Mesoscale Convective Systems over Ecuador: Climatology, Trends and Teleconnections

Abstract

Research on Mesoscale Convective Systems (MCSs) in Ecuador has focused on regional studies. However, it lacks a thorough and general examination of their relationship with the nation’s diverse orography and large-scale phenomena. This study conducts a climatological analysis of MCS occurrence throughout Ecuador’s natural regions. We perform this study using Sen’s Slope and the Mann–Kendall test. Teleconnections from the Pacific and Atlantic Oceans are studied through wavelet decomposition between time series and Pacific and Atlantic oceanic indices. The main factors that control MCS formation depend on the region. The Intertropical Convergence Zone (ITCZ) at the large scale affects the entire territory. In western Ecuador, MCS formation is mostly related to the El Niño current and the Chocó Low-Level Jet (CLLJ). The Orinoco Low-Level Jet (OLLJ) and evapotranspiration and nocturnal convection display the largest roles in the east. A progressive intensification of activity from Highlands-North in SON is detected (0.143 MCSs per year). MCSs contribute 26% of total precipitation on average, with regional variations from Coast-South (16.41%) to Amazon-North (44.13%). The research confirms existing knowledge about El Niño’s strong relationship (ρ = 0.7) with MCS occurrence in coastal areas while uncovering new complex patterns. The Trans-Nino Index (TNI) functions as a critical two-sided modulator that conventional analysis methods fail to detect. It produces null correlations over conventional time series of MCS occurrence yet emerges as a primary driver of low-frequency variability in the proposed six natural zones of Ecuador. Wavelet decomposition reveals contrasting TNI responses: Amazon-North shows positive correlation (0.73) while Amazon-South exhibits negative correlation (−0.70) at low frequencies. This affects Walker circulations dynamics over the Pacific Ocean. This research establishes fundamental knowledge about MCSs in Ecuador. It builds on a database with strong methodology as a backbone. The research provides essential information about the factors leading to convection in the country. This will help improve seasonal forecast accuracy and risk management effectiveness.

1. Introduction

Large, organized Mesoscale Convective Systems (MCSs) produce extensive high-intensity precipitation over large areas according to [1]. MCSs affect rainfall patterns which makes them essential for maintaining water cycle dynamics and hydrological equilibrium [2]. These systems receive worldwide research attention because they function as primary seasonal rainfall drivers in tropical areas [3]. Research indicates that MCSs generated 70% of Colombia’s precipitation during 1998 [4]. The strategic sectors of agriculture and hydroelectric power generation and water supply experience significant effects from these phenomena [5]. The heavy rainfall from MCSs creates floods which create substantial dangers to human populations and infrastructure [6].

Despite the comprehensive scientific research about MCSs from South America and subtropical regions, which enabled researchers to find appropriate research methods and establish a strong theoretical base [7], in Ecuador the study of MCSs is just beginning to emerge. Reference [8] established that MCSs generate 57% of precipitation in the Pacific Ocean region of Colombia. The Amazon Basin provides an optimal environment for MCS formation according to [9,10] because trade winds enable moisture transport and subsequent convection which leads to more than 7000 MCS events annually. The research by [11] studied MCS patterns in Peru through simulation methods to describe these systems in the central Andes region. The research by [1,12] studied MCS behavior in northern South America to understand their movement patterns and precipitation distribution.

Research has proven MCSs’ effect on precipitation in specific regions of South America, but, in Ecuador, scientists have not been able to determine the exact amount of annual precipitation that MCSs generate, in part because there are large regions that are under-monitored. Reference [13] investigated the Paute river basin in southern Ecuador through a localized study which revealed that MCS events mostly occur during rainy nights in eastern areas of the basin because of strong convective potential energy from the Amazon region. However, any risk to infrastructure and human well-being is evaluated. The National Secretariat for Risk Management (SNGR) of Ecuador documented 12,859 hydrometeorological events from 2010 to 2020 which included floods, landslides, and flash floods according to [14]. These examples demonstrate the intense nature of rainfall phenomena. The Coast-North region of Ecuador experienced a catastrophic flood on 24 January 2016, when a single MCS produced 245 mm of rain throughout 24 h in Esmeraldas’s city according to [15]. The residential area of La Gasca in Quito located in the Highlands-North of Ecuador has experienced multiple alluvial fan events because of extreme precipitation, probably triggered by the incidence of MCSs after several days of constant rainfall in the region. The 31 January 2022 flashflood event caused 28 fatalities and 52 injuries along with one person missing while resulting in 8.8 million USD worth of economic losses according to [16,17].

The study of MCS occurrence under climate change requires analysis because the generalized warming of the atmosphere may affect both the frequency and intensity of MCS events [18]. The scientific community continues to study the direction of these potential changes, and furthermore, it may be location-dependent. The study by [10] showed that MCSs across the Amazon Basin may have decreased by 50% due to the interannual variability processes. The research by [19] used past, present, and future scenario simulations to show that MCS events decreased steadily in the tropical Andes and Amazon regions. Research conducted in the southern United States indicates that MCSs will become more frequent and produce more intense storms with heavier precipitation because of changes in their convective cores [20].

The findings of MCS occurrence showing opposite trends may in turn be related to the complex response of teleconnections to climatic change. Thus, the quantitative assessment of large-scale atmospheric variations in specific areas through macro-climatic indices serves as climate change indicators [21]. The effect of such indicators can be analyzed through teleconnections or distant interactions usually using correlation methods [22,23]. In Ecuador, Ref. [24] established that South Pacific Sea surface temperatures affect Andean precipitation, but the western region receives stronger influences than the eastern region which responds more to the Tropical Atlantic Ocean dipole. The occurrence of MCSs in the Paute river basin shows sensitivity to El Niño-Southern Oscillation (ENSO) according to [13]. Ref. [25] applied regression models to establish annual relationships between precipitation and the difference in normalized SST anomalies between the sea surface temperature (SST) in the Pacific Ocean within −80°, −90° and 0°, −10° (Niño 1+2) and in the Pacific Ocean within 160°, −150° and 5°, −5° (Niño 4) regions using the Trans-Nino Index (TNI) [26]; this demonstrates the pressure difference between the Icelandic Low and Azores High as North Atlantic Oscillation (NAO) [27] in southern Ecuador. The Bayesian network analysis by [22] revealed that precipitation in the Coast and Amazon regions receives diverse impacts from both Pacific and Atlantic Ocean influences. Ref. [15] determined through their research that ENSO conditions together with an oceanic Kelvin wave led to the severe MCS in Esmeraldas, Ecuador, during January 2016. The existing research on precipitation regimes has not included a complete examination of MCS teleconnections throughout Ecuador using macro-climatic indices. The research by [28] examined MCSs in the Ecuador–Peru border region but the regional studies by [13,15] remain focused on specific areas.

Despite much progress having been achieved, a unified examination of MCS patterns at the national level in Ecuador is missing. This examination is fundamental to increasing our ability in predicting and managing extreme weather events in the current and future climates. In addition, the analysis of MCS climatology in Ecuador’s natural regions is a WMO recommendation to improve threat response capabilities and climate change detection [29,30]. Therefore, the present study aims to fill this vital gap through the extensive nationwide investigation on MCSs’ spatial and temporal patterns to the diverse mountainous landscape of the country [31]. Regions and time periods with elevated MCS activity are identified through monthly and seasonal climatological assessments. Statistical trend analysis with Sen’s slope [32] and the Mann–Kendall [33] test are used to detect changes between 2001 and 2020. Wavelet decompositions of time series data with their raw time series were applied to investigate how MCS formation is affected by both low-frequency background conditions and high-frequency bursts of Pacific and Atlantic oceanic activity. Scientific and technical knowledge are derived from this study, enhancing warning system capabilities and optimizing hydroelectric plant operations at the Coca Codo Sinclair reservoir, over the North-West of Ecuador, ultimately supporting national climate adaptation and risk reduction initiatives [34,35]. This research reveals peak MCS contribution of 44% during MAM, with an annual mean 26% of total precipitation, alongside the first documented positive trend of MCS occurrence in the country. Finally, the heterogeneous Ecuadorian orography creates opposing MCS responses to TNI in adjacent regions, where Amazon-North enhanced MCS genesis (0.73 correlation on low frequency) while Amazon-South exhibits convective suppression (−0.70 correlation on low frequency and 0.63 correlation with omega and TNI).

2. Materials and Methods

Research on MCSs during the twenty-year period from 2001 to 2020 follows a three-stage approach. The first stage required data collection and preprocessing to establish the MCS database. The second stage analyzed MCS occurrence patterns throughout the six natural regions of continental Ecuador. These include Coast-North, Coast-South, Highlands-North, Highlands-South, Amazon-North, and Amazon-South (Figure 1). The third stage determines yearly and seasonal patterns of MCS occurrence. Subsequently, we conducted analyses of trends in occurrence and teleconnections. We used open-source software R [36], and its integrated development environment RStudio (version 2025.09.0+387) served as the tool for performing all data processing and analysis.

2.1. Study Area

The research area is the continental region of Ecuador, spanning 2° N and 5° S latitude and 82° W and 75° W longitude (Figure 1). Rainfall patterns in Ecuador show complex spatial and temporal variations, along with various climatic and orographic factors which also shape MCS development [13]. The climate receives its modulation from various large-scale features. The Pacific Ocean Intertropical Convergence Zone (ITCZ) controls precipitation patterns when it moves through the country twice per year from December to May [37] while the South Pacific anticyclone creates coastal dryness through its strengthening of cool dry winds [38]. The Amazon Basin produces substantial moisture through vegetation evapotranspiration which generates copious amounts of water vapor [39,40]. The trade winds carry this moisture toward the west which creates strong Andes–Amazon interactions that control precipitation patterns in these areas [41].

The updated Köppen–Geiger classification identifies eight natural zones in Ecuador based on rainfall, temperature, and vegetation [42], but this study uses a six-zone division based on the three well-defined geographical regions of the country: Coast, Highlands (Andes), and Amazon, arranged from west to east [43]. The Andes Mountain range is also analyzed, whose complex topography and altitudinal variations trigger enormous climatic variability and complex weather patterns [44]. The latitudinal division is marked by the asymmetry present in the Equatorial Andes described by the different orographic profiles in the zone [45]. The northern zone of the eastern hill is greater than the western [46] and in the south zone the western zone is greater than the eastern [47]. Therefore, the middle point of the Equatorian region is taken as a divisional point. These regions are shaped by several key factors in their climatology. The biannual passage (April–October in the South) of the Pacific Ocean ITCZ over Ecuador is fundamental, modulating the pluviometry regimes [24]. This is compounded by orographic rainfall, which is generated by the interaction of moist air masses with the local topography [48]. These factors together produce the bimodal rainfall regime of the Andes, with rainy seasons occurring between January–April and October–December [15,49]. In contrast, the Coast exhibits a unimodal regime with a single rainy season (December–April), which results from the passage of the ITCZ combined with the weakening of the South Pacific anticyclone, leading to warmer water temperatures and consequently higher humidity [38]. Contrarily, the dry season (May–November) in the Coast is driven by the cold Humboldt Current which suppresses evaporation of the sea and stabilizes the atmosphere, inhibiting convection [50]. Finally, the Amazon region is characterized by a monomodal regime with year-round rainfall, sustained by moisture from both local vegetation evapotranspiration and the Atlantic Ocean [51].

The ENSO and the Pacific Decadal Oscillation (PDO) are macro-climatic phenomena that affect precipitation in Ecuador [52,53] by changing trade winds and, consequently, Pacific Ocean temperatures [54]. This influence, however, manifests differently across the country’s distinct geographical regions. The coast typically experiences increased precipitation during El Niño phases and drier conditions during La Niña [15,55]. Conversely, in the Highlands and Amazon regions, El Niño phases are often associated with elevated temperatures and droughts, while La Niña tends to bring cooler temperatures and increased rainfall [56,57]. Furthermore, the Pacific Decadal Oscillation (PDO) plays a significant modulating role of ENSO, affecting Coastal precipitation in Ecuador; specifically, its warm phase can intensify the effects of both El Niño and La Niña, while its cold phase may increase the uncertainty of their impacts [58]. In parallel, the PDO can also influence the moisture exchange between Amazon and the Andes, thereby altering local precipitation patterns [59].

2.2. MCSs Database

Atmospheric convection functions as a heat transfer mechanism that uses airflow exchanges to make warm moist air rise through density differences which then condenses into clouds [60,61]. A convective system consists of multiple organized convective cells which produce precipitation according to [62]. They also distribute entropy through atmospheric instability and moisture conditions [63,64]. A MCS classification requires convective clusters to span at least 100 km horizontally while lasting between several hours and multiple days [65].

In the present study, we used the MCS dataset from [66,67], which was obtained through a rigorous process and validated in [68]. To identify cloud systems that could potentially be classified as MCSs, Ref. [66] used brightness temperature from infrared (~10 μm) satellite imagery, tracking all objects with at least one pixel colder than 241 K. To confirm that the tracked cloud systems were indeed MCSs, Ref. [66] applied the following criteria:

• The MCS cloud shield must cover at least 40,000 km² with a brightness temperature of 225 K or colder for a minimum of 4 consecutive hours.
• Precipitation underneath the cloud shield must include at least one pixel with a peak precipitation rate ≥ 10 mm/h, persisting for at least 4 continuous hours.
• A minimum rainfall volume of 20,000 km²·mm/h must occur at least once during the MCS lifespan.

The MCSs were tracked for the entirety of South America and contain 1,412,688 records and 80,817 genesis events with sub-daily frequency between 2001 and 2020. This database has the advantage of covering the entire lifecycle of each MCS, from genesis to dissipation, allowing for analyses of either the complete evolution or exclusively the initiation phase.

Each record in the database provides descriptive information for every MCS, including local and UTC time, centroid longitude and latitude, duration up to that point, spatial extent in km² and pixels, lifecycle phase, and data integrity. Over our region and period (2001–2020) of interest, the database has 71,379 records and 4042 genesis events per year, exhibiting a stable trend over time (Figure 2). The year with the highest number of events was 2008 (4303), while the lowest was 2020 (3762). Similarly, 2018 had the most records (76,373), and 2002 had the fewest (65,201).

Database Source

The present analysis used the database from [66] which is explained in Section 2.2. This dataset stands out because it tracks MCSs from start to finish from their activity, size, and strength [69]. The MCS lifecycle consists of three phases according to [70], which include genesis followed by maturation and then dissipation [71], although [72] has identified a fourth preceding stage known as convective initiation. The three main stages described by [70] are considered for this study. The analysis of the MCS dataset was complemented by daily precipitation records obtained from INAMHI Ecuador’s National Institute of Meteorology and Hydrology station network used to obtain multi-annual monthly mean of observed rainfall and satellite-based information from the Integrated Multi Satellite Retrievals For Global Precipitation Measurement (GPM IMERG) [73]. The oceanic indices are retrieved from Physical Sciences Laboratory of National Oceanic and Atmospheric Administration (NOAA) [26,74,75,76]. The climatic variables are derived from ERA5 monthly averaged data on pressure levels from 1940 to present [77].

2.3. Data Preparation

Preprocessing Data

A dataset focused on Ecuador required applying spatial filters together with type and quality filters. The geographical boundaries of Ecuador were established through previous research [13,43,78] between −82° and −75° longitude and −5° to 2° latitude. The spatial filtering process allowed researchers to select and validate unique storm records. The discrimination criteria established by [66] served as the basis for this evaluation.

• The selection of events occurred exclusively during their genesis phase based on life phase = “genesis”.
• The criterion MISSING > 0 was used to remove events containing corrupt or missing data.

The filtered events received classification based on their geographical position within the six natural zones which were described in the study area (Section 2.1). Each event received its natural zone assignment through the process of spatial intersection between the event point location and the corresponding regional polygon [79].

2.4. MCS Characterization

2.4.1. Monthly and Seasonal Aggregation

The MCS database underwent filtering and partitioning before monthly and seasonal frequency analysis throughout 2001–2020. This evaluation enabled detection of intra-annual cycles which show seasonal patterns in the study area [80]. The analysis started by counting all events that occurred throughout each month of every year across the six regions. The validation of these findings used multi-year monthly averages from local stations during the same period to confirm the ability to represent local seasonality [78]. Seasonal frequency analysis involved counting events during December–January–February (DJF), March–April–May (MAM), June–July–August (JJA), and September–October–November (SON) seasons. Seasonal division effectively captures the climatic patterns of the tropical Andes. These regions lacks pronounced difference in air temperature or moisture that defines traditional seasons like winter or summer [9,81]. Additionally, the contribution of MCSs to the Ecuadorian total volumetric rainfall was assessed using precipitation data from MCS cloud areas compared to the total precipitation estimates from GPM IMERG data.

2.4.2. Hourly Patterns

Sub-daily data enabled us to study the temporal state at one-hour intervals within the MCS lifecycle. Hourly resolution adequately determines MCS development and persistence during diurnal cycles [82,83]. The diurnal cycle characterization stands as a fundamental element for understanding convective mechanisms in the region [82,84]. The data received initial processing through natural zone segmentation followed by monthly separation. The events from each zone were grouped according to local hourly and monthly periods throughout the entire 12-month period. The analysis followed the same procedure for four distinct time intervals: 00:00–06:00, 06:00–12:00, 12:00–18:00, and 18:00–24:00 local time. The results appear as heat maps which serve well to show hourly patterns [85].

The database considers MCS duration from genesis to dissipation. The time classification follows [86] to detect solar radiation effects on atmospheric stability. MCS events were sorted into their corresponding time intervals to study their lifespan distribution. Boxplots were used to evaluate and compare the diurnal cycle [87] along the four seasons (DJF, MAM, JJA, SON). This analysis helps to evaluate how the MCS diurnal cycles are affected by seasonal convective conditions [9].

2.5. Trend Analysis

Monthly and Seasonal Trend Estimation

The trend analysis was conducted on monthly and seasonal time series to determine possible climate change trends influenced by MCS occurrence in the six natural regions throughout the 20-year study period. A non-parametric Mann–Kendall test ($S$) and Sen’s Slope (β) [88] were applied for this analysis. The Mann–Kendall test detects monotonic trends through Equation (1). It shows positive values for increasing trends and negative values for decreasing trends [33]. Where

$$S=\sum _{k=1}^{n-1}\sum _{j=k+1}^{n}sgn\left(x_j-x_k\right)$$

(1)

• $x_j,x_k$ data sequence at time steps j and k, respectively.
• $sgn\left(x_j-x_k\right)$ is the function sign, +1 if sgn > 0, 0 if sgn = 0, −1 if sgn < 0.
• $n$ sample number.

Equation (2) represents the occurrence–time relationship magnitude by the tau ($\tau$) value. This shows monotonic strength and direction between −1 (decreasing) and +1 (increasing) [89]. Where

$$\tau ={\displaystyle \frac{S}{\left(n\left(n-1\right)/2\right)}}$$

(2)

Parameter β in Equations (3) and (4) determines the strength of the trend pattern in a series. The sign of the value indicates whether the trend increases (positive) or decreases (negative) during each time unit [32]. Sen’s Slope overall slope equals the median slope value from all calculated slopes [90]. The S value indicates trend existence while β measures the rate of trend change throughout time [91]. Where

$$Q_i={\displaystyle \frac{\left(x_j-\mathrm{ }{x}_k\right)}{\left(j\mathrm{ }-\mathrm{ }k\right)}}$$

(3)

$$\mathsf{\beta }=\mathrm{Median} \left(Q_i\right)$$

(4)

• $Q_i$ trend in time i when $j\ne k$.
• $\beta$ trend median.

These two methods are widely used in climate studies due to their robustness with non-normal distributions [92]. We executed them using the “Trend” (version 1.16) package in R [93]. Monthly series were formed by grouping the data by year–month (YM). Seasonal analysis used standard DJF, MAM, JJA, and SON trimesters. The null hypothesis (H0) assumed the absence of a trend, whereas the alternative hypothesis (H1) assumed the presence of a significant trend.

2.6. Teleconnection Analysis

Teleconnection analysis proceeded through Spearman’s correlation in two distinct stages. The first step involved calculating correlations between oceanic indices and MCS occurrence time series (S) for each region [94]. The second stage assessed potential teleconnections between the indices and the low frequency and high frequency of MCS occurrence. This used wavelet decomposition analysis at the first level, i.e., trend (V1) and detail (W1) components, respectively. Wavelet decomposition proved to be a useful method to reveal multiscale patterns in climatic indices [95]. Additionally, we conducted a second analysis of temporal series with 3-month lag (S lag). This analysis aimed to detect if an index has relevance on the next station within the observation of the delay in the ocean to alter the circulation pattern at a large scale [96]. Lag–correlation have been successfully used by [24] over the Andes, establishing physics conditions between Equatorian rainfall and oceanic anomalies. Analysis focused on Niño 1+2 and SST in the Pacific Ocean within −120°, −170° and 5°, −5° [°C] (Nino 3.4), TNI [°C], Tropical South Atlantic Index (TSA) [°C], Tropical North Atlantic Index (TNA) [°C], and Atlantic Meridional Mode (AMM) [°C] indices, because they represent major Pacific and Atlantic Ocean SST changes [75,97,98], to study their effects across the six natural zones.

2.6.1. Wavelet Decomposition

The analysis of MCSs from a non-stationary perspective allows us to understand how these phenomena interact with macro-climatic indices across different time scales [99]. The Maximal Overlap Discrete Wavelet Transform (MODWT) serves as the fundamental tool for decomposing time series data. It represents data in the time–frequency domain [100,101]. MODWT functions as a non-decimated version of the DWT and represents a robust solution for real-world climate series. It preserves the original signal length regardless of its length [102,103]. The “wavelets” package in R (version 0.3) [104] performed decomposition using the least asymmetric filter. We applied this procedure to the monthly, seasonal (DJF, MAM, JJA, SON), and annual time series. MODWT produces two main components, which include detail coefficients (W) that detect fast and sudden events like heavy rainfall and approximation coefficients (V) that show low-frequency patterns such as seasonal and interannual cycles [105]. Equations (5), (6) and (7) describe how W, V, and the reconstructed signal S are calculated, respectively [106].

$$W_j\left(t\right)=\sum _{l=0}^{L_j-1}\widetilde{g_{j,l}}{X}_{in,j}\left(\left(t-l\right)\backslash mod\mathrm{ }\left(N\right)\right)$$

(5)

$$V_j\left(t\right)=\sum _{l=0}^{L_j-1}\widetilde{h_{j,l}}{X}_{in,j}\left(\left(t-l\right)\backslash mod\left(N\right)\right)$$

(6)

$$S\left(t\right)=\mathrm{ }\sum _{j=1}^J{W}_j\left(t\right)\mathrm{ }+\mathrm{ }{V}_J\left(t\right)$$

(7)

• $W_j\left(t\right)$ detail component for level j.
• $V_j\left(t\right)$ approximation component for level j, same that goes for the next decomposition level or final component if $j>J$.
• $S\left(t\right)$ reconstructed series for level j.
• $X_{in,j}\left(t\right)$ entry level data for level j. If, $X_{in,1}\left(t\right)=X\left(t\right)$ (original signal), if $j>1$, $X_{in,j}\left(t\right)=V_{j-1}\left(t\right)$ previous level approximation.
• $g_{j,l}$ wavelet filter coefficients for high pass.
• $h_{j,l}$ wavelet filter coefficients for lower pass.
• $L_j$ filter length at level j.
• N sample length.
• $mod\left(N\right)$ circular convolution.

2.6.2. Spearman’s Rank Correlation

Spearman’s rank correlation (ρ) measures the degree of association between two variables [107]. As a non-parametric method, ρ assesses the monotonic relationship between two variables, which does not need to be linear [108]. For this reason, it has been widely used in teleconnection studies that involve non-normally distributed series, such as precipitation and drought [22,24,25,57]. The ρ coefficient ranges from −1 to +1 and is described in Equation (8), where

$$\rho =1-{\displaystyle \frac{6\mathrm{ }\sum d_i^2\mathrm{ }}{n\mathrm{ }\left(n^2-\mathrm{ }1\right)}}$$

(8)

• $d_i$ range difference of each variable.
• $n$ sample size.

The [109] adjustment was applied to Spearman’s rank correlation to avoid inflated values and obtain significant values [110]. The “atrocron” package (version 2.0.5) [111] in R performed the correlation analysis. The analysis were performed on both the complete data series (without and with lag) and their respective wavelet decompositions (only for original series without lag) against the various macro-climatic indices (Niño 1+2, Niño 3.4, TNI, TSA, TNA, and AMM). The null hypothesis (H0) assumed the absence of a significative correlation, whereas the alternative hypothesis (H1) assumed the presence of a significative correlation.

2.6.3. Analysis of TNI

The annual series of TNI was analyzed via Spearman’s rank correlation with Ebisuzaki’s adjustment and climatological variables and their respective wavelet components. The variables zonal wind (u) [m/s], meridional wind (v) [m/s], vertical velocity (ω) [Pa/s], specific humidity (q) [kg/kg] at 500 hPa and 800 hPa, Convective Available Potential Energy (CAPE) [J/kg], and Convective Inhibition (CIN) [J/kg] at single level were selected due to the relation with MCS formation [2,70,72]. Additionally, derived variables that quantify the vertical wind shear were selected, which is fundamental in MCS formation [112,113,114]. Zonal wind difference ($\Delta u$) [m/s], meridional wind difference ($\Delta v$) [m/s], and wind shear magnitude ($WSM$) [s⁻¹] were calculated in Equations (9)–(11), where

• u₅₀₀, u₈₀₀ represent the zonal wind component at 500 hPa and 800 hPa, respectively.
• v₅₀₀, v₈₀₀ are the corresponding meridional components.
• Δz is the geopotential height difference between both levels, 500 hPa and 800 hPa, of 3500 m.

$$\Delta u=u_{500}-u_{800}$$

(9)

$$\Delta v=v_{500}-v_{800}$$

(10)

$$WSM={\displaystyle \frac{\sqrt{\Delta u^2+\Delta v^2}}{\Delta z}}$$

(11)

The null hypothesis (H0) assumed the absence of a significative correlation, whereas the alternative hypothesis (H1) assumed the presence of a significative correlation.

3. Results

3.1. Spatio-Temporal Analysis Occurrence of MCS over Ecuador

3.1.1. Applied Filter to Ecuador Database

The database contains 1906 MCS events after applying the selection criteria for MCS occurrence. The MCSs maintain an average duration of 12 h starting at 14:00 local time. The MCS events have an average size of 15,319 km² while their mean location exists at −76.50° longitude and −0.69° latitude. The spatial distribution of the filtered database appears in Figure 3 which shows MCS development across Ecuadorian territory while indicating areas with the highest convective frequency during the study period.

3.1.2. Natural Zones Classification

The accumulated number of MCS events for the six sub-zones during the 2001–2020 period is presented in

Table 1. Total cumulative number of MCS genesis events by natural region of Ecuador (2001–2020).

Region	Coast	Highlands	Amazon
North	112	136	207
South	17	74	93

. The Amazon-North region shows the highest activity with 207 events, followed by Highlands-North (136) and Coast-North (112). On the other hand, the Coast-South zone recorded the lowest number of events (17). This spatial distribution provides the basis for determining the monthly and seasonal frequency, interannual variability, and existing patterns, as well as for analyzing trends in occurrence.

3.1.3. Monthly Frequency

The time series of monthly average MCSs from 2001 to 2020 is shown in Figure 4 for each of the six natural zones. The same period of monthly accumulated precipitation is shown in each region. The northern and southern regions show a clear seasonal pattern where the early months of the year have the highest frequency of MCS events which then decrease throughout the rest of the year. The Coast-North region has the highest number of events with 35 occurrences in April while the Coast-South region has 10 events in March, but in April drops to 0.

The Highlands-North and Highlands-South zones demonstrate that their highest activity occurs between January and May and between September and December. The highest number of events occurs in March and November with 20 occurrences in Highlands-North and 13 occurrences in Highlands-South in March. The occurrence in both zones decreased from June to August with July showing the lowest activity at one event in each zone. Amazon-North experiences its highest activity in March with 33 events, and April and October and November each record 24 events. The Amazon-South zone experiences its active months with 10–12 recorded events for January, February, October, and November.

3.1.4. Seasonal Frequency of MCS Occurrence

The accumulated number of MCS events for the 2001–2020 period with a total of 639 is presented in

Table 2. Total cumulative number of MCS genesis events and their contribution to estimated GPM IMERG precipitation, by season and natural region of Ecuador (2001–2020).

Station	Region	Coast		Highlands		Amazon
		Quantity of MCS	Percentage of Rainfall *	Quantity of MCS	Percentage of Rainfall *	Quantity of MCS	Percentage of Rainfall *
DJF	North	22	21.21	35	22.51	50	39.35
	South	5	15.23	22	20.91	28	36.46
MAM	North	80	26.90	53	27.94	79	44.09
	South	12	19.32	29	22.42	23	37.98
JJA	North	6	20.94	5	19.89	20	24.59
	South	0	24.58	7	18.75	12	20.77
SON	North	4	18.32	43	27.71	58	44.13
	South	0	18.25	16	26.64	30	39.71
Annual	North	112	21.12	136	24.59	207	38.09
	South	17	16.41	74	21.94	93	33.65

* The percentage of precipitation was calculated using GPM IMERG data.

according to season and natural zone. The MAM season has the highest number of MCS events with 276 occurrences, containing precipitation contribution periods with 44.09% in Amazon-North, 37.98% in Amazon-South, 27.94% in Highlands-North, and 26.90% in Coast-North. The JJA season has the lowest number of MCS events with only 50 occurrences and reduced precipitation contribution. Amazon-North drops to 24.59%, while Highland regions reach their annual minima (19.89% north, 18.75% south). The SON and DJF seasons have similar occurrences with 151 and 162 events, respectively, though SON shows notable recovery in precipitation contribution, particularly in Amazon-North (44.13%) and Highlands-North (27.71%).

The Coast-North and Coast-South zones have their highest activity during the MAM season with 80 and 12 events, respectively, contributing 26.90% and 19.32% to regional precipitation. Both coastal zones have a significant decrease in MCS occurrence during the JJA and SON seasons to the point where the Coast-South zone has no MCS events, yet they maintain precipitation contributions of 24.58% (JJA) and 18.25% (SON) from foreign MCS systems originating elsewhere.

The MAM season remains the most active period in both the Highlands-North and Highlands-South zones with 53 and 29 events recorded, respectively, contributing 27.94% and 22.42% to regional precipitation. Both zones display significant MCS activity during the SON (43 and 16 events, contributing 27.71% and 26.64%, respectively) and DJF seasons (35 and 22 events, contributing 22.51% and 20.91%, respectively). The JJA season demonstrates minimal MCS activity throughout both the Highlands-North and Highlands-South regions, with only five events in the north (19.89% precipitation contribution) and seven events in the south (18.75% contribution). The MCS activity remains high throughout all seasons in the Amazon-North and Amazon-South zones, with annual precipitation contributions of 38.09% and 33.65%, respectively. The Amazon-North region experiences the second highest number of MCS events with 79 occurrences during MAM (44.09% contribution), 50 during DJF (39.35%), 58 during SON (44.13%), and 20 during JJA (24.59%), while the Amazon-South shows similar patterns with 23 events during MAM (37.98% contribution) and reaches its maximum during SON with 30 events (39.71% contribution).

3.2. Diurnal Cycle Patterns of MCS

3.2.1. Hourly Patterns of Occurrence

The identification of MCS formation processes required an hourly analysis of their occurrence. The frequency of MCS occurrence in their genesis phase is presented in Figure 5 where the X-axis represents the month and the Y-axis represents the local time range. The panels presented correspond to the previously mentioned natural zones.

The Coast-North region experiences numerous events which start at 00:00 to 12:00 LTS in the early morning. The Coast-South region shows minimal activity from 06:00 to 12:00 LTS. The rainy season months from January through April mark the time when this activity takes place in the region. Highlands-North shows scattered activity throughout the time span from 00:00 to 12:00 LTS while events decrease from 12:00 to 18:00 LTS during the two rainy seasons of the inter-Andean region (February–May and September–November). The Highlands-South region experiences MCS events during nighttime periods of the two rainy seasons but with significantly fewer occurrences than the northern region. The Amazon-North region shows uniform activity distribution with most events happening between 06:00 and 12:00 LTS. The Amazon-South region experiences most of its events during the initial part of the day. The two rainy seasons of the Highlands which the ITCZ drives show their peak months according to the results from the monthly frequency analysis. Figure A1, shows a similar analysis by 1 h range.

3.2.2. Duration of MCS as a Function of Time of Occurrence

The lifespan in hours of MCSs in natural zones is shown in Figure 6. The MCSs that start between 15:00 and 22:00 LTS in the Coast-North and Coast-South zones tend to last longer than MCSs that initiate at other times. Coast-North shows greater dispersion in MCS duration compared to Coast-South. The Highlands-North zone experiences its longest MCS events between 11:00 and 18:00 LTS with duration reaching 14 h. The Highlands-South zone experiences its longest MCS events during the morning hours from 07:00 to 10:00 LTS with durations reaching 20 h. The region displays significant variability because different time ranges produce contrasting results. The Amazon-North region maintains uniformity because its median duration stays at 12 h throughout all time periods. Amazon-South exhibits similar heterogeneity to Highlands-South, but its longest duration reaches 15 h when events start between 07:00 and 10:00 LTS.

3.2.3. Duration of MCS as a Function of Time of Occurrence and Station

The seasonal patterns of MCS duration in natural zones are shown in Figure 7. The Coast-North region shows that MCSs with durations of 22 h and 14 h occur in the early morning hours (03:00–06:00 LTS) during the DJF and MAM seasons. However, in the remaining JJA and SON seasons, MCSs of similar duration only occur in the afternoon hours. In the Coast-South region, the data is so scarce that it can only be stated that MCSs exist in the afternoon during the DJF season and at night during MAM. The Amazon-North region maintains equivalent durations throughout all time ranges (mean of 10 h) for all seasons except the 03:00–06:00 LTS range during the DJF season, which has a mean duration of 20 h. The Amazon-South region during DJF matches Amazon-North pattern but with a different period from 07:00 to 10:00 LTS where the median duration reaches 25 h. The MAM season reveals MCS events that last 12 h on average between 11:00 and 22:00 LTS. The JJA season demonstrates minimal MCS occurrence together with short durations. The SON season exhibits extended MCS durations during the 07:00–14:00 and 19:00–22:00 LTS time intervals.

The Highlands-North region experiences MCS events that last approximately 20 h during the 15:00–18:00 LTS period throughout DJF, JJA, and SON seasons. The MAM season features the longest-duration MCS which occurs between 11:00 and 14:00 LTS. The Highlands-South region demonstrates identical patterns as Highlands-North during both DJF and MAM seasons. The JJA and SON seasons produce MCS events with a wider range of lifespans. The Amazon-North region maintains equivalent durations throughout all time ranges at 10 h during all seasons except when the 03:00–06:00 LTS range reaches 20 h during DJF. The Amazon-South region matches Amazon-North patterns during DJF but extends its behavior to the 07:00–10:00 LTS period with a 25 h median duration. The MAM season reveals equal event durations of 12 h throughout the 11:00 to 22:00 LTS period. The JJA season shows minimal MCS occurrence and duration. The MCS duration extends during the SON season across the 07:00–14:00 and 19:00–22:00 LTS time periods.

3.3. Trends on the Occurrence of MCS

3.3.1. Interannual Variability by Region

The time series of MCS occurrences in each natural region from 2001 to 2020 is depicted in Figure 8. The described trend remains stable throughout most zones with 2003 and 2014 marking the highest points. Highlands-North shows a small positive trend, but Amazon-North displays a small negative trend. The visual trends required additional investigation through Mann–Kendall (τ) and Sen’s Slope tests. The results of these tests for the six natural zones appear in

Table 3. Results of the Mann–Kendall (τ) test and Sen’s slope for annual series of MCS occurrence (2001–2020).

Region	Coast		Highlands		Amazon
	$\mathsf{\tau }$	β	$\mathsf{\tau }$	β	$\mathsf{\tau }$	β
North	0.024	0	0.065	0	−0.086	0
South	0.025	0	0.044	0	0.091	0

. Significant values (p-value < 0.05) are underlined.

The Mann–Kendall τ coefficient values in Table 3 show positive trends in five out of six zones (except Amazon-North), but these trends are not statistically significant. Additionally, the Sen’s Slope values are 0, which indicates that in case the trends were significant there is no consistent year-over-year trend in the study period. For this reason, trends are explored on a seasonal basis in the following section.

3.3.2. Seasonal Variability by Region

Table 4. Seasonal trend statistics for MCS occurrence in the natural zones of Ecuador (2001–2020). The values of the Mann–Kendall test (τ) and the slopes estimated by Sen’s Slope (β) are presented for each climatic season: DJF, MAM, JJA, and SON. Significant values (p-value < 0.05) are underlined.

Station	Region	Coast		Highlands		Amazon
		$\mathit{\tau }$	β	$\mathit{\tau }$	β	$\mathit{\tau }$	β
DJF	North	0.309	0.000	0.114	0.000	−0.309	−0.106
	South	0.168	0.000	0.103	0.000	0.166	0.000
MAM	North	0.078	0.000	0.047	0.000	0.012	0.000
	South	−0.092	0.000	−0.006	0.000	0.254	0.000
JJA	North	−0.188	0.000	−0.260	0.000	−0.068	0.000
	South	NA	0.000	0.218	0.000	0.102	0.000
SON	North	−0.054	0.000	0.463	0.143	0.017	0.000
	South	NA	0.000	0.115	0.000	0.000	0.000

presents seasonal trend analysis for the six natural zones over the 2001–2020 period. Seasonal trends provide more information for identifying patterns than the annual analysis in this specific study.

Overall trends in data do not show any clear pattern of change. Seasonal signals are localized to specific seasons and are not systematic. Highlands-North region exhibits a statistically significant positive trend (p-value = 0.009) during the SON season, with a τ of 0.463 and a β of 0.143 MC per year. These values indicate a noteworthy strengthening of convective activity for this season. This suggests a 0.143 increase in events each year. Results suggest a progressive intensification of activity from Highlands-North in SON which can be seen as 1 MCS more every seven years (1/β). In contrast, no statistically significant negative trends were observed. All the southern zones show no significant changes in trend for any season. The same analyses were also conducted for the area of the MCS, but no relevant findings emerged.

3.4. Teleconnections on the Occurrence of MCS

3.4.1. Correlation Between Time Series of Occurrence of MCS and Oceanic Indices

Table 5. Analysis of teleconnections and occurrence of Mesoscale Convective Systems (MCSs) in Ecuador (2001–2020). The table presents the results of Spearman correlations with [109] adjustment between the occurrence of MCS events and macro-climatic indices (Niño 1+2, Nino 3.4, TNI, TNA, TSA, and AMM) in different natural regions of Ecuador (Coast, Highlands, and Amazon), distinguishing between the north and south of the country. Both the original series (S) and their components derived from wavelet decomposition, specifically high-frequency (W1) and low-frequency (V1), are considered. Correlations are presented for each climatic season (DJF, MAM, JJA, SON), as well as for monthly and annual analyses. Relevant values (ρ ≤ −0.4 y ρ ≥ 0.4) and significant values (p-value < 0.05) are underlined.

Study Period	Region	Index	Coast				Highlands				Amazon
			S	S lag	V1	W1	S	S lag	V1	W1	S	S lag	V1	W1
DJF	North	Niño 1+2	0.69	0.25	−0.17	0.18	0.09	0.36	0.19	0.03	−0.16	−0.15	−0.09	−0.13
		TNA	−0.01	0.14	−0.34	0.03	0.33	−0.17	−0.48	−0.19	−0.08	−0.12	0.32	0.12
	South	Niño 3.4	0.30		0.12	0.43	−0.27	−0.13	−0.09	−0.43	−0.05	0.08	0.29	0.17
		TNA	0.23		−0.38	0.24	0.00	0.26	0.49	−0.29	−0.16	−0.13	−0.06	−0.29
MAM	North	Niño 1+2	0.45	0.36	0.20	0.19	0.09	−0.35	0.56	0.05	0.32	0.13	0.33	−0.34
		Niño 3.4	0.05	0.13	0.50	−0.10	0.06	0.00	0.39	−0.34	0.02	−0.12	0.32	−0.41
	South	Niño 1+2	0.30	0.48	−0.06	0.21	0.37	0.05	0.02	0.05	0.54	0.13	0.18	−0.01
		TNI	0.34	0.13	−0.38	0.44	0.61	0.25	0.01	0.38	0.49	0.15	−0.23	0.12
JJA	North	TNI	0.43	−0.04	0.52	0.18	−0.51	−0.04	0.17	0.17	−0.18	0.02	0.40	−0.33
	South	Niño 1+2		−0.17			−0.16	0.42	−0.44	0.06	−0.10	0.41	−0.23	0.16
		TNI		−0.22			0.02	0.25	−0.58	0.13	−0.02	0.23	−0.42	0.03
		TSA		0.25			0.56	0.27	−0.28	0.28	−0.14	0.11	−0.13	−0.38
SON	North	Niño 1+2	0.61	0.28	0.07	0.26	0.47	0.19	0.20	0.49	−0.18	−0.22	−0.09	−0.36
		Niño 3.4	0.41	0.07	0.27	0.13	0.43	0.33	0.16	0.40	−0.17	−0.01	−0.11	−0.20
		AMM	−0.24	0.16	−0.09	0.04	−0.19	−0.27	−0.51	−0.12	0.03	−0.26	−0.18	0.09
ANNUAL	North	Niño 1+2	0.59	0.54	0.01	0.27	0.21	0.06	0.47	0.10	−0.03	0.06	0.18	−0.35
		TNI	0.32	0.33	−0.38	0.17	0.10	−0.10	0.26	0.05	0.07	0.15	0.73	−0.01
	South	TNI	0.26	0.26	−0.60	0.14	0.38	0.34	0.03	0.36	0.26	0.21	−0.70	0.08

shows Spearman correlations between oceanic indices and MCS occurrence time series and their first-level decomposition for the six natural zones. DJF season shows the highest correlation value (0.69) of the original time series with the Nino 1+2 index in the Coast-North region. The relationship continues across the other seasons with significant values of 0.61 SON and also occurs on an annual basis (0.59). The TNI showed relevant correlations of 0.43 that affect JJA. Coast-South has a direct link between El Niño 1+2 (0.54) and the S lag during annual scale.

The Highland-North region shows a higher complexity in the incidences. This region shows positive correlations with Niño 1+2 (0.47) and Niño 3.4 (0.43) during SON. In contrast, the Highland-South region has a strong direct correlation of 0.61 with TNI and TSA (0.56) and S during MAM and JJA, respectably. S lag has direct relation with 1+2 (0.42) over JJA. Amazon-North does not show any correlation significative or relevant with S or S lag. In parallel, Amazon-South has a relation with Niño 1+2 (0.54) over MAM and S. The S lag does not have any significant or relevant correlation.

3.4.2. Correlation Between Wavelet Decomposed MCS Occurrence Time Series and Oceanic Indices

Coast-North shows a significant response in the low-frequency component through a positive relationship with Niño 3.4 (0.50) over MAM and TNI (0.52) in JJA. Coast-South is only related to Niño 3.4 (0.43) during DJF and TNI (0.44) over MAM for high-frequency events. However, the low frequency is strongly anti-correlated (−0.60) in the annual period to the TNI and AMM (−0.54) over DJF.

The Highlands-North region demonstrates advanced complexity in its teleconnection patterns. The region shows positive correlations with Niño 1+2 (0.56) and V1 during DJF. The AMM demonstrates a negative connection (−0.51) to MCS events in the SON season. The Highlands-North region displays negative connections with TNA (−0.48) during DJF for low-frequency signals. The high-frequency component demonstrates positive correlations with Niño 1+2 (0.49) during the SON season. The Highlands-South region demonstrates a 0.49 with TNA values and V1 during DJF. The Highlands-South region displays negative connections in the lower frequency with the TNI (−0.56) index during JJA. The high frequency only interacts with Niño 1+2 (−0.43) in DJF for the region.

The low-frequency component of Amazon-North shows a strong positive connection with TNI (0.73) during the annual scale. The high frequency displays a negative correlation only with Niño 3.4 (−0.41) during MAM. On the other hand, the negative relationship in V1 is manifested through a strong TNI (−0.70) in annual scale. The high-frequency component does not have any relevant correlation for the region. The table of all the correlations is shown on

Table A1. Complete analysis of teleconnections and occurrences of Mesoscale Convective Systems (MCSs) in Ecuador (2001–2020). The table presents the results of Spearman correlations [109] with adjustment between the occurrence of MCS events and macro-climatic indices (Niño 1+2, Nino 3.4, TNI, TNA, TSA, and AMM) in different natural regions of Ecuador (Coast, Highlands, and Amazon), distinguishing between the north and south of the country. The original series (S), lag series (S lag), and S components derived from wavelet decomposition, specifically high frequency (W1) and low frequency (V1), are considered. Correlations are presented for each climatic season (DJF, MAM, JJA, SON), as well as for monthly and annual analyses. Relevant values (ρ ≤ −0.4 y ρ ≥ 0.4) and significant values (p-value < 0.05) are underlined.

Study Period	Region	Index	Coast				Highlands				Amazon
			S	S lag	V1	W1	S	S lag	V1	W1	S	S lag	V1	W1
DJF	North	Niño 1+2	0.69	0.25	−0.17	0.18	0.09	0.36	0.19	0.03	−0.16	−0.15	−0.09	−0.13
		Niño 3.4	0.18	0.35	−0.20	0.24	−0.10	0.17	−0.17	0.10	−0.32	−0.12	−0.09	−0.04
		TNI	0.17	−0.10	0.03	−0.26	0.15	−0.10	0.25	−0.05	0.32	0.02	0.24	0.17
		TNA	−0.01	0.14	−0.34	0.03	0.33	−0.17	−0.48	−0.19	−0.08	−0.12	0.32	0.12
		TSA	0.28	0.27	0.00	0.26	0.01	0.15	0.02	0.05	−0.24	0.12	0.02	−0.02
		AMM	−0.18	−0.11	−0.22	−0.17	0.27	−0.42	−0.54	−0.11	0.09	−0.09	0.38	0.18
	South	Niño 1+2	0.33		0.25	0.29	0.09	0.02	−0.02	−0.19	0.12	0.23	0.40	−0.02
		Niño 3.4	0.30		0.12	0.43	−0.27	−0.13	−0.09	−0.43	−0.05	0.08	0.29	0.17
		TNI	0.02		−0.29	−0.26	0.35	0.22	0.23	0.36	0.03	0.02	−0.18	−0.15
		TNA	0.23		−0.38	0.24	0.00	0.26	0.49	−0.29	−0.16	−0.13	−0.06	−0.29
		TSA	0.23		0.11	0.32	−0.12	−0.23	−0.15	−0.24	−0.10	−0.02	0.13	0.24
		AMM	0.04		−0.54	−0.04	0.07	0.23	0.40	−0.05	−0.18	−0.19	−0.12	−0.23
MAM	North	Niño 1+2	0.45	0.36	0.20	0.19	0.09	−0.35	0.56	0.05	0.32	0.13	0.33	−0.34
		Niño 3.4	0.05	0.13	0.50	−0.10	0.06	0.00	0.39	−0.34	0.02	−0.12	0.32	−0.41
		TNI	0.28	0.27	−0.24	0.40	0.15	−0.13	0.32	0.29	0.32	0.08	0.17	0.09
		TNA	−0.20	0.10	0.09	−0.13	0.04	0.24	0.20	−0.29	−0.28	0.04	0.21	0.14
		TSA	0.07	0.28	−0.24	0.23	0.14	−0.20	−0.11	0.24	−0.04	0.15	0.07	0.16
		AMM	−0.27	0.04	−0.02	−0.14	0.04	0.30	0.06	−0.40	−0.20	0.10	0.09	0.10
	South	Niño 1+2	0.30	0.48	−0.06	0.21	0.37	0.05	0.02	0.05	0.54	0.13	0.18	−0.01
		Niño 3.4	0.07	0.40	0.35	−0.19	−0.12	−0.40	0.14	−0.26	0.16	−0.07	0.32	−0.19
		TNI	0.34	0.13	−0.38	0.44	0.61	0.25	0.01	0.38	0.49	0.15	−0.23	0.12
		TNA	0.23	0.06	0.31	−0.07	−0.05	−0.33	0.31	−0.22	0.09	−0.12	−0.09	−0.31
		TSA	0.36	−0.01	0.07	0.04	0.16	0.09	−0.07	−0.10	0.14	0.12	0.13	−0.05
		AMM	0.12	−0.02	0.20	−0.07	−0.04	−0.20	0.21	−0.14	−0.09	−0.22	−0.32	−0.35
JJA	North	Niño 1+2	0.44	0.09	0.37	0.08	−0.21	0.13	−0.06	0.23	−0.19	0.23	0.18	−0.24
		Niño 3.4	−0.03	0.30	0.06	−0.14	0.23	−0.01	−0.33	−0.08	−0.11	0.06	−0.37	0.16
		TNI	0.43	−0.04	0.52	0.18	−0.51	−0.04	0.17	0.17	−0.18	0.02	0.40	−0.33
		TNA	−0.10	−0.31	0.30	−0.14	−0.35	−0.10	0.33	0.10	0.02	−0.39	0.00	−0.30
		TSA	−0.04	−0.04	0.02	0.01	−0.13	−0.08	0.06	0.32	−0.01	0.04	0.09	−0.32
		AMM	−0.14	−0.29	0.32	−0.06	−0.29	−0.17	0.15	−0.11	0.05	−0.28	−0.15	−0.15
	South	Niño 1+2		−0.17			−0.16	0.42	−0.44	0.06	−0.10	0.41	−0.23	0.16
		Niño 3.4		−0.02			−0.21	0.24	0.23	−0.12	0.00	0.05	0.20	0.39
		TNI		−0.22			0.02	0.25	−0.58	0.13	−0.02	0.23	−0.42	0.03
		TNA		0.03			0.38	0.07	−0.09	0.24	0.29	0.02	−0.23	−0.20
		TSA		0.25			0.56	0.27	−0.28	0.28	−0.14	0.11	−0.13	−0.38
		AMM		0.04			0.17	−0.07	0.04	0.08	0.26	0.00	−0.31	−0.03
SON	North	Niño 1+2	0.61	0.28	0.07	0.26	0.47	0.19	0.20	0.49	−0.18	−0.22	−0.09	−0.36
		Niño 3.4	0.41	0.07	0.27	0.13	0.43	0.33	0.16	0.40	−0.17	−0.01	−0.11	−0.20
		TNI	0.24	0.13	−0.34	0.19	0.07	−0.15	−0.30	0.11	−0.07	−0.19	0.18	−0.01
		TNA	0.07	0.18	0.02	0.38	0.04	−0.29	−0.37	0.15	−0.05	−0.31	−0.12	0.12
		TSA	−0.11	−0.27	0.42	0.27	−0.12	−0.09	0.37	0.32	0.32	0.21	−0.08	0.23
		AMM	−0.24	0.16	−0.09	0.04	−0.19	−0.27	−0.51	−0.12	0.03	−0.26	−0.18	0.09
	South	Niño 1+2					−0.15	−0.22	0.32	−0.30	0.04	−0.43	−0.24	0.15
		Niño 3.4					−0.12	−0.41	0.36	−0.16	−0.05	−0.28	−0.33	0.05
		TNI					0.16	0.23	−0.16	−0.03	−0.03	−0.27	0.00	0.05
		TNA					0.21	0.00	0.07	−0.10	−0.06	−0.09	−0.32	0.12
		TSA					0.15	0.19	−0.34	−0.08	0.03	−0.25	0.18	−0.11
		AMM					0.20	0.11	−0.07	−0.16	0.12	0.19	−0.05	−0.03
MONTHLY	North	Niño 1+2	0.29	0.09	0.05	0.05	0.08	0.04	0.08	−0.02	0.04	−0.05	−0.04	0.01
		Niño 3.4	0.02	0.10	0.11	0.00	0.02	0.06	0.05	−0.04	−0.03	−0.11	−0.12	0.00
		TNI	0.22	−0.02	−0.09	0.03	0.09	−0.01	0.04	−0.01	0.09	0.04	0.05	0.00
		TNA	−0.26	−0.01	0.00	−0.02	−0.03	−0.14	−0.21	−0.02	−0.07	−0.16	−0.19	0.05
		TSA	0.16	0.08	0.11	0.05	0.02	0.02	0.08	0.03	0.02	0.08	0.12	−0.04
		AMM	−0.31	−0.23	−0.21	−0.01	0.06	−0.21	−0.32	−0.01	0.05	−0.17	−0.21	0.00
	South	Niño 1+2	0.15	0.01	−0.05	0.07	0.01	0.08	0.12	0.01	0.05	0.01	−0.01	−0.02
		Niño 3.4	0.05	0.02	0.01	0.02	−0.20	−0.11	−0.11	0.01	−0.07	−0.11	−0.14	0.00
		TNI	0.09	−0.03	−0.05	0.04	0.27	0.18	0.19	0.01	0.13	0.07	0.08	0.01
		TNA	0.00	−0.01	0.04	−0.02	−0.07	−0.14	−0.13	0.06	0.03	−0.08	−0.15	−0.02
		TSA	0.11	0.11	0.04	−0.04	0.04	0.08	0.12	−0.02	−0.02	0.10	0.13	0.01
		AMM	−0.02	−0.09	0.03	0.04	0.05	−0.18	−0.19	0.06	0.07	−0.12	−0.18	0.00
ANNUAL	North	Niño 1+2	0.59	0.54	0.01	0.27	0.21	0.06	0.47	0.10	−0.03	0.06	0.18	−0.35
		Niño 3.4	0.29	0.23	0.38	0.16	0.18	0.14	0.08	0.13	−0.23	−0.04	−0.31	−0.22
		TNI	0.32	0.33	−0.38	0.17	0.10	−0.10	0.26	0.05	0.07	0.15	0.73	−0.01
		TNA	−0.16	−0.13	0.16	0.14	0.02	−0.18	−0.27	−0.04	−0.10	0.03	0.18	−0.16
		TSA	0.09	0.10	0.15	0.24	−0.19	−0.16	0.26	0.26	−0.01	−0.13	−0.08	−0.02
		AMM	−0.23	−0.15	−0.05	0.03	0.02	−0.01	−0.39	−0.14	0.05	0.12	0.30	−0.05
	South	Niño 1+2	0.13	0.13	−0.26	0.08	0.26	0.37	−0.08	0.19	0.26	0.34	0.04	0.13
		Niño 3.4	−0.02	−0.02	0.28	−0.13	−0.19	−0.12	−0.01	−0.16	−0.13	−0.02	0.43	0.05
		TNI	0.26	0.26	−0.60	0.14	0.38	0.34	0.03	0.36	0.26	0.21	−0.70	0.08
		TNA	0.14	0.14	0.24	0.06	0.05	0.02	0.32	−0.06	0.10	−0.07	−0.24	−0.13
		TSA	0.18	0.18	−0.15	−0.04	0.18	0.09	−0.31	0.01	0.12	0.09	0.07	0.01
		AMM	0.16	0.16	0.06	0.09	0.14	0.06	0.24	−0.02	0.03	−0.18	−0.42	−0.28

. The series of S, S lag, V1, and W1 with relevant values of Spearman’s correlation and their corresponding indices are shown in

Table A2. Complete analysis of the relationship between the Trans-Niño Index (TNI) and atmospheric variables from ERA5 reanalysis in Ecuador (2001–2020). The table presents Spearman correlations between the occurrence of Mesoscale Convective Systems (MCSs) and key dynamic and thermo-dynamic variables at 800 hPa and 500 hPa (u, v, w, q, Δu, Δv, WSM, CIN, and CAPE), evaluated across the six natural regions of Ecuador (Coast, Highlands, and Amazon), distinguishing between north and south sectors. Both the original series (S) and their components derived from wavelet decomposition, low frequency (V1) and high frequency (W1).

Study Period	Region	Variable	Coast			Highlands			Amazon
			S	V1	W1	S	V1	W1	S	V1	W1
ANNUAL	North	u800	0.16	−0.27	−0.14	0.43	0.31	0.11	−0.07	−0.48	0.19
		v800	0.11	0.05	−0.12	−0.15	−0.20	−0.04	−0.18	−0.48	0.02
		w800	0.01	−0.04	0.01	0.56	0.35	−0.07	0.55	0.46	0.11
		q800	−0.28	−0.31	−0.09	−0.16	−0.34	−0.27	−0.19	−0.32	−0.19
		u500	−0.26	0.27	0.01	−0.06	0.46	−0.01	−0.09	0.58	−0.06
		v500	0.08	−0.30	0.00	−0.16	−0.48	−0.01	−0.29	−0.60	−0.03
		w500	−0.61	0.28	0.00	−0.73	−0.45	−0.12	−0.31	0.14	−0.34
		q500	0.58	−0.08	0.08	0.44	−0.06	−0.01	0.14	−0.23	−0.08
		∆u	−0.17	0.40	0.00	−0.12	0.46	−0.01	0.01	0.74	−0.02
		∆v	−0.25	−0.24	0.07	−0.19	−0.40	0.08	−0.02	0.01	−0.04
		WSM	0.15	−0.42	−0.03	0.05	−0.50	−0.01	0.32	−0.16	0.01
		CIN	−0.05	0.17	0.16	0.00	0.02	0.08	0.13	0.40	0.01
		CAPE	0.26	−0.31	0.06	0.23	−0.23	0.19	0.30	0.38	0.10
	South	u800	0.26	−0.25	−0.26	0.51	0.40	0.14	0.14	−0.48	0.10
		v800	−0.08	0.30	−0.10	0.21	0.14	0.14	−0.13	−0.44	0.03
		w800	0.34	0.58	−0.11	0.08	0.36	−0.05	0.63	0.43	0.10
		q800	−0.32	−0.51	−0.01	−0.21	−0.45	−0.12	−0.21	−0.35	−0.22
		u500	−0.25	0.23	0.04	−0.08	0.45	0.02	−0.05	0.57	−0.03
		v500	0.09	0.08	0.03	0.11	0.03	−0.01	−0.12	−0.41	−0.03
		w500	−0.62	0.24	−0.10	−0.59	−0.48	−0.31	−0.18	0.21	−0.11
		q500	0.51	−0.05	0.05	0.37	−0.16	0.09	0.16	−0.30	−0.07
		∆u	−0.31	0.25	0.09	−0.16	0.31	0.02	−0.06	0.64	0.01
		∆v	0.09	0.01	0.14	0.09	0.03	−0.04	0.09	0.16	0.00
		WSM	0.33	−0.22	−0.04	0.15	−0.28	−0.07	0.38	−0.13	−0.05
		CIN	−0.28	−0.14	−0.09	0.18	−0.17	0.01	0.28	0.53	0.05
		CAPE	0.14	−0.35	0.08	0.15	−0.07	0.13	0.22	0.24	−0.11

. The corresponding plots of relevant and significant correlations are shown in Figure A2.

3.4.3. TNI Influence on Climatic Variables

The TNI influence shows a heterogenic influence according to the natural zones in the annual scale (

Table 6. Analysis of the relationship between the Trans-Niño Index (TNI) and atmospheric variables from ERA5 reanalysis in Ecuador (2001–2020). The table presents Spearman correlations between the occurrence of Mesoscale Convective Systems (MCSs) and key dynamic and thermodynamic variables at 800 hPa and 500 hPa (u, v, w, q, Δu, Δv, WSM, CIN, and CAPE), evaluated across the six natural regions of Ecuador (Coast, Highlands, and Amazon), distinguishing between north and south sectors. Both the original series (S) and their components are derived from wavelet decomposition, low frequency (V1) and high frequency (W1). Relevant values ($\rho$ <= −0.4 y $\rho$ >= 0.4) and significant values (p-value < 0.05) are underlined.

Study Period	Region	Variable	Coast			Highlands			Amazon
			S	V1	W1	S	V1	W1	S	V1	W1
ANNUAL	North	u500	−0.26	0.27	0.01	−0.06	0.46	−0.01	−0.09	0.58	−0.06
		w500	−0.61	0.28	0.00	−0.73	−0.45	−0.12	−0.31	0.14	−0.34
		q500	0.58	−0.08	0.08	0.44	−0.06	−0.01	0.14	−0.23	−0.08
		∆u	−0.17	0.40	0.00	−0.12	0.46	−0.01	0.01	0.74	−0.02
	South	w800	0.34	0.58	−0.11	0.08	0.36	−0.05	0.63	0.43	0.10
		u500	−0.25	0.23	0.04	−0.08	0.45	0.02	−0.05	0.57	−0.03
		w500	−0.62	0.24	−0.10	−0.59	−0.48	−0.31	−0.18	0.21	−0.11
		q500	0.51	−0.05	0.05	0.37	−0.16	0.09	0.16	−0.30	−0.07
		∆u	−0.31	0.25	0.09	−0.16	0.31	0.02	−0.06	0.64	0.01

). Coastal regions (Coast-North and Coast-South) increased humidity at 500 hPa with q500 (0.58) and q500 (0.51), respectably. Additionally, a negative influence over the vertical motion at 500 hPa with w500 (−0.61) and w500 (−0.62) is noted. The Highlands’ region is characterized by an enhanced upward motion, over the 500 hPa layer with w500 (−0.73) and w500 (−0.59) over Highlands-North and Highlands-South, respectably. The Amazon region has the most complex response to TNI. In parallel, Amazon-North and Amazon-South have a big zonal wind influence at 500 hPa with u500 (0.58), ∆u (0.74), and u500 (0.57), and with ∆u (0.64), respectably. However, Amazon-South has the influence of w800 (0.63) over 800 hPa which is unique among all the regions. The complete analysis is shown in Figure A2.

4. Discussion

4.1. Database of the Occurrence of MCS

This research considered only data from MCSs in the genesis stage. The database from [66] provides South American MCS data through multistep methodology. This involves tracking using ForTraCC [115,116], implementing thermal and morphological thresholds from [68]. Validation with observational data which ensure both quality and extent for MCS characterization in Ecuador is also involved. This makes MCS spatial and temporal identification reliable. The database benefits from its satellite imagery origin. This approach has been used in multiple MCS tracking studies [19,68,72].

The database demonstrates spatial coherence by showing strong convective activity in areas with high moisture and thermal gradients. Amazon-North and Highlands-North zones have the highest concentration of MCSs while the Coast-South zone has the lowest occurrence, as shown in Figure 3 and presented in Table 1. This spatial distribution occurs because Amazon-North acts like a host spot, receiving benefits from moisture overlap from Atlantic Ocean humidity transported via OLLJ, and local evapotranspiration meets eastern Andes orographic lifting, creating ideal thermodynamic conditions for genesis of MCSs. Highlands-North experiences enhanced convection due to its narrower cordillera configuration. This allows better moisture penetration from Amazon-North and stronger nocturnal cooling gradients. In contrast, Coast-South shows minimal activity. The cold Humboldt Current creates stability, while the CLLJ core trajectory remains too distant to provide significant moisture transport to this region.

The proposed regionalization does not conflict with the division by [42]; the higher concentration of MCSs in the Highlands-North and Amazon-North is consistent with the findings of several authors who highlight the strong convective activity in these regions [9,19,117]. There is a clear agreement with the patterns observed in satellite sources such as TRMM and CHIRPS [78,94].

The focus on MCSs in this research provides advantages over precipitation studies by isolating convective mechanisms from other rainfall sources. While precipitation studies aggregate all rainfall types, our methodology threshold only includes MCSs. The detection of specific teleconnections and trends over convection is unseen and remain hidden in bulk precipitation analyses.

4.2. Monthly and Seasonal Behavior on the Occurrence of MCSs

The monthly and seasonal analysis reveal unprecedented sub-national variability in MCS contribution to precipitation shown in Table 2 and illustrated in Figure 4. This research is the first quantitative assessment of MCS spatial heterogeneity distribution within Ecuador, differing from precipitation studies to isolate convective mechanisms and their drivers. While, previous studies have identified general connections between total precipitation and climate patterns [10,11,118,119], our MCS analysis demonstrates that convective system contributions vary dramatically from 16.41% to 44.13% across regions, detailed in Table 2.

The Coastal zone follows a unimodal regime which produces its maximum MCS event accumulation during the DJF season because of ITCZ and Pacific anticyclone effects that enable favorable convection [13,120,121] and the influence of the cold Humbolt Current on the stabilization and inhibition of convection for the rest of the year [50]. This unimodal pattern occurs because the seasonal ITCZ migration controls moisture availability. During the first part of the year (DJF-MAM), the ITCZ is positioned over the north of Ecuador and enhances convergence at a low level, creating optimal convective conditions. During the second part of the year, the ITCZ’s movement to the north removes this convergence; meanwhile, the cold Humboldt Current strengthens coastal stability and, as a result, convective development is capped.

Coast-North exhibits the highest convective activity in the first months of the year with 35 events in April, yet Coast-South experiences only 10 events in March as illustrated in Figure 4, which demonstrates the distinct weather patterns between these two Pacific regions. This sharp contrast exists because the CLLJ trajectory passes directly through Coast-North, delivering massive moisture fluxes and creating low-level wind that create confluence with trade winds. In parallel, Coast-South remains far from the CLLJ’s influence while experiencing the already-explained Humboldt Current effects that create more stable atmospheric conditions and suppress convective initiation. The Chocó Low-Level Jet (CLLJ) may function as the main driver which controls the strong convective activity observed in the Coast-North region. Reference [122] identifies the jet as a major factor that produces extensive sustained cool and moist westerly winds to collide with the Andes range and trade winds resulting in atmospheric instability which [123] documented.

CLLJ effects in continental precipitation studies [122,123] are treated as spatially uniform. Our scope over MCS analysis reveals sharp contrast alongside the north and south within Ecuador’s coast detailed in Figure 4. While Coast-North shows 80 MCS events during MAM, contributing 26.90% to regional precipitation, Coast-South exhibits only 12 MCS events, contributing 19.32% as presented in Table 2. This key aspect reveals the geographic limits on the CLLJ influence over Ecuador’s coast. Additionally, this demonstrates that MCS genesis is geographically constrained and, rather than following general precipitation patterns, reveals mechanisms invisible to rainfall analyses.

The SNGR natural disaster database shows intense convective activity through three documented flood events that occurred due to rainfall in Rioverde on 5 May 2010, Quinindé on 6 March 2019, and San Lorenzo on 18 February 2023 [14]. Ref. [15] established a link between the severe MCS which triggered the January 2016 Esmeraldas flood through the CLLJ–Andes interaction that produced a genesis zone before the system shifted towards Coast. Coast-South shows lower levels of activity throughout its entire territory. The Humboldt Current generates atmospheric stability that leads to fog and drizzle formation instead of convection in this zone [50]. During its peak season, the CLLJ core remains distant from this area which prevents it from influencing the climate of this region.

The Highlands-North, Highlands-South, Amazon-North, and Amazon-South regions display bimodal MCS occurrence patterns which matches previous precipitation research findings [13,124]. The convective accumulation in DJF, MAM, and SON seasons reaches its highest levels during MAM because the ITCZ reaches its peak position over the equatorial zone in April [125] while JJA shows the lowest MCS activity when the ITCZ reaches its northernmost position. The seasonal convection patterns emerge from the combination of Amazonian moisture and atmospheric circulation with topographic interactions [11,121,126]. The Peru distinction helps explain MCS formation in the Andes through dynamic factors yet thermodynamic factors control Amazonian convection [11]. The nocturnal convection process described by [28] combines Andean katabatic flows with Amazonian warm air to produce ideal conditions for nocturnal convection. The highest MCS accumulation occurs in this zone according to [127], who used radar data to show that 63.5% of precipitation-generating air masses originate from the Ecuadorian Amazon. The MCS-producing Andes–Amazon interaction exists throughout the mountain range according to [9], who studied the Amazon Basin as part of this pattern.

The contribution to precipitation from MCSs over Ecuador is 26% across all the regions. The contribution is comparable to regional estimates in the 40–46% range found in [8], where a less restrictive size threshold of 2000 km² was used instead of the 40,000 km² threshold. Despite the marked difference, the results of this research provide the first detailed quantification of MCS contribution to precipitation at a sub-national scale for Ecuador.

The spatiotemporal heterogeneity is pronounced at the longitudinal level (Amazon 33–38% vs. Coast 16–21% annually) and demonstrates the critical role of orography in MCS formation [13,126,128]. The latitudinal difference is also evident and reflected in Table 2, where, for example, Amazon-North (207, 38.09%) has consistently higher occurrence and percentage contribution to precipitation compared to Amazon-South (93, 33.65%). Finally, Amazon is ratified as a convective hotspot [19], with consistent 35–40% MCS contribution throughout most seasons.

The MAM season stands as the period with the most significant national MCS occurrence during the years 2001–2020 with 43% of the total events (276). Thus, in Figure 9 the main climatic features during this season are shown, primarily the ITCZ which extends its southern-most point to reach the equatorial region [120,125,129]. The Pacific coast moisture fluxes receive their influence from the Niño current’s warm phase which occurs between December and May [50]. The CLLJ maintains its convective activity in northwestern zones of Ecuador, although it has not reached its peak strength during SON [122,123,130]. The eastern part of the country experiences convective conditions because of vegetation evapotranspiration [121,131]. The vapor produced through this process added to the Atlantic Ocean moisture transport, and possibly the wind shear provided by the OLLJ, although its strength remains lower than during its peak in DJF [132,133]. The abrupt vertical motion of the region’s orography forces available moisture to rise, which produces nocturnal convection [28]. The atmosphere of MAM becomes so unstable and full of moisture that any local convergence can initiate MCS formation. It is important to observe that the moisture transport studied from the rainwater isotopic composition for Ecuador [134] may influence rainfall, but further studies are necessary to link its influence to the development of MCSs.

4.3. Diurnal Cycle and Duration of MCSs

MCS occurrence follows a daily pattern which shows how different atmospheric conditions at local and regional levels interact [135], detailed in Figure 5. The Coastal and Highlands’ regions show standard patterns of daytime and nighttime variations. Coast-North shows a specific pattern from 04:00 to 08:00 LTS (Figure 5a), which results from a nighttime land and sea temperature gradient that transports oceanic moisture through sea breeze flows and nocturnal cooling inland, creating convective conditions [2,82,136]. The observed behavior matches the findings of [8], who found that MCS events on the Pacific coast of Colombia reach their highest frequency during the period between 00:00 and 06:00 LTS. Highlands (Figure 5b,e) show a peak in the morning hours between 04:00 and 06:00 LTS that matches the short precipitation events observed in the Andes during the morning [128]. The Coast-North and Highlands patterns stem directly from nocturnal convection which the CLLJ triggers. The system transports Pacific moisture into coastal convection areas while creating an orographic lift that fuels Andean convection [15,122,130,137].

The Amazon-North (Figure 5c) zone demonstrates that MCS events occur frequently because of its flat terrain combined with its humid climate. The uniformity of conditions that support convection leads to this homogeneity according to [9,138]. The Amazon-South (Figure 5f) region displays a unique bimodal pattern because MCS events occur most frequently during early morning hours and midday hours (00:00–02:00 and 10:00–12:00 LTS). The different MCS occurrence times in these Amazonian zones stem from variations in Atlantic Ocean moisture transport and persistence which affect the timing of MCS events [10].

MCSs exist as structures which maintain direct connections to energy sources and the environmental factors that influence their duration [70]. The study reveals different patterns of MCS behavior. The longest-duration MCS occurs in Coastal and Highlands regions (North–South) after midday until the early evening hours (15:00–22:00 LTS) according to [139]. The Highlands region shows morning MCS origins, but the study confirms this pattern through [9] who state that MCS maturation takes 2.2 to 5.2 h after their initial formation. The time-lag phenomenon emerges because of strong Convective Available Potential Energy (CAPE) values that develop through diurnal heating processes. Research shows that high CAPE values directly lead to stronger MCSs [140], and tropical mountainous regions produce the highest global CAPE values [141]. The study areas demonstrate suitable conditions for high CAPE values which explains the existence of such long-lasting MCSs.

A pronounced contrast exists in Amazon-North, where a homogeneity in MCS durations across all time slots, often exceeding 12 h, confirms that the Amazon, with its humid and warm conditions (independent of the solar cycle), is an ideal scenario for the formation and maintenance of long-duration MCS [9]. Another observed behavior is the interaction between the adjacent Highlands–Amazon (South) zones, which have their highest activity during the early morning and night, particularly in the early morning (01:00–04:00 LTS). This behavior is generated by late-night katabatic convergence in concave terrain, which manifests as the collision of warm air from the Amazon with the freezing air from the slopes of the Andes, capable of sustaining convection for hours [28,139]. Another larger-scale mechanism that fuels convection in the Ecuadorian Amazon is the Orinoco Low-Level Jet (OLLJ), which transports copious amounts of moisture from the Atlantic Ocean or the Caribbean Sea to the Amazon-North Basin during the night [133]. Indeed, [132] demonstrated that the OLLJ reaches its maximum intensity during the DJF season, where it fuels the longest-duration MCS in a staggered manner, affecting both Amazon-North (03:00–06:00 LT) and Amazon-South (07:00–10:00 LT).

The seasonal patterns of the zones create additional challenges for understanding how long MCS events last and how intense they become. Highlands–Amazon (South) regions experience two daily periods of long-lived MCS generation during the DJF season which is the most active period for most regions between 07:00–10:00 and 15:00–18:00 LTS. The nocturnal convergence studied by [28] becomes stronger because of the moisture exchange between the Amazon and Andes region which [121] discovered; the afternoon window corresponds to the highest CAPE values that support convection. The seasonal conditions enable MCSs to move between different areas and affect surrounding regions [142]. Highlands-North experiences its peak precipitation during afternoon hours during DJF and MAM and SON seasons according to multiple studies that show annual precipitation peaks between 13:00 and 19:00 LTS [128,143,144], which demonstrates both seasonal and regional differences in Ecuador.

4.4. Interannual Variability and Trends on the Occurrence of MCSs

The MCS occurrence trend throughout the 2001–2020 period does not present an unmistakable pattern in the monthly analysis (Figure 8). The different convective dynamics in Ecuador stem from its transitional Andes–Amazon zones and orography, together with influences from both the Pacific and Atlantic Oceans. In contrast, seasonal analysis reveals distinct patterns of convective system occurrence, providing more detailed observations than monthly perspective. While the Highlands-North region shows no significant trend on the monthly analysis, it exhibits substantial positive trends on the seasonal scale during SON (τ = 0.463, β = 0.143) as presented in Table 4.

This unique positive trend occurs because warming conditions in Highlands-North enhance atmospheric thermodynamics. The instability creates higher CAPE values that favor MCS occurrence. The observed increase in MCS events provides the convective mechanism underlining precipitation and extreme temperature patterns documented by [145] in Quito Ecuador. While Ref. [145] identified changes in precipitation, our scope is on MCSs and reveals that these changes in precipitation are directly result of increased organized convective activity. MCS occurrence and intensity directly correlate with changes in atmospheric thermodynamics and stability alongside increased moisture transport from wind (advection or convection) [18,140,146]. This convective-focused approach advances beyond bulk precipitation studies by isolating the mesoscale processes driving regional climate changes that precipitation-based analyses cannot distinguish. The remaining regions do not show significant trends in their respective seasonal periods.

The trend must be considered within a broader context of climate change over MCSs. The increase in MCS occurrence stands in opposition to the findings of [10] and the projected decline of MCS occurrence in Amazon basin according to [19]. However, it is aligned to documented intensification of MCS activity and area in future climate projection, related to warming condition and higher CAPE values in South America [20]. Additionally, [18] shows increased MCS frequency and duration in convection-permitting models related to bigger moisture amounts available in lower levels of the atmosphere. Ref. [12] finds similar drivers and triggers on the organization of convective pattern in the northwest of South America. These contrasting findings, alongside our MCS-specific results, highlight the importance of knowing the orographic and local climate factors in control of MCS formation at each location.

4.5. MCS Teleconnections Patterns

4.5.1. Temporal Series Teleconnection

The extracted teleconnections match previous climatological findings on rainfall for Ecuador and demonstrate multiple oceanic forcing mechanisms, presented in Table 5. The Pacific Ocean functions as the primary climate driver because of its close location to the coastal region. The Niño 1+2 sea surface temperature anomalies have a positive correlation with MCS activity during SON or annually (0.61 or 0.59) in the Coast-North region. Furthermore, the strongest connection between MCS frequency and the Niño 1+2 (Figure 10a) index occurs during DJF over Coast-North (0.69). The Coast-North region of Ecuador experiences increased MCS occurrences and elevated rainfall during the warm phase of Niño 1+2 according to multiple studies, including [15,54,94,147].

Teleconnections vary according to the study region; indeed, the Atlantic and Pacific Oceans exhibit complex signals in Highlands and Amazonia. Specifically, the Highlands-North, Highlands-South, and Amazon-South regions display complex signals influenced by the Atlantic Ocean, with a strong TSA (0.56) during the dry season of JJA (Figure 10b). Our TSA finding is consistent with [25], who revealed the TSA controls over rainfall patterns in the southern Andes. In parallel, these zones are also influenced by the Pacific Ocean through positive influences of TNI (0.61) (Figure 10c) during the rainy seasons of SON. Although the Pacific Ocean lies close to the Highlands and Amazon, the heterogeneous division of indicators is marked by the disruptive presence of the Andes cordillera, which acts as a physical wall. In addition, Highlands and Amazon regions are also under the influence of moisture transport generated on the eastern slope by the immense vegetation in the Amazon Basin and by the OLLJ, which carries moisture from the Atlantic Ocean [117,132]. In parallel, the CLLJ transports moisture from the Pacific along the western slope [122,123]. The granularity in the teleconnections and the Atlantic Ocean’s influence is consistent with the literature [24] in pioneering fashion, which observed that the eastern Andes were clearly influenced by its SSTs.

Lag correlations of 3 months reveal a delay in the oceanic atmospheric mechanism for Ecuador. The delay occurs because SST anomalies take time to alter regional circulation patterns (one season approximately) through atmospheric wave propagation. Coast-North remains persistent with influence from El Niño 1+2 (0.54) (Figure 10d); therefore, the Pacific Ocean has a massive influence over the littoral. Additionally, Amazon-South also has influence from the Pacific Ocean. The influence with delay can be explained by the propagation of atmospheric waves generated during ENSO and rainfall [54] and now can be related to MCS formation.

4.5.2. Wavelet Decomposition Teleconnection

Wavelet decompositions enable us to overcome traditional analysis constraints. This achieves better comprehension of MCS occurrence patterns in Ecuador. Wavelet decomposition reveals hidden teleconnection patterns. Conventional correlation analysis aggregates both short term meteorological events and long-term climatic signals. This adds noise to the original signal. Coast-North during DJF demonstrates these limitations through its time series. This shows a strong positive 0.69 correlation. Meanwhile, V1 and W1 signals reveal negative and positive correlations of −0.17 and 0.25 with Niño 1+2, respectively.

The same phenomenon happened during the SON and annual periods. The strong connection between Niño 1+2 and MCS occurrence emerges from short-term events that span weeks to months, such as Pacific SST peaks. The negative correlation at lower frequencies indicates that atmospheric stability together with subsidence work to reduce MCS activity. The signal decomposition reveals hidden complexities through its low-frequency component which conventional methods cannot detect, according to research that decomposes climatic records in the country [22,25]. The difference between the complete series and its low-frequency signal can be understood through the cross-scale interference concept described by [15]. This shows that V1 represents long-term behavior controlled by W1. W1 detects triggers at shorter time scales. Low frequencies in the South Coast and the TNI (−0.60) show a dichotomy in the North and South of the same region. While Coast-North responds to the high frequency through the modulation of the temporal series, Coast-South part responds to the background conditions during La Niña.

The non-traditional links between oceanic teleconnections and MCS activity become detectable through wavelet analysis at both high and low frequency components. Implementing wavelet decomposition on non-stationary series is an adequate approach [101]. The signal decomposition technique reveals the Atlantic Ocean’s influence on Ecuador’s eastern mountain slopes. This is denoted by the two seasonal influences of the V1 TNA (−0.49) in Highlands-North (Figure 11a) or TNA (0.48) in Highlands-South (Figure 11b) during DJF. The physical basis for our observed correlations emerges from studying the formation locations of MCSs in the southeastern regions of the country. The research of [13] shows that Atlantic SST increases create favorable convective conditions in this region, while [117] explains that the OLLJ transports this warmth into the interior to enhance Amazonian moisture advection and now can be related to MCS genesis on Ecuador.

Previous studies document how the Pacific Ocean affects environmental conditions throughout Ecuador. However, our MCS’s approach over wavelet decomposition reveals hidden opposing responses illustrated in Figure 11c,d. Specifically, TNI demonstrates its major influence on MCS occurrence through clear evidence, yet this influence manifests differently across regions and time scales and frequency components, shown in Table 5. The Amazon is vivid proof of the TNI contrasting influence shown in Table 6; the Amazon-North region has a strong direct relationship between TNI and V1 (0.73) (Figure 11c). In contrast, the Amazon-South region has an opposite value of −0.70 (Figure 11d) for V1. When the Eastern Pacific warms (indicated by a positive TNI), it alters large atmospheric circulation patterns including zonal wind and vertical motions (omega).

The Amazon-North region undergoes zonal circulation reorganization and atmospheric instability enhancement during TNI-positive or Niño events. The significant correlations, detailed in Table 6, in zonal wind components u500 (0.58) and zonal wind difference between 500 hPa and 800 hPa ∆u (0.74), within low-frequency variability, indicate alterations caused by Walker circulation modifications. Westerly wind enhanced at high altitude facilitates moisture advection from the Pacific Ocean, which converges with local atmospheric conditions. The combination of favorable zonal wind patterns and optimal vertical wind shear establishes conditions for MCS genesis. In contrast, Amazon-South demonstrates convective suppression during TNI-positive periods. Paradoxically, the region also experiences enhanced zonal wind components u500 (0.57) and zonal wind difference between 500 hPa and 800 hPa ∆u (0.64). However, other factors in MCS genesis predominate, especially the subsidence manifested by w800 (0.63).

The difference in response is explained by the asymmetric topography [45]. The narrow Northen Ecuadorian Andes facilitates zonal flow penetration (Amazon–Andes, Andes–Amazon) and allows the development of pronounced climatic gradient in short distances [148,149]. Meanwhile, in the south exists formations like the “Nudo de Loja” at 4° S, which is a rough and eroded high mass of mountains that marks the transition between Peruvian and Ecuadorian Andes [126]. This type of formation generates compensatory subsidence that creates an atmospheric barrier that competes with optimal shear conditions, generating stability in the zone. These finding challenges continental scale precipitation assumptions and reveals that the topographic asymmetry creates opposing responses that precipitation cannot detect; this can be evidenced in Table 2’s quantity of MCS occurrence and percentage of precipitation contribution of MCS differences among both regions.

The opposite relationship between TNI’s low-frequency signal and this region creating synergic behavior with La Niña. As a matter of fact, considering that La Niña intensifies the OLLJ [150], and an intense OLLJ transports more humidity to the slopes of the Andes [132,133], in these conditions the nocturnal convection described by [28] is potentiated. Therefore, the genesis of MCSs in Amazon-South could be modulated by background conditions established by La Niña, as some studies suggest its effects for rainfall [24].

It should be emphasized that TNI’s importance is not accidental but causal; in fact, ref. [25] described it as a “striking discovery” for the area. Indeed, ref. [13] went even further, establishing that, in the southwestern Andes of Ecuador, TNI is the Pacific index with the greatest influence. Therefore, decomposing the MCS occurrence series helps us understand TNI, revealing that beneath a zero correlation lies an uneven, sometimes opposite effect in the low or high frequency ranges, depending on the region under analysis.

4.6. Study’s Limitations

One limiting factor may be the threshold size of MCSs selected in satellite imagery. While the present study applies a minimum MCS area threshold of 40,000 km², using a lower threshold may increase both the number of identified MCS events and the proportion of MCS-derived precipitation relative to total rainfall. For instance, Ref. [8] employed a less restrictive size threshold of 2000 km², resulting in a higher proportion of precipitation attributed to mesoscale convective systems. Future works may include smaller size thresholds to evaluate its impact on seasonality, diurnal cycle, and so on. This may lead to under-counting events, which may create subsequent analysis limitations. However, the methodology from [66] identifies the appearance of the MCSs in the infrared satellite image, but it also validates the strength and significance of the event through observed precipitation data which ensure real convection. Undoubtedly, researchers have successfully applied similar approaches to multiple studies throughout the region [19,68,117].

While the analysis concentrates on genesis, this may represent a bias because it ignores the complete life cycle of MCSs. Concentration on the genesis stage enables the identification of triggers that would be diffused by precipitation-focused studies or mature stage dynamics. The threshold in size may lead to smaller number of MCS events but ensures focus on mesoscale organization, as demonstrated by the 26% average of the contribution to precipitation from MCSs. The database construction and documentation of genesis-phase teleconnections with Pacific and Atlantic indices produces reliable and valuable insights. Research should progress to study MCS evolution because it will reveal how nearby moisture transport processes impact the region. In addition, future research could attempt to understand the spatial-temporal patterns of MCSs to create probabilistic forecasting or nowcasting using Bayesian network or deep leaning techniques to improve prediction of MCSs in Ecuador.

5. Conclusions

The research provides the first-ever comprehensive look at MCS in Ecuador from 2001 to 2020. Using a solid database built from infrared satellite imagery, we identified MCS events with established filters and confirmed them with actual rainfall data. The findings reveal significant variations across the country, both east-to-west and north-to-south. For instance, there is a twelve-fold difference between the highest activity in Amazon-North (207 events) and the lowest in Coast-South (17 events). This highlights the crucial role played by the Andean slopes and the availability of moisture from both the Pacific and Atlantic Oceans in how these storm systems form.

This study advances beyond traditional precipitation climatology by focusing specifically on MCSs. MCSs contribute substantially to Ecuador’s total precipitation, with an annual average of 26% across all regions. The highest contributions occur during MAM, reaching 44% in Amazon-North, while the lowest contribution is observed in Coast-South with 16%. This quantitative assessment reveals the critical role of organized convective systems in Ecuador’s hydrological cycle, providing convective-specific insights unavailable in precipitation studies.

The MAM season stands out as the period with the most convective activity, accounting for 43% of all events during the study period. In these months, the ITCZ passes directly over the country. Moisture flows in from the west, carried from the Pacific Ocean by a weaker CLLJ, while from the east, moisture comes from Amazon Basin evapotranspiration, supplemented by Atlantic Ocean humidity transported by the OLLJ, which is near its peak strength. This discovery confirms that MCS in Ecuador are the result of complex interactions between forces at different scales, from local nighttime convection to global teleconnections.

The patterns of MCS occurrence are distinct for each region, driven by different convective mechanisms. In Coast-North, storms tend to form in the early morning (4:00–8:00 AM), which is linked to the CLLJ, land–sea breezes, and nighttime cooling. In contrast, Highlands-North and Highlands-South experience their longest-lasting events where they start around noon or in the afternoon (12:00–6:00 PM), likely fueled by accumulated CAPE. Amazon-North and Amazon-South, however, show consistent storm occurrence throughout the day, indicating a state of permanent atmospheric instability. Nighttime convection is a signature phenomenon here, where cool air flowing down the eastern Andean slopes (katabatic winds) meets the warm air of the Amazon. This process is key for forming late-night MCSs, especially during DJF, when the OLLJ is at its annual maximum and the longest-lasting storms (over 20 h) begin in the early morning.

Using the Mann–Kendall and Sen’s Slope methods to analyze trends, we found a statistically significant increase in MCS frequency in Highlands-North during the SON season (p-value= 0.009, τ= 0.463, β= 0.143 MCSs per year). This result is clear evidence of climate change impacting convective activity in the Ecuadorian Andes during this season, translating to one additional MCS every 7 years. The trend points to a strengthening of the conditions that favor MCS formation, which aligns with global warming and the rise in extreme rainfall events, such as the Quito flash flood. In stark contrast, no significant monthly or seasonal trends were found in other regions. This lack of a widespread trend, which differs from regional studies, shows that convective conditions in Ecuador are highly localized and shaped by the unique interplay between the country’s topography and circulation patterns.

The observed positive trend in Highlands-North during SON provides a testable hypothesis for future climate model validation. This regional intensification trend could be examined through convection-permitting models. CMIP6 downscaling experiments using high-resolution models to determine whether current generation climate models reproduce similar convective activity changes under historical forcing scenarios. Such validation studies would strengthen climate change attribution and provide insights into projected future changes in organized convection across Ecuador’s complex topography under different warming scenarios.

A standard analysis of teleconnections confirms the Pacific Ocean’s dominant influence on the coast of Ecuador. This is evident from the connection between MCS frequency and Niño 1+2 occurring at several time stages like DJF with Coast-North (0.69), SON (0.61), and annually (0.59). The Highlands regions display complex dual oceanic influences, characterized by competition over this region. Highlands-South is affected by both the Pacific through TNI (0.61 in SON) and the Atlantic through TSA (0.56 in JJA). The interior part of the regions needed wavelet decomposition analysis to reveal hidden teleconnection that traditional methods miss. This method showed striking contrasting behaviors, like the opposing responses of Highlands-North and Highlands-South to TNA during DJF (−0.49 vs. 0.48 on low frequency). Specifically, the remarkable dual role of TNI in the Amazon is demonstrated. TNI directly enhances Amazon-North MCS genesis (0.73 correlation on low-frequency V1) and suppresses Amazon-South convection (−0.70 correlation on low-frequency V1) at the annual scale. Ecuador’s asymmetric topography over North and South may explain these opposing MCS responses to TNI, though the exact mechanisms are not fully understood. In the north, the narrower Andes potentially allow easier atmospheric flow between the Pacific and Amazon, which could enhance convection in Amazon-North during positive TNI phases. In the south, massive mountain formations like the "Nudo de Loja" appear to create different atmospheric conditions, possibly contributing to convective suppression in Amazon-South. However, the precise physical processes linking topographic differences to these contrasting regional responses require further investigation, as other factors such as local moisture sources, atmospheric stability, and regional circulation patterns may also play significant roles.

In summary, these results confirm that MCS in Ecuador are highly complex weather events that respond to multiple processes depending on the region and time of year. Their occurrence is shaped by the interaction of local topography, nighttime convection, moisture from Amazonian evapotranspiration, and humidity from the tropical Pacific and Atlantic Oceans transported by the CLLJ and OLLJ. By using wavelet decomposition, we identified teleconnections missed by conventional methods, such as the dual role of the TNI in the Amazon throughout the year and TSA over the Highlands in DJF. We also identified a rising trend in MCS frequency in Highlands-North during the SON season, a clear sign of climate change related to warming conditions that trigger higher CAPE. This knowledge provides a scientific foundation for improving early warning systems, helping to predict extreme events that impact strategic sectors like hydroelectric power operations. Furthermore, this comprehensive MCS database provides a valuable resource for evaluating the capability of high-resolution climate models to represent extreme precipitation events and for training artificial intelligence models to simulate extreme weather phenomena. These applications directly support the operational sectors, including hydroelectric plant management, risk assessment, and climate change adaptation planning for the country.