Spatiotemporal variability of monthly precipitation concentration in Argentina

3.1. Annual accumulated precipitation distribution and PCI

In the north, the distribution of annual precipitation presents a marked east–west gradient. In the east, the annual accumulations exceed 2200 mm, but in the west do not exceed 700 mm, with minima in San Juan and Mendoza (21 and 31 respectively, Fig. 1). In the south of the country, the gradient changes and the relative maximum (800 mm) is in the Andes region, and the coast has values below 300 mm.

Fig. 1. 

Annual accumulated precipitation (mm) (1991–2021).

The accumulated precipitation is distributed throughout the year in different forms, according to the region. Once again, there is a marked east–west gradient. The maximum values are in the north-west, presenting a strong irregularity of precipitation distribution (PCI > 20). The west presents the absolute maximum in San Juan station (PCI = 27.3). In the north-east, the PCI values are less than 15, associated with a moderate distribution (Fig. 2). The northern region of Patagonia shows an irregular distribution (PCI of 15–20). These regions with a distribution irregularity are associated with the arid diagonal observed in rainfall (Bruniard 1982). In the south, the Ushuaia station (number 64) has the lowest PCI value (PCI = 10.6), the only locality with a uniform annual distribution.

Fig. 2. 

Annual PCI field.

Analysing the PCI values joined with the precipitation shows different particularities. Ushuaia and La Quiaca (number 1) have approximately the same annual accumulations, but the PCIs are very different: Ushuaia has the lowest PCI (10.6), and La Quiaca is a high PCI (23.9). This difference is due to the precipitation distribution (Fig. 3): in Ushuaia this is homogeneous throughout the 12 months, mainly due to the passage of frontal systems throughout the year, whereas La Quiaca has a strongly irregular distribution, with rain only during summer. Ushuaia can also be compared with Bernardo de Irigoyen station (number 12) with a similar PCI (11.1), which station has the maximum annual accumulation (2225 mm) perfectly distributed throughout the year. These three examples show how similar PCI values can be associated with different precipitation accumulations, just as comparable precipitation values can have very different PCIs. The magnitude of PCI is independent of the accumulated precipitation. Fig. 3 shows the three extremes cases: minimum and maximum PCI and maximum precipitation.

Fig. 3. 

Mean monthly precipitation (mm) in La Quiaca, Ushuaia and Bernardo de Irigoyen.

3.2. Wet and dry years

To determine any closer relationship between PCI values and annual accumulation, wet and dry years are selected based on how much they deviate from a standardised value. Fig. 4 shows the percentage of change in PCI between wet and dry years. The spatial distribution has two big centres: the Patagonian coast has a change of ~25%, always in the category of irregular distribution, but wet years have a PCI approximately equal to 21, whereas in the dry years, the value is 16. The north mountain range, La Rioja station (number 15) has PCI of 18 in wet years and 27 in dry years, which represents a 50% increase; Chamical (number 22) shows a 30% increase, going from 17 to 22 between wet and dry. Another case to mention is the east region with two relative centres: in the north, the dry years have lower PCI (~10%); in the south, the PCI is greater in dry years (20%). For example, PCI in Olavarria (number 49) changes between 11 (wet) and 14 (dry).

Fig. 4. 

Percentage (%) change in PCI between wet and dry years.

3.3. Seasonal PCI

The PCI is calculated for each season in particular (Fig. 5). Winter is the season that shows the four PCI categories: in the southern mountain range, the distribution is uniform, as well as the extreme north-east of the country. These two regions coincide in that they accumulate the highest precipitation at a seasonal level (Fig. 6). The irregular distribution category is relegated to the north-west, presenting only three station PCI values corresponding to a highly irregular distribution (La Quiaca, 1; La Rioja, 15; and San Juan, 21) with precipitation below 50 mm in the season (low rainfall and poorly distributed). The intermediate categories cover the rest of the territory. The transition seasons (autumn and spring) have very similar behaviour to the east–west gradient, with most of the country within the category of moderate distribution of precipitation (11 < PCI < 15). No stations have a highly irregular distribution. In the central-north region of the country, the distribution of the isolines of PCI shows a similar disposition to the isohyets (where the precipitation is high, the distribution is more homogeneous, and vice versa). The months that comprise these seasons have approximately the same accumulation. Finally, during summer, the north stations have a uniform distribution while the rest of the country shows moderate PCI values. The 250-mm isohyet seems to mark the limit between the two categories of precipitation distribution.

Fig. 5. 

Seasonal PCI fields (autumn, winter, spring and summer).

Fig. 6. 

Seasonal accumulated precipitation (mm) (autumn, winter, spring and summer).

3.4. Temporal changes in PCI

As a first point, the variability of the PCI is analysed. It is noteworthy that the highest standard deviations are for the stations where PCI represents an irregular distribution of precipitation. The precipitation has a marked seasonality, but not all years are the same. There are particular situations, for instance, San Juan station, where most years present a strong irregular distribution. The most extreme year is 1994 (PCI = 69), mainly due to extreme rain in February that doubled the monthly average value (Perucca and Paredes 2005). On some occasions, the PCI indicates a moderate distribution – as for 2000 when, atypically, the months of autumn and winter accumulated rainfall outside the normal values (Fig. 7). San Juan shows the most extreme case, but the situation is similar in the entire Andean region.

Fig. 7. 

Monthly precipitation (mm) in San Juan and Trelew, mean value and extreme PCI years.

Another point to highlight occurs on the coast of northern Patagonia, with Trelew and Comodoro Rivadavia (station numbers 59 and 60) usually presenting a PCI with an irregular distribution, but in some years the distribution is moderate and in others normal. These cases of strong irregular distribution are caused by extreme events such as the one that occurred in Trelew in 1998, when April precipitation was 10 times the mean value (Chachero 2012), giving a PCI of 50.3. The north-eastern stations, given their regular precipitation regime, show less variation.

The PCI behaviour is analysed using the M–K test. On an annual scale (Fig. 8), only three stations present a significant positive trend, indicating that precipitation is distributed increasingly irregularly in Bernardo de Irigoyen, Corrientes (number 13, in the north-east) and San Antonio Oeste (number 56) in the north of Patagonia. Of the remaining stations, 75% present a positive trend, but not significant, and only in 18 stations are the values of M–K negative, mainly in coast stations and in a band in the central country. At the same time, the analysed trend in the annual precipitation records (data not shown) shows a negative trend for 45 stations, of which only 9 are significant at 5%. No significant positive increases are observed.

Fig. 8. 

Mann–Kendall test results for annual PCI.

The three stations with significant changes in the PCI present negative variations in the accumulated precipitation (but not significant). Over time the amount of precipitation is increasingly less evenly distributed.

On a seasonal scale, the trends in accumulated precipitation and PCI are also analysed jointly. In the four seasons, the PCI values present numerous positive trends (although not significant), and precipitation behaves oppositely. Winter is the most extreme case, with 49 stations showing an increase in PCI, giving rise to a more irregular distribution over the years. However, winter precipitation is less each year (48 of the 64 stations), repeating the behaviour seen on an annual scale (i.e. less precipitation and more irregular).

3.5. Correlation between PCI and climate indices

The PCI values are correlated with different climatic indices (PDO; El Niño 3.4; Antarctic Oscillation, AAO; and SOI) to understand the possible teleconnection of PCI with large-scale circulations. As climate indices may influence the PCI values of a current year (season) and the following year (season), the correlations are calculated for 0 and 1 year ahead and 0 and 1 season ahead (Table 1) between the climatic indices and the PCI values of each station.

On an annual scale, non-significant negative correlations (0 years) with PDO and El Niño predominate, whereas for SOI, the correlations are mainly positive (significant in six stations located in the central north of the country). In summer and autumn, this pattern of behaviour is repeated (0 seasons ahead – no lagged correlation). In winter, there is no defined pattern, and the amounts of positive or negative correlations are similar. In spring, the signs of the correlations between PCI and the different indices change: with PDO and El Niño, there is a greater quantity of positive correlations, whereas with AAO and SOI, negative correlations dominate.

The correlations with 1-year lag (climatic indices in the previous year or season) are analysed to find a pattern that allows a forecast. At the annual level, the correlations with El Niño are mainly positive. Between AAO and PCI, the correlations are negative, and with PDO and SOI, the number of stations with positive correlation is similar to that of negative ones. In all cases, significant correlations are few.

On a seasonal level, with a lag of 1 season between the climatic indices and the PCI, the results are dissimilar. In the autumn, between PDO and El Niño, negative correlations dominate – similar to what was found by Zhang et al. (2019). These negative correlations may be for two reasons. (1) Positive El Niño events (associated with higher precipitation in this region). This higher precipitation can be distributed along the months, giving rise to a lower PCI value. (2) Negative El Niño (events associated with less rain). If this precipitation has a poor distribution, it can give rise to higher PCI. For winter, PDO, El Niño and AAO present more negative correlations; depending on the index, between 15 and 25% of the cases are significant, mainly in stations located in the centre of the country.

In spring, the pattern change is that positive correlations dominate. The AAO and precipitation during spring behave oppositely (Silvestri and Vera 2003). High AAO is associated with less rainfall, if this has an irregular distribution, then PCI is high, whereas for the SOI, more than 50 stations show negative correlations. In summer, the amounts of positive or negative correlations are very similar in all cases.

In all cases, the geographical distribution of significant relationships is highly variable, and does not define a clear pattern of behaviour. However, this does not mean that the annual and seasonal PCI are influenced by the climate indices studied.