Variations in Seasonal Precipitation Regime According to the Elevation

1. Introduction

According to the World Meteorological Organisation (WMO), extreme weather in Europe and the Mediterranean Basin in 2025 was marked by significant droughts, extreme wildfires, floods, and severe windstorms [1]. The report also indicates an increasing number of extreme weather cases in the region, associated with the rising average temperature observed over the last few decades. These adverse conditions require taking immediate action. For instance, it is necessary to collect as much information as possible about the current meteorology. Therefore, this allows the comparison of those values with historical data, facilitating the management of hydro-meteorological resources in one of the most densely populated areas [2,3,4].

The Mediterranean Basin countries have historically experienced floods, caused by the meteorological and geographical conditions [5,6,7,8]. Among the meteorological causes, we can find extreme weather conditions. In this way, long dry periods produce warm air masses. The combination of these with some moist advections changes the nature of the lower-level atmosphere. Then, the instability produced by cold air at high levels or the irruption of an extra-tropical cyclone is the triggering element that develops highly efficient thunderstorms [9,10,11]. The large rainfall cumulated values combined with arid soil conditions and small, rapid response basins cause dramatic floods that affect infrastructure, such as highways or buildings, but can also impact the population [12,13,14].

Several authors have shown that topography has been revealed as one of the conditionings that modulate rainfall regimes [15,16,17,18]. Precipitation usually increases with height, but only until a certain altitude, which is considered the breaking point in the rainfall trend. This behaviour does not coincide in all regions, and it depends on the climatic conditions [19,20,21]. Besides, other factor that directly participates in the water cycle is the oceanic or sea waters, particularly in those cases of warm waters, such as the Caribe, the Chinese Sea, the Australian Seas, or the Mediterranean [22,23,24,25,26].

The aim of this Research is to show the changes on the precipitation regime along the year in Catalonia. To do this, it has been selected different radar and lightning parameters. These parameters have been evaluated for different points considering the altitude to categorize them. The variables have been evaluated daily for each category, considering the mean value for the different points.

2. Materials and Methods

2.1. Area of Study

Catalonia is in the northeastern Iberian Peninsula, covering 32,000 km2 (Figure 1a). The region's topography is abrupt, ranging from 0 m at the coast to over 3000 m ASL (Above Sea Level) in the north (Figure 2a). Several valleys and plains are surrounded by different ranges. These features are like those of other Mediterranean areas [27,28,29,30]. From a hydro-meteorological viewpoint, heavy rainfall events are frequent in Catalonia, producing flashfloods or floods, depending on the nature of the rainfall event [31,32]. Catastrophic episodes have historically caused large damage and casualties throughout the years, with important affectation in hydrologic structures [33,34]. Furthermore, long drought episodes have recurrently implied societal problems [35,36].

2.2. Data

As many previous works have shown [37,38], the Region of Study is well-covered by two remote-sensing networks, used in the present work (Figure 1b). Both networks belong to the Meteorological Service of Catalonia (SMC, Servei Meteorològic de Catalunya in Catalan), which has operated both since the early 2000s [39]. First, the weather radar network (XRAD) is composed of four C-band single-pol systems that produce volumetric scans every 6 minutes, with 15 elevation angles from 0.6 to 30º, with a range of 130 km from the radar. The composite of all four radars provides a complete vision of the atmospheric conditions from surface to nearly 20 km ASL, with a horizontal grid of 1km x 1km.

The second network is a passive remote sensing system, which only provides information in the case of lightning activity [40]. The Lightning Location System (XDDE, Xarxa de Detecció de Descàrregues Elèctriques) operates in two frequency ranges (Low Frequency, LF, and Very High Frequency, VHF) to generate solutions of the locations of Cloud-to-Ground (CG) and Intra-Cloud (IC) flashes produced by thunderstorms. The detection efficency (DE) is close to 95% of CG and 80% of IC over the Catalan terrain, decreasing the quality of detection as far is the thunderstorm from the network. The precision in the location is close to 500 m in the case of CG, while in the case of IC ranges between 1000 and 2000 m, over the Catalan territory. Figure 1b shows the location of the four detectors.

The Research has consisted of the analysis of different radar products [41,42,43,44], from reflectivity at the surface (ZSF) and the maximum value for the complete column (ZTO), the echo top (or the maximum height where an echo exceeds a certain reflectivity threshold) for 12 dBZ, 35 dBZ, and 45 dBZ (T12, T35, and T45), the quantitative precipitation estimation at daily and 30-minute time resolutions (RNN and R30), and the VIL (Vertically Integrated Liquid) density (VILD). Reflectivity provides an estimation of the instantaneous rainfall intensity, echo tops give information about the vertical development of the precipitation structures, rainfall estimation produces an accurate map of the precipitation distribution, and VIL density estimates whether hail is possible in a thunderstorm. Except for rainfall products, the time resolution is 6 minutes (10 fields hourly). The period of analysis go from 2016 to 2025, both included (a total of 10 years).

In the case of lightning data, it has considered CG flashes exclusively to reduce the location problems associated with IC lightning. Therefore, thunderstorm activity has been analyzed considering those flashes reaching ground level, similarly to Taszarek et al. (2019) [45].

2.3. Methodology

To evaluate the different radar and lightning fields depending on the topography, a series of points (Figure 2b) has been selected, placed at heights of eight different ranges:

* Zone 1: between 0 and 5 m ASL (coastal plain);

* Zone 2: between 5 and 100 m ASL (coastal line or valleys);

* Zone 3: between 100 and 250 m ASL (low prelitoral ranges or internal planes);

* Zone 4: between 250 and 500 m ASL (high prelitoral ranges or elevated planes);

* Zone 5: between 500 and 1000 m ASL (very high prelitoral ranges);

* Zone 6: between 1000 and 1500 m ASL (mid internal ranges);

* Zone 7: between 1500 and 2000 m ASL (mid Pyrenees and high internal ranges);

* Zone 8: over 2000 m ASL (high Pyrenees)

We considered different segments (see Figure 1b), ranging from the Mediterranean Sea to the Pyrenees (A to E), and included an additional segment in the Southern area (F). This selection ensures that we obtain enough points from the various zones to validate the Research goal. We obtained a daily value for each point for each of the different radar and lightning products from the period 2016 to 2025. Therefore, each field produces 1336 outputs for 51 selected points (3 from zone 1, 5 from zone 8, 6 from zones 2 and 3, 7 from zones 4 and 7, 8 from zone 5, and 9 from zone 6, respectively). The selection of fewer points for zones 1 and 8 corresponds to the smaller area of those regions. However, it was important to consider the rest because of their nature.

The general behaviour of each variable has been evaluated globally for the different regions to identify some discriminating characteristics associated with the changing altitude. Then, the monthly evolution of some parameters has helped to reach important conclusions regarding the changes for the different variables and zones. This variability has been related to the precipitation regimes and topography.

3. Results

3.1. General Results

The different panels of Figure 3 are associated with the precipitation regime: surface reflectivity (a) can be associated with the ground precipitation intensity, total reflectivity (b) is the maximum intensity in the full radar column, daily quantitative precipitation estimation (c) estimates the total precipitation during a day, and 30-minute quantitative precipitation estimation (d) measures the semi-hourly intensity. The violin plot [46] from R software [47] visualizes the combination of a box plot with a kernel density estimate. Therefore, it is like a box plot, but the different widths at various heights of the sample marks indicate the more typical values. It can be observed that in the case of the reflectivity fields (top panels), the values increase with the height until zone 4. This is the breakpoint, and from here, reflectivity values gradually decrease. Similarly, quantitative precipitation estimation also increases. However, the breakpoint is zone 6 for the daily field and zone 5 for the 30-minute product.

Figure 4 shows variables associated with the development of the precipitating systems (Echo top for three different reflectivity thresholds: 12 dBZ for all precipitation, 35 dBZ for mid-intense rainfall, and 45 dBZ for high-rain rate precipitation, according to the experience of the Catalan Meteorological Service staff) and with the probability of hail in thunderstorms (VIL density). In all cases, higher TOP values occur in zone 3. However, the maximum median value for the violin plot is located between zones 7 and 8. This means that the higher development of thunderstorms occurs when the systems are in low regions, but the systems observed in higher regions tend to be more developed in general. Finally, the diagnosis of hail shows the largest values of DVIL between zones 3 and 4, while the largest median values are recorded in zones 5 and 6. Then, mid-topography has a main role in hail occurrence, particularly at the highest peaks.

Figure 5 provides a summary of the electrical activity for the different zones. In this case, two principal regions are distinguished: the first is the flattest one, zone 1, indicating the high contribution of the Mediterranean Sea in the electrification of precipitation systems. The second one corresponds to the top zones (6 to 8), where the electrical activity is notably more important than in the mid-topographic regions.

To go more in depth with the previous results, table 1 shows the upper level of the internal black box around the median (white dot), corresponding to the 90^th quantile. This parameter has been estimated for the eight zones for the nine products. In general, higher values are concentrated in the upper regions (zones 6 to 8) for most variables (TOP12, TOP35, TOP45, RNN, RN30, and ZTO). On the other hand, XDD (electrification) is more important in flat areas, while VIL density and surface reflectivity have the predominant range between zones 3 and 6.

3.3. Outliers

Outliers identify those extreme values for a data sample. This identification is very important to determine the limits associated with extreme events. To select the outliers of the different samples by zones, the “extremevalues” R package has been used [48]. It is important to note that the analyzed values correspond to sets with several samples per day, smoothing the daily value for each zone. This is more evident for fields of variables that usually are concentrated in relatively small areas, caused by deep convection (VIL Density, heavy rainfall, TOP45 or lightning activity), meanwhile other parameters (mainly TOP35 and TOP12) are less affected by this circumstance.

One of the first points of the comparison between previous Table 1 and the following Table 2, which contains the outlier upper limits, is that in some cases (Echo top for 12 dBZ, and the surface and total reflectivity) the outlier is below the third quantile of the sample used. In those cases, the probability of observing extraordinary values is higher than for the rest of the parameters. So, very short-term intensities (reflectivity) and very large vertical development of precipitating systems are more frequent that the other variables: tops associated with mid and high intense reflectivity, lightning activity, or 30-minute and daily quantitative precipitation estimation.

Another point of interest resides in the high fitting of the datasets for those three variables (TOP12, ZSF, and ZTO), in comparison to the rest, according to the R-squared value (values inside the parentheses). In those cases, the value exceeds 0.97 in all cases, only comparable with the TOP45 ones (near 0.96). For the rest of variables fit is worst, in special for lightning activity (R² between 0.12 and 0.54).

Regarding the breaking point, there are many similarities between both tables, predominating zones 5, 6 and 7 (ZSF, ZTO, DVIL, RN30, RNN, TOP35 and TOP12). This confirms that highest convective activity occurs in the immediate areas to the most elevated ranges, at altitudes between 500 and 2000 m ASL.

3.3. Monthly Behavior

The combination of the Mediterranean Sea and the different Catalan ranges play a principal role in the modulation of the precipitation regime along the different seasons of the year. In this section the different variables from a monthly point of view are analyzed, based on boxplot graphics.

3.3.1. Surface Reflectivity

Figure 6 shows the variation between the different zones regarding the surface reflectivity: meanwhile for zones 1 to 4 (altitude below 500 m) the variation between months is not evident and only hidden maxima can be detected (June and November in zone 1, October in zone 2, and September in zones 3 and 4). In the opposite case, zones 5 to 8 present a clear maximum for the Summer months, mainly for July and August. The maximum is more evident for the highest zones. The same behaviour is observed for total reflectivity (ZTO, not shown).

3.3.2. Echo Top

Figure 7 and Figure 8 show the evolution of Echo Top for 12 and 45 dBZ along the year for the different zones, respectively. In the first case, TOP12, there is a total agreement between all the regions, with a similar monthly pattern: a maximum between July and August, while the minimum occurs during the period between November and March. This is an indicative that deep convection occurrence is more usual during the summer months than in the cold season. In the same line, TOP45 shows highest values during the same period of the year. This parameter is indicative of the presence of hail, or large hail, especially when values are close or over 4 km. This threshold is exceeded generally during the warmest period (June to September), and more usually for the more elevated zones (4 to 8).

3.3.3. VIL Density

VIL density can be associated with hail occurrence. For instance, values exceeding 2.5 g/m³ have been revealed as severe hail (diameter over 2 cm) diagnostician in the region of study, meanwhile 1 g/m³ is highly correlated with hail of diameter exceeding 0.5 cm. Figure 9 shows how the probability of hail occurrence in zone 1 is concentrated between August and November. From here, the length of the hail season (this is, with probability of hail occurrence) increases: from April to November (zone 5), decreasing again from this point, with a new minimum in zone 8 (from May to August). Besides, it can be observed as the height increases the influence of the warm season (June to August) is also higher.

3.3.4. Quantitative Precipitation Estimation

The last figures of the monthly analysis refer to precipitation estimation, daily (Figure 10) and 30-minute (Figure 11). Regarding the first one, the pattern is similar in all cases, with a bimodal distribution, but with differences depending on the zone (this is, the altitude): zones 1 to 5 show values generally below 5 mm and the peaks in January and November, with very scarce precipitation in summer months (June to August). Even the pattern is similar for zones 6 to 8, the maxima shift to February or March on one hand, and October or November on the other hand. Besides, the values are larger for all months.

In the case of 30-minute precipitation, which is the cause of flash-floods in many events, the behavior differs respect on the daily field. Again, there are two clear differentiated patterns, with the break point again between zones 5 and 6. While for the zones below 500 m ASL of altitude there are two peaks (on March and September/November, respectively) and the values are below 2.5 mm, in the case of top zones there is only one maximum, in the summer months (June to August), where the value is over 2.5 mm.

4. Discussion

The daily analysis of different points for eight zones with similar heights across several radar fields and lightning data for the period 2016-2025 has revealed important conclusions regarding the yearly precipitation regime associated with sea and topographical influences. Gnann et al. [15] made an extensive review of the factors associated with topography that contribute to the precipitation regime of a mountainous region. Similarly, Schneider et al. [17] demonstrated that low-mountain zones are highly influenced by topography. In the present analysis, from a global perspective, rainfall intensity and cumulation increased with height up to an altitude of 500 or 1000 m ASL, respectively. Virts et al. [19] showed the discrimination between regions below and over 500 m in the case of thunderstorm activity. Moreover, Tan et al. [20] observed that 1000 m was a breaking point in the precipitation regime of the South-Eastern Asian region. Taszarek et al. [45] have demonstrated that lightning activity varies throughout the year, affecting coastal or mountainous regions depending on the season. A similar result was observed in the current study, with maxima in the flattest area (zone 1) and in the upper ranges (zones 6-8).

Taszarek et al. [45] also demonstrated single-mode or bi-modal monthly behaviour, depending on the region, for convection. The analysed variables associated with convection in the present study also showed this variability, with a single mode for ZSF, TOP12, TOP45, or DVIL; meanwhile, for daily cumulated rainfall, a double mode was observed. A case in point is the 30-minute cumulated rainfall, which exhibits a bimodal pattern at 500 m and a unimodal pattern at higher altitudes.

Abel et al. [16] showed that precipitation in coastal areas has a high dependence on the elevation gradient of the region and, besides, how the topography blocks precipitation in some circumstances, allowing high amounts of rainfall in elevated areas, coinciding with Godart et al. [21]. This is coincidental with the current results for the 90th quantile or the outliers analyses, with breakpoints at 1000 and 2000 m, respectively.

Miglietta et al. [9] analysed the influence of the Mediterranean Sea on cyclogenesis, which has a peak in the region of study during the Summer and Autumn, according to Campins et al. [11]. However, this Mediterranean cyclogenesis is very variable depending on the year (Almazouri et al. [10]), coinciding with the high variability of the Mediterranean precipitation regime associated with the atmospheric conditions that occurred in the region (Dayan et al. [23]). Water conditions have been revealed as one of the main ingredients in the precipitation regime in a coastal region (Caine et al. [26]). All those factors influence the previous results because of the high variability of the weather conditions. However, considering an analysis such as the current one, it is possible to understand better the situations that can lead to a critical scenario, with an excess of precipitation or, on the other hand, with a dramatic drought.

To sum up, this type of Research is important to combine with the results of historical analysis of floods (such as Benito et al. [12]) and of the input of hydrological models (Tramblay et al. [5]; Zoccatelli et al. [6]; Dottori et al. [7]), to avoid or reduce the high impact of floods in the region of study (Hochman et al. [4]; Vinet et al. [8]; Llasat et al. [13]; Stamos and Diakakis [14]). Besides, it is important to bear in mind that droughts and floods are expected to increase in the area of analysis in the future due to climate change because of the meteorological conditions (Granata et al. [2]; Raymond et al. [3]). Furthermore, these conditions are applicable to other regions, such as the Caribbean Sea (Martinez et al. [24]) or South China (Peng et al. [25]; Ummenhofer et al. [22]).

5. Conclusions

Catalonia, like many other Mediterranean Regions, has a climate with a high dependance on the atmospheric conditions, but modulates by the combination of the complex topography and the Mediterranean Sea. These factors make the precipitation regime is highly variable and can produce very long dry or wet seasons. Some reports verified by recent observations have indicated that Global Warming has a major impact on the Mediterranean Basin, with an intensification of droughts and high rain, with the increase also of extreme wildfires and floods and flash floods. Therefore, the current analysis intends to give some clues to the current state of the precipitation regime to help on the management of the water resources of the region.

From the study of daily values of different radar and lightning parameters for different altitudes, it has been observed that: first, there is a high relationship between the precipitation and the height of the observation, until an altitude between 500 m and 1000 m. Moreover, lightning occurrences are bigger in coastal areas and very high zones. Topography also plays a major role in the development of deep convection, which initiates at highest altitudes to move to moderate height zones.

Considering the monthly data, it has observed double behaviour: on one hand, a single maximum mode, generally concentrated on warm months for convective parameters (T45, ZSF, DVI or RN30, only for heights over 1000 m). On the other hand, a double maxima pattern with peaks near March and November, mainly.

Identifying disruptions of this behaviour for a period of three or more months can warn of anomalous atmospheric conditions that can anticipate a long wet or dry period.