3.1. The 2010–2015 Hydrological Drought on a Recent Historical Context (1971–2016)
In order to put the 2010–2015 hydrological drought in a historical context, we analyze, over the last 46 years, the variations in the SSI3 for the Mendoza and Agrio Rivers as representative records of CAA and NP, respectively. Figure 3
shows the variations in monthly streamflow and SSI3 for the Mendoza River from July 1971 to June 2016. Based on Figure 3
, we observed that above-average streamflows are mainly recorded in the periods 1977–1995 and 2006–2010, whereas below-average streamflows were registered in 1971–1977, 1995–2001 and 2010–2016. Wet and dry periods are not only based on streamflow peaks during the warm season, but also along the annual low flows. For instance, an increase (decrease) in the streamflow during the low-flow season is observed together with an increase (decrease) in the streamflow peaks in summer wet (dry) periods. Regarding the hydrological drought for CAA in the recent period, we noted that severity reached the extreme category between December 2010 and January 2011, being the lowest SSI3 value on record. Even when drought duration seems to be comparable to the event between 1974 and 1977, the number of months with SSI3 values lower than −1.0 is remarkable larger in the most recent drought. When comparing the Mendoza River with the most rivers in the CAA, similar patterns regarding severity, duration and timing of the recent hydrological drought emerge (not shown), consistent with the homogeneous behavior of streamflow variations across the CAA [21
], and particularly in line with the regionalization performed by [31
] based on the SSI for the period 1961–2006.
Streamflow and SSI3 variations across NP, based on the records from Agrio River are depicted in Figure 4
. Both monthly streamflow and SSI3 records show a more “noisy” pattern in comparison with the Mendoza River (Figure 3
). This result is consistent with Caragunis et al. [41
], who showed that the contribution of low frequency variations to the streamflow variability accounts for 40% and ~15% of the total variance across the CAA and NP, respectively. Main wet periods are observed during the period 1979–1983, whereas hydrological droughts are identified in 1996, 1999 and 2013 (Figure 4
). Comparing with records from CAA, there are several differences in timing, duration and occurrence of the maximum severity during the recent hydrological drought. Several drought events were recorded since 2007, with the maximum severities between 2012 and 2014 (Figure 4
). This latter hydrological drought event shows the lowest SSI3 for the Agrio River record, and is comparable in duration and magnitude with the droughts of 1996/1997 and 1998/1999. Analyzing the spatial pattern of the recent hydrological drought across NP, we noted that most basins experienced extreme hydrological droughts between 2012 and 2013, with durations ranging from seven to over 24 months. The extreme hydrological drought in 1998/1999 shows the lowest SSI3 value on record in three of the five basins south of 38° S.
Visual inspection of the SSI3 series for each of the selected rivers indicate that, in the basins of NP, the timing of the maximum severity of the 2010–2015 hydrological drought, its duration, onset and demise were not consistent with those recorded for the drought across the CAA (Figure 3
and Figure 4
). The spatial representativeness of the 2010–2015 hydrological drought is limited to the basins located north of 38° S. Nevertheless, the intensification of hydrological drought over NP during 2012/2013 is associated with the same physical mechanisms that sustained the drought conditions over the CAA. Details on this aspect are provided in next sections.
The percentage of stations within each SSI3 category is useful to quantify the spatial extension of drought events, even when it is limited by the lack of information in the immediate surroundings. Nevertheless, streamflow integrates the hydrological processes over a river basin and the selected stations belong to similar and spatially contiguous basins over the study area, i.e. providing an acceptable estimation of the proportion of area under drought conditions. The temporal evolution of this index, represented in Figure 5
, was obtained for each month during the period 1971–2016 by calculating the percentage of stations with SSI3 below the selected thresholds (see Table 2
). The percentage of stations in the moderate drought category includes those stations showing severe and extreme drought conditions. Likewise, the percentage of stations in the severe category includes those stations showing extreme drought. We used the percentage of stations instead of the number of stations given the length differences in the records between CAA and NP regions, particularly after 2015. In general, regional hydrological droughts are observed during 1976/1977, 1991/1992, 1996/1997, 1998/2000 and 2010–2016, with more than 50% of the study area under hydrological drought conditions at different severity levels (Figure 5
). During December 1996, January 1997 and between July and October 2015, all the analyzed rivers show hydrological droughts at different severity levels. The 2010–2015 hydrological drought is substantially longer than any other drought event in the last 46 years. On average, between April 2010 and December 2015 (69 months), 63% of the rivers along the study area were affected by moderate to extreme streamflow drought conditions. In terms of severity, however, the droughts of 1996/1997 and 1999/2000 affected a larger percentage of stations under extreme dry conditions, although with shorter durations.
In terms of severity and spatial extension, three of the most important hydrological droughts over the last 46 years were recorded in 1996/1997, 1999/2000 and 2010–2015 (Figure 5
). To compare the spatial extension of these droughts during the month showing the maximum intensity, Figure 6
displays the regional patterns of the hydrological drought categories for these three events. As expected from Figure 5
, the worst conditions in terms of severity and extension are registered during December 1996, with all the records under drought conditions, 18 reaching the severe category and 14 under extreme hydrological drought conditions. The spatial extension of the hydrological drought during January 1999 seems to be limited to the stations south of 33° S, although the two stations with normal conditions have negative SSI3 almost reaching the moderate category. In a regional perspective, [31
] showed that this drought event extended farther south, affecting the basins in Central Patagonia reaching 45° S. During August 2015, only 13 stations located across the CAA have available records. All rivers were under moderate hydrological drought, eight and five of them reaching the severe and extreme drought categories, respectively. Hydrological drought severity exhibits a more heterogeneous pattern, attributed by Rivera et al. [25
] to geomorphological factors within each basin. However, additional studies are needed to properly account for the difference between basins.
3.2. Spatial and Temporal Patterns of the 2010–2015 Hydrological Drought
For assessing the spatial and temporal patterns of the 2010–2015 hydrological drought, 15 stations from the CAA were analyzed. SSI3s from July 2009 to June 2016 are shown in Figure 7
. In most gauges, the onset of the hydrological drought started during the second half of 2010. SSI3s rapidly fall below −1.0, with extreme droughts in the Mendoza, Tupungato and Vacas Rivers during the summer of 2011 and severe drought conditions in the rest of the rivers in the CAA. Across the CAA the 2010–2015 hydrological drought is a continuous, long-lasting event. Only three stations recorded SSI3 positive values in October 2012 (Diamante River), October-November 2013 (Grande River) and November 2013 (Barrancas River). Three peaks in drought intensity are evident: the first between 2010 and 2012, the second between 2014 and 2015 and the last between 2015 and 2016. Stations reaching extreme hydrological droughts during the first intensity peak are located between 32° S and 34° S, whereas those related to the second intensity peak between 34° S and 37° S. A more heterogeneous spatial pattern is observed for the 2015–2016 intensity peak, as represented in Figure 6
. The hydrological drought demise is recorded between the end of year 2015 and the beginning of 2016, although some stations continued under hydrological drought conditions (i.e., Tunuyán, Vacas, Poti Malal and Pincheira, see Table 1
for more details).
The spatial distribution of the hydrological drought severity at different stages during the 2010–2015 drought event is shown in Figure 8
for the months of January, May and September in the years 2011 to 2015. As previously recorded for the temporal variability in SSI3, differences in drought spatial patterns arise from the comparison between the CAA and NP rivers. Less intense drought severities are observed over the NP basins, except for the period between September 2012 and May 2013. As shown in Figure 7
, the hydrological drought develops and intensifies quickly, with severe drought conditions in all CAA basins during January 2011. The hydrological drought event across the CAA decreases in severity since January 2012, with few stations showing moderate drought conditions in September 2012 and January 2013. This spatial pattern contrasts with the increase in severity over NP, suggesting that the intensification of drought could be related to reduced rainfalls south of 38° S. Over the period 2010–2015, higher drought severity was recorded in rivers located between 35° S and 36° S. Water demand for irrigation in the CAA is larger during spring-summer months; therefore, when comparing the hydrological drought categories during January it can be seen that the large number of stations under drought was observed during 2011 and 2015. This result is in line with drought intensity peaks in Figure 7
summarizes the main hydrological drought characteristics over the period 2010–2016. Except for the San Juan River, the remaining gauges show that the onset of the hydrological drought occurred between June and October 2010. The drought ended during the first part of 2016, although four rivers still remain under drought conditions at the end of the records. The mean hydrological drought duration was 67 months, being one of the longest dry periods in the CAA rivers. Six rivers registered the lowest SSI3 values (i.e., maximum drought severity) of the entire record (1971–2016). A large heterogeneity is observed in the date of the maximum hydrological drought severity, consistent with the findings on meteorological drought conditions reported by [42
3.3. Drivers of the Hydrological Drought
Since hydrological drought conditions over the CAA respond to lower than average accumulation of snow over the Andes [21
], we explore the relationship between drought and the mean regional snow water equivalent anomaly (SWEA) over the period 1989–2015. The annual (July to June across the CAA) percentage of stations with hydrological drought conditions highlights the strong link between large negative (positive) anomalies in SWEA and the occurrences of widespread hydrological droughts (excesses; r
= 0.67, p
< 0.01; Figure 9
). The largest negative anomalies in SWEA were registered in 1996 and 1998, leading to severe to extreme hydrological drought conditions over most of the CAA basins (Figure 5
and Figure 6
). A dry pattern was also observed over NP in those years (see Figure 4
and Figure 5
), but the small contribution of rainfall, instead of snow, likely is the main driver. The 2010–2015 drought was entirely consistent with six years in a row showing negative SWEA anomalies (Figure 9
) and a mean of 68% of the basins affected by drought between 2010/2011 and 2015/2016. Positive SWEA leads to a lower percentage of stations showing droughts (Figure 9
Previous research shows that La Niña events are related to lower than average snow accumulation over the CAA [21
]. The sea surface temperature (SST) anomaly field for the months with at least 70% of the gauges recording hydrological droughts is shown in Figure 10
for the period 1989–2014. A clear La Niña pattern is evident over the tropical Pacific Ocean, with cold SST over the equatorial ocean. This pattern emerges after discarding the year 2015, which recorded a very strong El Niño event. Warm SSTs over the subtropical Pacific Ocean east of Australia are produced by the advection of warm water from the tropics toward the subtropics (Figure 10
). Based on the Oceanic Niño Index (see [45
] for details), we noted that the years 2010/2011 and 2011/2012 were classified as La Niña years, while 2012/2013 and 2013/2014 also characterized by cold SSTs over the tropical Pacific Ocean not reached La Niña threshold. Including the year 2015, anomalies over the tropical Pacific Ocean still remain below zero, whereas the warm region in the subtropics show similar anomalous values (not shown).
shows the composite anomaly for the 500 hPa geopotential height (Z500) during the same months used for developing the SST anomaly composite. Figure 11
displays a strengthening and southward latitudinal location of the semi-permanent south Pacific subtropical anticyclone. This circulation pattern is associated with a decrease of the westerly zonal winds at subtropical latitudes and the decrease of the frontal activity over the study area.