Effect of extreme El Niño events on the precipitations of Ecuador

. Extreme El Niño events stand out not only for their powerful impacts but also because they are significantly different 20 from other El Niños. In Ecuador, such events are accountable for impacting negatively the economy, infrastructure


Introduction
El Niño is the positive (warm) phase of ENSO (El Niño Southern Oscillation), characterized as a complex phenomenon of variable extent and intensity and contrasted impacts, from regional to global.It is originated by unusual warming of Sea Surface Temperature (SST) from the center of the equatorial Pacific Ocean to the coasts of Peru and Ecuador, bringing anomalously heavy rainfall to this region (Gelati et al., 2014;Zambrano-Mera et al., 2018), and are associated with substantial two geographical regions are separated by the Andean Cordillera which crosses the country from north to south and constitutes a substantial topographic barrier (Morán-Tejeda et al., 2016).Regarding the presence of the Andean mountain chain, this most certainly modifies the ENSO signal (Vuille et al., 2000;Morán-Tejeda et al., 2016;Tobar and Wyseure, 2017;Quishpe-Vásquez et al., 2019).However, there is no clear explanation of how far into the Andes the effects of the ENSO are perceived (Morán-Tejeda et al., 2016).Detailed basin-wide assessments of the influence of ENSO in the transition from the coastal plain toward the western Andean Cordillera are also very scarce (Pineda et al. 2013).Various studies have considered precipitation in Ecuador (Rossel et al., 1999;Bendix and Bendix, 2006;Buytaert et al., 2006), or have focused on precipitation in specific areas of the country (Mora and Willems, 2012;Thielen et al., 2015 and2016).In general, a strong connection between the ENSO and precipitation in Ecuador has been found, but none of the previous studies analyzed stations throughout the entire country (Morán-Tejeda et al., 2016).According to these authors, a study of trends and variability in precipitation, based on up-to-date data for the entire country, is still lacking.A situation that is especially true when considering the different effects on spatial and temporal precipitation dynamics resulting from the occurrence of various types of extreme El Niño events (Thielen et al., 2021a).
In Ecuador, historical extreme El Niño events are accountable for generating very important, direct or indirect, negative effects on the economy, infrastructure and the population (OPS-OMS, 1999).Most economic costs are related to losses of agricultural production and damages to infrastructure (US$ 640.6 million in 1982/83, and US$ 2,882 million in 1997/98).Around 60% of the population of Ecuador may have their living conditions altered, directly or indirectly, by the occurrence of an extreme El Niño event.They have been responsible for generating mayor migratory waves (CEPAL, 1998).For instance, over one million Ecuadorians fled the country after affectation on Ecuador´s economy due to El Niño in 72/73 and 82/83 (Bernabé et al., 2014), as well as in 97/98 (OPS, 2000).Extreme El Niño events in Ecuador have also been accountable for important epidemics of diverse vector-transmitted and infectious diseases such as cholera, leptospirosis, dengue, chikungunya, zika, malaria, etc. (Gabastou et al., 2002;The World Bank Group, 2011).Now, in response to greenhouse warming, extreme El Niño events are projected to double their occurrence, while a less pronounced increase is projected for moderate El Niño events (Gulizia and Pirotte, 2022;Cai et al., 2014).Likewise, climate model projections also indicate an increase in the frequency of extreme Coastal El Niño (Peng et al., 2019).Of course, these impacts on climate extremes as well as the associated socioeconomic impacts would also take place much more frequently too (Gulizia and Pirotte, 2022;López et al., 2022).
Taking into account the implications for the Ecuador of such a forecast, the main objective of the present study was to analyze and compare, based on up-to-date data for the entire insular and continental territory, the dynamics of precipitation anomalies resulting from the various types of extreme El Niño events, including the Coastal El Niño.The results were discussed regarding the spatial and temporal variability generated by the topographic gradients of the dorsal of the Andes at both, the Pacific slope and the Amazon slope, as well as in terms of their principal basins or hydrological systems.The results provide solid and opportune evidence that can be used at different decision-making levels for identifying, in the context of global climate change scenarios, the most appropriate practices for reducing vulnerability and risks from a potential increase in extreme El Niño frequency and intensity.

Study area
The study area was defined as the totality of the continental and insular (offshore) territory of Ecuador.As for the continental territory, this was first divided into two main and distinctive zones: The Pacific slope (116,592 km 2 ) and the Amazon slope (131,948 km 2 ).Delineation of these zones was defined by the dorsal of the Andes, a dominant orographic barrier determining if runoff from rainfall is to be drained to the Pacific shore or the Amazon basin.Following CNRH (2002) classification system, each of these two continental zones was further divided into hydrographic systems or basins regarding their climate and spatial homogeneity.Through GIS applications, data from HydroSHEDS (http://hydrosheds.cr.usgs.gov)was used to delineate each basin.This resulted in 23 hydrographic systems for the Pacific slope, and 7 for the Amazon slope (see Fig. 1).Finally, regarding the insular territory, a unique hydrographic system was established encompassing all offshore islands, specifically the Galapagos Islands (8,233 km 2 ).

Data
Precipitation data was obtained from the Climate Hazards Group Infrared Precipitation with Stations (CHIRPS V2.0, https://iridl.ldeo.columbia.edu/SOURCES/.UCSB/.CHIRPS/.v2p0/.monthly/.global/).CHIRPS V2.0 is a quasi-global gridded rainfall time series dataset, spanning 50°S-50°N, from 1981 to near-present, 0.05° resolution satellite imagery with in situ station data, with great applications in monitoring precipitation extremes (Funk et al., 2015).Precipitation layers derived from interpolations of data from climate gauge networks has proven to have some limitations (Deblauwe et al., 2016).CHIRPS provides reliable precipitation observations with high accuracy and is particularly suitable for areas with few rainfall gauges (Paredes-Trejo et al., 2016;Beck et al., 2017), especially over montane (López-Bermeo et al. 2022) or arid regions (Paredes-Trejo et al. 2017;Ramoni-Perazzi et al. 2021) where extreme events may be rather common.According to Beck et al. (2017), in a global-scale evaluation of 23 precipitation datasets, CHIRPS V2.0 tends to perform the best in hydrological modeling of tropical regions, specifically in Central and South America.As for Ecuador, Thielen et al. (2021a) successfully tested its applicability in the spatial/temporal analysis of hydroclimatological extreme events in one of the most important and extended basins of the Ecuadorian Pacific slope.For the present study, monthly data for the time series Jan-1981/Dec-2018 were obtained from 456 rasters.Monthly and annual mean, as well as some other basic precipitation parameters, were obtained through GIS applications.

Calculation of the Standardized Pluviometric Drought Index -SPDI
In this study, the precipitation spatial-temporal dynamics was analyzed by the Standardized Pluviometric Drought Index (SPDI) developed by Pita (2001).The SPDI is a monthly rainfall index that is based on the calculation of cumulative monthly rainfall anomalies, similar to the well-known Standardized Precipitation Index (SPI) of McKee et al. (1993), more specifically, the 12-month SPI.As in this index, values ranging from +1 to +1.5 and +1.5 to +2.0 are associated with moderately humid and very humid episodes, respectively, and values exceeding +2 are representative of extremely humid episodes.Moderately dry, very dry, and extremely dry spells are characterized by the same ranges with a negative sign (see Table 1).One of the main advantages of using the SPDI is that it reflects precisely the beginning and end of each extreme precipitation event, as well as continuous information about its duration and intensity (Sanchez-Toribio et al. 2010).This ability makes it particularly suitable for characterizing long-lasting extreme events such as ENSO.Differently than the SPI, the SPDI does not require its application at multiple time scales to reflect the different durations of extreme events (Peña-Gallardo, 2016).The SPDI curves sometimes explain wet and dry periods not indicated by the SPI curves (Mega and Medjerab, 2021).The SPDI is calculated as follows: First stage, Eq. ( 1): Where  is the monthly precipitation anomaly,  is the monthly precipitation, and  is the median precipitation of the month for the series.As for this study: 1981-2010.
Second stage, Eq. ( 2): Where  is the accumulated precipitation anomaly of the month.
Third Stage, Eq. ( 3): Where  is the average value of accumulated precipitation anomalies of all the months of the series, and  is the standard deviation of accumulated precipitation anomalies of all the months of the series.
GIS applications allowed us to implement these equations to the 456 aforementioned CHIRPS V2.0 rasters and generate SPDI products such as images of 0.05° resolution or monthly zonal values, and at different space and/or time criteria.In the present study, monthly values of SPDI were estimated, based on the 1981-2010 climatology, for the three main zones: Pacific slope, Galapagos, and Amazon slope; as well as for each of the 30 continental hydrological systems (see Fig. 1).Analysis of monthly SPDI dynamics was performed in the two-year series comprising each extreme El Niño event.The significance of the statistical difference between monthly precipitation and/or SPDI values for any pair of extreme events, in each of the three zones, was identified by two-tailed Paired t-Tests and an α = 0.05.While similarities in spatial-temporal dynamics of SPDI between the Richman (1995) noted that nonhierarchical methods, such as the K-means algorithm, outperformed hierarchical methods (Ward's and the average linkage methods) when tested with precipitation data, as well as for SPI series (Santos et al., 2010).

Altitudinal dynamics of SPDI
As for continental Ecuador and through geoprocessing tools available from GIS software, results from SPDI estimations for each extreme El Niño event were combined with rasterized altitude data obtained from SRTM 1 Arc-Second Global (approx. 30m resolution, and freely available at https://earthexplorer.usgs.gov/),and then resampled at CHIRPS resolution (ie.0.05°).
The frequency of pixels with SPDI ≥2.0 was determined along the entire altitudinal gradient, for both the Pacific and the Amazon slope, and for each extreme El Niño event.The significance of the statistical differences between any pair of such SPDI spatial dynamics was identified by two-tailed Paired t-Tests and an α = 0.05.

Seasonality Index (𝑺𝑰)
The seasonality of precipitation in continental Ecuador was estimated by the Seasonality Index () (Walsh and Lawler, 1981) which quantifies the variability in monthly precipitation throughout the year.It is estimated by the sum of the absolute deviations of mean monthly precipitations from the overall monthly mean, divided by the mean annual precipitation, Eq. ( 4): Where ̅ is the mean precipitation of month n and  is the mean annual rainfall.The  can vary from zero (if all months have equal precipitation amounts) to 1.83 (if all the rainfall occurs in a single month).Thus, the higher the , the more seasonal or concentrated the precipitations are.The relationship between resulting seasonality index and the anomalies of precipitation (as SPDI) was evaluated at basin level.
2.6 Definitions of extreme El Niño events: The mega-El Niño and the Coastal El Niño The Oceanic Niño Index (ONI) is NOAA's primary indicator for monitoring ENSO.It is based on the monitoring of sea surface temperatures (SSTs) in the central Pacific Ocean and is used to identify the onset of an above-average SST threshold that persists for several months, encompassing both the beginning and end of an El Niño episode (Glantz and Ramirez, 2020).The ONI tracks the running 3-month average sea surface temperatures (SST) in the east-central tropical Pacific (120°-170°W, 5°S-5°N), specifically the NIÑO 3.4 region.El Niño occurs when the anomalies exceed +0.5°C for at least five consecutive months.
The threshold is further broken down into Weak (with a +0.5 to +0.9 SST anomaly), Moderate (+1.0 to +1.4), Strong (+1.5 to +1.9), and Very Strong (≥ 2.0) events.As for Very Strong or mega-El Niño events, the SST anomalies may be +2.0°C for several months (Chen et al., 2017).According to these parameters, three mega-El Niño events were identified since 1951:  NIÑO 3.4 (Larkin and Harrison, 2005;Ashok et al., 2007;Kug et al., 2009).Although all aforementioned mega-El Niño events have been EP El Niño events, the 2015/16 event should be considered as a mixed regime of both EP and central Pacific (CP) El Niño (Santoso et al., 2017;L'heureux et al., 2017;Vicente-Serrano et al., 2017;Wang et al., 2020), mainly due to an erratic response of SST anomalies peaking of SST (Xie and Fang, 2019).As for the present study, the three mega-El 3 Results

Precipitation
Historical mean annual precipitation for the Pacific slope resulted in 1348mm (series 1981-2010).About 70% of annual precipitations occurs during the first four months of the year (Jan-Apr).According to Walsh and Lawler (1981), precipitations here are markedly seasonal with a long drier season ( = 0.96).As appreciated in Fig. 2 altitudes rather high (2500-4000masl), while for COA-EN 17 such extreme events encompassed a much wider altitudinal gradient (1000-4000masl) (Fig. 4).Extreme precipitation anomalies were most spatially restricted during EP-EN 82/83 for the Amazon slope: 64% comprised altitudes between 300 and 400masl and the remaining 39% from elevations >2500m.In the lowlands of the Amazon slope, the presence of precipitation anomalies observed in Figure 3 (positive as in EP-EN 82/83, or negative as in EP-EN 97/98 and COA-EN 17) was preexistent long before the beginning of any of these extreme El Niño events, thus unrelated to their dynamics (CAF, 2000).

Discussion
In this study, the application of the SPDI was most appropriate when analyzing and comparing temporal and spatial dynamics of precipitation extremes among different extreme El Niño events.Likewise, CHIRPS V2.0 was confirmed to be a valuable source of monthly precipitation data for monitoring extreme events and contributed to the understanding of the spatial and temporal variability of monthly rainfall in Ecuador, as demonstrated for other extended South American regions (Paredes-Trejo et al., 2016;Baez-Villanueva et al., 2018;Rivera et al., 2019;Thielen et al., 2020 and2021b).
On the other hand, for both, Pacific slope and Galapagos, it is during the second year of EP-EN 82/83, EP-EN 97/98 and COA-EN 17) (Year 2: 1983(Year 2: , 1998(Year 2: and 2017, correspondingly), correspondingly), and more specifically during the first half of these years, coincidentally encompassing the rainy season, when most precipitation extremes occur.According to Cai et al. (2020), the rainfall impacts on the coast of Ecuador and Peru occur mainly in the rainy months of February, March, and April, when regional SSTs are seasonally at their highest, and the threshold for deep convection is more likely to be reached.As for Year 2 of MIX-EN 15/16 (2016), there was no evidence of any significant precipitation anomaly generated on Ecuadorian territory by the occurrences of the mixed (EP-CP) type of extreme El Niño.
Regarding overall SPDI dynamics (series 1981-2020) at both, the Pacific slope and the Galapagos, 85.7% of the months showed any degree of positive precipitation anomaly (SPDI >1.0), and 100% of the months showed an extremely wet condition (SPDI ≥2.0), were associated to Year 2 of an extreme El Niño event (Fig. 2).Such extreme rainfall conditions were concomitant, with a lag of zero months, to the presence of warm SST temperatures in the easterly most Niño region (ie.Niño 1+2), and for both types, the EP El Niños (Thielen et al., 2015;Bravo de Guenni et al., 2016;Morán-Tejeda et al., 2016;Quishpe-Vásquez et al., 2019), and the coastal El Niño (Thielen et al., 2021a;Rollenbeck et al., 2022).Through transferring heat from the ocean to the atmosphere, this anomalous warming elevates air temperatures in the coastal region, triggering localized atmospheric convection and heavy rainfall (Cai et al., 2020).Based on SST at Niño 1+2 dynamics, Thielen et al.
(2016) predicted that precipitation anomalies in the Ecuadorian coast generated by the mixed type El Niño 2015/16 would not be as significant as those from the El Niño 82/83 and 97/98, forecast that was fully corroborated in the present research.This lack of response in coastal and insular precipitation is most certainly true for CP El Niño events, as well as the mixed (EP-CP) El Niño type, such as MIX-EN 15/16.As for the Amazon slope, even though the number of months showing any degree of positive precipitation anomaly (SPDI >1.0) doubled that of the coastal and insular zones, less than 4% of the months occurred during an extreme El Niño event, from which none reached the extremely wet condition (SPDI ≥2.0).According to Kiefer and Karamperidou (2019), during EP and COA warm events, the coastal region is prone to extreme precipitation associated with convective bursts originating from the Pacific, while during a warm CP El Niño, as well as during a cold La Niña, moisture originates from the Atlantic and may reach the area as broader-scale less-intense precipitation.
At the Pacific slope, there are no significant differences (P>0.05) on SPDI values resulting from EP-EN 82/83 and EP-EN 97/98, when considered on annual bases.For both of these Eastern Pacific mega-El Niño events, precipitation anomalies lasted 10 months, reaching mean SPDI values of 2.09 for 1983, and 2.39 for 1998 (Tables 2 and 3).Differences between these two events become extremely significant when comparing the first 6 months -Precipitation anomalies during EP-EN 97/98 occurred sooner and reached faster maximum SPDI values during the first 6 months of 1998 than during 1983, or any other extreme El Niño event.SPDI dynamics for the first six months of Year 2 between EP-EN 82/83 and COA-EN 17 tended to be similar (P>0.01)(Tables 2 and 3).
Regarding how far the extreme El Niño events influence extends in the Pacific slope, the present study identified three most relevant facts: 1.For any extreme El Niño event, over 50% of all extreme anomalies occurred at elevations under 150m (Fig. 4).This represents over 40% of Ecuadorian coastal surface, and involves a most strategic zone for Ecuador since it comprises almost all lowland agriculture and aquaculture, which after petroleum oil production, represent the main activities generating export products.This zone also holds, besides a high density of rural population as well as numerous small to medium size towns, the largest city in Ecuador: Guayaquil, with a little more than 2.5 million inhabitants.
2. The difference between extreme El Niño events was more significant (P<0.05) when considering how far into the Andes the precipitation anomalies are perceived.For instance, during the long-lasting EP-EN 97/98, 80% of all extreme anomalies (SPDI ≥2.0) occurred at elevations up to 500m, while for the relatively less lasting extreme events, such as COA-EN 17 and EP-EN 82/83, this value was reached at altitudes much higher: at 800m and 1000m, respectively (Fig. 4).
3. The difference between the three extreme El Niño events disappears at around 3000m asl when reaching the accumulation of 97% of all extreme anomalies (SPDI ≥2.0).At an altitude of 4000m, all extreme El Niño events reach the mean (series 1981-2010), which is a little before all reach 100% at the maximum height of 4300masl (Fig. 4).
Several authors have also investigated the ENSO influence extends inside continental Ecuador.Bendix and Bendix (2006) and Kiefer and Karamperidou (2019), for instance, showed that positive rainfall anomalies during ENSO mainly affect the coastal plain of Ecuador to the western slope of the Andes at altitudes <1800m; while Pineda et al. (2013), observed ENSO signals at locations as high as 2700m.Regarding the presence of ENSO signal at high altitudes in the Pacific slope, relief plays a twofold role in the control of ocean-atmospheric forcing: It can modulate the atmospheric circulation, leading to a dissipation of the signal, or can favor meteorological processes, leading to enhancement of orographic precipitation (Pineda et al., 2013).Now, there is no easy answer about the difference between the results from such studies regarding how far the extreme El Niño events influence extends on the Pacific slope.
Preexisting studies limit their analysis to specific areas of Ecuador or may confront severe data limitations due to discontinuities in space and in time.Such limitations were overcome in the present study.the Andes, which correspond to the highest sections of the southernmost hydrographic system, 22-CHIRA, any ENSO signal disappeared.At this point, there is no clear pattern regarding extreme El Niño events and the generation precipitation anomalies in the highest sections of the Andean Cordillera.In any case, the ENSO signal was observed at mean altitudes ranging from 3200 to 3900m.Other physical determinants such as distance to coastline and steepness of the Cordillera may play an important role in determining the degree of ENSO signal.Coincidentally, the aforementioned 320 km transect that did reach precipitation anomalies was located at a distance from the seashore of 120 km or less, also showing a dominant steep relief (Figs. 1 and 5).Now, from the results of the spatial-temporal analysis of precipitation dynamics, it is evident that the degree of seasonality also conditions the magnitude of the ENSO signal in entire continental Ecuador. Figure 6, for instance, shows that it is in the most extremely seasonal regions ( 1.0 -1.2) where precipitation anomalies are the strongest, while regions with low or no seasonality ( 0.0 -0.6) show no precipitation anomaly during the event of an extreme El Niño.According to Carréric et al., (2019), the strong EP El Niño events peak in boreal winter is extended by two months, which results in significantly more events peaking in February-March-April, the season when the climatological Inter-Tropical Convergence Zone is at its southernmost location.The Pacific slope shows strong seasonality, while the Amazon slope exhibits mild to no seasonality and the Sierra with a moderate seasonality (Tobar and Wyseure, 2017).
From Figures 5 and 6, a list of the most vulnerable hydrographic systems regarding affectation in the event of an extreme El (1.87 and 1.15).

Conclusion
The present study generates reliable information about the most relevant aspects of spatial-temporal extreme precipitation dynamics resulting from the various types of extreme El Niño events, including the Coastal El Niño.Information that becomes most valuable and highly strategic considering that these extreme climatic events are expected to double their occurrence in the foreseeable future.
 For both, the Pacific slope and Galapagos, it is during the first half of the second year of an extreme El Niño event, coincidentally encompassing the rainy season, when most precipitation extremes occur, and it is during this time when any difference between extreme El Niño events become more evident.
 There was no evidence of any significant precipitation anomaly generated on Ecuadorian territory by the occurrences of the mixed (EP-CP) type of extreme El Niño.Likewise, there was no evidence of any significant precipitation anomaly generated on the Amazon slope by the occurrences of any type of extreme El Niño: eastern Pacific, Central Pacific, mixed or Coastal.
 For any extreme El Niño event, over 50% of all extreme anomalies (SPDI ≥2.0) occurred at elevations under 150m.
But, differences between events become significant when considering how far into the Andes the precipitation anomalies are perceived.For instance, during EP-EN 97/98, 80% of all extreme anomalies occurred at elevations up to 500m, while for COA-EN 17 and EN 82/83, this was 800m and 1000m, respectively.Any difference between extreme El Niño events disappears again around 3000 m asl, when accumulative extreme anomalies reach 97%.
Finally, at an altitude of 4000m, all extreme El Niño events reach the historical mean (series 1981-2010).
 Nevertheless, the ENSO signal is variable, not only along the lowlands of the Pacific slope but also in the highlands and along the dorsal of the Andes.Here, the ENSO signal can be observed, in continuous sections of several hundred kilometers, and at mean altitudes ranging from 3200 to 3900m.Other physical determinants such as distance to coastline and steepness of the Cordillera may play an important role in determining the degree of ENSO signal on the Andean Cordillera.
 Finally, the degree of seasonality also conditions the magnitude of the ENSO signal in entire continental Ecuador: It is in the regions showing the highest seasonality index where the most severe precipitation anomalies from extreme El Niño events occur.In these terms, 13 hydrographic systems from the Pacific slope showing strong seasonality resulted to be the most vulnerable to extreme precipitations generated by extreme El Niño events.Both the north of Ecuador and the Amazon slope exhibits mild to no seasonality.Concomitantly, hydrographic systems from these regions show no significant precipitation anomalies regardless of the type or strength of El Niño event.
Results from present research allowed us to generate most valuable information regarding similarities and differences between the effects on precipitation from types of extreme El Niño events, as well as highlights spatially and quantitatively, those regions or hydrographic systems where most extreme precipitations anomalies are most likely to occur in the event of an extreme El Niño, either eastern Pacific or coastal El Niño.Access to information like this is most strategic when designing and incorporating disaster-risk analyses and policies (Ward et al., 2014).For instance, the Figure 5 shows where the most negative direct effects from such anomalies are expected, as well as where such extreme events may exert strong and widespread influences on both flood hazard and risk.Because extreme El Niño events have some predictive capacity, mainly the eastern Pacific type, these specific results represent a solid contribution toward developing a risk-predictive model with applications for improved disaster planning (Ward et al., 2014).The results also provide solid and opportune evidence for identifying, in the context of global climate change scenarios, an increase in the frequency and intensity of extreme climatic events, the most appropriate management practices aimed to achieve sustainability of ongoing anthropogenic activities in one of the most climatic vulnerable regions from the Pacific coast of South America, either by adapting or mitigating the direct effects such as flooding and mudslides, as well as by reducing the risk of indirect effects such as the case of the emergence of important infectious diseases in a region that, historically and linked to the occurrence of extreme climatic events, has shown to be most vulnerable to significant epidemics of cholera, leptospirosis, dengue, chikungunya, zika, malaria, etc. (OPS-OMS, 1999;The World Bank Group, 2011), and more recently to COVID-19, one of the worst pandemics known by humankind in recent history, about which there is currently no clue about what to expect and to control the spread of such disease in the event of an extreme El Niño event, in the Ecuador or elsewhere.
Table 4. Standardized Precipitation Drought Index (SPDI) temporal and spatial dynamics at the different hydrographic systems for year 2017 of Coastal El .Cluster analysis (K-means clustering using Euclidean distance) was performed on both Figure5is the spatial representation, for the entire territory of Ecuador, of the mean annual (Year 2) precipitation anomalies (as SPDI) generated by the most important extreme El Niño events since 1981, that is the mega-El Niños EP-EN 82/83 and EP-EN 97/98, and the Coastal El From this figure, it is evident that the ENSO signal is variable, not only along the lowlands of the Pacific slope but also along the highlands and the dorsal of the Andes.From north to south, the first half of the 1,030 km of the dominant orographic barrier of the dorsal of the Andes, does not show any effect or signal from ENSO -That is, no precipitation anomalies are generated by extreme El Niño at the highest sections of the hydrographic systems 02-MIRA, 06-ESMERALDAS and part of 13-GUAYAS.From this point on, and for 320 km along the dorsal of the Andes, the highest sections of systems 13-GUAYAS, 15-CANAR, 16-NARANJAL PAGUA, and 17-JUBONES, showed moderate to high precipitation anomalies (SPDI 1.0 -1.5) during an extreme El Niño event.But then again, in the last 165 km of the dorsal of

Figure 1 :
Figure 1: Map of study area, defined as the totality of the territory of Ecuador.The continental territory is divided by the dorsal of the Andes into two main and distinctive zones: The Pacific slope (116,592 km2) and the Amazon slope (131,948 km2).Following CNRH (2002) classification system, each of these two continental zones was further divided into 30 hydrographic systems: 23 for the Pacific slope, and seven for the Amazon slope.Regarding the insular territory, a unique hydrographic system was established 5

Figure 5 .
Figure 5. Potential affectation from precipitation anomalies generated by extreme El Niño events, as determined from mean annual 10

Figure 6 .
Figure 6.Relationship at basin level between the Seasonality Index (, Walsh and Lawler, 1981) and the mean annual (Year 2) SPDI resulting from the mega-El Niño events EP-EN 82/83 and EP-EN 97/98, and the Coastal El Niño COA-EN 17. Extremely humid Niño events are referred toTakahashi and Martínez, 2017;Ramírez and Briones, 2017;98, for El Niño 1997/98; and MIX-EN 15/16, for El   Niño 2015/16.Besides tropical Pacific El Niño events (EP or CP), the study area is also affected by a more local type of El Niño event: The Coastal El Niño, a very rare and unique event which develops differently from either CP or EP El Niño events.To distinguish the Coastal El Niño from the warm ENSO phase, ENFEN (2012) operationally defines the Coastal El Niño based on the seasonal NIÑO 1+2 SST anomaly: 3-month running-mean Niño-1+2 SST above 0.4°C for three or more consecutive months.A strong Coastal El Niño developed off the coast of Peru from January to April 2017(ENFEN, 2017; WMO, 2017a,b;Takahashi and Martínez, 2017;Ramírez and Briones, 2017; Garreaud, 2018) (hereafter COA-EN 17), and has been the strongest on record, and developed rather fast and unexpectedly from the warming of SST specific to far eastern tropical Pacific.
Walsh and Lawler (1981)oral dynamics between the different extreme El Niño events are given in Fig.2-Ib.As a result of precipitation dynamics during Year 1 (Fig.2-Ia), none of the mega-El Niño events and neither the Coastal El Niño showed SPDI values different than the "near normal" condition (-0.99 -0.99).On the other hand, the Pacific slope experienced "extremely humid" (SPDI ≥2.00) during Year 2 of EP-EN 82/83 and EP-EN 97/98, with mean SPDI values not significantly different: 2.02 and 2.19, respectively (P=0.431).As for EP-EN 82/83, precipitation anomalies started in March 1983 (SPDI =1.64) and lasted for ten months until December of that year (SPDI =1.12).For seven consecutive months (Apr-Oct), EP-EN 82/83 presented a sustained extremely humid condition (SPDI ≥2.00).In 1983, 54.9% of continental Ecuador was affected by this extreme precipitation anomaly, from which, 93.8% comprised the Pacific slope (Fig.3-I).In this easterly slope, about 90% of such extreme precipitations occurred at altitudes of 1900 m or less, and over half at less than 200 m asl (Fig.4).Historical annual mean precipitation for Galapagos resulted in 89mm (series 1981-2010), much drier than that of the Pacific slope (Fig.2-IIa).About 82% of annual precipitation occurs from February to April, reflecting an extreme seasonality ( = 1.28,Walsh and Lawler, 1981).As in the Pacific slope, precipitations in Galapagos during Year 1, for any of the extreme El Niño events considered, did not differ significantly (P>0.05) from that of the historical mean.On the other hand, during Year 2 of mega-El Niño events EP-EN 82/83 and EP-EN 97/98(1983 and 1998, respectively), rainfalls about tripled that of the monthly mean value for the 30-years series 1981-2010 (267 and 284mm, respectively), an increase that tends to be significantly higher than the historical mean (P≈0.06).Annual precipitations among mega-El Niño events EP-EN 82/83 and EP-EN 97/98, as for Year 1 and Year 2, did not differ significantly (P=0.1390 and 0.616, respectively).As for MIX-EN 15/16, this mega-El Niño event did not generate precipitations significantly different from those of the historical mean, neither in 2015 (96mm, P=0.260) nor in 2016 (74mm, P=0.641).Likewise, the Coastal El Niño of 2017 (COA-EN 17) did not generate precipitations any different from that of the historical mean (131mm, P=0.205) for the Galapagos Islands.On the other hand, in 1998, a very significant difference (P=0.003) in SPDI dynamics was observed between these two contrasting geographical regions.Such a significant difference was also observed in the resulting SPDI from Coastal El Niño 2017 between the Galapagos and the continental Ecuador.Historical annual mean precipitation (series 1981-2010) for the Amazon slope was 2824mm, more than double that of the Pacific slope (1348mm).The monthly precipitation from March to July is about 10%, that is 50% of annual total amount.As for the rest of the year, that is from August to February, precipitation discretely drops to around 7% per month (see Fig.2-IIIa).This results in a Seasonality Index of 0.26 which, according toWalsh and Lawler (1981), is referred to places where precipitation spread throughout the year, but with a definite wetter season.As for precipitation in Year 1, both EP-EN 97/98 and MIX-EN 15/16, showed values significantly drier (P<0.05)than the historical mean (2538 and 2269mm, respectively).While for Year 2, none of the extreme ElNiño events, showeda significate value different from the historical (P>0.05).MIX-EN 15/16 was significantly drier (2519mm, P<0.05) than that ofEP-EN 82/83, EP-EN 97/98, and COA-EN 17 (2978, 2801mm, and 3134mm, respectively).None of the three mega-El Niño events, nor the Coastal El Niño of 2017 generated SPDI values different than "near normal" (-0.99 -0.99, Fig.2-IIIb) in the Amazon slope.Mean SPDI values for Year 2 of EP-EN 97/98, MIX-EN 15/16, and COA-EN 17 were significantly drier (-0.64, -0.69, and -0.19, respectively; P<0.0001) than EP-EN 82/83 (0.89).No significant difference (P>0.05) was observed between SPDI values of Year 1 and Year 2 of either EP-EN 82/83 (0.92 and 0.89) or EP-.On the other hand, a very high significant difference was detected between SPDI values for the -Ia, mean monthly precipitation during the first year (Year 1) of any extreme El Niño event does not differ significantly from that of historical mean (P<0.05).It is during the second year (Year 2) of El Niño event, specifically the first half, when precipitations significantly differ from that of historical values.For example, annual precipitation for Year 2 of mega-El Niño eventsEP-EN 82/83 (ie.1983)andEP-EN    97/98 (ie.1998)resulted significantly higher (2483mm, P=0.003; and 2590mm, P=0.022; respectively) than the historical mean.Although precipitation values between mega-El Niño events EP-EN 82/83 and EP-EN 97/98 were not significantly different (P=0.194),values for Year 1 was significantly drier in EP-EN 82/83 than in EP-EN 97/98 (1234 vs. 1609mm, P=0.042).On the other hand, during the other mega-El Niño event, MIX-EN 15/16, neither Year 1 nor Year 2 presented annual precipitations significantly different than those of the historical mean (1368mm, P=0.868; and 1299mm, P=0.666; respectively).As for the Coastal El Niño, COA-EN 17, annual precipitation for the year 2017 tended to be significantly different from historical mean (2072mm, P=0.097).1998(Fig.2-Ib).It is also evident from this figure that extremely wet SPDI values were obtained for the Pacific slope earlier in EP-EN 97/98 than in EP-EN 82/83 mega-El Niño.About 75.6% of continental Ecuador was affected by precipitation anomalies of SPDI ≥2.00 during Year 2 of EP-EN 97/98.As in previous mega-Niño, the most comprised area was the Pacific slope (98.8% of total area, Fig.3-II).On this slope, during 1998, 90% of extreme precipitations occurred at altitudes of 1300m or lower, and about 50% of such precipitations, at coastal areas with elevations less than 150masl (Fig.4).Precipitation during mega-El Niño MIX-EN 15/16 resulted in SPDI dynamics for Year 2 (ie.2016) being extremely different (P<0.0001)fromYear 1 and Year 2 of both, EP-EN 82/83 and EP-EN 97/98.During this mixed mega-El Niño, the SPDI value for 2015 tended to be similar to that of 2016 (0.41 and 0.33, respectively; P=0.051), for a resulting "near normal" precipitation condition for the entire duration of this extreme El Niño event (Fig.2-Ib).During 2017, Coastal El Niño COA-EN 17 generated a precipitation anomaly lasting five months (Mar-Jul), with an SPDI maximum of 2.23 for about two months (Fig.2-Ib).SPDI dynamics during 2017 was significantly different from that of Year 2 of EP-EN 82/83 and EP-EN 97/98 (P<0.05),affectingonly5.7% of continental Ecuador with extreme precipitation anomalies of SPDI ≥2.00.Still, as in the two mega-El Niño EP events, over 88.7% of this area comprised the Pacific slope (Fig.3-III).Extreme precipitation anomalies in COA-EN 17 reached the breakpoint of 90% at an altitude of about the same as EP-EN 82/83, 1700masl.As in EP-EN 97/98, over 50% of all extreme anomalies occurred at elevations under 150 m (Fig.3.2.2 SPDIAs for Year 1 in Galapagos, all of the mega-El Niño events generated an SPDI value "near normal" (-0.99 -0.99).A situation that changed very significantly in Year 2 of both, EP-EN 82/83 and EP-EN 97/98, when the mean condition turned to "extremely humid": 2.13 in 1983, and 3.64 in 1998 (see Fig. 2-IIb).From this Figure, as well as from Fig. 3-I and II, it is evident for Year 2 an extremely significant difference (P<0.0001) in the temporal dynamics of SPDI values between EP-EN 82/83 and EP-EN 97/98 mega-El Niño events.For instance, in 1983, the "extremely humid" condition (SPDI ≥2.0) was reached abruptly in March and lasted until August that year when reached a maximum SPDI value of 3.46.From April to August 1983, 99% of the Galapagos was affected by an SPDI mean value of 3.33.As for 1998, overall affectation by excessive rainfall lasted 12 months.It started in February when precipitation generated a "very humid" condition (SPDI =1.64), and for the next 11 months persisted an "extremely humid" condition with SPDI values ranging from 3.01 to 4.31, and affecting more than 98% of the Galapagos surface.Mega-El Niño EP-EN 97/98 affectation lasted until the first months of 1999, after a sudden drop in SPDI values: from 3.82 in January to 0.06 in February.No significant effects on SPDI value dynamics were associated with the mega-El Niño MIX-EN 15/16 event.As for the Coastal El Niño 2017, from April to June, precipitations generated an SPDI value reaching 1.01 to 1.03, a value that barely denotes a moderately wet condition (see Fig. 2-IIb and Fig. 3-III).SPDI monthly values dynamics of 1983 in the Galapagos was not significantly different from that in the Pacific slope (P=0.606)(seeFig. 2-Ib and IIb).years 2015 and 2016 (0.33 and -0.69, respectively; P<0.0001).SPDI for Year 2 of EP-EN 82/83 was significantly higher than those of EP-EN 97/98, MIX-EN 15/16, and COA-EN 17 (P<0.0001,see Fig. 2-IIIb).As for areas of continental Ecuador with extreme precipitation anomalies of SPDI ≥2.00, around 6.2% occurred in the Amazon slope during Year 2 of EP-EN 82/83, 1.2% during EP-EN 97/98, and 11.3% during COA-EN 17 (Fig. 3-I, II and III).No such extreme events were observed during MIX-EN 15/16 for this slope during 2016.About 90% of all extreme precipitation during EP-EN 97/98 occurred at From cluster analysis of 1983 (i.e.Year 2) of this mega-Niño´s SPDI monthly data, four distinctive groups of hydrological systems are evident (Table2).A first group conformed by 11 basins (Cluster 1: 16-NARANJAL PAGUA, 23-ISLA PUNA, 17-JUBONES, 14-TAURA, 13-GUAYAS, 15-CANAR, 12-ZAPOTAL, 21-PUYANGO, 18-SANTA ROSA, 19-ARENILLAS, and 20-ZARUMILLA), all belonging to the Pacific slope, having an extremely high collective monthly mean During 1998, that is Year 2 of EP-EN 97/98 mega-El Niño event, two distinctive groups of hydrological systems showed prolonged and extremely high precipitation anomalies (Table3).A first large group, conformed by fourteen Pacific slope Jul mean monthly SPDI condition as extremely humid as Cluster 1 (3.62 vs. 3.68).But, differently than Cluster 1, this group of three basins steadily prolonged their extremely humid condition until Nov, for a total span of ten months, resulting in a collective SPDI mean value significantly higher than for any cluster and for any other extreme El Niño event (Table3-a), steadily affecting, fairly homogeneously, about 93% of the basins areas (Table3-b).On the other hand, the mega-El Niño event EP-EN 97/98 did not appear to have generated humid precipitation anomalies for the rest of continental Ecuadorian basins.Cluster 3, for example, during 1998 a nine-basin group from both Amazon and Pacific slope (03-MATAJE, 02-MIRA, 29-MORONA, 28-PASTAZA, 27-CUNAMBO, 30-SANTIAGO, 04-CAYAPA, 05-VERDE, Table4-a).While Cluster 2 mean SPDI tended to be positive, and Cluster 3 negative, none showed precipitation dynamics that resulted, spatially and temporarily, in a mean condition different than normal.The last cluster, conformed exclusively by a rather small and southernmost Pacific slope basins (Cluster 4: 19-ARENILLAS), showed a unique SPDI dynamic when, from Mar to Dec, reached a sustained extremely humid condition (mean SPDI =2.8), spatially affecting 98.9% of basin´s area from Mar to Jun, and 66.2% from Jul to Dec (Table4-a and b).
Two additional clusters of hydrological systems resulted from the spatiotemporal analysis of 1983 SPDI monthly data.Cluster 3 is a group of three southerly Amazon slope basins (29-MORONA, 28-PASTAZA, and 30-SANTIAGO), where none of them showed significant precipitation anomalies (i.e.SPDI>1.0)(Table2-a).Cluster 4, on the other side, showed SPDI dynamics that generated moderately to very humid conditions throughout the entire year, and were not confined only to the March/October pulse observed for Clusters 1 and 2. Cluster 4 comprises eleven basins from both slopes: four from the Amazon slope (31-CHINCHIPE, 26-NAPO, 27-CUNAMBO, and 25 SM PUTUMAYO); six from the northernmost section of the Pacific slope (01-CARCHI, 02-MIRA, 05-VERDE, 03-MATAJE, 04-CAYAPA and 06-ESMERALDAS), and one to the southernmost basin of the Pacific slope (22-CHIRA, see Table 2-a).In this cluster, the most extreme precipitation anomalies as well as a most extended area of affection occurred from Jun until Oct, and even until the end of 1983.This was mainly due to the SPDI dynamics of the Pacific slope basins (Table 2-b).3.4.3 MIX-EN 15/16As for the effects of mega-El Niño event MIX-EN 15/16 on the hydrological systems´ precipitations, overall monthly SPDI values for 2016 were well defined as near normal (-0.99 -0.99).During this event, only five systems (08-JAMA, 09-CHONE, 19-ARENILLAS, 21-PUYANGO, and 20-ZARUMILLA) showed a short lasting (2 months, Mar/Apr) and discrete increase of SPDI values, barely reaching a moderately humid condition, with a collective SPDI mean of 1.12.While the Pacific slope´s systems tended to have positive SPDI values during 2016, the Amazon slope´s systems tended rather negative ones.As for 31-CHINCHIPE, this southernmost Amazon basin showed an overall 2006 SPDI value of -1.40, a moderately dry condition.and22-CHIRA),showedacollectivesustainedSPDImeanvalue about normal (>-1, <1; Table3-a).Moreover, during this mega-El Niño event, Cluster 4, a four-basins group (26-NAPO, 25-SM PUTUMAYO, 31-CHINCHIPE, and 01-CARCHI), showed a moderately dry mean precipitation condition (SPDI =-1.23,Table 3-a).3.4.4 COA-EN 17 During the Coastal El Niño of 2017, a group of 15 Pacific slope basins (Cluster 1: 20-ZARUMILLA, 14-TAURA, 09-CHONE, 08-JAMA, 16-NARANJAL PAGUA, 15-CANAR, 13-GUAYAS, 17-JUBONES, 11-JIPIJAPA, 10-PORTOVIEJO, 12-ZAPOTAL, 23-ISLA PUNA, 22-CHIRA, 21-PUYANGO, and 18-SANTA ROSA) showed an extremely humid condition from Mar to Jun (mean SPDI =2.63) (Table4-a).During these four months, the Coastal El Niño event affected 78.4% of the area of the basins of Cluster 1.By July, the affected area was 47.0%, and then lowered to 13.3% for the rest of the year (Table4-b).Precipitation anomalies of varying intensities extended until Sep for some of these basins of Cluster 1.In the case of 18-SANTA ROSA, the anomalies lasted until Dec.From cluster analysis of SPDI dynamics during COA-EN 17, two other groups of basins from both Pacific and Amazon slopes Niño becomes easily identifiable.Besides the insular system of 24-GALAPAGOS, the other 13 are continental systems from the Pacific slope that show the highest seasonality in their precipitation: 19-ARENILLAS (SPDI 2.47 and  1.18), 08-JAMA

Table 2 .
Standardized Precipitation Drought Index (SPDI) temporal and spatial dynamics at the different hydrographic systems for 1983 (Year 2) of mega-El Niño event EP-EN 82/83.Cluster analysis (K-means clustering using Euclidean distance) was performed to both rows and columns, and with the statistical tool ClustVis