state.

Fig. 5 Dots on the map are stations that recorded recent drought events. On the right side, the gray shaded area indicates the most recent drought period (2015-2019). 

3.3 Identification of extreme daily precipitation events

Figure 6 shows the ratio (x/y) between the number of extreme events identified (x) and P99 per stations (y) for the 1980-2019, July-September period in each weather station. The total number of EDP events was 57 for 28 summer series analyzed.

Fig. 6 Ratio (x/y) between the number of extreme events identified (x) over the P99 value (y [mm]) per station. Calculations were performed in July-September 1980 -2019. Terrain elevations are given by blue (400 m), green (1000 m), orange (1600 m), and purple (2200 m) lines. 

Stations near the coast (14036 and 14044) recorded the most frequent EDP events (five and seven, respectively), representing nearly 20% of the extreme events identified in the entire state. In contrast, the lowest number of extreme events (14023, 14053, and 14144) was recorded over the northern region, with only two events in the three stations (Fig. 6), representing 3.5% of the extreme events identified in the entire state.

3.4 ENSO-conditions

3.4.1 Drought events

A time correlation analysis between SPI and ONI shows that most stations are positively correlated, particularly for correlations with SPI-3 and lag 0 (Fig. 7), while SPI-12 showed negative correlations close to the coast and in the northwestern region. This relationship would indicate that El Niño conditions are subject to wetter conditions, while those of La Niña are subject to drier conditions. The largest correlation coefficient between ONI and SPI-3 is 0.36 and 0.37 for SPI-12, both for station 14144 located in the northern region; station 14044 exhibits the most negative correlation, which is greater in SPI-12. The statistical relationship between ONI and SPI showed a strong association in the northern and northeastern regions (inland) of the state, and lower near the coast.

Fig. 7 Correlation between the Oceanic Niño Index (ONI) and the Standardized Precipitation Index (SPI-3 and SPI-12) (lag 0). 

Table IV shows the correspondence of the drought events identified on a short time scale (SPI-3) with the different ENSO phases. Stations located in the coastal region (14036 and 14044) only present drought conditions when the ENSO phase is in transition from El Niño to La Niña and La Niña to El Niño. When La Niña conditions occur, the most affected region is the northern part of the state (14023, 14053, and 14144). Finally, for El Niño conditions, the most affected by droughts in the state is the central region in the Chapala Lake basin (14040 and 14156).

Table IV Drought events in the state of Jalisco, Mexico, with large coverage identified in the Standardized Precipitation Index (SPI-3) in the state of Jalisco and their relationship with the different El Niño Southern Oscillation (ENSO) phases. 

Drought	ENSO-conditions	Region
1982-1983	El Niño	Central
1989-1990	La Niña	North
1994-1995	El Niño to La Niña	Coastal
1997-1998	El Niño	Central
2005-2006	La Niña to El Niño	Coastal
2011-2012	La Niña	North

Figure 8 shows the time series of SPI-3 (Fig. 8a) and SPI-12 (Fig. 8b) of all stations compared to ONI. Periods with drought events with large coverage are characterized by ENSO cold conditions in short and long-term SPI. This relationship is not observed in 1982-1983 and 1997-1998, when 64.3 and 50% of the stations showed drought. However, in this case, a strong positive ENSO event occurred, preceded by a negative ONI, although without exceeding the threshold for a minimum of five consecutive overlapping seasons with negative conditions.

Fig. 8 Standardized precipitation index (SPI) and Oceanic Niño Index (ONI) series: (a) SPI 3 and (b) the SPI 12. The red line indicates the SPI mean across stations and shaded areas indicate ±2.0 times the standard deviation (σ). Percentages correspond to values indicated in Table III. 

3.4.2 Extreme precipitation events

Four years were identified as the times when the most frequent EDP events and different phases of ENSO occurred (Fig. 9). For example, in 1987, El Niño conditions were observed; in 1999, negative SST anomalies led to La Niña conditions; in 2010, a transition from El Niño to La Niña; finally, in 2013 neutral conditions were observed (Fig. 9). Of the 57 EDP events identified for the state of Jalisco, 43.9% (25 events) corresponded to the ENSO neutral phase (Fig. 9), followed by 17.5% (10) El Niño and 17.5% (10) La Niña conditions, 15.9% (9) under El Niño to La Niña transition conditions, and finally 5.3% (3) from La Niña to El Niño (Fig. 9).

Fig 9 Oceanic Niño Index (ONI) series anomalies (black lines). Shaded boxes indicate years in which the most frequent extreme daily precipitation events occurred. (+) El Niño conditions; (-) La Niña; (N) neutral, and (+)/(-) transition from El Niño to La Niña. 

4.1 Drought events

Long-term scale series, such as 12-month series (Fig. 8b), reflect the cumulative precipitation variability, resulting in smoother series. In contrast, short time scales, such as SPI-3 (Fig. 8a), are more sensitive to changes in total precipitation. This property explains why SPI-12 hides some features of a dry episode found when short time scales are analyzed. In this sense, temporal time scales (SPI-3) were identified over the coast; drought events with large coverage only occur under ENSO transition scenarios, as in 1994-1995 and 2005-2006 (Table IV). Some factors that may have contributed to the drought episodes in ENSO transition phases are reported in ^{Vega-Camarena et al. (2023)}, who found that during the transition years (for example 2006) an anticyclonic circulation center was located south of the Gulf of California, reducing water vapor mixing ratio at 200 mb over the coast of Nayarit and Jalisco. Low water vapor mixing ratio values were also observed at 500 mb. Water vapor reduction in the middle and upper levels implies that moisture is insufficient for precipitation generation, causing negative precipitation anomalies. The result is consistent with that identified in the present study, where greater coverage of droughts was identified in ENSO transition phases on the coast of Jalisco. The rest of the state was less affected by drought events (Table IV), largely attributed to TC passages that contribute significantly to seasonal rainfall accumulation in the region, as reported by ^{Farfán et al. (2013)}. Under La Niña conditions, large rainfall accumulations have been recorded after a TC passage (^{Morales et al., 2013}), which coincides with the results of the present study, since for the coastal region, droughts were identified under El Niño to La Niña and La Niña to El Niño transition conditions. This situation largely derived from different moisture sources, such as TCs that contribute large amounts of rainfall, especially in Sinaloa, Nayarit, and Jalisco (^{Farfán et al., 2013}; ^{Martínez-Sánchez and Cavazos, 2014}). TCs have the greatest influence on Mexican rainfall in September at their maximum northeasterly extent in the eastern Pacific. In addition, easterly waves crossing the Atlantic into the eastern Pacific are associated with strong vertical coupling due to adiabatic heating, influencing rainfall patterns over western Mexico (^{Giovannettone and Barros, 2008}; ^{Ladwig and Stensrud, 2009}).

Under La Niña conditions, drought events with large coverage have occurred mainly in the northern part of the state, such as the one that occurred in 2011-2012, classified as an exceptionally well-documented drought (^{Sánchez-Rodríguez and Cavazos, 2015}; ^{Vega-Camarena et al., 2018}; ^{Cabral-Alemán et al., 2022}). According to the results in the present study, approximately 60% of Jalisco was affected by this drought. The atmospheric flow is depicted in Figure 10, where it shows geopotential height for the 2011-2012 drought case and the mid-level climatological anticyclone (^{Castro et al., 2001}; ^{Turrent and Cavazos, 2012}), which is displaced southward at 200 mb in northwestern Jalisco. The position and strength of the mid-level anticyclone are related to wet and dry summers in northern regions (^{Castro et al., 2001}; ^{Turrent and Cavazos, 2012}). In the 2011-2012 drought, the mid-level anticyclone was stronger and stretched to the north (strong positive anomalies) and south, strengthening subsidence in the study region.

Fig. 10 Geopotential height anomalies (m) for 2011-2012 and wind fields for (a) 500 mb and (b) 200 mb. Black lines represent climatological values (1981-2022), and blue lines show values of the 2011-2012 case study. Regions with well-defined high pressure are marked with the letter H. 

According to ^{Alcalá et al. (2005)}, El Niño conditions tend to make the midsummer drought in Jalisco longer and more intense. This is consistent with the results of the present study, which found that during the midsummer periods, the central region of the state was the most affected. Likewise, between 2015 and 2019, drought events were concentrated in the central part of Jalisco, with only one station (14036) in the coastal region reporting such conditions.

Likewise, the Pacific coast of Mexico, with its latitudinal expanse, characterized by arid and semiarid ecosystems along a narrow corridor, receives moisture from the Pacific Ocean and the Gulf of California. However, much of the Atlantic moisture is blocked due to the Sierra Madre Occidental rain shadow effect, creating a gradient of dryland ecosystems, including Mediterranean scrubs, deserts, subtropical thorn scrubs, and tropical dry deciduous forests. These ecosystems are highly sensitive to water availability, with significant water-limited parts of the year. Positive rainfall anomalies, often driven by climatic oscillations, are essential for ecosystem renewal, allowing plant recruitment and the regeneration of biological communities. For instance, studies have highlighted the role of moisture pulses in enabling vegetation recovery and nutrient cycling, particularly in arid coastal regions (^{Caso et al., 2007}). Inland areas, however, rely more on consistent seasonal precipitation, vital for agricultural productivity and groundwater recharge.

The results in the present study demonstrate that the relationship between ENSO and drought conditions is more pronounced inland (central, northern, and northeastern regions), where ONI correlations with SPI-3 and SPI-12 are the strongest, suggesting that inland regions are more sensitive to ENSO-driven climatic variability with El Niño, generally associated with wetter conditions, and La Niña, with drier conditions. By identifying stronger ENSO correlations in inland regions, the present study provides new insights into how climatic oscillations influence drought impacts, offering a clearer understanding of regional vulnerabilities across Jalisco.

4.2 Extreme precipitation events

According to the results in the present study, EDP events occured under different ENSO conditions during the 1980-2019 period; however, more than 30% of the extreme events occurred in four different years (1987, 1999, 2010, and 2013), corresponding to different ENSO phases (Fig. 9). This result indicates that no particular ENSO phase favored the development of EDP events in Jalisco, a climatic pattern widely related to precipitation in Mexico (^{Magaña et al., 2003a}; ^{Pavia et al., 2006}; ^{Bravo et al., 2018}). Thus, our results are consistent with those reported by ^{León-Cruz et al. (2022)} for the state of Colima in the western region of Mexico, where EDP events do not show a clear pattern with ENSO.

The precipitation regime of Jalisco is largely affected by the influence of the trade winds and TCs (^{García-Oliva et al., 1991}). According to ^{Morales-Hernández et al. (2013)}, large rainfall accumulations have been recorded under La Niña conditions after the passage of TCs. An example was observed in 1998, when more than 200 mm of rainfall were recorded after the passage of tropical storm Javier along Jalisco’s coast. The results in the present study are consistent with those reported in the literature, since EDP events identified at different stations (14036, 14044, and 14118) on the same or nearby days were generated by TC passage. To illustrate the amount of precipitation produced by these systems, Figure 11 shows TC Greg in September 5-6, 1999 along with its track. Note that the strongest convection occurs in the coastal region, off the coast of Jalisco.

Fig. 11 (a) Track positions of tropical cyclone Greg from September 5 to 9, 1999. Numbers indicate day/UTC each six hours. The black dot indicates landfall position. (b) Time series of sustained wind speed (m s^-1) associated with the development of tropical storm Greg. (c) GOES-8 infrared image from 23:15 UTC on September 5, 1999. The vertical bar indicates the associated calibrated scale of cloud-top temperatures (ºC). 

In addition to tropical cyclones in Mexico, another source of well-organized conditions favorable for generating extreme precipitation events are MCSs. As an example, Figure 12 shows the MCS displayed for the event identified on July 23-24, 2010, over the states of Jalisco and Nayarit with associated extreme precipitation of more than 150 mm recorded at station 14044. For the period 1980-1987, ^{Howard and Maddox (1988)} examined the typical life cycle, track, and duration of MCSs, finding a distinct group of nocturnal systems that initiated over western Jalisco, Nayarit, and southern Sinaloa in July and August. These features are consistent with our findings, in which extreme precipitation events are related to MCSs in July.

Fig. 12 GOES-13 infrared image at 07:15 UTC (01:15 LT) on July 24, 2010. The blue cross represents the location of station 14044, as shown in Table III. The vertical bar indicates the associated calibrated scale of cloud-top temperatures (ºC). Cloud tops cooler than -60 ºC (above 14 km) are associated with a mesoscale convective system over western Jalisco, southern Nayarit, and a portion of the Pacific Ocean. 

^{Farfán et al. (2021)} examined weather events associated with intense MCSs that developed along coastal Colima, Jalisco, and Nayarit in July from 2009 through 2018, identifying the occurrence of deep inland convection and over the Gulf of California’s entrance when large convective values of available potential energy, precipitable water, and lifted indexes were available. Finally, ^{Vega-Camarena et al. (2023)} found that on the coast of Nayarit and Jalisco during the 2010 season, a higher frequency of cloud cover was observed compared to 2006, consistent with above-average precipitation anomalies. These results are consistent with those reported in the present study, in which 2010 (along with 1987, 1999, and 2013) experienced more than 30% of extreme precipitation events.