Progressive Water Deficits During Multi-Year Droug PDF

https://doi.org/10.
5194/hess-2020-249
Preprint. Discussion started: 18 June 2020
c Author(s) 2020. CC BY 4.0 License.
Progressive water deficits during multi-year droughts in central-

south Chile
Camila Alvarez-Garreton1,2, Juan Pablo Boisier1,3, René Garreaud1,3, Jan Seibert4, Marc Vis4
1
Center for Climate and Resilience Research (CR2, FONDAP 15110009), Santiago, Chile
2
5 Department of Civil Engineering, Universidad de La Frontera, Temuco, Chile
3
Department of Geophysics, Universidad de Chile, Santiago, Chile
4
Department of Geography, University of Zurich, Zurich, Switzerland
Correspondence to: Camila Alvarez-Garreton (calvarezgarreton@gmail.com)

10
Abstract. A decade-long (2010-2019) period with precipitation deficits in central-south Chile (30-41ºS), the so-called
megadrought (MD), has led to larger than expected hydrological response and water deficits, indicating an intensification in
drought propagation. We used the CAMELS-CL dataset and simulations from the HBV hydrological model to explore the
causes of such intensification. Across 124 basins with varying snow/rainfall regimes, we compared annual rainfall-runoff (R-
15 R) relationships and runoff generation mechanisms before and during the MD, and identified those catchments where
drought propagation was intensified. We show that catchments’ hydrological memory -mediated by groundwater flows- is a
key control of drought propagation intensity, and that baseflow contribution to runoff is positively correlated with snow
accumulation preceding the year affected by a drought. Hence, under persistent drought conditions, snow-dominated
catchments progressively generate less water, compared with their historical behaviour, notably affecting the semi-arid
20 regions in central Chile. Finally, we addressed a general question: what is worse, an extreme single year drought or a
persistent moderate drought? In semi-arid regions, where water provision strongly depends on both the current and previous
precipitation seasons, the worst scenario would be an extreme meteorological drought following consecutive years of
precipitation below average. In temperate regions of southern Chile, where catchments have more pluvial regimes,
hydrologic memory is still an important factor, but water supply is more strongly dependant on the meteorological conditions
25 of the current year, and therefore an extreme drought would have a higher impact on water supply than a persistent but
moderate drought.
1 Introduction
Persistent climatic anomalies may cause alterations in catchment response to precipitation. Catchment dynamics under
unusually multi-year precipitation deficits might, thus, not be correctly predicted based on the interannual variability over the
30 last decades. This applies even when past decades include severe, but shorter dry conditions (Saft et al., 2016a). In other
words, stationarity as commonly assumed for streamflow projections under climate change might not exist (Blöschl and
Montanari, 2010), which poses challenges for hydrological model calibration (Duethmann et al., 2020; Fowler et al., 2016).
1
https://doi.org/10.5194/hess-2020-249
Non-stationary catchment response modulates hydrological functioning. This applies particularly to drought propagation,
i.e., the process leading to soil moisture droughts and hydrological droughts (streamflow and groundwater deficits) under
35 meteorological dry conditions (Van Loon et al., 2014). The severity (duration and deficit) of meteorological, soil moisture
and hydrological droughts vary depending on climate and catchment characteristics. While meteorological droughts are
mainly controlled by regional precipitation, soil moisture and hydrological droughts are additionally controlled by catchment
characteristics. Therefore, under similar meteorological conditions, the severity of hydrological droughts can vary
significantly within a climatic region (Van Lanen et al., 2013). Most drought-related impacts on, for instance, agriculture,
40 ecosystems, energy, industry, drinking-water and recreation depend primarily on groundwater and streamflow deficits (Van
Loon, 2015). Therefore, understanding the geographical variation in hydrological droughts provides critical information for
drought-hazard adaptation and mitigation (Van Loon and Laaha, 2015). In addition, non-stationary catchment responses to
precipitation and precipitation deficits result in a temporal variation of potential drought propagation.
This temporal aspect is becoming increasingly important since many regions around the globe are experiencing
45 unprecedented long dry spells due to global warming and circulation changes, causing unforeseen impacts on water supply
(e.g., Schewe et al., 2014). Recent evidence has shown that under persistent dry conditions, droughts may propagate
differently within the same catchment (i.e., same landscape characteristics and governing runoff mechanisms) under similar
precipitation deficits and temperature anomalies than other years from the historical records. For example, studies in south-
eastern Australia, where the Millennium drought took place for more than a decade (1997-2010), reported changes in
50 catchment functioning (Fowler et al., 2018; Saft et al., 2015; Saft et al., 2016b; Yang et al., 2017). More recently, Garreaud
et al. (2017) reported an unprecedented decrease in annual runoff during of a multi-year drought in central-south Chile —the
so-called megadrought (MD). This amplified response of streamflow to a drought signal may be due to variations of
drainage density related to depleted groundwater levels within the catchment (Eltahir and Yeh, 1999; Van De Griend et al.,
2002), a factor also emphasised by Saft et al. (2016b).
55 In this context, the MD experienced in central-south Chile since 2010 (continuing up to date) offers a unique opportunity to
understand the potential impacts of global changes on hydrology and water supply over wide ranges of hydro-climatic
regions and landscape characteristics. The persistency and extension of the MD have few analogues in the last millennia, and
its causes have been partially attributed to anthropogenic climate change (Boisier et al., 2016, 2018; Garreaud et al., 2017,
2019). This uninterrupted sequence of years with precipitation deficits has led to water scarcity problems affecting various
60 sectors of the socio-ecological system, including impacts on coastal ecology (Masotti et al., 2018), fire regimes (Gonzalez et
al., 2018) and water supply (Muñoz et al., 2020).
To deepen the understanding of the impacts of multi-year dry conditions on water supply, we explore the mechanisms
causing the larger-than-expected hydrological deficits in central-south Chile, with a particular focus on the MD. We
complement previous analyses of the MD in Chile (Garreaud et al., 2017; Muñoz et al., 2020) by incorporating three more
65 years to the MD period, and by focusing on drought propagation over 124 catchments located between 30ºS and 41ºS. The
conceptual framework of our analysis is based on the water balance within a catchment, where the water sourced from
2
precipitation takes different flow pathways and is temporally retained in various stores. For a dry year within a long drought,
we can distinguish three cases: i) stationary drought propagation, when the streamflow deficits are similar to those observed
in isolated years (single year drought) with similar precipitation deficits; ii) intensified drought propagation, when
70 streamflow deficits are larger than those observed in years with similar precipitation deficits; and iii) attenuated drought
propagation, when streamflow deficits are lower than those observed in years with similar precipitation deficits. Based on
previous studies relating groundwater dynamics to non-stationary catchment response to droughts (Carey et al., 2010; Eltahir
and Yeh, 1999; Saft et al., 2016b) we hypothesise that in catchments with longer hydrological memory (i.e., catchments
where water is retained for longer time in different storages such as aquifers, snowpack and glaciers), the propagation of
75 drought during multi-year precipitation deficits is intensified (i.e., larger streamflow deficits than those observed in years
with similar precipitation deficits), when compared to single dry years.
To test this hypothesis, we characterised the historical precipitation and streamflow deficits at the catchment scale, and
followed Saft et al. (2015) to evaluate annual rainfall-runoff (R-R) relationships over the last decades and identified those
catchments where drought propagation during the MD was maintained, intensified or attenuated with respect to their
80 historical behaviour. We analysed catchment memory from observed hydrometeorological data and from the hydrological
processes simulated by a bucket-type model calibrated for each basin. Finally, we related catchment hydrological memory
with shifts in R-R relationships during the MD, and with drought propagation for different types of drought, from extreme
single year droughts to moderate but persistent droughts (including the MD). Furthermore, we addressed a general question
with practical implications: what is worse in terms of water supply, a single year with extreme precipitation deficits, or
85 several consecutive years with moderate deficits?
2 Study region and data
The study area corresponds to 124 catchments in central-south Chile, spanning 9 out of 16 administrative regions between
30ºS and 41ºS (Fig. 1). Hydrometeorological data was obtained from the CAMELS-CL dataset (Alvarez-Garreton et al.,
2018), including catchment-scale daily precipitation, potential evapotranspiration and streamflow time series for the period
90 1979-2018. This dataset also includes catchment characteristics such as topography, land cover, and boundaries.
Hypsometric curves for each catchment were processed based on ASTER GDEM (Tachikawa et al., 2011).
Gaps in monthly streamflow time series were filled up based on a procedure previously used for monthly precipitation data
(Boisier et al., 2016). The method uses multivariate regression models, taking advantage of the streamflow co-variability
among multiple gauging stations in the study region (within or across basins). Thresholds on the model performance
95 (determination coefficient, r2, larger than 0.75) and on the record length (75% of valid data within the target period) defines
the locations and months to be filled-up. From the 516 CAMELS-CL catchments, we selected those with at least 15 years of
streamflow data (n=343).
3
The CAMELS-CL dataset provides snow-data (daily snow water equivalent time series) only for basins located between
25ºS and 37ºS, and only for 1985-2015. Therefore, we computed catchment-scale solid precipitation fractions and daily
100 estimates for snowmelt over the period 1981-2018 based on the ECMWF ERA5-Land dataset, a land-surface reanalysis
driven by the ERA5 atmospheric reanalysis. ERA5-Land results from an offline, post-simulation carried out with land
surface component used in ERA5, including a higher spatial resolution (9 km) and other improvements (Copernicus Climate
Change Service, 2019).
105 Figure 1: Study catchments with mean annual precipitation (panel a), mean annual temperature (panel b) and mean annual snow
fraction (panel c) for the period 1979-2018. Panel a presents the 9 administrative regions covered by the study.
Given the latitudinal extent and terrain complexity, the study region features very different hydroclimate regimes (Fig. 1).
Annual precipitation ranges from less than 100 mm to the north to more 3000 in the southern part. Precipitation also increase
substantially towards the west due to the orographic effect exerted by the Andes on the predominantly westerly atmospheric
110 flow. The elevation of the Andes range also modulates the snowfall/rainfall partition, being larger in the semi-arid, northern
4
part of the study region, and leading to contrasted precipitation and runoff seasonal cycles. In central Chile precipitation has
a marked Mediterranean-type regime, with most storms occurring during the winter and with very dry summer conditions.
Streamflow, instead, peaks on spring and early summer in snow-dominated basins. Based on this range of climatic
conditions within the study catchments, which in turn affect hydrologic behaviour and drought propagation (as analysed in
115 the following sections), we classified the study catchments into two groups. We refer as semi-arid catchments to the basins
located between 30ºS and 34ºS, characterised by long-term annual precipitation values of 448 mm on average (ranging
between 253 to 980 mm). The basins located between 34ºS and 41ºS are referred as temperate catchments and feature long-
term annual precipitation values of 1824 mm on average (ranging between 672 to 3379 mm).
3 Methods
120 3.1 Drought characterization
Droughts were characterised by annual precipitation and streamfllow anomalies at the catchment-scale. We computed
relative anomalies of hydrological years (April-March) as deviations from climatological means (period 1979-2009, i.e.,
excluding the MD), normalised by the climatological mean of each time series. These relative anomalies are easy to interpret
and commonly used in drought impact planning (Van Loon, 2015), but also have some limitations. Very large anomalies are
125 obtained when the long-term mean is small. Furthermore, neither absolute nor relative deviations provide information about
how unusual the deficits are at specific locations. Therefore, we also computed the annual deviations normalised by the
standard deviation of the complete time series for the period 1979-2009 (i.e., z-scores).
To characterise the MD and asses how unusual this period has been in comparison to previous decades, we computed the 8-
year mean for each water flux (precipitation and streamflow) during the MD. We compared them with historical eight-year
130 average values. To increase sample size, we computed 100 synthetic values by randomly sampling and averaging eight
annual values from the historical records.
3.2 Changes in catchment response during multi-year droughts
Stationarity in drought propagation under several consecutive years of precipitation deficits (during the MD) was assessed by
following the procedure suggested by Saft et al. (2015) to identify significant shifts in annual R-R relationships over
135 Australian catchments during the Millennium drought. These authors showed that prolonged rainfall deficits resulted in
shifts in R-R relationships at the catchment scale, and Saft et al. (2016b) related the shifts to catchment characteristics
(aridity index and rainfall seasonality) and soil and groundwater dynamics. The physical mechanisms likely associated with
these factors were discussed by Saft et al. (2016b), but not explicitly modelled.
In this study, for each catchment, we linearized annual R-R relationships by applying log-transformation to annual
140 precipitation and runoff time series. We performed a global test to validate linear model assumptions with the R-package
5
gvlma (Peña and Slate, 2006). From the initial sample (n=343), we selected 124 basins where the linear assumptions were
fulfilled and where the annual rainfall explained more than 50% of the variance in annual runoff (r2 larger than 0.5).
For each catchment, we tested if the R-R relationship during the MD (2010-1018) was different to the R-R relationship
during the previous 40 years (1979-2009), by performing the analysis of variance model from R-package aov (Chambers et
145 al., 2017) to the intercept parameter from the linear regressions (see Eq. 1 from Saft et al. 2015). From this analysis, we
defined two types of cases: i) catchments with a significant shift in R-R relationship at a 5% significance level, and ii)
catchments that did not experience a significant shift (test p-value greater than 0.05). For those catchments experiencing a
significant shift, we computed the degree of change in R-R regressions, as the relative difference between streamflow
estimations from both linear regressions, given the same precipitation value. This characteristic precipitation value was
150 defined for each basin as the average precipitation during the MD period (2010-2018).
3.3 Hydrological model
Since observations of groundwater storages and fluxes were not available, we run the HBV model (Bergström, 1972;
Lindström et al., 1997) to simulate streamflow and other fluxes for each of the 124 study catchments. The HBV is a bucket-
type model that simulates the main hydrological process in a catchment through a number of routines. In the snow routine,
155 snow accumulation and melt are simulated based on a simple degree-day approach. A variable fraction of all melted and
rainfall water is retained in the soil depending on the current soil water level. The remaining part is transferred to the
groundwater routine. In this routine, groundwater storage is represented by two boxes, both with linear outflows. Finally, the
simulated outflows from the groundwater stores are routed using a simple routing scheme, leading to the total streamflow.
Besides streamflow, also time series of a number of other fluxes and storages can be obtained from the model, such as actual
160 evaporation, soil water storage, or the different streamflow components (baseflow, intermediate flow and peakflow).
As a result of the continuous simulation of the different storages within a catchment, the HBV model allows considering
memory effects. However, it has to be noted that the model is a simplified representation of actual hydrological processes,
which limits its capability to simulated long-term memory effects. Memory effects in the HBV model are caused by soil
water storage as this store can accumulate precipitation deficits. The groundwater stores in HBV cannot drain below the
165 level where streamflow ceases and, thus, represent only the dynamic storage in a catchment (Staudinger et al., 2017), which
means that memory effects caused by groundwater stores are limited in the model. This makes the model a useful tool to
distinguish between memory effects caused by soil water storage and groundwater storage. The HBV model has been
implemented in several software packages. Here we used the version HBV light (Seibert and Vis, 2012).
The HBV model was calibrated using a genetic algorithm (Seibert, 2000) with parameter ranges similar to those suggested
170 earlier (e.g., Seibert and Vis, 2012). The 14 free parameters values were derived after 3500 model runs. For each catchment,
100 independent calibration trials were performed based on the non-parametric variation of Kling-Gupta efficiency (Pool et
al., 2018), which resulted in ensembles with 100 parameter sets.
6
3.4 Runoff mechanisms and hydrological memory
The hydrological memory of a catchment is a composite of the response times associated to the different physical
175 mechanisms transferring and storing water through the basin. Thus it has been qualitatively related to the presence of
aquifers, lakes and snow (Van Loon and Van Lanen, 2012). As a consequence, there is no unique way to quantify
hydrological memory. For example, catchment memory has been assessed based on soil moisture and groundwater dynamics
(Agboma and Lye, 2015; Peters et al., 2006), on streamflow recession curves (Rodríguez-Iturbe and Valdes, 1979), on lag-
correlations between soil moisture and other fluxes within the catchment (Orth and Seneviratne, 2013), and based on
180 recovery times from droughts (Yang et al., 2017).
In this study, we analysed the main runoff generation mechanisms within the 124 study catchments and then assessed
hydrological memory by computing different lag-correlation statistics. The mechanisms were explored by characterising the
hydrological regimes of the study basins and identifying snow-dominated and pluvial catchments. Snow-dominated
catchments were defined as those where the long-term snow fraction, computed as the average of ERA5-L solid to total
185 precipitation ratio for the period 1981-2018, was larger than 0.3. We assessed the importance of groundwater mechanisms on
streamflow generation based on the baseflow index (BFI), computed for each basin as the mean annual baseflow (defined as
flow from the lower groundwater box in the HBV model) normalised by mean annual streamflow for the entire 1979-2018
period.
To assess the hydrological memory, we computed the correlation between precipitation (concentrated mainly in winter of a
190 certain year n) and the generated streamflow and baseflow of the following seasons (up to fall of year n+1). Further, to
remove the effect of the precipitation of the current year in annual R-R regressions, we computed the correlation (r2) of the
R-R regression residuals from Sect. 3.2 (i.e., annual streamflow not explained by the current precipitation) to the
precipitation from the previous year. This represents the information gained when the precipitation from the previous year
(lagged precipitation) is incorporated in the R-R relationships. We discuss the physical mechanisms explaining hydrological
195 memory, and relate catchments memory with the shifts in R-R relationships and the propagation from meteorological to
hydrological droughts during extremes dry years.
4 Results
4.1 Droughts over the last decades
Heatmaps in Figure 2 illustrate the deviations of precipitation and streamflow with respect to their climatological mean
200 values for the period 1979-2009, normalised by mean values (relative anomalies, panels a and b) and normalised by standard
deviations (z-scores, panels c and d). The unprecedented dry conditions during the last decade is evident. The spatial pattern
of precipitation deficits for single years during the MD is characterized by general negative anomalies for the 124
catchments between 30°S and 41°S (Fig. 2a). Over the last four decades, we see few analogues to this spatial pattern,
7
210
205

Lat Lat
-40
-39
-38
-37
-36
-35
-34
-33
-32
-31
-40
-39
-38
-37
-36
-35
-34
-33
-32
-31
1979 1979 1979 1979
1980
1980 1980 1980 1980
1980
1981 1981 1979 1981 1981
1982
1982 1982 1982 1982
1982
1983 1983 1980 1983 1983
1984
1984 1984 1981 1984 1984
1984
1985 1985 1985 1985
1986
1986 1986 1982 1986 1986
1986
1987 1987 1987 1987
1988
1988 1988
1983 1988 1988
1988
1989 1989 1984 1989 1989
1990
1990 1990 1990 1990
1990
1991 1991 1985 1991 1991
1992
1992 1992 1986 1992 1992
1992
0
1993 1993
50
1993 1993
50
Pa_def
Pa_def
−50
−50
1994
1994 1994 1987 1994 1994
1994
1995 1995 1988 1995 1995
1996
1996 1996 1996 1996
c) Annual precipitation
1996
1997 1997 1989 1997 1997
1998
1998 1998 1998 1998
1998
1999 1999
1990 1999 1999
2000
2000 2000 1991 2000 2000
2000
2001 2001 2001 2001
2002
2002 2002 1992 2002 2002
0
2002
50
2003 2003 2003 2003
Pa_def
1993
−50
2004
2004 2004 2004 2004
2004
z-score (-)
2005 2005 1994 2005 2005
2006
2006 2006 2006 2006
2006
2007 2007
1995 2007 2007
a) Annual precipitation anomalies (%)
2008
2008 2008 1996 2008 2008
2008
2009 2009 2009 2009
2010
2010 2010 1997 2010 2010
2010
2011 2011 1998 2011 2011
2012
2012 2012 2012 2012
2012
2013 2013 1999 2013 2013
2014
2014 2014
2000 2014 2014
2014
2015 2015 2015 2015
2016
2016 2016 2001 2016 2016
2016
2017 2017 2017 2017
2018
2018 2018
2002 2018 2018
2018
2003
1979 1979 1979 1979
1980
1980 1980 2004 1980 1980
1980
8
1981 1981 2005 1981 1981
1982
1982 1982 1982 1982
1982
over the study period and the last century (Garreaud et al., 2017).
1983 1983 2006 1983 1983

1984
1984 1984 1984 1984
1984
1985
2007 1985
1985 1985
1986
1986 1986 2008 1986 1986
1986
1987 1987 1987 1987
1988
1988 1988
2009 1988 1988
1988
1989 1989 2010 1989 1989
1990
1990 1990 1990 1990
1990
1991 1991 2011 1991 1991
1992
1992 1992 2012 1992 1992
1992
0
1993 1993
50
1993 1993
50
Qa_def
Qa_def
−50
−50
1994
1994 1994 2013 1994 1994
1994
d) Annual streamflow
1995 1995 1995 1995
1996
1996 1996
2014 1996 1996
1996
corresponds to one study catchment and the catchments are sorted from north to south.
1997 1997 2015 1997 1997
1998
1998 1998 1998 1998
1998
1999 1999 2016 1999 1999
2000
2000 2000 2017 2000 2000
2000
2001 2001 2001 2001
2002
2002 2002 2002 2002
2002
2003 2003 2003 2003 1979
z-score (-)
2004
2004 2004
1979 2004 2004
2004 1980
2005 2005 1980 2005 2005
2006
2006 2006 2006 2006
2006 1981
b) Annual streamflow anomalies (%)
2007 2007 1981 2007 2007

2008
2008 2008 2008 2008
2008
1982
1982
2009 2009 2009 2009 1983
2010
2010 2010 1983 2010 2010
2010
2011 2011 2011 2011 1984
2012 2012
1984 2012
2012
2012 2012 1985
2013 2013 1985 2013 2013
2014
2014 2014 2014 2014
2014 1986
2015 2015 1986 2015 2015
2016 2016 2016
2016
1987
2016 1987 2016
2017 2017 2017 2017 1988
2018
2018 2018 1988 2018 2018
2018
1989
1989
1990
1990
1991
1991
1992
0
z-score
1992
50
0.0
0.5
1.0
1.5
1993
−50
−1.5
−1.0
−0.5
1993
(-)
1994
Anomalies
Pa_zscore
1994
1995
1995
Pa_def (%)
1996
millennium, Garreaud et al., 2017) and 2007. The multi-year persistence of the MD spatial pattern, however, has no analogue
including the 1988-1990 three-year drought, and the single-year droughts in 1996 and 1998 (one of the driest years in the last
normalised by standard deviation). The MD period (2010-2018) is highlighted in a grey box). Each row in the heatmaps
Figure 2: Annual deviations from climatology (1979-2010) for catchment-scale precipitation (panel a), streamflow (panel b). Panels
c and d present the z-score of the annual precipitation and streamflow, respectively (computed as deviations from mean
1996
1997
1997
1998
1998
1999
1999
2000
2000
2001
2001
2002
230
225
220
215

8-year average (mm)
8−year mean Q (mm)
8−year mean Q (mm) 8−year mean P (mm)
8−year mean P (mm)
10
100
1000
300
1000
3000
10
300
10
100
1000
300
1000
3000
100
1000
1000
3000
●
●
●
●
●
●
●
●
−40.9333 −40.9333 −40.9333 −40.9333
●
●
●
●
−40.8856 −40.8856 −40.8856 −40.8856
●
●
●
●
−40.7889 −40.7889 −40.7889 −40.7889
●
●
●
●
−40.7136 −40.7136 −40.7136 −40.7136
●
●
●
●
−40.6181 −40.6181 −40.6181 −40.6181
●
●
●
●
−40.3842 −40.3842 −40.3842 −40.3842
●
●
●
●
−40.2667 −40.2667 −40.2667 −40.2667
●
●
●
●
−40.25 −40.25 −40.25 −40.25
-40
●
●
●
●
−39.8583 −39.8583 −39.8583 −39.8583
●
●
●
●
ones (north of 34ºS).
−39.7667 −39.7667 −39.7667 −39.7667
●
●
●
●
−39.6675 −39.6675 −39.6675 −39.6675
●
●
b) Runoff
●
●
−39.55 −39.55 −39.55 −39.55
●
●
−39.3833 −39.3833 −39.3833 −39.3833
−39.36 −39.36 −39.36 −39.36
● ●
● ●
−39.3333 −39.3333 −39.3333 −39.3333
● ● ● ●
●
● ● ● ●
●
−39.2667 −39.2667 −39.2667 −39.2667
−39.2561 −39.2561 −39.2561 −39.2561
● ●
● ●
−39.15 −39.15 −39.15 −39.15
●
● ● ●
●
● ● ●
−39.1 −39.1 −39.1 −39.1
a) Precipitation
●
●
−39.0833 −39.0833 −39.0833 −39.0833
●
●
−39.0144 −39.0144 −39.0144 −39.0144
-39
●
● ● ●
●
● ● ●
−38.9833 −38.9833 −38.9833 −38.9833
−38.8653 −38.8653 −38.8653 −38.8653
● ●
● ●
−38.85 −38.85 −38.85 −38.85
● ● ●
●
● ● ●
●
−38.78 −38.78 −38.78 −38.78
●
●
●
●
−38.6867 −38.6867 −38.6867 −38.6867
●
●
●
●
−38.6833 −38.6833 −38.6833 −38.6833
●
●
●
●
−38.6167 −38.6167 −38.6167 −38.6167
●
●
●
●
−38.6081 −38.6081 −38.6081 −38.6081
●
●
●
●
−38.4333 −38.4333 −38.4333 −38.4333
●
●
●
●
−38.4303 −38.4303 −38.4303 −38.4303
●
●
●
●
−38.3 −38.3 −38.3 −38.3
●
●
−38.299 −38.299 −38.299 −38.299
●
● ●
●
● ●
−38.2167 −38.2167 −38.2167 −38.2167
●
●
●
●
−38.15 −38.15 −38.15 −38.15
●
●
●
●
−38.149 −38.149 −38.149 −38.149
●
●
●
●
−38.1167 −38.1167 −38.1167 −38.1167
●
●
●
●
−37.9647 −37.9647 −37.9647 −37.9647
●
●
●
●
−37.8636 −37.8636 −37.8636 −37.8636
●
●
-38
●
●
−37.85 −37.85 −37.85 −37.85
●
●
●
●
−37.7756 −37.7756 −37.7756 −37.7756
●
●
−37.7333 −37.7333 −37.7333 −37.7333
●
● ●
●
● ●
−37.7157 −37.7157 −37.7157 −37.7157
●
●
−37.7106 −37.7106 −37.7106 −37.7106
●
● ●
●
● ●
−37.5503 −37.5503 −37.5503 −37.5503
●
●
−37.55 −37.55 −37.55 −37.55
●
● ●
●
● ●
−37.5422 −37.5422 −37.5422 −37.5422
●
●
−37.2333 −37.2333 −37.2333 −37.2333
● ●
●
● ●
●
−37.0667 −37.0667 −37.0667 −37.0667
●
●
●
●
−36.9333 −36.9333 −36.9333 −36.9333
●
●
●
●
−36.9244 −36.9244 −36.9244 −36.9244
●
●
●
●
−36.8667 −36.8667 −36.8667 −36.8667
-37
−36.8386 −36.8386 −36.8386 −36.8386
● ●
● ●
● ●
● ●
−36.8167 −36.8167 −36.8167 −36.8167
−36.7 −36.7 −36.7 −36.7
● ●
● ●
● ●
● ●
−36.6653 −36.6653 −36.6653 −36.6653
9
during the MD (2010-2018). Please note the log-scale of the Y-axes.
−36.6583 −36.6583 −36.6583 −36.6583
● ●
● ●
● ●
● ●
−36.65 −36.65 −36.65 −36.65
●
●
●
●
−36.6167 −36.6167 −36.6167 −36.6167
●
●
−36.55 −36.55 −36.55 −36.55
●
● ●
●
● ●
−36.3758 −36.3758 −36.3758 −36.3758
●
●
●
●
−36.2558 −36.2558 −36.2558 −36.2558
●
●
●
●
−36.2333 −36.2333 −36.2333 −36.2333
●
●
●
●
−36.2303 −36.2303 −36.2303 −36.2303
●
●
●
●
−36.05 −36.05 −36.05 −36.05
●
●
●
●
−36.0167 −36.0167 −36.0167 −36.0167
●
●
●
●
−36.0028 −36.0028 −36.0028
Latitude
−36.0028
●
●
●
●
−35.9667 −35.9667 −35.9667 −35.9667
●
●
●
●
−35.9086 −35.9086 −35.9086 −35.9086
●
●
●
●
−35.9 −35.9 −35.9 −35.9
occurring (lower z-score values, indicating higher exceedance probabilities).
●
●
●
●
−35.8167 −35.8167 −35.8167 −35.8167
●
●
−35.75 −35.75 −35.75 −35.75
● ●
●
● ●
●
−35.6169 −35.6169 −35.6169 −35.6169
●
●
●
●
−35.5575 −35.5575
-36
−35.5575 −35.5575
●
●
●
●
−35.55 −35.55 −35.55 −35.55
●
●
●
●
−35.4856 −35.4856 −35.4856 −35.4856
●
●
●
●
−35.4069 −35.4069 −35.4069 −35.4069
●
●
●
●
−35.2783 −35.2783 −35.2783 −35.2783
●
●
−35.2744 −35.2744 −35.2744 −35.2744
●
● ●
●
● ●
−35.1783 −35.1783 −35.1783 −35.1783
●
●
●
●
−35.1731 −35.1731 −35.1731 −35.1731
●
●
●
●
−34.9961 −34.9961 −34.9961 −34.9961
●
●
●
●
−34.9844 −34.9844 −34.9844 −34.9844
●
●
●
●
−34.7186 −34.7186 −34.7186 −34.7186
●
●
●
●
−34.6867 −34.6867 −34.6867 −34.6867
●
●
●
●
−34.4922 −34.4922 −34.4922 −34.4922
●
●
●
●
−34.4311 −34.4311 −34.4311 −34.4311
●
●
●
●
−34.2467 −34.2467 −34.2467 −34.2467
●
●
●
●
−34.0617 −34.0617
-35
−34.0617 −34.0617
●
●
●
●
−33.8111 −33.8111 −33.8111 −33.8111
●
●
●
●
−33.8056 −33.8056 −33.8056 −33.8056
●
●
●
●
−33.7347 −33.7347 −33.7347 −33.7347
●
●
●
●
−33.7219 −33.7219 −33.7219 −33.7219
●
●
●
●
−33.6614 −33.6614 −33.6614 −33.6614
●
●
●
●
−33.5939 −33.5939 −33.5939 −33.5939
●
●
●
●
−33.5875 −33.5875 −33.5875 −33.5875

−33.4961 −33.4961 −33.4961 −33.4961
● ●
● ●
● ●
● ●
−33.3703 −33.3703 −33.3703 −33.3703
●
●
●
●
−33.3414 −33.3414 −33.3414 −33.3414
●
●
●
●
−33.3256 −33.3256 −33.3256 −33.3256
-34
●
●
−32.9164 −32.9164 −32.9164 −32.9164
●
● ●
●
● ●
−32.9072 −32.9072 −32.9072 −32.9072
●
●
●
●
−32.8625 −32.8625 −32.8625 −32.8625
●
●
−32.8572 −32.8572 −32.8572 −32.8572
●
●
−32.8503 −32.8503 −32.8503 −32.8503
●
● ● ●
●
● ● ●
−32.7572 −32.7572 −32.7572 −32.7572
●
●
−32.5017 −32.5017 −32.5017 −32.5017
● ●
● ●
−32.33 −32.33 −32.33 −32.33
● ●
●
● ●
●
−32.2253 −32.2253 −32.2253 −32.2253
●
●
●
●
−32.0708 −32.0708 −32.0708 −32.0708
●
●
●
●
−31.9667 −31.9667 −31.9667 −31.9667
●
●
●
●
−31.9481 −31.9481 −31.9481 −31.9481
●
●
−31.6958 −31.6958 −31.6958 −31.6958
●
● ●
●
● ●
−31.6647 −31.6647 −31.6647 −31.6647
●
●
●
●
−31.5608 −31.5608 −31.5608 −31.5608
-33 -32
●
●
●
●
−31.2281 −31.2281 −31.2281 −31.2281
●
●
●
●
−31.1119 −31.1119 −31.1119 −31.1119
●
●
●
●
−31.0325 −31.0325 −31.0325 −31.0325
●
●
●
●
−31.0119 −31.0119 −31.0119 −31.0119
●
●
●
●
−30.8422 −30.8422 −30.8422 −30.8422
●
●
●
●
−30.8128 −30.8128 −30.8128 −30.8128
●
●
●
●
−30.8006 −30.8006 −30.8006 −30.8006
-31
●
●
●
●
−30.7047 −30.7047 −30.7047 −30.7047

The relative anomalies in precipitation are consistently higher in semi-arid catchments (north of 34ºS) (Fig. 2a), which is
the first decile for 87% of the study catchments (108 out of 124). The average runoff during the MD has been more extreme
streamflow for each basin (Sect. 3.1), with the mean values during the MD plotted in red dots. These plots indicate that the
precipitation, which is due to the strong dependency of streamflow on local terrestrial characteristics. If we compare the
where several precipitation events occur during the year. During the MD in particular, annual precipitation in temperate
of having high anomalies is larger in semi-arid catchments (z-scores closer to zero in Fig. 2c), where few meteorological
Figure 3: Boxplots of 8-year mean precipitation (panel a) and runoff (panel b). The red points correspond to the mean values
MD has been extremely unusual in terms of precipitation and streamflow. The average precipitation during the MD is within
To further characterize the MD anomaly, Fig. 3 presents the frequency distribution of 8-year mean precipitation and
For the case of streamflow, annual anomalies (Fig. 2b) present larger values and larger variations in space and time than
spatial patterns of anomalies (Fig. 2b) and z-scores (Fig. 2d) during the MD, we see that consistently with precipitation,
events contribute to most of the annual accumulation. Thus, the interannual variability is higher than in southern basins,
57%). Still, the moderate deficits experienced in the southern region (median deficit of 25%) have a lower probability of
larger anomalies are observed in semi-arid catchments (extreme dry years up to 90% of streamflow deficits with a median of
exceedance probability of 94.3%. Z-scores during the MD are usually lower for these catchments, compared to semi-arid
basins decreases up to 1.5 times below the historical standard deviation of annual precipitation, which represents an
partly due to the very low annual precipitation values in the northern region compared to the southern region. The probability
−30.4622 −30.4622 −30.4622 −30.4622

than precipitation deficits, with 94% of catchments (117 out of 124) presenting mean runoff values within the first decile.
These values represent the minimum value over the last four decades for some basins located north to 32ºS and correspond
235 to an outlier of the historical distribution for the rest of the catchments.
4.2 Shifts in R-R relationships during the MD
Given the relatively small scale of catchments in Chile, runoff in most of them has a strong dependency to the interannual
precipitation variability, explaining typically ~ 75% of the streamflow variances. Yet, the annual R-R relationships observed
during the MD changed significantly for some of the catchments within the study region (examples in Fig. 4, a–c), while
240 other catchments showed no significant change (Fig.4, d-f). Table 1 summarises the number of catchments that experienced
a significant shift in R-R relationship during the MD, and their associated sifts. This table also summarises the results for
those catchments with no change. From the 124 studied catchments, 66.9% had a significant change in the R-R relationship
during the MD and 33.1% had not (Table 1).
For the catchments experiencing a change, the historical R-R regressions consistently underestimate the runoff deficits
245 during the MD. 99% (82 out of 83) of the catchments with significant change in R-R relationship during the MD, had a
negative shift. That is, an intensification in drought propagation. For similar precipitation deficits as other dry years,
observed streamflow during the MD in semi-arid catchments was up to 77% lower than those predicted by the historical R-R
relationship. In temperate catchments, these shifts reached up to 43% (Table 1).
While these results provide insights about changes in R-R relationship during a multi-year period, it does not indicate if the
250 changes are progressive, i.e., if the basins progressively generate less water for a given precipitation amount, compared with
the historical behaviour. We address this in the following sections, where runoff generation mechanisms, hydrological
memory, and drought propagation are analysed in detail.
10
500 ●
1250
Lat −31.8 ●● ● Lat −
https://doi.org/10.5194/hess-2020-249 400 ●
change chan
Preprint. Discussion started: 18 June 2020 1000
300 ●
● ●
200 750 ●●
●●●
●●
●
100 ●● ●● ● ●
● ● ●●
●●
● ●● ● 500 ●● ●
● ●●●● ●
0 ●
●● ● ●
●
200 400 600 500
500 ● 500 ● ● ● 3000

500 ●
1250 3000 ●● 3500
Annual runoff (mm)
Annual runoff (mm)

1250 ●● 3500
Lat −31.8 ●● ● Lat −31.8 Lat −33.6 ● ● ●● ●
Lat −33.6Lat −35.3 Lat −35.3Lat ●−39.4
Lat−33.3 3000● Lat −
● 400 ● ● ● ● ● ● ● ●
400 change change change ● change change● ● ● ● no change
change change ●
● ●●
●● betw
chan
● 2500 ● 2500
400 ●● ●●● ●
● 3000
●
1000 ● ● 1000 ● ● ● ● 3000 ●
● ● ●● ●●● ●●●
300 ● 300 ●●
● ● ●●● ●●●
● ● ●
● ●●● ●
● ●● ● 2000 ● ● ● ● 2000
● ● ● ● ● ●● 2000 ● ● ●
300 2500 ● ●
● ● ● 2500
200 750 ● ●
●●●●●
200 750 ●●● ● ● ● ●● ● ●●
● ● ●● ● ●
●●●
●● ●● ● ●● ● ● ● ●● ●●● ●
-75
● ● ● ●
1500 ● ● ●● ●● ●
●● ●● ● 100 ●● ●● ●● ● 1500●●●
● ●●
●● ● ●
●●●● 2000 ●
100 ●● ● ● ●● ●● ● 200 ●● ●
2000 ●● ●● 1000
● ● ●
● ●
● ● ●
500 ●
●●
●● 500 ●● ● ●
●● ● ●● ●● ● ●●
●● ●
● ●● ● ● ●●●● ●● ● ● ● ● ● ●● ● ●
d)
0
● ●
●
●●
●
●
●
● a)
0 ●
●● ●
●
● ●
●1000 ● ●●
●● ● ●
●
1000 ●● ● ●
1500 ● ●
1500
200 400 600 200 400
500 600 1500 2000
1000 500 1000 1500 2000
100015002000250030003500 300 600 1200
900 3000
1200
3500 4000 1000
100015002000250030003500
2000 2500 200
● ● 500 3000 ●
1250● ●● 800
500 800 ●
Annual runoff (mm)
Annual runoff (mm)

1250
3500 ●
● ● ●500
1250
Lat ●● Lat −33.3Lat ● ● ● ●● Lat −36.4 ● ● Lat −38.6 ●● Lat −
Lat −33.3
−33.6
● 3000 Lat −36.4
−35.3 ●
●● ● 3000
3000
●
Lat
Lat −38.6
●●
−39.4 ●
● ● ● 3500 Lat −40.6 ● ● ● ●
● 1250no change ● ● ● ● ●●●
Lat −31.8
no change● ●
change ●● ● ●●● Lat −33.6
between
change
● ●● ●
● between
Lat −35.3 ● ●
no change
change
● ●
● ●●
● no
Latchange
−39.4
no change ● ●● ● no ch
400 2500 ● ● ●● ● ●●
400
400 1000 ●●●● ●
●●
●
● ●● 3000
1000 ●● ● ● ●●● ●
● 1000
● ● ●● ●● ●
change ● change ● ● ●● change ●
●● ● ● change ● ● ●●
● ● ● 600
● ●● ● ● ●●● ● 2500 ● ●
●
●●● 3000 600 ● ● ● ●● ●●●
● ● 1000 ● ● ● ● ●● ● ● ●● ●● ● ● ●
300 ● ●
● ●
●● 2000● ● ●● ●●●
2000 ● ●
●●
●● 2000 ●● ● ●●● ● ●●●●
●
● ●●●●●
● ●
●●● ●
● ●
● ● 300 ● ● 300 ● ● ●● ● 2500 ●
●
●●
● ●●● ●●
●
● 750 ● ● ●
● ●
●● ●
750 ●● ● ● ● ●●● ● ●
●●●
750 ●● ●
●●●● ●● ● ● ●● ●●● 2000 ● ●●● 2500
●
● ●●●●● ●
●●● ● ● ●●● ●● ● ●●●●
● 200 ●●●● ● 750 ● ●● ● ● ●●● ●● ●●● ●
● ● ● ● 400 ● ● ● ● ● ●●
●●●●
400
●● ●●●●● ●
●●
● ●
1500 ● ●● ● ●
● ●
●
●
● ● ● ● ● ● ● ●●
● ● ● ● ●
●
200 ●● ●●● 200 ● ● ● ● ● ● 1000 2000 ●●●● ● ● 500
●
●●● ●
● 1000
●● ● ● ● ●● 1500 500
● ● ●●● ● ● ●● ●●●●
100 500 ●
● ● ●
●
●
●●●●●●●
● ●
●
b) ● ●
500 ●
●
●●
●
●● ●●● ● ●●
●
● ● ● ●● ●
● ●● ●
● 2000
● ●
●●● ●●
●
e) ● ●
● ● ●● ● ●1000● ● ●●● ●● ●
● ●● ● ● ● ●
● ●●●● 1500 ● ● 200 200
0 ●● ●
● ● 1000 ● ●● ●
● 1500 750 1000 1250
400 600 300
500 1000 600 1500 1200 300
900 2000 600
1000 90020001200
1500 2500 3000 1000 1500
100015002000250030003500 750 2000
2000 10002500
2500 12503000
3000 1500
35001750
4000 1250 1500
15001750 1750 2000 12
200 400 600 500 1000 1500 2000 100015002000250030003500 2000 2500 3000 3500 4000
● ●
● ● 1250 ● 800 ● ●
● 3000 Lat −36.4
500 3000 ●● 3500 Lat −38.6 1250 Lat −40.6 ●● 800 ●
●
Annual runoff (mm)
●
●● ●● ● ●
Annual runoff (mm)
● ● ● ● ● ●
.6 ● ● Lat −35.3
Lat −33.3 Lat −39.4
Lat −36.4 ● Lat −38.6 ●● Lat −40.6 ● ●
e● ● between ●
●●● ● ●
3000
● ●● ●
no change ●●●●● no change ●● ● ●
●●
● ●● no change
change ● change ● ● ● ●
between ● ● ●●
●●● no change no change
● ● 1000 ●● ● ●
● 400 2500 ●● ●
● ●●● 3000 ● ● ●●
● ●● ● ●
1000 600 ●
●
● ●
●● ●●● ●
●
●
●●● ● ●●●●●● ● ● ●
●● ● ● 600 ●
●
●● ● ●● ● ●
● ●●
●
● 2000 ● ● ●●●
●●● ● ●● ●●●●●● ●● ● ●●●● ●
●●
●● 2000 ● ● ●
● ●
●●● 2000 750 ●●● ● ● ●●●●●
●●● ●●●
●● 300 2500 ●●
● ● ●
●
●●● ●●
●
●● ●● ●●●●●●●●● 750 ● ● ●
●●● ●● ● ●●● ●● ●●● ● ● 400 ● ●
● ●● ● ● ●●● ● ● ● ●●●
● ●●
● 400
● ●●●●● ●
1500
1000 ●●●●
● ●● ●●
● ● ● ●●● ● ●
●
500● ● ● ● ●● ● ●
200 ●● ● ●
● ●
● 1000 2000 ● 500 ●●
●● ●●
● ● ● ● ●
● ●
1000
● ●
● ● ● ●●
●● ●●
● c) ●
●●
200● ●
● ●●
● f)
1500 200
600 900 1200 1000 1500 2000 2500 3000 750 1000 1250 1500 1750 1250 1500 1750 2000
1000 1500 2000 300 100015002000250030003500
600 900 1200 1000 1500
20002000 2500
2500 3000
3000 3500 4000 750 1000 1250 1500 1750 1250 1500 1750 2000
1000 2000 3000
Annual precipitation (mm)● ● Annual precipitation (mm)

● 1250 800 ● ●
.4 ● Lat −38.6 Lat −40.6 ● ●
●● ● ● ● ●
●●
n ● ● no change ●● no change ● ●
● ● 1000 ●
● ● ● ●
● ●
●●● ●●●●
600 ●● ●
●● ● ● ●●
●
● ●●
● ● ●
● 255 Figure 4: Annual runoff ●(y-axes) and precipitation (x-axes)
● ● for 6 selected catchments with a significant change in R-R relationship
●● ●
●●
●
● 750
●● ●
● ● ●
●●
●
●
●● during the megadrought ● ●●●
●
(panels
● a-c) and without
400 change (panels d-f). The years within the MD are highlighted in magenta (panels
● ● ●
●
a-c) and grey (panels
● ● d-f). The map shows the● catchments ● with change, coloured by their shift in R-R relationship. The 41
500 with no
catchments ●●
significant change are outlined in blue.
● ● ● ●
200
00 2000 2500 3000 750 1000 1250 1500 1750 1250 1500 1750 2000
260
11
Table 1: Number of catchments with significant shift and no shift in R-R relationship during the MD. The shift magnitude in
catchments with change is also presented.
Semi-arid Temperate
Total
(30ºS-34ºS) (34ºS-41ºS)
Change 31 52 83
Mean shift (min, max) -38% (-77%, -13%) -16% (-43%, 27%) -24% (-77%, 27%)
0.25 0.25
No change 5 36 41
0.20 0.20
Total 36 89 124
0.15 0.15
4.3 Hydrological memory
0.10 0.10
265 The typical hydrological regimes of the study basins are presented in Fig. 5. Based on a long-term snow fraction threshold of
0.05 0.05
0.3, we classified 49 catchments as snow-dominated and the remaining 75 as pluvial catchments. Most snow-dominated
basins are located towards the north of the study region, given the higher
0.00 elevation of the Andes in central Chile (Fig.1). 83% 0.00
apr maywhile
(30 out of 36) of semi-arid catchments were classified as snow-dominated, jun jul aug sep
78.4% (69 oct
out nov dec of
of 89) jantemperate
feb mar ap
catchments were classified as pluvial. Towards the south, temperate catchments are controlled by winter precipitation with a
0.20
270 seasonal synchronicity between precipitation and streamflow. If solid precipitation does occur, it melts within the same
season.
0.15 varname
0.25 a) 49 snow-dominated basins 0.25 b) 75 pluvial basins P
0.10 Q
Normalised water flux (-)
BF
0.20 0.20 snowmelt
0.05
0.15 0.15
0.10 0.00
0.10
aprmayjunjulaugsepoctnovdecjanfebmar
0.05 0.05
0.00 0.00
apr may jun jul aug sep oct nov dec jan feb mar apr may jun jul aug sep oct nov dec jan feb mar
Figure 5: Hydrologic regimes for snow-dominated catchments (panel a) and pluvial catchments (panel b). The mean monthly
fluxes0.20
of precipitation (P), streamflow (Q), baseflow (BF) are normalised by the annual P from CAMELS-CL dataset for the
275 period 1979-2018. Since snowmelt was obtained from a different dataset, the mean monthly snowmelt fluxes are normalised by the
mean annual precipitation from ERA5-Land.
0.15 varname
To explore the importance of groundwater within the study catchments, in Fig. 6 we present the BFI computed from HBV
P
simulations. Q
0.10 The HBV model calibration resulted in overall acceptable simulation performance, with values of 0.72, 0.89 and
BF
snowmelt
0.05
12
0.00
aprmayjunjulaugsepoctnovdecjanfebmar
1.00 ● 1.00 ●
● ●● ●● ●
● ● ●
●
● ●
● ●
● ●
●
●● ●●
●
●
0.75 ●
●
● ●●
● ● ● ●
0.75 ●
●
● ● ● ●
●● ●
● ● ●
●
●●●●
●
●● ● ●
●
● ● ●
● ●● ●
●
●● ● ●
https://doi.org/10.5194/hess-2020-249 ●
● ●
● ●●●●
●● ●
BFI
BFI
0.50 ●
●
●
●●
●
●●
●●
●
●
●●
●
●
●● ●
●
● ●
● ●
● ●● ● ●● ● 0.50
● ●● ● ●●
Preprint. Discussion started: 18 June 2020 ●
●●
●
0.25 0.25
●
0.00 0.00
0.00 0.25 0.50 0.75 −75
SF
1.00 ● 1.00
● ● ●●●
● ● ●●●
●●●
●
●
●
●●
●
● ● ●
● ● ● 0.75 ● ●●●
●
●● ● 0.75
●● ● ●● ● ●●
●
0.95 for the 10th, 50th and 90th percentile, respectively. For comparison, the corresponding Kling-Gupta●●●●●●●efficiency
● ● ●●
●● ● ●●values ● ●●●
0.50 ● ●
●●
●
● ●●● ●● ● ● ● ●●●
●●●
●●
●●
● ● ● ● ●
●
BFI
BFI
● ● ●●
●
●
●
●
●● ●●●●
●●●●
● ●● ●● ●
● ●●
● ●●
●● ●
●
● ● 0.50 ●
●
●
●
●●
●●
●●
●
●●
●
●●●●
●
280 (Gupta et al., 2009) were, respectively, 0.57, 0.8 and 0.92. ● ●●
●●
●●
●● ● ●
● ●●
●● ●
0.25 0.25
1.0 ● 1.0 ●
a) ● ●●
●●
●
●
● ● b) ●● ● ● ● ●
0.00 ●
0.00
●● ●
●● ●
● 0.00 0.25 0.50 0.75 0.0
●
● ● ●
●●●
●
● ●
● ●
● ●● ●
0.8 ●
●
●
● ●●
●●
●
●
●
● ●● 0.8 ●
●
●
●
●● ● ●
● ●●●
Hydrological memory (r2 lagged BF−Pw) Hydro
●●●●
● ●
●
● ●●
●
● ●
● ●
●●
BFI
BFI
●● ●
BFI
BFI
●●●
0.6 ●
●
●
●
●
●
●
● ● ● ●
●●
●●
●
●●
●●
●●●
●●
●● ●
●
●
● ●●
● ●● ●
● ●
●
●●
● ● ● ● ● ●●
● ● 0.6
●
●●
●
●
● ●●
●●
●
● ●●
●
● ●
●●
●
●
●
●
Lat. (º)
LAT
0.4 ● ●
● ● ●
●● ● ●
● ●
● ●
● ●
●●
●
●●
●
●●
● ●
●
●
●
●
● 0.4 ●● ●
●● ● −32.5
● r2 = 0.51 ● r2 = 0.40 −35.0
0.2 ● ●
−37.5
0.00 0.25 0.50 0.75 −75 −50 −25 0 25
1.00 ●●
SnowSF ShiftShift
in R−R (%) −40.0
fraction R-R (%) ●●●●
●
●● ●●●●●●
● ●
●●●●●●●
● ●
●
●● ●
0.75 ● ●
●
● ●● ●
●
●●
●
● ●● ●●●
●●
●
●
●●●● ●● ● ●
●●
● ●
1.0 ● ● 1.0 ●
●
●
0.50 ●
●
● ● ● ●●●● Area
size (km2)
BFI
●
●
●●●●
●●●
●●●
c) ● d) ●●●
●
●
●
●
●
●
●
●
●
●
● ●
●●● ●● ●
●
● ●
● ●
● ●●
●
● ●
●
●
● ●● ●●●
●
●●●
● ●
●
●
● ● ● ●●
●●
● ● ● 5000
●● ●● ●
● ● ● ●
● ● 0.25
0.8 ●
● ● ● ● ●● 0.8 ●
●
●
●
● ● ● ● 10000
●
● ●
● ●
● ●● ● ● ● ● 0.00 ●
● ●
●
●● ●
●
●
●
●
●
● ●●
●●●
●
●
●● ● ● 0.00
0.25
0.50
0.75
●
● 15000
BFI
BFI
●● ●● ●
BFI
BFI
●
● ●● ●●
● ●
0.6 ● ● ● ● ● 0.6 ● ● ● ● ●● ●
● ●
SF
● 20000
●
● ●
●
●●
●
●●●
● ●
●●●
● ●●●●
●
●
●
●●● ●
●
●
●
●
●●
●
● ●●● ●
●
●
● ● ● ●● ●
● ● ● ●
●
● ●
●
●● ●
● ● ● ●●
●
●● ● ● ● ●
● ● ● ● ● ●● ●
● ●●●
●● ● ●
●● ● ●
●
●●
● ●●
●● ●● ●
0.4 ● ● ●●
● 0.4 ●● ● ●● ●
● ●
● ●● ● ● ●
● ●
r2 = 0.48 r2 = 0.25
0.2 ● 0.2 ●
0.00 0.25 0.50 0.75 1.00 0.0 0.1 0.2 0.3

HydrologicalHydrological
memory (r2 lagged
memory
BF−Pw) Hydrological memory
Hydrological (delta
memory
r2)
r2 (BF, lagged P) r2 (R-R residual, lagged P)
Figure 6: BFI as a function of catchments snow fraction (panel a), and the shift in R-R relationships (panel b). Panel c relates BFI
with hydrological memory quantified as the r2 between winter precipitation and fall-baseflow from the following year. Panel d
relates BFI with hydrological memory quantified as the r2 between the residuals in R-R regressions, with the annual precipitation
285 from previous year.
LAT
Although the BFI does not provide explicit information about the processes behind the slow flow contribution to runoff (e.g.,
subsurface routing and drainage density, storages), it is a proxy of the overall baseflow contribution to runoff. Groundwater
−32.5
mechanisms depend on several factors that vary across basins, including topography, soil properties, geology, drainage area
and water table levels (Robinson and Ward, 2017). Observations from most of these components are commonly not
−35.0
290 available, so here we focus on the water entering the soil routine in HBV (total precipitation minus evapotranspiration) and
−37.5
the water leaving the system as baseflow (represented by the BFI). These two components are related (Fig. 6a), with snow-
dominated
1.0 catchments contribution of baseflow to runoff (r2 = 0.51). In Fig. 6b we show that, for those
● having a larger −40.0
●●●●●●●●● ●● 2
catchments with
● significant
● ● ●●
● change (Sect. 4.2), higher BFI values are associated to larger shifts in R-R relationships (r =
●
0.8 ●
0.51), and thus●
●
●
●
●
to ●the
●●●
●
●●
● ●
●intensification of drought propagation during the MD, with respect to historical annual responses to
●●
●
● ●●
● ●
●
●
●
BFI
● ●
0.6 ●
●●
●●
●●
●
●
● ●● size 13
●
●
●
●
●
●
●
●●
●
●
●●●
●●
●
●●
●
●
●
●●●●
●
● ●
● ● ●●
●
●
●
●
●
●
●●●
●● ●
●●● ● ● 5000
0.4 ●
● ●
●
●
● ●
●
● 10000
Correlation coeff.
r2 r2
0.00
0.25
0.50
0.75
0.00
0.25
0.50
0.75
1.00
0.00
0.25
0.50
0.75
1.00
−40.9333
−40.8856
−40.7889
−40.7136
−40.6181 −40.9333 −40.9333
−40.3842 −40.8856 −40.8856
−40.2667 −40.7889 −40.7889
−40.25 −40.7136 −40.7136
−39.8583 −40.6181 −40.6181
−39.7667 −40.3842 −40.3842
−39.6675 −40.2667 −40.2667
−39.55 −40.25 −40.25
−39.3833 −39.8583 −39.8583
−39.36 −39.7667 −39.7667
−39.3333 −39.6675 −39.6675
−39.2667 −39.55 −39.55
−39.2561 −39.3833
−39.36 −39.3833
−39.36
−39.15
−39.1 −39.3333
−39.2667 −39.3333
−39.2667
−39.0833
−39.0144 −39.2561
−39.15 −39.2561
−39.15
−38.9833 −39.1 −39.1
−38.8653 −39.0833 −39.0833
−38.85 −39.0144 −39.0144
−38.78 −38.9833 −38.9833
−38.6867 −38.8653 −38.8653
−38.6833 −38.85 −38.85
−38.6167 −38.78 −38.78
−38.6081 −38.6867 −38.6867
−38.4333 −38.6833 −38.6833
−38.4303 −38.6167 −38.6167
−38.3001 −38.6081 −38.6081
−38.3 −38.4333 −38.4333
−38.2167
310
305
300
295
−38.1501 −38.4303
−38.3 −38.4303
−38.3
−38.15 −38.2167 −38.2167
−38.1167 −38.15 −38.15
−37.9647 −38.1167 −38.1167
−37.8636 −37.9647 −37.9647
−37.85 −37.8636 −37.8636
−37.7756 −37.85 −37.85
−37.7333 −37.7756 −37.7756
−37.7167 −37.7333 −37.7333
−37.7106 −37.7167 −37.7167
−37.5503 −37.7106 −37.7106
−37.55 −37.5503 −37.5503
−37.5422 −37.55 −37.55
−37.2333 −37.5422 −37.5422
−37.0667 Correlation coeff. −37.2333 −37.2333
−36.9333 −37.0667 −37.0667
−36.9244 −36.9333 −36.9333
−36.8667 −36.9244 −36.9244
−36.8386 −36.8667 −36.8667
−36.8167 −36.8386 −36.8386
−36.7 −36.8167 −36.8167
0.00
0.25
0.50
0.75
−36.6653 −36.7 −36.7
−36.6583 −36.6653 −36.6653
−36.65 −36.6583 −36.6583
−36.6167−40.9333
−36.55
−36.65
−36.6167 −36.65
−36.6167
−36.3758−40.8856 −36.55 −36.55

−40.7889
−36.2558−40.7136 −36.3758 −36.3758
−36.2333−40.6181 −36.2558
−36.2333 −36.2558
−36.2333
LAT
−36.2303−40.3842
−36.05 −36.2303 −36.2303
−36.0167−40.2667 −36.05 −36.05
LAT
LAT
−40.25
−36.0028−39.8583 −36.0167 −36.0167
−35.9667−39.7667 −36.0028
−35.9667 −36.0028
−35.9667
Correlation with −35.9086−39.6675
−35.9 −39.55
−35.8167−39.3833
−35.75 −39.36
−35.6169−39.3333
−35.9086
−35.9
−35.8167
−35.75
−35.9086
−35.9
−35.8167
−35.75
−35.5575−39.2667 −35.6169
−35.5575 −35.6169
−35.5575
−35.55 −35.55 −35.55
−35.4856−39.2561
Correlation with fall-winter P (r2) −35.4069 −39.15
−39.1
−35.2783−39.0833
−35.2744−39.0144
−35.1783−38.9833
−35.4856
−35.4069
−35.2783
−35.2744
−35.1783
−35.4856
−35.4069
−35.2783
−35.2744
−35.1783
−35.1731−38.8653 −35.1731 −35.1731
annual −34.9961 −38.85
−34.9844 −38.78 −34.9961
−34.9844 −34.9961
−34.9844
2)
−34.7186−38.6867
−34.6867 −34.7186
−34.6867 −34.7186
−34.6867
Correlation
Correlation coeff.r2P (r
coeff.Correlation coeff. r2 r2 r2 r2 r2 −34.4922−38.6833
−34.4311−38.6167
−38.6081
−34.2467−38.4333
−34.4922
−34.4311
−34.2467
−34.4922
−34.4311
−34.2467
r2
−34.0617−38.4303
−33.8111−38.3001 −34.0617
−33.8111 −34.0617
−33.8111
−33.8056 −38.3
−33.7347−38.2167 −33.8056
−33.7347 −33.8056
−33.7347
−33.7219 −33.7219 −33.7219
−33.6614−38.1501
−38.15 −33.6614 −33.6614
−33.5939−38.1167 −33.5939
−33.5875 −33.5939
−33.5875
explained
−33.5875−37.9647
−33.4961−37.8636
−33.3703 −37.85 −33.4961
−33.3703 −33.4961
−33.3703
−33.3414−37.7756
−33.3256−37.7333
−33.3414
−33.3256 −33.3414
−33.3256
−32.9164−37.7167 −32.9164 −32.9164
more than
−32.9072−37.7106
−32.8625−37.5503
−32.9072
−32.8625 −32.9072
−32.8625
−32.8572 −37.55
−32.8503−37.5422
−32.8572
−32.8503 −32.8572
−32.8503
−32.7572−37.2333
−32.5017−37.0667
−32.7572
−32.5017 −32.7572
−32.5017
0.00
0.25
0.00
0.50
0.25
0.75
0.50
1.00
0.75
1.00
0.00
0.25
0.00
0.50
0.25
0.75
0.50
1.00
−32.33
0.75
1.00
0.00
0.25
0.00
0.50
−32.33
0.25
0.75
0.50
1.00
indicates0.75
1.00
0.00
0.25
0.50
0.75
0.00
0.25
0.50
0.75
1.00
0.25
100% of0.75
−32.33 −36.9333
−32.2253−36.9244
−32.0708−36.8667
−32.2253
−32.0708 −32.2253
−32.0708
−31.9667−36.8386
−31.9481−36.8167
−31.9667
−31.9481 −31.9667
−31.9481
−31.6958 −36.7 −31.6958 −31.6958
precipitation
−31.6647 −31.6647 −31.6647
−31.5608−36.6653 −31.5608 −31.5608
−31.2281−36.6583
−36.65
−31.2281 −31.2281
−31.1119−36.6167 −31.1119
−31.0325 −31.1119
−31.0325
−31.0325 −36.55
−31.0119−36.3758 −31.0119
−30.8422 −31.0119
−30.8422
−30.8422−36.2558
−40.9333 −40.9333 −40.9333 −30.8128−36.2333 −40.9333 −30.8128 −40.9333 −40.9333 −30.8128
-41
−40.9333 −30.8006−36.2303 −40.9333 −30.8006
−30.7047 −30.8006
−30.7047
−40.9333
LAT
−30.7047 −36.05
−30.4622−36.0167 −30.4622
38.3001 −30.4622
38.3001
−40.8856 −40.8856 −40.8856 −40.8856 −36.0028 −40.8856 −40.8856 −40.8856 −40.8856 −40.8856
−35.9667
1.00 c)
0.00 a)
b)
−35.9086
−35.9
−35.8167
−40.7889 −40.7889 −40.7889 −40.7889 −35.75 −40.7889 −40.7889 −40.7889 −40.7889 −40.7889
−35.6169
−35.5575
−35.55
−40.7136 −40.7136 −40.7136 −35.4856 −40.7136 −40.7136 −40.7136
−35.4069
−40.7136 −35.2783
−40.7136 −40.7136
−35.2744
−35.1783
−40.6181 −40.6181 −40.6181 −40.6181 −35.1731 −40.6181 −40.6181 −40.6181 −40.6181 −40.6181
−34.9961
−34.9844
−34.7186
−40.3842 −40.3842 −40.3842 −34.6867 −40.3842 −40.3842 −40.3842
−34.4922
var_name
var_name
−40.3842 −34.4311
−40.3842 −40.3842
var_name
−34.2467
−34.0617
−40.2667 −40.2667 −40.2667 −40.2667 −33.8111 −40.2667 −40.2667 −40.2667 −40.2667 −40.2667
−33.8056
−33.7347
−33.7219
−40.25 −40.25 −40.25 −33.6614
−33.5939
−40.25 −33.5875
−40.25 −40.25 −40.25 −40.25 −40.25
-40
−33.4961
−33.3703
−39.8583 −39.8583 −39.8583 −39.8583 −33.3414 −39.8583 −39.8583 −39.8583 −39.8583 −39.8583
−33.3256
−32.9164
−32.9072
cor2_Pw_GW_JFM
cor2_Pw_GW_JFM
cor2_Pw_GW_AMJ
cor2_Pw_GW_AMJ
−32.8625
cor2_Pw_GW_OND
cor2_Pw_GW_OND
−39.7667 −39.7667 −39.7667 −39.7667 −32.8572 −39.7667 −39.7667 −39.7667 −39.7667 −39.7667
−32.8503
−32.7572
r2_Pa_Qres_next_yr
−32.5017
−39.6675 −39.6675 −39.6675 −39.6675 −32.33 −39.6675 −39.6675 −39.6675
−32.2253
−39.6675 −39.6675
r2_Pa_Qres_current_yr
−32.0708
−31.9667
−31.9481
−39.55 −39.55 −39.55 −39.55 −31.6958 −39.55 −39.55 −39.55 −39.55 −39.55
−31.6647
−31.5608
−31.2281
−39.3833 −39.3833 −39.3833 −31.1119 −39.3833 −39.3833 −39.3833
−39.3833 −31.0325 −39.3833 −39.3833
−31.0119
−30.8422
Annual Q
−30.8128
−39.36 −39.36 −39.36 −39.36 −30.8006 −39.36 −39.36 −39.36 −39.36 −39.36
−30.7047
−30.4622
−39.3333 −39.3333 −39.3333 −39.3333 −39.3333 −39.3333 −39.3333 −39.3333 −39.3333
−39.2667 −39.2667 −39.2667 −39.2667 −39.2667 −39.2667 −39.2667 −39.2667 −39.2667
−39.2561 −39.2561 −39.2561 −39.2561 −39.2561 −39.2561 −39.2561 −39.2561 −39.2561
−39.15 −39.15 −39.15 −39.15 −39.15 −39.15 −39.15 −39.15 −39.15
−39.1 −39.1 −39.1 −39.1 −39.1 −39.1 −39.1 −39.1 −39.1
var_name
−39.0833 −39.0833 −39.0833 −39.0833 −39.0833 −39.0833 −39.0833 −39.0833 −39.0833
−39.0144 −39.0144 −39.0144 −39.0144 −39.0144 −39.0144 −39.0144 −39.0144 −39.0144
-39
−38.9833 −38.9833 −38.9833 −38.9833 −38.9833 −38.9833 −38.9833 −38.9833 −38.9833
−38.8653 −38.8653 −38.8653 −38.8653 −38.8653 −38.8653 −38.8653 −38.8653 −38.8653
r2_Pa_Qres_next_yr
−38.85 −38.85 −38.85 −38.85 −38.85 −38.85 −38.85 −38.85 −38.85
r2_Pa_Qres_current_yr
−38.78 −38.78 −38.78 −38.78 −38.78 −38.78 −38.78 −38.78 −38.78
−38.6867 −38.6867 −38.6867 −38.6867 −38.6867 −38.6867 −38.6867 −38.6867 −38.6867
−38.6833 −38.6833 −38.6833 −38.6833 −38.6833 −38.6833 −38.6833 −38.6833 −38.6833
−38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167

−38.6081 −38.6081 −38.6081 −38.6081 −38.6081 −38.6081 −38.6081 −38.6081 −38.6081
−38.4333 −38.4333 −38.4333 −38.4333 −38.4333 −38.4333 −38.4333 −38.4333 −38.4333

−38.4303 −38.4303 −38.4303 −38.4303 −38.4303 −38.4303 −38.4303 −38.4303 −38.4303
−38.3001 −38.3001 −38.3001 −38.3001 −38.3001 −38.3001 −38.3001 −38.3001 −38.3001

−38.3 −38.3 −38.3 −38.3 −38.3 −38.3 −38.3 −38.3 −38.3
−38.2167 −38.2167 −38.2167 −38.2167 −38.2167 −38.2167
Q JFM same hydrological year

−38.2167 −38.2167 −38.2167
Q OND same hydrological year
BF JFM same hydrological year

−38.1501 −38.1501 −38.1501 −38.1501 −38.1501 −38.1501 −38.1501 −38.1501 −38.1501
BF OND same hydrological year

−38.15 −38.15 −38.15 −38.15 −38.15 −38.15 −38.15 −38.15 −38.15
−38.1167 −38.1167 −38.1167 −38.1167 −38.1167 −38.1167 −38.1167 −38.1167 −38.1167
-38
Q AMJ following hydrological year
−37.9647 −37.9647 −37.9647 −37.9647 −37.9647 −37.9647 −37.9647 −37.9647 −37.9647
BF AMJ following hydrological year

−37.8636 −37.8636 −37.8636 −37.8636 −37.8636 −37.8636 −37.8636 −37.8636 −37.8636
−37.85 −37.85 −37.85 −37.85 −37.85 −37.85 −37.85 −37.85 −37.85
−37.7756 −37.7756 −37.7756 −37.7756 −37.7756 −37.7756 −37.7756 −37.7756 −37.7756
−37.7333 −37.7333 −37.7333 −37.7333 −37.7333 −37.7333 −37.7333 −37.7333 −37.7333
−37.7167 −37.7167 −37.7167 −37.7167 −37.7167 −37.7167 −37.7167 −37.7167 −37.7167
−37.7106 −37.7106 −37.7106 −37.7106 −37.7106 −37.7106 −37.7106 −37.7106 −37.7106
−37.5503 −37.5503 −37.5503 −37.5503 −37.5503 −37.5503 −37.5503 −37.5503 −37.5503
−37.55 −37.55 −37.55 −37.55 −37.55 −37.55 −37.55 −37.55 −37.55
relationship, whereas baseflow is inherent to all basins.
−37.5422 −37.5422 −37.5422 −37.5422 −37.5422 −37.5422 −37.5422 −37.5422 −37.5422

−37.2333 −37.2333 −37.2333 −37.2333 −37.2333 −37.2333 −37.2333 −37.2333 −37.2333
−37.0667 −37.0667 −37.0667 −37.0667 −37.0667 −37.0667 −37.0667 −37.0667 −37.0667
-37
−36.9333 −36.9333 −36.9333 −36.9333 −36.9333 −36.9333 −36.9333 −36.9333 −36.9333
Annual R-R residual following year

−36.9244 −36.9244 −36.9244 −36.9244 −36.9244 −36.9244 −36.9244 −36.9244 −36.9244
−36.8667 −36.8667 −36.8667 −36.8667 −36.8667 −36.8667 −36.8667 −36.8667 −36.8667
−36.8386 −36.8386 −36.8386 −36.8386 −36.8386 −36.8386 −36.8386 −36.8386 −36.8386
−36.8167 −36.8167 −36.8167 −36.8167 −36.8167 −36.8167 −36.8167 −36.8167 −36.8167
−36.7 −36.7 −36.7 −36.7 −36.7 −36.7 −36.7 −36.7 −36.7
−36.6653 −36.6653 −36.6653 −36.6653 −36.6653 −36.6653 −36.6653 −36.6653 −36.6653
14
−36.6583 −36.6583 −36.6583 −36.6583 −36.6583 −36.6583 −36.6583 −36.6583 −36.6583
−36.65 −36.65 −36.65 −36.65 −36.65 −36.65 −36.65 −36.65 −36.65
−36.6167 −36.6167 −36.6167 −36.6167 −36.6167 −36.6167 −36.6167 −36.6167 −36.6167
−36.55 −36.55 −36.55 −36.55 −36.55 −36.55 −36.55 −36.55 −36.55
−36.3758 −36.3758 −36.3758 −36.3758 −36.3758 −36.3758 −36.3758 −36.3758 −36.3758
Latitude
−36.2558 −36.2558 −36.2558 −36.2558 −36.2558 −36.2558 −36.2558 −36.2558 −36.2558
−36.2333 −36.2333 −36.2333 −36.2333 −36.2333 −36.2333 −36.2333 −36.2333 −36.2333
LAT
LAT
LAT
correlation between fall-winter precipitation and the streamflow inLAT

LAT
LAT
LAT
LAT
LAT
−36.2303 −36.2303 −36.2303 −36.2303 −36.2303 −36.2303 −36.2303 −36.2303 −36.2303
that snowmelt governs runoff generation in these catchments.
−36.05 −36.05 −36.05 −36.05 −36.05 −36.05 −36.05 −36.05 −36.05

−36.0167 −36.0167 −36.0167 −36.0167 −36.0167 −36.0167 −36.0167 −36.0167 −36.0167
−36.0028 −36.0028 −36.0028 −36.0028 −36.0028 −36.0028 −36.0028 −36.0028 −36.0028
-36
−35.9667 −35.9667 −35.9667 −35.9667 −35.9667 −35.9667 −35.9667 −35.9667 −35.9667
−35.9086 −35.9086 −35.9086 −35.9086 −35.9086 −35.9086 −35.9086 −35.9086 −35.9086
−35.9 −35.9 −35.9 −35.9 −35.9 −35.9 −35.9 −35.9 −35.9
−35.8167 −35.8167 −35.8167 −35.8167 −35.8167 −35.8167 −35.8167 −35.8167 −35.8167
−35.75 −35.75 −35.75 −35.75 −35.75 −35.75 −35.75 −35.75 −35.75
−35.6169 −35.6169 −35.6169 −35.6169 −35.6169 −35.6169 −35.6169 −35.6169 −35.6169
−35.5575 −35.5575 −35.5575 −35.5575 −35.5575 −35.5575 −35.5575 −35.5575 −35.5575
−35.55 −35.55 −35.55 −35.55 −35.55 −35.55 −35.55 −35.55 −35.55
−35.4856 −35.4856 −35.4856 −35.4856 −35.4856 −35.4856 −35.4856 −35.4856 −35.4856
−35.4069 −35.4069 −35.4069 −35.4069 −35.4069 −35.4069 −35.4069 −35.4069 −35.4069
−35.2783 −35.2783 −35.2783 −35.2783 −35.2783 −35.2783 −35.2783 −35.2783 −35.2783
−35.2744 −35.2744 −35.2744 −35.2744 −35.2744 −35.2744 −35.2744 −35.2744 −35.2744
−35.1783 −35.1783 −35.1783 −35.1783 −35.1783 −35.1783 −35.1783 −35.1783 −35.1783
−35.1731 −35.1731 −35.1731 −35.1731 −35.1731 −35.1731 −35.1731 −35.1731 −35.1731
-35
−34.9961 −34.9961 −34.9961 −34.9961 −34.9961 −34.9961 −34.9961 −34.9961 −34.9961
−34.9844 −34.9844 −34.9844 −34.9844 −34.9844 −34.9844 −34.9844 −34.9844 −34.9844
−34.7186 −34.7186 −34.7186 −34.7186 −34.7186 −34.7186 −34.7186 −34.7186 −34.7186
−34.6867 −34.6867 −34.6867 −34.6867 −34.6867 −34.6867 −34.6867 −34.6867 −34.6867
−34.4922 −34.4922 −34.4922 −34.4922 −34.4922 −34.4922 −34.4922 −34.4922 −34.4922
−34.4311 −34.4311 −34.4311 −34.4311 −34.4311 −34.4311 −34.4311 −34.4311 −34.4311
−34.2467 −34.2467 −34.2467 −34.2467 −34.2467 −34.2467 −34.2467 −34.2467 −34.2467
−34.0617 −34.0617 −34.0617 −34.0617 −34.0617 −34.0617 −34.0617 −34.0617 −34.0617
-34
−33.8111 −33.8111 −33.8111 −33.8111 −33.8111 −33.8111 −33.8111 −33.8111 −33.8111
−33.8056 −33.8056 −33.8056 −33.8056 −33.8056 −33.8056 −33.8056 −33.8056 −33.8056
−33.7347 −33.7347 −33.7347 −33.7347 −33.7347 −33.7347 −33.7347 −33.7347 −33.7347
−33.7219 −33.7219 −33.7219 −33.7219 −33.7219 −33.7219 −33.7219 −33.7219 −33.7219
−33.6614 −33.6614 −33.6614 −33.6614 −33.6614 −33.6614 −33.6614 −33.6614 −33.6614
−33.5939 −33.5939 −33.5939 −33.5939 −33.5939 −33.5939 −33.5939 −33.5939 −33.5939
−33.5875 −33.5875 −33.5875 −33.5875 −33.5875 −33.5875 −33.5875 −33.5875 −33.5875
−33.4961 −33.4961 −33.4961 −33.4961 −33.4961 −33.4961 −33.4961 −33.4961 −33.4961
−33.3703 −33.3703 −33.3703 −33.3703 −33.3703 −33.3703 −33.3703 −33.3703 −33.3703
regressions), and the variance of the R-R residuals explained by the precipitation from the preceding year.
−33.3414 −33.3414 −33.3414 −33.3414 −33.3414 −33.3414 −33.3414 −33.3414 −33.3414
the following year runoff as baseflow, since snowmelt process is mostly finished by January (Fig. 5a).
−33.3256 −33.3256 −33.3256 −33.3256 −33.3256 −33.3256 −33.3256 −33.3256 −33.3256
-33
−32.9164 −32.9164 −32.9164 −32.9164 −32.9164 −32.9164 −32.9164 −32.9164 −32.9164
−32.9072 −32.9072 −32.9072 −32.9072 −32.9072 −32.9072 −32.9072 −32.9072 −32.9072
−32.8625 −32.8625 −32.8625 −32.8625 −32.8625 −32.8625 −32.8625 −32.8625 −32.8625
−32.8572 −32.8572 −32.8572 −32.8572 −32.8572 −32.8572 −32.8572 −32.8572 −32.8572
−32.8503 −32.8503 −32.8503 −32.8503 −32.8503 −32.8503 −32.8503 −32.8503 −32.8503
−32.7572 −32.7572 −32.7572 −32.7572 −32.7572 −32.7572 −32.7572 −32.7572 −32.7572
−32.5017 −32.5017 −32.5017 −32.5017 −32.5017 −32.5017 −32.5017 −32.5017 −32.5017
−32.33 −32.33 −32.33 −32.33 −32.33 −32.33 −32.33 −32.33 −32.33
−32.2253 −32.2253 −32.2253 −32.2253 −32.2253 −32.2253 −32.2253 −32.2253 −32.2253
−32.0708 −32.0708 −32.0708 −32.0708 −32.0708 −32.0708 −32.0708 −32.0708 −32.0708
-32
−31.9667 −31.9667 −31.9667 −31.9667 −31.9667 −31.9667 −31.9667 −31.9667 −31.9667
−31.9481 −31.9481 −31.9481 −31.9481 −31.9481 −31.9481 −31.9481 −31.9481 −31.9481
−31.6958 −31.6958 −31.6958 −31.6958 −31.6958 −31.6958 −31.6958 −31.6958 −31.6958
−31.6647 −31.6647 −31.6647 −31.6647 −31.6647 −31.6647 −31.6647 −31.6647 −31.6647
−31.5608 −31.5608 −31.5608 −31.5608 −31.5608 −31.5608 −31.5608 −31.5608 −31.5608
−31.2281 −31.2281 −31.2281 −31.2281 −31.2281 −31.2281 −31.2281 −31.2281 −31.2281
−31.1119 −31.1119 −31.1119 −31.1119 −31.1119 −31.1119 −31.1119 −31.1119 −31.1119
−31.0325 −31.0325 −31.0325 −31.0325 −31.0325 −31.0325 −31.0325 −31.0325 −31.0325
−31.0119 −31.0119 −31.0119 −31.0119 −31.0119 −31.0119 −31.0119 −31.0119 −31.0119
−30.8422 −30.8422 −30.8422 −30.8422 −30.8422 −30.8422 −30.8422 −30.8422 −30.8422
−30.8128 −30.8128 −30.8128 −30.8128 −30.8128 −30.8128 −30.8128 −30.8128 −30.8128
−30.8006 −30.8006 −30.8006 −30.8006 −30.8006 −30.8006 −30.8006 −30.8006 −30.8006
−30.7047 −30.7047 −30.7047 −30.7047 −30.7047 −30.7047 −30.7047 −30.7047 −30.7047
−30.4622 −30.4622 −30.4622 −30.4622 −30.4622 −30.4622 −30.4622 −30.4622 −30.4622
-31 -30
between fall-winter precipitation and seasonal baseflow of the following year (Fig. 7b). The relatively low r2 values (below
the fall season of the following year (green bars in Fig. 7a
In semi-arid catchments, hydrological memory beyond one hydrological year (April to March) is represented by the
following seasons allows exploring the hydrological memory further. Fall-winter precipitation was used instead of annual
The correlation between fall-winter precipitation and the resulting streamflow (Fig. 7a) and baseflow (Fig. 7b) over the
For the temperate catchments, there is some memory effect beyond one hydrological year, represented by the correlation
1.00 since it is more directly related to snow dynamics. Precipitation in fall and winter represents between 50 and
snow processes control the hydrologic response of a subset of catchments experiencing significant change in R-R
the total precipitation volume (Garreaud et al., 2017). Figure 7a indicates that fall-winter precipitation explains
by fall-winter precipitation (P). Panel c shows the r2 between annual precipitation and annual streamflow (r2 of R-R
Figure 7: Catchment memory represented by the variance in seasonal runoff (panel a) and seasonal baseflow (BF, panel b)
droughts. The relationship between snow fraction and R-R shifts is not strong (r2 = 0.11, not shown here), probably since
and 7b). The high r2 values (up to 0.75 in some northern catchments) indicate that fall-winter precipitation is contributing to
0.50half of the variance in spring (OND) and summer (JFM) streamflow in catchments located north of 36°S. This
0.25 in most basins) are due to fall-winter precipitation of the current year explaining most of the generated streamflow
315 (these are mostly pluvial catchments), which overshadows the dependency with previous seasons. To remove the effect of
the precipitation of the current year, we computed the dependency of the R-R regression residuals from Sect. 4.2 (i.e., annual
streamflow not explained by the current precipitation) to the precipitation from the previous year. The R-R residuals r2
values in Fig. 7c (corresponding to positive correlation coefficients, not shown here) are also an indicator of hydrological
memory since they show that the precipitation explains a part of the annual streamflow from the previous year. Interestingly,
320 although the residual r2 is larger in semi-arid catchments (mostly snow-dominated), there are also significant values in the
most humid and pluvial basins southward, suggesting that baseflow in pluvial catchments may also contribute to more
extended catchment memory.
The relationship between hydrological memory and groundwater dynamics was further explored in Fig. 6c and Fig. 6d.
These plots show that larger BFI values are associated with stronger longer-than-a-year hydrological memory, represented
325 by larger increments in r2 from Fig. 7c, and larger r2 values between winter precipitation and baseflow from the following
year (green bars in Fig. 7b). In the next section we provide a more detailed analysis of the relationship between hydrological
memory and drought propagation.
4.4 Extreme versus persistent droughts
Figure 8 presents drought propagation for 25 years with negative anomalies over the last four decades (see Fig. 2). Drought
330 propagation is represented by the difference between runoff and precipitation deficits (red markers in Fig. 8). In the semi-
arid basins, the difference increases in the second year of precipitation deficits, showing that in catchments with longer
hydrological memory consecutive years with precipitation deficits are associated to intensified drought propagation. This
plot also provides insights about hydrological recovery, understood as the end of a hydrological drought after a
meteorological drought has ceased (Yang et al., 2017). While 2016 had above-average precipitation in semi-arid basins,
335 probably there was not enough water entering the system over enough time to recharge groundwater systems up to levels
such as those before the MD (similarly to the conceptual drought propagation illustrated in Fig. 3 from Van Loon, 2015).
This is reflected by the larger streamflow deficits in 2016 compared to 2008, even when the above-mean precipitations in
2008, following the deficits in 2007 are comparable to those in 2016 and 2015, respectively. This can be related to the
catchments’ memory (Sect. 4.3) and the 7-year (2009 to 2015) precipitation deficits prior 2016, which probably prevented
340 full hydrological recovery after a single year of above-average precipitation. These results are consistent with large recovery
times reported for semi-arid Australian catchments following extreme droughts (Yang et al., 2017). In this way, hydrological
memory would be an explanatory factor for both the intensification in drought propagation as well as a delayed hydrological
recovery.
For temperate catchments (Fig. 8b), prior to 2010, drought propagation has been around 10% (i.e., streamflow deficits are,
345 on average, 10% larger than precipitation deficits), independently of the precipitation deficits of previous years. Such
propagation was observed even in the driest two years of the historical record, 1996 and 1998. After 2010, there are two
15
years (2012 and 2016) where drought propagation was intensified (red markers below 20%). These larger streamflow
deficits during the MD may be due to different factors, including the hydrological memory of these basins (Sect. 4.3) and the
large ET in 2012 and 2016 (positive anomalies, and z-scores above 1.5, as can be seen in Fig. S1) combined with large
350 precipitation deficits. Since R-R relationships do not explicitly account for variations in ET, the anomaly values of ET during
the MD might explain the R-R shifts identified in some temperate catchments (Fig. 4). Furthermore, towards the south of the
study region, basins move from water-limited to energy-limited (Alvarez-Garreton et al., 2018). Therefore, ET in temperate
catchments is modulated by both the available water and the available energy.
a) Semi−arid catchments (Lat 29S − 34S)
a)Semi−arid
a) Semi-arid catchments
catchments (30ºS
(Lat 29S - 34ºS)
− 34S)
● ●
● ●
0 ● drought
(%)
● ● ●
● ●
00 ● ● ●
(%)
● ●
● ● ● ● ● ●
●
● drought
(%)
● ● ●
● ● ● ● ● Pa
Deficit
● ● ●
● ● ●
−40 ● ● ● ●
● ● ● ●
● ● ● Pa
Deficit
●
Deficit
●
-40
−40 ● ● Q
−80 Q
-80
−80
1979 1981 1985 1988 1989 1990 1993 1994 1995 1996 1998 1999 2003 2004 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
1979 1981 1985 1988 1989 1990 1993 1994 1995 1996 1998 1999 2003 2004 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
b)Mediterranean
b) Temperate catchments (34ºS
catchments (Lat 34S −- 41S)
41ºS)
b) Mediterranean catchments (Lat 34S − 41S)
●
(%)
●
00 ●
●
● ●
● ●
● ● drought
(%)(%)
● ● ● ● ● ●
● ● ● ●
● ● ● ● ● ● ●
drought
Deficit
0 ● ● ● ● ● ●
●
● ● ● ●
a) Semi−arid catchments (Lat 29S − 34S)
● ● ●
● Pa
Deficit
●
a) Semi−arid
● catchments● (Lat 29S − 34S) ● ● ● ●
-40
−40
drought_type
●
●
●
Pa
Deficit
a) Semi−arid catchments (Lat 29S − 34S) Q

0 drought_typePa_def
Deficit (%)
−50
-80
−80 Q
0 PPa_def QQa_def
● Difference between P and Q deficits
Deficit (%)
−40 ●
0 Qa_def Qpred_def_HBV_obs
●
−40 1979 1981 1985 1988 1989 1990 1993 1994 1995 1996 1998 1999 2003 2004 2007
● 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Deficit (%)
−80 ● ●
1979 1981 1985 1988 1989 1990 1993 1994 1995 1996 1998Extreme
1999 2003 dry
2004years 2010 2011 2012 2013 2014 2015 2016 2017 2018 ●
Qpred_def_HBV_obs
2007 2008 2009Qpred_def_logLR_obs
●
355
−80 Figure
1 8: Precipitation and runoff deficits for dry years for northern semi-arid catchments (panel a)
2 3 4 Extreme dry5 yearsQpred_def_logLR_obs
● and southern temperate
●
catchments (panel b). The difference between runoff and precipitation deficits are represented by red points and consecutive years
1 2 3 4 5 −40
are connected with red lines.
b) Mediterranean catchments (Lat 34S − 41S)
b)Finally, drought_type
to analyse
Mediterranean catchment
catchments responses
(Lat 34S − 41S)during short extreme droughts and persistent moderate droughts, Fig. 9 presents
0
−80 drought_type
Deficit (%)
Pa_def
drought propagation for three cases: i) single-year extreme droughts (mean precipitation anomalies below -50% in semi-arid
0
Deficit (%)
−40 Pa_def Qa_def

360 basins and mean precipitation anomalies below -25% in temperate basins) preceded by a wet year (mean precipitation above
Qa_def Qpred_def_HBV_obs
−40
−80
1979
average); ii) single-year extreme droughts preceded by a dry year (mean 1981Qpred_def_logLR_obs
1985
precipitation 1988average);
below
Qpred_def_HBV_obs 1989 1990 1993 1994
iii) persistent 1995 1996 1
years
−80
1 2 3 4 5
with precipitation anomalies below dry
# consecutive -5%.
years (P deficit > 5%)
Qpred_def_logLR_obs
1 2 3 4 5
# consecutive dry years (P deficit > 5%)
b) Mediterranean catchments (Lat 34S − 41S
● ●
16 0 ● ●
eficit (%)
● ●
● ●
●
−50
a)Extreme and persistent

Extreme droughts (30ºS - 34ºS) a)
droughts b) Semi−arid
− Persistent
north
Extreme catchments
and persistent
droughts (Lat 29S − 34S)
droughts
(30ºS - 34ºS) − south b) Mediterran
Extreme and persistent drou
n=600 n=600 n=257 n=175 n=103 n=65 model
0 0 0 0 0
(%)
Q deficit (%)
Q deficit (%)
Deficit (%)
Deficit (%)
Pobs
deficit(%)
−25 −25 −25 −25 −25

Qobs
QDeficit
−50
Extreme −50
and persistent droughts − north −50
Extreme and persistent droughts − south−50
Extreme −50
and persistent droughts −
−75 −75 −75 −75 model−75 Qpred HBV
0 0 0
Q deficit (%)
Q deficit (%)
Q deficit (%)
−100 −100 −100 −100 Precip.
−100
Pobs Qpred R−R
−25 Following
Extreme Pdef afterawet

year
Extreme Following
Pdef
−25
after adry year Extreme Pdef after
Extreme
wet year
Pdef after dry−25
year
ExtremeExtreme
Pdef Pdef
after after
wet year
dry year
1 2 3 4 1
wet year
dry year
Consecutive years with precipitation−50
deficits > 5% Qobs
Qobs
−50 (1988, 1998, 2007) (1990, 1996) −50 # consecu
−75 −75 −75 34S Qpred
QpredHBV
HBV
mi−arid catchments
ent droughts (Lat 29S
− north
c)Extreme
Extreme − 34S)
and persistent
droughts droughts
(34ºS - 41ºS) −b)Persistent
d) Mediterranean
south
Extreme catchments
and persistent
droughts (Lat
droughts
(34ºS - 41ºS) −− 41S)
north
−100 −100 model −100 Qpred
QpredR-R
R−R
0 after
Extreme Pdef
0 0
(%)
Extreme
wet year
Pdef after dry year Extreme Pdef after
Q deficit (%)
Extreme
wet year
Pdef after dry year
Extreme
Extreme
Pdef Pdef
after after
wet year
dry year
Deficit (%)
Pobs
deficit(%)
−25 −25 −25

Qobs
QDeficit
−50 −50 −50

−75 −75 −75 Qpred HBV
n=1230 n=1230 n=314 Qpred R−R

−100 −100 −100 n=636 n=169 n=111
1after dry year2 Extreme Pdef

3 after
Extreme
wet
Following a
year
4Pdef after dry
Following a
year
Extreme
Extreme
Pdef
1 Pdef
after after
wet year
dry2year 3 4
wet year
dry year
Consecutive years with precipitation deficits > 5%
(1988, 1998, 2007) (1990, 1996)
# consecutive dry years (P deficit > 5%)
365 Figure 9: Observed and simulated annual drought propagation for consecutive dry years. Panel a presents the results for semi-
arid catchments, and panel b for humid catchments. The number of cases is presented for each set of boxplots.
5 Discussion
Longer-than-a-year hydrological memory is related to slow groundwater and subsurface flow which is responding to
precipitation from previous seasons (Fig. 7b). In catchments where snow processes contribute significantly to runoff
370 generation, the importance of low-flow contribution to runoff is positively correlated with snow accumulation (Fig. 6a).
A higher baseflow in basins with higher proportion of solid precipitation can be interpreted by analysing the groundwater
recharge mechanisms when the snow melts. Snowmelt infiltrates in situ, where ET is low compared to precipitation, the soil
storage is shallow and there is a low permeability bedrock underneath (Carroll et al., 2019). Infiltrated snowmelt is routed
through the steep topography as shallow ephemeral interflow, which supports large recharge rates in topographic
375 convergence zones, where ET is still moderate and ephemeral stream channels in the upper basin appear (Anderson and Burt,
1978). At the high elevations where snowmelt starts, topography and soil permeability would have a larger control in
groundwater recharge than precipitation. That is, the rates at which snowmelt infiltrates are more sensitive to catchment
characteristics than to the snowpack volume (Carroll et al., 2019). The large snowmelt infiltration rates at high elevations
during snowmelt season, combined with the absence of precipitation events (precipitation is concentrated in winter season)
17
380 would explain the higher BFI in snow-dominated basins and is consistent with the positive correlation between BFI and per
cent of precipitation falling as snow reported in Upper Colorado River Basin (Rumsey et al., 2015).
Further, high snow fraction ratios indicate an important spatial variation of precipitation within the basin, with most
precipitation falling in the upper part of the basin and therefore, traveling longer paths to reach the outlet where streamflow
is recorded. Longer paths lead to longer travel times, which may support groundwater recharge from interflow. The role of
385 snow in hydrological memory and its relation to water provision during droughts reveal the need to understand seasonal
characteristics of drought propagation further. The effects of spring-summer precipitation deficits may be different from
those of fall-winter winter deficits (Berghuijs et al., 2014; Jasechko et al., 2014). This is particularly important in central-
south Chile, where expected (modelled) shifts in precipitation and streamflow seasonality due to anthropogenic climate
change have already been detected in observations (Boisier et al., 2018; Bozkurt et al., 2018; Cortés et al., 2011).
390 In addition to the climatic characteristics (e.g., the ratio between solid and liquid precipitation), groundwater contribution to
runoff depends on physical factors (e.g., geology, topography, soil properties, soil drainage density), which are not explored
in this paper, and which may explain the large scatter in Fig. 6a.
Our results also indicate that hydrological memory is related to the intensification of drought propagation during multi-year
droughts. This can be seen in Fig. 6b, where catchments with longer hydrological memory (represented by higher BFI, as
395 shown in Fig. 6c and Fig. 6d) present larger shifts in R-R relationships. This is consistent with the conceptualisation
proposed by Carey et al., (2010), who classified basins that can store water over long time periods (i.e., long hydrological
memory) as catchments with higher resistance, and showed that such catchments featured lower resilience, defined as a
lower capacity to keep their normal functioning (translating inputs to outputs) under changing inputs.
When we analyse isolated extreme droughts, and consistently with the large hydrological memory in semi-arid catchments,
400 we observe that drought propagation in these basins is highly dependent on the meteorological conditions from the previous
year (Fig. 9a), which define the condition of groundwater reservoirs. This is also observed in persistent but moderate
droughts, where under similar precipitation deficits, the surface water supply in semi-arid regions consistently decreases
after one year with below-average conditions. This result also indicate that the HBV model represents catchment response
for extreme and persistent droughts well, consistently outperforming the prediction from R-R regressions. It should be noted
405 though that HBV allows memory effects only to a certain extent given the groundwater storage capacity defined by
calibration.
Wetter and more rainfed basins in the study region generally show a weaker longer-than-a-year hydrological memory than
semi-arid catchments, which may be explained by their pluvial or mixed hydrologic regimes, and their lower groundwater
contribution to runoff. The weaker hydrological memory leads to a more similar behaviour under extreme meteorological
410 droughts occurring after a wet and a dry year (Fig. 9c), which is in line with the concept of resilient catchment proposed by
Carey et al., (2010). However, even when these basins are largely controlled by precipitation during the same year, there is
some over-one-year memory (Fig. 7c), which is reflected by a decrease in streamflow generation after one year of
consecutive precipitation deficits (Fig. 9d). This effect is well captured by the HBV model, while the annual R-R
18
relationship tends to overestimate observed runoff. This indicates that a good representation of fluxes such as ET, soil
415 moisture and groundwater dynamics, allow foreseeing catchment response to persistent droughts and to extreme droughts
6 Conclusions
The MD in central-south Chile has been extraordinary because of its persistence (10 years to date) and extended spatial
domain (~1000 km). Annual precipitation anomalies during the MD have been larger in semi-arid catchments compared to
temperate catchments. However, observing such large deficits at a single year in the northern region is not unusual given its
420 highly variable rainfall regime. The precipitation anomalies experienced in the southern region, on the other hand, were
highly unusual considering the last four decades. When we aggregated the entire 8-year MD period, precipitation and
streamflow anomalies are shown to be extremely unusual across the complete domain.
The persistent precipitation deficits have led to larger than expected water deficits in 68.5% of the study catchments when
compared to their historical (1979-2009) streamflow response to precipitation (R-R relationship). This indicates that there
425 has been an intensification in drought propagation over the last decade. We argue that this intensification during multi-year
droughts reflects the hydrological memory within a catchment. Hydrological memory is directly related to groundwater
storages, which respond to liquid and solid precipitation of the previous seasons. Within this study region, catchments are
more snow-dominated towards the north due to the higher elevation of the Andes in central Chile, and present a larger
contribution of groundwater to runoff.
430 Catchments with more extended hydrological memory showed larger shifts in R-R relationships, which is related to the
intensification of drought propagation during multi-year droughts. Water deficits during a meteorological drought in basins
with longer hydrological memory are closely tied to previous meteorological conditions. For semi-arid catchments (30ºS-
34ºS), after one year of precipitation deficits, the surface water supply –under equivalent precipitation deficits– significantly
decreases. That is, the basins progressively generate less water for a given precipitation amount, compared with the historical
435 behaviour. Catchments in the southern part of the study region (34ºS-41ºS) generally feature a shorter memory, since they
have a more pluvial regime, with less predominant groundwater contribution to streamflow. In these catchments, the annual
streamflow is mostly explained by the precipitation of the current year, and this is reflected by a slighter decrease in water
supply after consecutive years of precipitation deficits, compared to semi-arid catchments.
Our results also reveal the limitations of predicting annual streamflow based only on precipitation (R-R relationship).
440 Explicitly accounting for other processes and fluxes (ET, snow, soil moisture and groundwater dynamics), in particular those
representing hydrological memory, improves the predictability of water deficits, and this becomes particularly relevant under
persistent climate anomalies, such as the MD.
What is worse, an extreme single year drought or a persistent moderate drought? We have shown that for any type of
drought, hydrological memory, and initial storages conditions are key factors modulating catchment responses. In semi-arid
445 regions of Chile, catchments strongly depend on both the current and previous precipitation seasons. Soil water levels can be
19
slowly (more than a year) and rapidly (during the same year) depleted, both cases leading to large water deficits. The worst
scenario would be an extreme meteorological drought following consecutive years of below average precipitation, as
occurred in 2019 (not analysed in this paper). In more pluvial regimes, initial conditions and hydrologic memory are still
important factors to represent catchment response fully. However, water supply is more strongly dependant on the
450 meteorological conditions of the current year, and therefore an extreme drought would have a higher impact on water supply
than a persistent but moderate drought.
This work provides new evidence and knowledge about drought propagation. We argue that hydrological memory plays a
key role in catchments response to persistent droughts, and show that extreme and persistent droughts impact water supply in
different ways depending on catchment characteristics. Since the propagation of persistent droughts is modulated by
455 catchment memory, which in turn is related to snow dynamics, future work should focus on seasonal characteristics of
drought propagation.
Snow-dominated catchments store water that is then released during dry seasons, a characteristic particularly valuable in
regions with limited water supply and severe risk to droughts, such as in central Chile. However, snow and baseflow
dynamics also lengthen hydrological memory of a basin and therefore makes them prone to intensify the propagation of
460 persistent droughts. For adaptation strategies, we recommend focusing future work on linking drought propagation and
hydrological memory to water scarcity, which requires to incorporate human activities and the long-term use of water
resources.
Data availability. CAMELS-CL catchment dataset was obtained from the Center for Climate and Resilience Research
website (http://www.cr2.cl/camels-cl/). ECMWF ERA5-Land dataset was downloaded from https://www.ecmwf.int/en/era5-
465 land. ASTER GDEM elevation data were downloaded from the NASA Earthdata website
(https://search.earthdata.nasa.gov/search/).
Author contributions. This research was conceived and design by CAG, JPB and RG. JS and MV implemented the HBV
model. CAG wrote the manuscript with input from all co-authors. All the authors have been involved in interpreting the
results, discussing the findings, and editing the paper.
470 Competing interests. The authors declare that they have no conflict of interest.
Acknowledgements. This research has been developed within the framework of Center for Climate and Resilience Research
(CR2, ANID/FONDAP/15110009).
References
Agboma, C. O. and Lye, L. M.: Hydrologic memory patterns assessment over a drought-prone Canadian prairies catchment,
475 J. Hydrol. Eng., 20(7), 1–11, doi:10.1061/(ASCE)HE.1943-5584.0001106, 2015.
Alvarez-Garreton, C., Mendoza, P. A., Boisier, J. P., Addor, N., Galleguillos, M., Zambrano-Bigiarini, M., Lara, A., Puelma,
C., Cortes, G., Garreaud, R., McPhee, J. and Ayala, A.: The CAMELS-CL dataset: catchment attributes and meteorology for
20
large sample studies – Chile dataset, Hydrol. Earth Syst. Sci., doi:https://doi.org/10.5194/hess-22-5817-2018, 2018.
Anderson, M. G. and Burt, T. P.: Role of Topography in Controlling Throughflow Generation., Earth Surf. Process., 3(4),
480 331–344, doi:10.1002/esp.3290030402, 1978.
Berghuijs, W. R., Woods, R. A. and Hrachowitz, M.: A precipitation shift from snow towards rain leads to a decrease in
streamflow, Nat. Clim. Chang., 4(7), 583–586, doi:10.1038/nclimate2246, 2014.
Blöschl, G. and Montanari, A.: Climate change impacts—throwing the dice?, Hydrol. Process., 24, 374–381, doi:DOI:
10.1002/hyp.7574, 2010.
485 Boisier, J. P., Rondanelli, R., Garreaud, R. and Muñoz, F.: Anthropogenic and natural contributions to the Southeast Pacific
precipitation decline and recent megadrought in central Chile, Geophys. Res. Lett., 43(1), 413–421,
doi:10.1002/2015GL067265, 2016.
Boisier, J. P., Alvarez-Garretón, C., Cordero, R. R., Damiani, A., Gallardo, L., Garreaud, R. D., Lambert, F., Ramallo, C.,
Rojas, M. and Rondanelli, R.: Anthropogenic drying in central-southern Chile evidenced by long-term observations and
490 climate model simulations, Elem Sci Anth, 6(1), 74, doi:10.1525/elementa.328, 2018.
Bozkurt, D., Rojas, M., Boisier, J. P. and Valdivieso, J.: Projected hydroclimate changes over Andean basins in central Chile
from downscaled CMIP5 models under the low and high emission scenarios, Clim. Change, 150(3–4), 131–147,
doi:10.1007/s10584-018-2246-7, 2018.
Carey, S. K., Tetzlaff, D., Seibert, J., Soulsby, C., Buttle, J., Laudon, H., McDonnell, J., McGuire, K., Caissie, D., Shanley,
495 J., Kennedy, M., Devito, K. and Pomeroy, J. W.: Inter-comparison of hydro-climatic regimes across northern catchments:
Synchronicity, resistance and resilience, Hydrol. Process., 24(24), 3591–3602, doi:10.1002/hyp.7880, 2010.
Carroll, R. W. H., Deems, J. S., Niswonger, R., Schumer, R. and Williams, K. H.: The Importance of Interflow to
Groundwater Recharge in a Snowmelt-Dominated Headwater Basin, Geophys. Res. Lett., 46(11), 5899–5908,
doi:10.1029/2019GL082447, 2019.
500 Chambers, J. M., Freeny, A. E. and Heiberger, R. M.: Analysis of variance; designed experiments, in Statistical Models in
S., 2017.
Copernicus Climate Change Service: ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate,
Copernicus Clim. Chang. Serv. Clim. Data Store (CDS), accessed 2019-01-01, 2019.
Cortés, G., Vargas, X. and McPhee, J.: Climatic sensitivity of streamflow timing in the extratropical western Andes
505 Cordillera, J. Hydrol., 405(1–2), 93–109, doi:10.1016/j.jhydrol.2011.05.013, 2011.
Duethmann, D., Blöschl, G. and Parajka, J.: Why does a conceptual hydrological model fail to predict discharge changes in
response to climate change?, Hydrol. Earth Syst. Sci. Discuss., (January), 1–28, doi:10.5194/hess-2019-652, 2020.
Eltahir, E. A. B. and Yeh, P. J. F.: On the asymmetric response of aquifer water level to floods and droughts in Illinois,
Water Resour. Res., 35(4), 1199–1217, doi:10.1029/1998WR900071, 1999.
510 Fowler, K., Peel, M., Western, A. and Zhang, L.: Improved Rainfall-Runoff Calibration for Drying Climate: Choice of
Objective Function, Water Resour. Res., 54(5), 3392–3408, doi:10.1029/2017WR022466, 2018.
21
Fowler, K. J. A., Peel, M. C., Western, A. W., Zhang, L. and Peterson, T. J.: Simulating runoff under changing climatic
conditions: Revisiting an apparent deficiency of conceptual rainfall-runoff models, Water Resour. Res., 52(3), 1820–1846,
doi:10.1002/2015WR018068, 2016.
515 Garreaud, R., Alvarez-Garreton, C., Barichivich, J., Boisier, J. P., Christie, D., Galleguillos, M., LeQuesne, C., McPhee, J.
and Zambrano-Bigiarini, M.: The 2010-2015 mega drought in Central Chile: Impacts on regional hydroclimate and
vegetation, Hydrol. Earth Syst. Sci., 21, 6307–6327, doi:10.5194/hess-21-6307-2017, 2017.
Garreaud, R. D., Boisier, J. P., Rondanelli, R., Montecinos, A., Sepúlveda, H. H. and Veloso-Aguila, D.: The Central Chile
Mega Drought (2010–2018): A climate dynamics perspective, Int. J. Climatol., (May), 1–19, doi:10.1002/joc.6219, 2019.
520 Gonzalez, M. E., Gómez-González, S., Lara, A., Garreaud, R. and Díaz-Hormazábal, I.: The 2010 – 2015 Megadrought and
its influence on the fire regime in central and south-central Chile, Ecosphere, 9(August), 1–17, doi:10.1002/ecs2.2300, 2018.
Van De Griend, A. A., De Vries, J. J. and Seyhan, E.: Groundwater discharge from areas with a variable specific drainage
resistance, J. Hydrol., 259(1–4), 203–220, doi:10.1016/S0022-1694(01)00583-2, 2002.
Gupta, H. V., Kling, H., Yilmaz, K. K. and Martinez, G. F.: Decomposition of the mean squared error and NSE performance
525 criteria: Implications for improving hydrological modelling, J. Hydrol., 377(1–2), 80–91, doi:10.1016/j.jhydrol.2009.08.003,
2009.
Jasechko, S., Birks, S. J., Gleeson4, T., Wada, Y., Fawcett, P. J., Sharp, Z. D., McDonnell, J. J. and Welker, J. M.: The
pronounced seasonality of global groundwater recharge, Water Resour. Res., 50(11), 8845–8867, doi:10.1002/
2014WR015809, 2014.
530 Van Lanen, H. A. J., Wanders, N., Tallaksen, L. M. and Van Loon, A. F.: Hydrological drought across the world: Impact of
climate and physical catchment structure, Hydrol. Earth Syst. Sci., 17(5), 1715–1732, doi:10.5194/hess-17-1715-2013, 2013.
Van Loon, A. F.: Hydrological drought explained, Wiley Interdiscip. Rev. Water, 2(4), 359–392, doi:10.1002/wat2.1085,
2015.
Van Loon, A. F. and Laaha, G.: Hydrological drought severity explained by climate and catchment characteristics, J.
535 Hydrol., 526, 3–14, doi:10.1016/j.jhydrol.2014.10.059, 2015.
Van Loon, A. F. and Van Lanen, H. A. J.: A process-based typology of hydrological drought, Hydrol. Earth Syst. Sci., 16(7),
1915–1946, doi:10.5194/hess-16-1915-2012, 2012.
Van Loon, A. F., Tijdeman, E., Wanders, N., Van Lanen, H. A. J., Teuling, A. J. and Uijlenhoet, R.: How climate
seasonalitymodifies drought duration and deficit, J. Geophys. Res., 119(8), 4640–4656, doi:10.1002/2013jd020383, 2014.
540 Masotti, I., Aparicio-Rizzo, P., Yevenes, M. A., Garreaud, R., Belmar, L. and Farías, L.: The Influence of River Discharge
on Nutrient Export and Phytoplankton Biomass Off the Central Chile Coast ( 33◦ – 37◦ S ): Seasonal Cycle and Interannual
Variability, Front. Mar. Sci., 5(November), 1–12, doi:10.3389/fmars.2018.00423, 2018.
Muñoz, A. A., Klock-barría, K. and Alvarez-garreton, C.: Water crisis in Petorca basin , Chile : the combined effects of a
mega-drought and water management, submitted, 1–19, 2020.
545 Orth, R. and Seneviratne, S. I.: Propagation of soil moisture memory to streamflow and evapotranspiration in Europe,
22
Hydrol. Earth Syst. Sci., 17(10), 3895–3911, doi:10.5194/hess-17-3895-2013, 2013.

Peña, E. A. and Slate, E. H.: Global validation of linear model assumptions, J. Am. Stat. Assoc.,
doi:10.1198/016214505000000637, 2006.
Peters, E., Bier, G., van Lanen, H. A. J. and Torfs, P. J. J. F.: Propagation and spatial distribution of drought in a
550 groundwater catchment, J. Hydrol., 321(1–4), 257–275, doi:10.1016/j.jhydrol.2005.08.004, 2006.
Pool, S., Vis, M. and Seibert, J.: Evaluating model performance: towards a non-parametric variant of the Kling-Gupta
efficiency, Hydrol. Sci. J., 63(13–14), 1941–1953, doi:10.1080/02626667.2018.1552002, 2018.
Robinson, M. and Ward, R.: Hydrology: principles and processes, 2017th ed., IWA., 2017.
Rodríguez-Iturbe, I. and Valdes, J. B.: The geomorphologic structure of hydrologic response, Water Resour. Res., 15(6),
555 1409–1420, 1979.
Rumsey, C. A., Miller, M. P., Susong, D. D., Tillman, F. D. and Anning, D. W.: Regional scale estimates of baseflow and
factors influencing baseflow in the Upper Colorado River Basin, J. Hydrol. Reg. Stud., 4(PB), 91–107,
doi:10.1016/j.ejrh.2015.04.008, 2015.
Saft, M., Western, A. W., Zhang, L., Peel, M. C. and Potter, N. J.: The influence of multiyear drought on the annual rainfall-
560 runoff relationship: An Australian perspective, Water Resour. Res., 51(4), 2444–2463, doi:10.1002/2014WR015348, 2015.
Saft, M., Peel, M. C., Western, A. W., Perraud, J. M. and Zhang, L.: Bias in streamflow projections due to climate-induced
shifts in catchment response, Geophys. Res. Lett., 43(4), 1574–1581, doi:10.1002/2015GL067326, 2016a.
Saft, M., Peel, M. C., Western, A. W. and Zhang, L.: Predicting shifts in rainfall-runoff partitioning during multiyear
drought: Roles of dry period and catchment characteristics, Water Resour. Res., 52(12), 9290–9305,
565 doi:10.1002/2016WR019525, 2016b.
Schewe, J., Heinke, J., Gerten, D., Haddeland, I., Arnell, N. W. and Clark, D. B.: Multimodel assessment of water scarcity
under climate change, , 111(9), doi:10.1073/pnas.1222460110, 2014.
Seibert, J.: Multi-criteria calibration of a conceptual runoff model using a genetic algorithm, Hydrol. Earth Syst. Sci.,
doi:10.5194/hess-4-215-2000, 2000.
570 Staudinger, M., Stoelzle, M., Seeger, S., Seibert, J., Weiler, M. and Stahl, K.: Catchment water storage variation with
elevation, Hydrol. Process., doi:10.1002/hyp.11158, 2017.
Tachikawa, T., Hato, M., Kaku, M. and Iwasaki, A.: Characteristics of ASTER GDEM version 2, in International
Geoscience and Remote Sensing Symposium (IGARSS), pp. 3657–3660., 2011.
Yang, Y., McVicar, T. R., Donohue, R. J., Zhang, Y., Roderick, M. L., Chiew, F. H. S., Zhang, L. and Zhang, J.: Lags in
575 hydrologic recovery following an extreme drought, Water Resour. Res., 4821–4837, doi:10.1002/2017WR020683.Received,
2017.
23

Progressive Water Deficits During Multi-Year Droug PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Progressive Water Deficits During Multi-Year Droug PDF

Uploaded by

Copyright:

Available Formats

https://doi.org/10.

Progressive water deficits during multi-year droughts in central-

Correspondence to: Camila Alvarez-Garreton (calvarezgarreton@gmail.com)

2 Study region and data

120 3.1 Drought characterization

3.2 Changes in catchment response during multi-year droughts

3.3 Hydrological model

3.4 Runoff mechanisms and hydrological memory

4.1 Droughts over the last decades

2003 2003 2003 2003

1983 1983 2006 1983 1983

2007 2007 1981 2007 2007

−38.299 −38.299 −38.299 −38.299

occurring (lower z-score values, indicating higher exceedance probabilities).

−35.55 −35.55 −35.55 −35.55

−35.4069 −35.4069 −35.4069 −35.4069

−35.2783 −35.2783 −35.2783 −35.2783

−35.1783 −35.1783 −35.1783 −35.1783

−35.1731 −35.1731 −35.1731 −35.1731

−34.9961 −34.9961 −34.9961 −34.9961

−34.9844 −34.9844 −34.9844 −34.9844

−34.7186 −34.7186 −34.7186 −34.7186

−34.6867 −34.6867 −34.6867 −34.6867

−34.4922 −34.4922 −34.4922 −34.4922

−34.4311 −34.4311 −34.4311 −34.4311

−34.2467 −34.2467 −34.2467 −34.2467

−33.8111 −33.8111 −33.8111 −33.8111

−33.8056 −33.8056 −33.8056 −33.8056

−33.7347 −33.7347 −33.7347 −33.7347

−33.7219 −33.7219 −33.7219 −33.7219

−33.6614 −33.6614 −33.6614 −33.6614

−33.5939 −33.5939 −33.5939 −33.5939

−33.5875 −33.5875 −33.5875 −33.5875

−33.3703 −33.3703 −33.3703 −33.3703

−33.3414 −33.3414 −33.3414 −33.3414

−33.3256 −33.3256 −33.3256 −33.3256

−32.9072 −32.9072 −32.9072 −32.9072

−32.8625 −32.8625 −32.8625 −32.8625

−32.7572 −32.7572 −32.7572 −32.7572

−32.33 −32.33 −32.33 −32.33

−32.2253 −32.2253 −32.2253 −32.2253

−32.0708 −32.0708 −32.0708 −32.0708

−31.9667 −31.9667 −31.9667 −31.9667

−31.9481 −31.9481 −31.9481 −31.9481

−31.6647 −31.6647 −31.6647 −31.6647

−31.5608 −31.5608 −31.5608 −31.5608

−31.2281 −31.2281 −31.2281 −31.2281

−31.1119 −31.1119 −31.1119 −31.1119

−31.0325 −31.0325 −31.0325 −31.0325

−31.0119 −31.0119 −31.0119 −31.0119

−30.8422 −30.8422 −30.8422 −30.8422

−30.8128 −30.8128 −30.8128 −30.8128

−30.8006 −30.8006 −30.8006 −30.8006

−30.7047 −30.7047 −30.7047 −30.7047

−30.4622 −30.4622 −30.4622 −30.4622

4.2 Shifts in R-R relationships during the MD

200 400 600 500

500 ● 500 ● ● ● 3000

Annual runoﬀ (mm)

Annual runoﬀ (mm)

Annual precipitation (mm)● ● Annual precipitation (mm)

0.00 0.25 0.50 0.75 1.00 0.0 0.1 0.2 0.3

−38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167 −38.6167