The anomalous 2017 coastal El Niño event in Peru

Abstract

Remarkably heavy and devastating rainfalls affected large parts of Peru during the austral summer 2016–2017. These rainfalls favoured widespread land sliding and extensive flooding and generated one of the most severe disasters of Peru since the 1997–1998 El Niño event. The amount of rainfall recorded between January and March 2017 only compares to the biggest El Niño events of the last 40 years (i.e. 1982–1983 and 1997–1998) and exceeded the 90th percentile of available records (1981–2017) over much of the northern and central coasts of Peru, the Andean region and Amazonia. The occurrence of these heavy rainfalls was highly anomalous as it occurred during the first austral summer following the development and decay of a very strong El Niño in 2015–2016. Here, we propose that the likely cause of the anomalous rainfalls is linked to the combination of an especially intense wet spell over the Central Andes related to a deep, long-lasting anticyclone located adjacent to the Chilean coast, and to the unusual development of warm water off the coast of Peru in the nominal El Niño 1 + 2 region. This warming has been related to an anomalous weakening of the mid-upper level subtropical westerly flow, which in turn led to a weakening of the southeasterly trades off the coast, thus hindering the upwelling near the Peruvian coast and favoring the eastern Pacific warming. This development is counter to the usual evolution of sea surface temperature in the eastern equatorial Pacific following very strong El Niño events, such as those occurred in 1982–1983, 1997–1998, and 2015–2016. This paper explores the unusual nature of this event in the observational record and illustrates its consequences.

Full text

Vol.:(0123456789)
1 3
Climate Dynamics
https://doi.org/10.1007/s00382-018-4466-y
The anomalous 2017 coastal El Niño event inPeru
C.Rodríguez‑Morata1· H.F.Díaz2· J.A.Ballesteros‑Canovas1,3· M.Rohrer1,4· M.Stoel1,3,5
Received: 20 March 2018 / Accepted: 22 September 2018
© Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract
Remarkably heavy and devastating rainfalls affected large parts of Peru during the austral summer 2016–2017. These rainfalls
favoured widespread land sliding and extensive flooding and generated one of the most severe disasters of Peru since the
1997–1998 El Niño event. The amount of rainfall recorded between January and March 2017 only compares to the biggest
El Niño events of the last 40years (i.e. 1982–1983 and 1997–1998) and exceeded the 90th percentile of available records
(1981–2017) over much of the northern and central coasts of Peru, the Andean region and Amazonia. The occurrence of
these heavy rainfalls was highly anomalous as it occurred during the first austral summer following the development and
decay of a very strong El Niño in 2015–2016. Here, we propose that the likely cause of the anomalous rainfalls is linked to
the combination of an especially intense wet spell over the Central Andes related to a deep, long-lasting anticyclone located
adjacent to the Chilean coast, and to the unusual development of warm water off the coast of Peru in the nominal El Niño
1 + 2 region. This warming has been related to an anomalous weakening of the mid-upper level subtropical westerly flow,
which in turn led to a weakening of the southeasterly trades off the coast, thus hindering the upwelling near the Peruvian coast
and favoring the eastern Pacific warming. This development is counter to the usual evolution of sea surface temperature in
the eastern equatorial Pacific following very strong El Niño events, such as those occurred in 1982–1983, 1997–1998, and
2015–2016. This paper explores the unusual nature of this event in the observational record and illustrates its consequences.
Keywords Coastal El Niño· Sea surface temperature· Rainfalls· Flooding· Peru
1 Introduction
During the 2016–2017 austral summer, Peru experienced
widespread and unusually heavy rainfall, which triggered
landslides and severe flooding, thereby leading to cata-
strophic damage to housing and infrastructure, and affect-
ing more than 6,60,000 people, and leaving more than 100
deaths (INDECI 2017). The areas most affected by the dis-
aster were located in the northern and central coastal regions
of Peru (Fig.1), as well as southeast of the Loreto region in
the Peruvian Amazonas basin. In addition to the death tolls
and physical injuries, these heavy rainfalls also resulted in
increased exposure of local populations to disease pathogens
and mental health problems associated with loss, disrup-
tion, and displacement (Hales etal. 2003; Waring and Brown
2005; Few 2007). In the context of the 2016–2017 extreme
rainfall season 6270 dengue cases were confirmed in Peru
(PAHO 2017), which triggered the declaration of a sanitary
emergency by the Ministry of Health (MINSA) in seven
regions of Peru. The heavy rainfalls of the 2016–2017 aus-
tral summer were highly anomalous because they occurred
Electronic supplementary material The online version of this
article (https ://doi.org/10.1007/s0038 2-018-4466-y) contains
supplementary material, which is available to authorized users.
* C. Rodríguez-Morata
Clara.Rodriguez@unige.ch
1 Climate Change Impacts andRisks intheAnthropocene
(C-CIA), Institute forEnvironmental Sciences, University
ofGeneva, Boulevard Carl-Vogt 66, 1205Geneva,
Switzerland
2 Department ofGeography andEnvironment, University
ofHawai‘i atManoa, Honolulu, USA
3 Dendrolab.ch, Department ofEarth Sciences, University
ofGeneva, Rue des Maraîchers 13, 1205Geneva,
Switzerland
4 Meteodat GmbH, Technoparkstrasse 1, 8005Zurich,
Switzerland
5 Department F.A. Forel forAquatic andEnvironmental
Sciences, University ofGeneva, Boulevard Carl-Vogt 66,
1205Geneva, Switzerland

C.Rodríguez-Morata et al.
1 3
during the first austral summer following the development
and decay of a very strong El Niño in 2015–2016. The rain-
fall recorded between January and March 2017 exceeded
the 90th percentile of available measurements for the period
1981–2017 over much of the northern and central coasts
of Peru, the Andean region and Amazonia. This event is
only comparable to the rainfall measured during the larg-
est El Niño events of the last 40years (i.e. 1982–1983 and
1997–1998). Interestingly, in these areas, rainfall totals dur-
ing the 2016–2017 austral summer exceeded those recorded
during the 2015–2016 El Niño event. The 2015–2016 El
Niño has been classified as a strong event by the interna-
tional community, even if it has had much less dramatic
consequences for Peru than the strong El Niño events in
1972–1973, 1982–1983, and 1997–1998 (L’Heureux etal.
2017; Sanabria etal. 2018).
Rainfall during the 2016–2017 austral summer occurred
in conjunction with largely negative (positive) SST anoma-
lies in the central (eastern) equatorial Pacific off the coast
of South America. Based on the monthly “ENSO Diag-
nostic Discussion Archive” from the Climate Predic-
tion Center (CPC 2018a), below normal SST anomalies
were registered in the Niño 3.4 region for November and
December 2016, suggesting a developing La Niña event
(NOAA 2017). The Oceanic Niño Index (ONI) however
extends La Niña status from July–September 2016 to
November-January 2017 (CPC 2018b). However, during
January 2017 and extending through March 2017, a strong
Fig. 1 Geographical distribution
of regions and rain gauges used
in this study. A digital elevation
model shows the topographical
characteristics and altitudes of
Peru

The anomalous 2017 coastal El Niño event inPeru
1 3
positive SST anomaly was observed in the 1 + 2 region,
leading to an event which has previously been referred to a
“Coastal El Niño” (Takahashi and Martínez 2017; hereaf-
ter CEN-2017). Therefore, a watch status was put in place
at the end of January 2017, and an alert status was issued
between February and May 2017 (ENFEN 2017a, b).
The observed SST anomaly event in the equatorial
Pacific during the 2016–2017 austral summer was highly
atypical because it occurred during the austral summer
season, and at a time when the 2015–2016 El Niño event
was in its last phase of decay (L´Heureux etal. 2017). The
situation in early 2017 differed substantially from previous
summers following strong El Niños in terms of the spatio-
temporal SST pattern. Thus, the 2016–2017 rainfall event
cannot be classified with classical ENSO definitions (i.e.
Rasmusson and Carpenter 1982; Ashok etal. 2007; Kug
etal. 2009; Yeh etal. 2009; Kao and Yu 2009; Takahashi
etal. 2011; Yu etal. 2011). The reasons for this may be
related to the fact that (1) the 2016–2017 austral sum-
mer event was not a clearly coupled ocean–atmosphere
phenomenon in the equatorial Pacific (Garreaud 2018),
and that (2) the warming was not present throughout the
Equatorial Pacific Basin.
The ENSO cycle may present strong variability in terms
of its amplitude, temporal evolution, and spatial pattern
(Larkin and Harrison 2005; Ashok etal. 2007; Takahashi
etal. 2011; Dommenget etal. 2013; Capotondi etal. 2015).
For instance, Takahashi and Martinez (2017) describe a pre-
vious coastal event in 1925 as the result of cold conditions in
the western-central equatorial Pacific that would have helped
to destabilize the ITCZ and generate strong northerly winds
across the equator. More recently, Garreaud (2018) devel-
oped a first approximation to the origin of the CEN-2017
providing a reasonable description of the atmospheric forc-
ing involved in the origin of the coastal SST overwarming.
This author proposed that during the recent coastal El Niño,
an external forcing provided by vacillations in the free-trop-
ospheric subtropical westerlies would have led to a weaken-
ing of the SE trades at lower levels and consequently to the
increase of coastal SST.
In this paper, we take advantage of the information con-
tained in these previous works to extend our understanding
about (1) the atmospheric and oceanic context during the
2016–2017 austral summer in Peru, and (2) the impacts
it has had on the magnitude, distribution, and timing of
extreme rainfalls that occurred over Peru. To this end, we
first describe the CEN-2017 event and related impacts in Peru
before we explore how different this event was compared to
others in the historical record. We use different reanalysis
data for SST and atmospheric variables as well as gridded
and observational rainfall data sources to contextualize the
2016–2017 austral summer precipitation and to describe the
spatio-temporal evolution of the event during this period.
2 Data andmethods
2.1 Precipitation analysis
We used rainfall information from 338 Peruvian gauge
stations, provided by the National Service of Meteorol-
ogy and Hydrology of Peru (SENAMHI), the MeteoDat
portal (Schwarb etal. 2011) as well as gridded data from
the GPM IMERG dataset (Huffman 2015) to characterize
the spatial distribution and intensity of accumulated rain-
fall during the 2016–2017 DJFM austral summer. Gauge
stations are distributed all along the country but major
density is founded in the most populated coastal regions
(Fig.1). Yet, rainfall records of these stations are not
always continuous and the number of gauge stations with a
full record for each month can vary substantially. Gridded
satellite data from GPM IMERG has a spatial resolution
of 0.1° × 0.1° but only covers the period between 2014 to
the present (extended information is available at https ://
pmm.nasa.gov/index .php?q=GPM). In fact, it is possible
that observed values from gauge stations can be different
from GPM Imerg values for the same location. This is due
to (1) the inherent error associated with the precipitation
model (Huffman etal. 2017) and (2) the 0.1° × 0.1° spatial
resolution of the GPM data, meaning that rainfall values
are in fact average values of an area of 121km2 for which
differences in precipitation must be expected at more local
scales.
To provide further context for the magnitude of the pre-
cipitation event, we have used gridded data (1° × 1°) from
GPCC Monitoring Product (Schneider etal. 2015) as well
as observational data from 73 Peruvian gauge stations cov-
ering the entire period (1982–2017), (i.e. 17 gauge stations
for the month of December; 40 for January; 24 for Febru-
ary and 34 for March). With this data base, we have com-
puted rank percentile maps of precipitation to compare the
2016–2017 DJFM monthly precipitation with rainfall data
for the same period over the last 36years (1982–2017).
2.2 Large‑scale atmospheric andoceanic synoptic
analysis ofthe2016–2017 coastal event
The domain of the synoptic analyses covers the entire
South American continent and part of the East Pacific and
West Atlantic oceans and is large enough to capture the
main features of climate variability over South America
[i.e., the Bolivian High (BH), the South American mon-
soon (SAH), Intertropical Convergence Zone (ITCZ), and
the South Atlantic Convergence Zone (SACZ); Garreaud
etal. 2009]. At high levels (i.e., 200hPa), the Bolivian
High (BH) is the major synoptic feature occurring during

C.Rodríguez-Morata et al.
1 3
austral summers (Lenters and Cook 1997). It consists in
a closed anticyclone developing over the central Andes,
which forms in response to latent heat released during
deep convection over the Amazon basin (Sulca etal.
2016).To describe the CEN-2017, we have used daily
data of geopotential height, wind velocity components,
outgoing longwave radiation (OLR) and sea surface tem-
perature (SST) from the NCEP/NCAR reanalysis (Kalnay
etal. 1996) and provided by the NOAA/OAR/ESRL PSD
(http://www.esrl.noaa.gov/psd/). This data is available on
a 2.5° × 2.5° grid at 17 pressure levels. Anomalies refer to
the 1981–2010 climatology. Outgoing long-wave radiation
(OLR) anomalies were used as a proxy for convective pro-
cesses (NCAR 2014). Negative (positive) OLR anomalies
are indicative of enhanced (suppressed) convection and
hence more (less) cloud coverage. More (less) convec-
tive activity in the central and eastern equatorial Pacific
implies higher (lower), colder (warmer) cloud tops, which
emit much less (more) infrared radiation into space.
Analysis of the dominance and changes of synoptic pat-
terns over time between December 2016 and March 2017 (at
the daily scale) was performed with Self Organizing Maps
(SOMs; Kohonen 2001; Hewitson and Crane 2002; Reusch
etal. 2005a, b, 2007; Cassano etal. 2006a). To this end, we
have used geopotential and zonal wind anomalies at 200,
500, and 850hPa as well as daily SST data, with a spatial
resolution of 2.5° × 2.5°, from the ERA-Interim reanalysis
climate data set (ECMWF 2016) with a time span from 1979
to present (Dee etal. 2011). ERA-Interim data was retrieved
from http://www.ecmwf .int/resea rch/era.
2.2.1 Self‑organizing maps
Self-organizing maps (SOM) represent a clustering tech-
nique that allows summarizing large, high-dimensional
records by treating data as a continuum. SOM identify pat-
terns using an iterative clustering algorithm (Hewitson and
Crane 2002), and produce a set of nodes (i.e., generic syn-
optic states directly interpretable as physical process states)
in a two-dimensional lattice with similar states close to each
other and the most extreme states at the opposite corners.
Analyses of these nodes allow the characterization of the
frequency of each synoptic state, the spatio-temporal transi-
tions between states as well as their dominance in a given
temporal horizon (Kohonen 2001; Hewitson and Crane
2002).
This technique has been used successfully in many
meteorological, climatological, and oceanic research
applications worldwide, either to characterize extreme
weather and rainfall events (Hong etal. 2005; Cassano
etal. 2006a, b; Morata etal. 2006; Zhang etal. 2006;
Uotila etal. 2007; Schuenemann etal. 2009), including in
the Peruvian Amazon and Andes regions (Espinoza etal.
2012, 2013; Paccini etal. 2017; Rodriguez-Morata etal.
2018), to visualize synoptic weather patterns over a region
(Hewitson and Crane 2002; Reusch etal. 2005a, b, 2007;
Johnson etal. 2008; Seefeldt and Cassano 2008; Wiseand
Dannenberg 2014), or to evaluate Global Climate Model
(GCM) results (Lynch etal. 2006; Cassano etal. 2007;
Skific etal. 2009a, b).
Initially, SOMs are formed by an arbitrary number of
clusters or nodes. Each cluster is associated with two vec-
tors. The first vector describes the position of the cluster
on the lattice, whereas the second (also referred to as ref-
erence vector) represents the position of the cluster cen-
troid in the data space. By using an iterative process, an
unsupervised algorithm is applied to adjust the reference
vectors representing the nodes based on the differences
between the reference vectors and each input value. In
each iteration, the Euclidean distance between input data
and reference vectors is calculated and the best match-
ing reference vector is identified for each input record.
Neighboring reference vectors of each best match are then
updated to result in adjacent nodes having the strongest
similarity. Iterations are ended when stable values of the
reference vectors are reached. The choice of the number
of nodes depends on the specific research context and
amount of data. Generally, smaller (larger) number of
nodes implies less (more) possibilities to characterize the
high-dimensional data space and therefore more (less)
generalization of the input data.
To analyze the 2016–2017 austral summer event, we
defined a 5 × 5 nodes lattice to best discriminate the main
synoptic South American summertime features including
those patterns associated with ENSO (positive and nega-
tive phases). The SOM analysis was carried out with the
MeteoLab toolbox for Matlab (http://grupo s.unica n.es/ai/
meteo /Meteo Lab.html) using a linear decay to zero for
both the learning rate and neighborhood amplitude after
5000 cycles. Input records (i.e. days) with common syn-
optic patterns were then linked to the same SOM node
or cluster. The resulting set of clusters (SOM) represents
meaningful subgroups (i.e. generic climatic patterns)
within the larger dataset. To identify these groups (i.e.
nodes) we used a coordinate system naming the rows with
numbers (i.e. 1–5) and the columns with letters (i.e. A–E).
Note that while several variables can be jointly analyzed
only one SOM grid is produced and each SOM node con-
tains the four variables. However, for clarity, SOM maps
for each variable are shown separately. The daily scale of
our study has allowed us to construct frequency maps of
the SOM grid for each austral summer (DJFM) since 1979,
thereby allowing us to track transitions between synoptic
states and the dominance (i.e. the number of days that a
synoptic pattern is present during the austral summer) of
each state.

The anomalous 2017 coastal El Niño event inPeru
1 3
2.3 Interannual comparison ofaustral summers
We have used the OLR anomaly of the NOAA/Monthly
Mean upward longwave flux at top of the atmosphere
dataset since 1979 (http://clime xp.knmi.nl/selec t.cgi?field
=umd_olr) as well as SSTs from the HadISST1 data set, and
derived El Niño 1 + 2 and 3.4 indices (Rayner etal. 2003)
to assess similarities/differences between the 2016–2017
rainfall event and other austral summers since 1870. Had-
ISST1 has a monthly resolution, from 1870 to date, and is
available on a 1° × 1° grid. The use of the 3.4 and 1 + 2
SST indices served to separate the different behavior in
the central (3.4; 5°N–5°S, 170°W–120°W) and eastern
(1 + 2; 0°–10°S, 90°W–80°W) Pacific, since SST in these
two regions modulate rainfall over Peru in different ways
(Lavado-Casimiro and Espinoza 2014; Sulca etal. 2017).
Indices were obtained from https ://www.esrl.noaa.gov/psd/
gcos_wgsp/Times eries /. We employed a Superposed Epoch
Analysis (SEA) to compare 1 + 2 and 3.4 El Niño indices
based on SST standardized monthly anomalies correspond-
ing to all summers after El Niño events since 1950 (this is
the temporal limitation of the ONI index). The selection of
El Niño events was based on the classification of the Oce-
anic Niño Index (ONI) (http://www.cpc.noaa.gov/produ cts/
analy sis_monit oring /ensos tuff/ensoy ears.shtml ). This index
is most commonly used to define El Niño and La Niña events
and is based on SST anomalies in the Niño 3.4 region, which
represents the average equatorial SSTs across the Pacific
from about the dateline to the South American coast (Tren-
berth 2016). The ONI uses a 3-month running mean, and
to be classified as a full-fledged El Niño or La Niña, the
anomalies must exceed + 0.5°C or − 0.5°C for at least five
consecutive months. Furthermore, we wanted to compare the
SST distribution in the equatorial Pacific during the CEN-
2017 event with its counterpart post-strong El Niño austral
summers in 1878–1879, 1983–1984 and 1998–1999. Statis-
tical analyses have been carried out using the t-test for one-
sided sample (Haynes 2013) at significance level of 0.05.
Fig. 2 Maps representing the DJFM 2016–2017 accumulated rainfall
using data from (1) 300 rain gauges distributed all along Peru and (2)
GPM IMERG gridded data of 0.1° × 0.1° spatial resolution. Black
star is indicating the location of Lima. Note that the scale range is
different for the accumulated rainfall for the entire period from DJFM
(a) and at monthly scale (b–e)

C.Rodríguez-Morata et al.
1 3
3 Results
3.1 December–March precipitation analysis
overPeru
From December 2016 to February 2017, accumulated pre-
cipitation over Peru (Fig.2) varied from extremely low
values (Obs.: 0mm; GPM: 31.76mm; Fig.2a) to unusu-
ally high totals (Obs.: 3291mm; GPM: 2142mm; Fig.2a).
By month, data from stations indicate that the highest
values of rainfall were recorded in March (1062mm;
Fig.2e) in the station of Pasaje Sur, in the north of Lam-
bayeque region. December 2016 (Fig.2b) exhibits the
second highest value (856.2mm) in El Boquerón station,
in the border between the Huánuco and Ucayali regions.
January 2017 (Fig.2c) recorded its maximum at the same
station (812.1mm) and February 2017 rainfalls (Fig.2d)
were highest at Quincemil station (831.1mm) located at
the border between Cuzco and Madre de Dios.
Rank percentile maps (Fig.3) show that many areas
of Peru received cumulative precipitation totals between
Fig. 3 Rank percentile maps for DJFM 2016–2017. The number of stations available for each month varies substantially. We indicate the number
of stations having a full record for the period 1982–2017 for each month

The anomalous 2017 coastal El Niño event inPeru
1 3
December 2016 and March 2017 that exceed all values
recorded since 1982. Regarding GPCC data by month,
cumulative precipitation exceeded the 80th percentile in
only some parts (7.9%) of the country in December 2016
(Fig.3a). In west Loreto and in the Ancash region precipi-
tation totals were over the 90th percentile or even unprec-
edented (i.e. 100th) in the case of the Ancash coast. By
contrast, in the other parts of Peru, rainfall totals remained
under the 50th percentile. During January 2017 (Fig.3b),
the 80th percentile was exceeded in 35.8% of the country,
whereas the 90th percentile was exceeded along the North
(i.e. Tumbes and Piura) and Central coast (i.e. Ancash and
Lima). Unprecedented values were seen along the South
coast (i.e. Arequipa), Central Andes (i.e. San Martin, Huá-
nuco, Pasto and Junín) as well as in the Amazonian lowlands
(i.e. south-west Loreto, Amazonas and Ucayali). In Febru-
ary 2017 (Fig.3c), cumulative precipitation exceeded the
80th percentile over 17% of the country but values above
the 90th percentile were still observed along the North coast
(i.e. Piura and Lambayeque) as well as west of the Loreto
and Huánuco regions. During March 2017 (Fig.3d), more
than one-fourth of the country (28.5%) shows percentile val-
ues above the 80th percentile and two spots – i.e. along the
North coast (Tumbes and Piura) and in the Central Andes
(Huánuco, Pasto, Junin and Ancash) exhibit 100th percentile
values.
From the 17 gauge station records available for Decem-
ber 2016, only two show rank percentiles above the 60th
percentile: Bambamarc (66th) and Querocotillo (97th). In
line with the GPCC data, the other stations present values
under the 50th percentile. In January 2017, out of 40 sta-
tions, 25 exhibit values above the 60th, 15 above the 90th
and 5 reached the 100th percentile, most of the latter being
located in the western part of the Peruvian Central Andes.
From the 24 stations available for February 2017, 17 exhib-
ited values above the 60th, 7 above the 90th and 3 reached
the 100th percentile, again on the western Andean slopes but
also along the north coast of Peru. In March 2017 (for which
34 stations are available), 27 stations show values above the
60th, 16 above the 90th and 8 reaching the 100th percentile,
all of the latter are located along the west Andean slope.
3.2 Large‑scale atmospheric andoceanic synoptic
analysis ofthe2016–2017 coastal event
3.2.1 Event description
From December 2016 to March 2017, consecutive geopoten-
tial anomalies at 200hPa occur adjacent to the Chilean coast
centered at 33°S (Fig.4a). The biggest anticyclonic anomaly
developed on 20th January 2017 and lasted about 2weeks
with values above 112 gpm reaching positive anomalies of
Fig. 4 Time-longitude plots for the period 1/12/2016 to 30/4/2017
averaged at latitude 33°S. a Geopotential height anomaly at 200hPa.
b Zonal wind anomaly at 500hPa. Several periods of positive geo-
potential anomalies are observed and concurrent with strong negative
anomalies in the zonal winds at the free troposphere (red colors in
both plots)

C.Rodríguez-Morata et al.
1 3
200 gpm in some periods. Furthermore, two other, simi-
lar events occur in February and March 2017 but were of
shorter duration and less intense with anomaly values
ranging between 90 and 158 gpm. Coincident with these
height anomalies, strong negative zonal wind anomalies
are observed at 500hPa with net velocity values ranging
between 14.4 and 24m/s (Fig.4b). The negative character
of these wind anomalies indicates a weakening of the west-
erly wind component and therefore a strengthening of the
easterly component at 33°S. The average of the 1000hPa
vector wind anomaly along the Peruvian coast between Janu-
ary 20 and March 25, 2017, displays a common west-east
vector wind anomaly pattern just off the north coast of Peru
(Fig.5a) with a maximum speed wind anomaly of 4.8m/s.
Figure5b–d represent the situation corresponding with the
strongest 5-day anomalies of specific events for January,
February, and March, with highest speed anomalies occur-
ring in February and March with 5.1 and 5.2m/s, respec-
tively, just off the north coast (Fig.5c,d).
Large-scale convection development during the CEN-
2017 event is represented by OLR anomalies in Fig.6. In
December, the main convection center is located in the
Amazon basin, while a secondary area starts to develop
over the south of Brazil (Fig.6a). Later in January, in this
second area an enhancement of the negative anomaly of the
OLR can be seen with a net maximum value of − 30W/m2,
indicating an increase of the convective activity (Fig.6b).
Additionally, the shape of this feature extends in a NW–SE
diagonal forming a band of convection typical of the South
Atlantic Convergence Zone (SACZ). Also during January,
negative OLR anomalies increase along the Peruvian and
north Chilean coast. In February, the convective band asso-
ciated with the SACZ is less intense (higher values of OLR)
and moves southward. At the same time an intense con-
vective band progress in the central-east equatorial Pacific
(Fig.6c). During March, the anomaly associated with the
SACZ practically disappears and the most extensive con-
vective center is located in the eastern equatorial Pacific
adjacent to the northern Peruvian coast with negative OLR
anomalies of 30W/m2 (Fig.6d).
The spatio-temporal propagation of the absolute SST in
the equatorial Pacific as well as along the Peruvian coast
from November 2016 to November 2017 is represented in
Fig.7. Thus, for 0°–10°S (El Niño region 1 + 2; Fig.7a)
SST values raised abruptly by 2°C (from 25 to 27°C) along
the Peruvian north coast around mid-January until they pro-
gressively reach SST values > 30°C between mid-February
and mid-March (Fig.7a). On the other hand, for 10°S–17°S
(Fig.7b), SST were more moderate along the south-central
coast of Peru (Fig.7b) with maxima of 28°C in mid-March.
Considering SST anomalies (Fig.8), December 2016 was
characterized by negative SST anomalies in the central-east-
ern Pacific (3 and 3.4 Niño regions) and close to neutral SST
prevailed in region 1 + 2 (Fig.8a). During January 2017,
Fig. 5 Near surface vector wind anomalies between January and
March 2017 indicating a trend from the west to the east in the equato-
rial band and along the South American coast thus indicating easter-
lies weakening. a Average anomaly from January 20th to March 25th.
b–d the situation for different moments is concurrent with the U wind
negative anomalies represented in the Fig.4b

The anomalous 2017 coastal El Niño event inPeru
1 3
Fig. 6 Monthly Outgoing Longwave Radiation (OLR) anomaly for December 2016 to March 2017
Fig. 7 Sea surface temperature (SST) time-longitude plot showing the
overwarming in the South Pacific averaged to a El Niño 1 + 2 region
(averaged between 0° and 10°S); and b along the center and south
Peruvian coast (averaged between 10°S and 17°S). Data provided by
the NOAA/ESRL Physical Sciences Division, Boulder Colorado from
their Web site at http://www.esrl.noaa.gov/psd/

C.Rodríguez-Morata et al.
1 3
negative SST anomalies started to become more intense in
the equatorial Pacific, but remained restricted to region 3.4.
At the same time, SST anomalies in the 1 + 2 region start
to rise (Fig.8b). Between February and March 2017, the
SST anomaly was at its maximum with values 2.8°C above
the average in region 1 + 2 (Fig.8c, d). Central Pacific SST
anomalies were close to neutral values.
3.2.2 SOM analysis
The SOM maps corresponding to the anomalies of the geo-
potential (Z) and zonal wind anomaly (U) at 200, 500 and
850hPa, as well as SST show 25 possibilities of synoptic
settings during the austral summer (an overview of all SOM
maps is provided in FiguresS1–S7, Supporting information).
The distribution of these 25 possibilities place contrasting
situations in opposite sides of the SOM map (i.e., El Niño
like patterns in the right side and La Niña in the left side)
and a range of intermediate possibilities in between. Thus we
find that strong El Niño events (i.e. 1982–1983, 1997–1998,
2009–2010, and 2015–2016) are mostly related with node
1E (Fig.9) representing a pattern dominated by (1) a full-
basin overwarming in the equatorial Pacific (Fig. S7) and
(2) a strong positive 850hPa zonal wind anomaly along
the equatorial Pacific, characteristic of El Niño events (Fig.
S6). In opposite, strong La Niña events (i.e. 1988–1989,
1998–1999, 1999–2000, 2007–2008 and 2010–2011) are
related with nodes of column A, linked to different exten-
sions of cooling in the equatorial Pacific (Fig. S7). Synoptic
states related with neutral conditions are scattered in the
center of the SOM grid (Fig.9) and are related with varia-
tions in the strength and position of pressure centers over the
continent, the south Pacific, and the south Atlantic (Fig S1,
2, 3). Weak-to-moderate ENSO extremes are also related to
scattered synoptic states but exhibit a clear trend toward El
Niño or La Niña sides of the SOM grid.
The temporal evolution and dominance (i.e. the number
of days of pattern activity) of each generalized pattern dur-
ing each austral summer since 1979 (considering 121days
between December to March) are represented in Fig.9. We
identify different pattern tracks depending on the ENSO
phase (i.e. El Niño, La Niña, or neutral conditions). For
strong El Niño and La Niña years, the path is restricted to
few a patterns with very high dominance in each setting. For
example, in the case of the strong El Niño in 1982–1983,
1997–1998 and 2015–2016, the dominance of the node 1E
was of 61, 86, and 88 days, respectively. However, in the
case of neutral and weak-moderate El Niño and La Niña
years, the path is more diverse with less time at each node.
For the 2016–2017 austral summer event, however, the path
was very much restricted to the nodes in the central columns
of the SOM map, and moves from nodes in the lower part of
Fig. 8 Plots a–d show the evolution of the SST monthly anomaly in the equatorial Pacific between December 2016 and March 2017. We use
HadISST1 data provided by the NOAA/OAR/ESRL PSD (http://www.esrl.noaa.gov/psd/)

The anomalous 2017 coastal El Niño event inPeru
1 3
the grid (associated with cold SST in the equatorial Pacific),
to nodes in the upper part (related with warmer SST in the
1 + 2 El Niño region than in the central Pacific). Upper-level
pressure anomalies were negative to neutral at the beginning
of the event and then evolve to very positive anomalies.
3.3 The 2017 coastal event inalarger context
Figure10 represents the JFM averaged anomaly of El Niño
3.4 and 1 + 2 index values since 1870 to the present. The
value corresponding to the 1 + 2 El Niño region (shaded in
red Fig.10 and indicated with a black arrow) during the
2016–2017 was only exceeded by years classified as El
Niño events (i.e. 1878*, 1889, 1897, 1900, 1926, 1983*,
1987, 1997*, and 2016*; asterisks indicate strong El Niño
events). The values corresponding to the summers following
the decay of an El Niño event (Table1) show that the 2017
JFM anomaly in the 1 + 2 El Niño region was significantly
higher (0.94) than average (− 0.15). The situation in the 3.4
region is different as it presents a value (− 0.15) slightly
above the average for 2017 but still negative and not signifi-
cant. Regarding the OLR anomalies in El Niño 1 + 2 region
since 1979 (Fig.11), we observe that the negative anomaly
during 2016–2017 austral summer only has been exceeded
during the El Niño years in 1982–1983, 1997–1998, and
2009–2010.
Results of the SEA (Fig.12) comparing the Niño SST
indices (i.e. Niño 1 + 2 and 3.4) of several austral summers
after El Niño events indicate that for the central Pacific (Niño
3.4 index; Fig.11a) 2016–2017 SST anomaly values were
not significantly far from the average for D (0) (i.e. index
value − 0.47, p-value 0.098), J + 1 (i.e. index value − 0.44,
p-value 0.177) and F + 1 (i.e. index value − 0.07, p-value
0.105). Since M + 1 (Mar 2017) the 3.4 SST anomaly was
statistically significantly high but below of 1992–1993 aus-
tral summer. In the 1 + 2 SST index case, (Fig.11b) results
indicate statistically significant high values during the
2016–2017 austral summer for all the months of the period.
Figure13 shows differences of the spatial distribution
of the SST anomaly along the equatorial Pacific between
the 2016–2017 austral summer and its counterparts in
1878–1879, 1983–1984 and 1998–1999. During the CEN-
2017, the west-central Pacific (Niño 4 and 3.4 regions) in
fact exhibited the lowest SST anomaly during December-
March with the only negative values of the period. This cool-
ing is coherent as such but still far from the negative values
observed in 1878–1879, 1983–1984 and 1998–1999. The
Pacific Niño 3 and 3.4 regions showed neutral to positive
anomalies, which were again significantly higher in 2017
than in 1878–1879, 1983–1984 and 1998–1999 when the
anomalies were clearly negatives. The east Pacific (Niño
1 + 2 region) displays by far the largest anomalies with maxi-
mum value of 2.1°C during the CEN-2017, very different
from the very negative values, which remained in this region
for the other years.
4 Discussion andconclusions
During the 2016–2017 austral summer, intense rainfalls
were recorded over large parts of Peru. The accumulated
measured precipitation exceeded all summer values recorded
since 1982, and led to catastrophic damage to housing and
infrastructure, affecting more than 660,000 people, and leav-
ing more than 100 deaths (INDECI 2017). Furthermore, the
spatial and temporal synchronicity of these extraordinary
rainfall values observed during the DJFM 2016–2017 in
various and very diverse geographic settings across Peru
can be described as unusual.
Our observations of the climatic background regarding
the large-scale oceanic and atmospheric setting of this cos-
tal event agree with Garreaud (2018), who pointed to the
role of extratropical climatic forcing. While it is true that
between January and March 2017, positive anomalies of the
sea surface height were detected along the Peruvian coast
(these have been related to warming Kelvin waves activity
impacting the north Peruvian coast), these waves contributed
only partially to the SST warming in this region (ENFEN
2017a, b). In fact, time series (not shown) of subsurface sea
temperatures and dynamic height of the sea surface obtained
from the Tropical Atmosphere Ocean (TAO) array of buoys
moored in the tropical Pacific do not provide clear evidence
of Kelvin activity, suggesting they were not the primary
mechanism associated with the strong coastal warming in
2017.
At the beginning of the 2016–2017 austral summer, the
strongest rainfall totals were found over some areas of the
Central Andes, in the Huánuco region. Even though seasonal
rainfalls over the Peruvian Central Andes (and Peru in gen-
eral) occur during the summer (IGP 2005; Sulca etal. 2016),
the accumulated rainfall in January was the highest since
1982 over many of the Central Andes regions (i.e. San Mar-
tin, Huánuco, Pasto and Junín) as well as along the South
coast (i.e. Arequipa) and over the Amazonian lowlands (i.e.
south-west Loreto, Amazonas and Ucayali). According to
our synoptic analysis, the long-lasting, upper-level anticy-
clone observed adjacent to the Chilean coast during January
(Fig.4a) led to an intensification of mid-upper level sub-
tropical easterly winds (Fig.4b) favoring a moisture flux
from the Amazon basin and thus intense precipitation in
the Central Andes. These observations agree with Garreaud
(2000) and Sulca etal. (2016) who describe wet events over
the Central Andes in relation with upper-level geopoten-
tial anomalies linked to the equatorward propagation of
mid-latitude wave trains. This is also in line with Garreaud
(1999, 2018), who found that the propagation of mid-latitude

C.Rodríguez-Morata et al.
1 3

The anomalous 2017 coastal El Niño event inPeru
1 3
Rossby wave trains drive rainfall variability on the Altiplano
on intra-seasonal time scales.
During February and March 2017, the extreme rainfall
values moved to the Peruvian north and central coast areas
as well as to the Central Andes with cumulative March pre-
cipitation exceeding any other value for the same period
since 1982 in many regions (Figs.2, 3). For this period,
OLR anomalies in the 1 + 2 El Niño region (Figs.6, 11) were
consistent with the exceptionally high values of SST in the
region (Figs.7, 8, 10, 13d), reaching levels that are normally
indicative of the development of an El Niño event (Rasmus-
son and Carpenter 1982; Trenberth and Stepaniak 2001).
Furthermore, Fig.4b shows that the upper-level subtropical
easterlies were reinforced, which, according to Garreaud
(2018), could have modified the subtropical circulation in
the lower troposphere and in turn may have directly affected
SST in the eastern equatorial Pacific (Fig.5). In fact, the
continued strength of the easterly trades in the equatorial
band is supported by the low-level equatorward flow along
the Pacific coast of South America, generated by descent
when the subtropical westerlies meet the Andes (Rodwell
and Hoskins 2001). If the subtropical westerlies are weak
or reversed, this equatorward flow does not provide the
mass continuity to the SE trade winds, thus hampering the
upwelling along the coast and leading to surface warming
in the eastern Pacific.
We also point to the potential role of SST in the South
Atlantic and the SACZ during the CEN-2017 event. As
observed in Fig.8, an intense positive anomaly is observed
in the South Atlantic centered at 40°S, 40°W, coincident
with a SST anomaly in the equatorial Pacific, OLR nega-
tive anomaly over the SACZ (Fig.6), and heavy rainfalls
over the Peruvian central Andes (Figs.2, 3). This situation
is coherent with findings of Rodrigues-Chaves and Nobre
(2004) and Carvalho etal. (2004), who stated that positive
correlation exists between SST over the South Atlantic (from
the 40°S–0°) and the SACZ. Furthermore, our observations
are in line with Lavado-Casimiro etal. (2013) who found
that higher values in the southern tropical Atlantic (0–20°S,
30°W–10°E) than in the north tropical Atlantic (5–20°N,
60–30°W) favor precipitation in the Peruvian Amazon-
Andes. On the other hand, the intensification of the con-
vective activity in the SACZ in January (Fig.6b) could be
related to the rainfall in the Central Andes by controlling
the position of the Bolivian High (Lenters and Cook 1999).
Fig. 9 Graphs showing the dominance of the synoptic patterns during
the austral summers since 1979. The ONI classification of each aus-
tral summer is represented by the colored squares and the abbrevia-
tions correspond to: VSEN Very strong El Niño; SEN Strong El Niño;
MEN Moderate El Niño; WEN Weak El Niño; SLN Strong La Niña;
MLN Moderate La Niña; WLN Weak La Niña
◂
Fig. 10 Niño 3.4 and 1 + 2 indices since 1870. The black arrow indicates the year 2017

C.Rodríguez-Morata et al.
1 3
Finally, the southward shift of the SACZ observed in Febru-
ary (Fig.6c) could be related with the enhanced tropical con-
vection over the central and eastern Pacific (Nogues-Paegle
and Mo 1997).
Striking rainfall totals and high rank percentiles in the
northern interior lowlands of Piura and Lambayeque as well
as in southern region up to La Libertad during March 2017
are consistent with studies stating that inter-annual rainfall
variability in these regions cannot be explained by El Niño
activity alone (Lagos etal. 2008; Lavado-Casimiro and
Espinoza 2014; Rau etal. 2017). For instance, Rau etal.
(2017) state that inter-annual variations of rainfall do not
necessarily correspond to strong El Niño years, and that an
important part of rainfall variability in the region may cor-
respond to local convective events associated with coastal
warm oceanic conditions related mainly to Kelvin waves
and the Madden and Julian Oscillation (MJO) (Bourrel etal.
2015). By contrast, inter-annual rainfall variability in the
northern highlands of Cajamarca was not clearly related with
SST in the east Pacific, but negatively correlated with SST in
the central Pacific (Bourrel etal. 2015; Rau etal. 2017). This
is consistent with the negative anomalies that we observe in
the west-central Pacific (Niño 4 and 3.4 regions in Fig.8).
Regarding rainfall activity in the Peruvian Amazonian
lowlands, rainfall totals reached their long-term maxima
mostly during January–February 2017 as well. We hypoth-
esize that this anomaly can be related with the positive SST
Table 1 JFM SST anomalies in the 1 + 2 and 3.4 El Niño regions cor-
responding to years following the decay of an El Niño event since
1870 and indicating either neutral or weak La Niña conditions
The 2017 year was significantly above the average in the 1 + 2 El
Niño region, indicating a clear El Niño situation. The situation in the
3.4 region reflects temperatures slightly under the average but they
are not significant. Note that the considered cases are those where the
SST values in the Niño region decay toward zero during the remain-
ing months of year + 1 of the El Niño event. Some Niño’s do not fol-
low that decay mode pattern, and so are not included here. Statistical
significance computed using a t-test for one-sided samples at a sig-
nificance level of 0.05 (below)
* Significant values
JFM El Niño 1 + 2 SST
anomaly
JFM El Niño
3.4 SST
anomaly
1879 − 0.36*− 0.32*
1890 − 0.42*− 1.92*
1898 − 0.51*− 0.6*
1904 − 0.44*− 0.8*
1927 − 0.013*0.05
1967 − 0.63*− 0.53*
1984 − 0.49*− 0.6*
1999 0.04*− 1.29*
2017 0.94*− 0.15
1870–2017 average − 0.15 − 0.073
Fig. 11 Time series of OLR anomaly in El Niño 1 + 2 region since 1979

The anomalous 2017 coastal El Niño event inPeru
1 3
anomalies in the South Tropical Atlantic observed in Fig.8c.
This hypothesis is consistent with Espinoza etal. (2009) or
Lavado-Casimiro etal. (2013) who concluded that intense
rainfalls over Peru are likely related with the positive SST
anomalies observed in the South Tropical Atlantic. Yet, fur-
ther research is needed to understand the role of the tropical
Atlantic during the CEN-2017 event.
The warming in the equatorial Pacific was not a full
basin event and the SOI index, with values close to zero
(not shown) indicates that the CEN-2017 event was not a
coupled ocean–atmosphere phenomenon. As such, it dif-
fers from a typical El Niño and defines it like an ocean-
coastal event (Takahashi and Martinez 2017). Consider-
ing the SST values in the eastern Pacific, this 2016–2017
summer also differs from its counterparts after El Niño
events since 1870 (Fig.10, 11, 12 and 13). Differences
with previous years are also noticed in the transition
between synoptic states throughout the summer (Fig.9).
Even if the synoptic states related with the 2016–2017
austral summer are commonly linked to the large-scale
summer dynamic over South America and are not exclu-
sive of the 2016–2017 summer, what is different this time
is the low dispersion and relatively low number of synoptic
states compare with previous years (except for the strong
El Niño and La Niña years). Thus, dominance of these
synoptic states was higher, favoring the climatic processes
that trigger the overwarming in the equatorial Pacific.
Consequently, the usual precipitation pattern after a strong
El Niño event—with slightly above-normal rainfalls in
the south of Peru and nearly normal to dry conditions in
the north—was altered after the end of the 2015–2016 El
Niño, as it did not result in the usual cooling of the central
equatorial Pacific (Lavado-Casimiro and Espinoza 2014).
We conclude that the coexistent SST anomalies in the
equatorial Pacific (and presumably also in the Tropical
Atlantic) have clearly favoured the development of the
extreme “Coastal El Niño” event and concomitant high
magnitude of precipitation over Peru in DJFM 2016–2017.
The approach shown in this paper, together with its inter-
pretation within a climatic context, demonstrates that the
DJFM 2016–2017 rainfall pattern over Peru was highly
anomalous, both in terms of its magnitude and timing after
a strong El Niño event. The rather severe consequences
and important death tolls cannot be explained by the
anomalous weather phenomenon only, but are also due to
the absence of El Niño early warnings, which in turn were
largely the result of the abrupt and unexpected warming
above the average in the Niño region 1 + 2. This suggests
that disaster management strategies in Peru should main-
tain the same level of vigilance across time, regardless of
the ENSO phase, taking into account the whole variability
of South American summers.
Fig. 12 Superposed epoch analysis for the anomalies of the 3.4 (a)
and 1 + 2 (b) El Niño indices during the year after strong El Niño
events since 1951. Following the nomenclature system of Rasmusson
and Carpenter (1982), analysis covers the months of December after
an El Niño event (D0) to June of the following year (J + 1) and they
are centered in March 2017 (M + 1). Shades years are statically sig-
nificant at 0.05 level

C.Rodríguez-Morata et al.
1 3
Acknowledgements The authors wish to thank Dr. Rene Garreaud and
the two anonymous reviewers whose comments and suggestions con-
siderably helped to improve the manuscript.
References
Ashok K, Behera SK, Rao SA, Weng H, Yamagata T (2007) El
Niño Modoki and its possible teleconnections. J Geophys Res
112:C11007. https ://doi.org/10.1029/2006J C0037 98
Bourrel L, Rau P, Dewitte B, Labat D, Lavado W, Coutaud A, Vera
A, Alvarado A, Ordoñez J (2015) Low-frequencymodulation and
trend of the relationship between ENSO and precipitation along
the northern to centre Peruvian Pacific coast. Hydrol Process
29(6):1252–1266
Capotondi A etal (2015) Understanding ENSO diversity. Bull Am
Meteorol Soc 96:921–938
Carvalho LMV, Jones C, Liebmann B (2004) The South Atlantic Con-
vergence Zone: intensity, form, persistence, and relationships with
intraseasonal to interannual activity and extreme rainfall. J Clim
17: 88–108. https ://doi.org/10.1175/1520-0442(2004)017%3C008
8:TSACZ I%3E2.0.CO;2
Cassano EN, Lynch AH, Cassano JJ, Koslow MR (2006a) Classifica-
tion of synoptic patterns in the western Arctic associated with
extreme events at Barrow, Alaska, USA. Clim Res 30(2):83–97
Cassano JJ, Uotila O, Lynch A (2006b) Changes in synoptic weather
patterns in the polar regions in the 20th and 21st centuries, part
1: Arctic. Int J Climatol 26:1027–1049
Cassano JJ, Uotilla P, Lynch AH, Cassano EN (2007) Predicted
changes in synoptic forcing of net precipitation in large Arctic
river basins during the 21st century. J Geophys Res. https ://doi.
org/10.1029/2006J G0003 32
CPC (2018a) ENSO Diagnostic Discussion Archive.http://www.cpc.
ncep.noaa.gov/produ cts/exper t_asses sment /ENSO_DD_archi
ve.shtml . Accessed 25 May 2018
CPC (2018b) http://www.cpc.ncep.noaa.gov/produ cts/analy sis_monit
oring /ensos tuff/ONI_v5.php. Accessed 25 May 2018
Dee DP etal (2011) The ERA-Interim reanalysis: configuration and
performance of the data assimilation system. Q J R Meteor Soc
137:553–597
Dommenget D, Bayr T, Frauen C (2013) Analysis of the non-linearity
in the pattern and time evolution of El Niño Southern Oscilla-
tion. Clim Dyn 40:2825–2847. https ://doi.org/10.1007/s0038
2-012-1475-0
ECMWF (2016) ERA-Interim daily data. http://apps.ecmwf .int/datas
ets/data/inter im-full-daily /levty pe=sfc/. Accessed 10 Jan 2016
ENFEN (2017a) Comunicados oficiales. https ://www.dhn.mil.pe/
comun icado _ofici al_enfen . Accessed 13 Sep 2017
ENFEN (2017b) Informe Técnico Extraordinario No 001-2017/ENFEN
EL NIÑO COSTERO 2017. Resumen ejecutivo. http://www.imarp
e.pe/imarp e/archi vos/infor mes/imarp e_inftc o_infor me__tecni
co_extra ordin ario_001_2017.pdf. Accessed 18 Sep 2018
Espinoza JC, Ronchail J, Guyot JL, Cochonneau G, Naziano F, Lavado
W, Oliveira ED, Pombosa R, Vauchel P (2009) Spatio-temporal
rainfall variability in the Amazon basin countries (Brazil, Peru,
Bolivia, Colombia, and Ecuador). Int J Clim 29:1574–1594
Fig. 13 Sea surface temperature (SST) anomaly maps for December
to March for the first austral summer after the strong El Niño events
in 1877–1878, 1982–1983, 1997–1998 and 2015–2016. We use Had-
ISST1 data provided by the NOAA/OAR/ESRL PSD (http://www.
esrl.noaa.gov/psd/)

The anomalous 2017 coastal El Niño event inPeru
1 3
Espinoza JC, Lengaigne M, Ronchail J, Anicot S (2012) Large-scale
circulation patterns and related rainfall in the Amazon Basin: a
neuronal networks approach. Clim Dyn 38(1–2):121–140. https
://doi.org/10.1007/s0038 2-011-1010-8
Espinoza JC, Ronchail J, Lengaigne M, Quispe N, Silva Y, Bettolli M,
Avalos G, Llacza A (2013) Revisiting wintertime cold air intru-
sions at the east of the Andes: propagating features from subtropi-
cal Argentina to Peruvian Amazon and relationship with large-
scale circulation patterns. Clim Dyn 41(7–8):1983–2002. https ://
doi.org/10.1007/s0038 2-012-1639-y
Few R (2007) Health and climate change hazards: framing social
research on vulnerability, response and adaptation. Glob Envi-
ron Change 17:281–295
Garreaud RD (1999) Multiscale analysis of the summertime precipi-
tation over the central Andes. Mon Weather Rev 127:901–921.
https ://doi.org/10.1175/1520-0493(1999)127,0901:MAOTS
P.2.0.CO;2
Garreaud RD (2000) Intraseasonal variability of moisture and rain-
fall over the South American Altiplano. Mon Weather Rev
128:3337–3346
Garreaud RD (2018) A plausible atmospheric trigger for the 2017
coastal El Niño. Int J Climatol 38(Suppl. 1):e1296 e1302
Garreaud RD, Vuille M, Compagnucci R, Marengo J (2009) Present-
day South American climate. Paleogeogr Palaeoclimatol Pal-
aeoecol 281:180–195
Hales S, Edwards S, Kovats RS (2003) Impacts on health of climate
extremes. In: McMichael A, Campbell-Lendrum D, Corvala´n
C, Ebi K, Githeko A, Scheraga J, Woodward A (eds) Climate
change and human health: risks and response. World Health
Organization, Geneva, pp79–102
Haynes W (2013) Student’s t-test. In: Dubitzky W, Wolkenhauer
O, Cho KH, Yokota H (eds) Encyclopedia of systems biology.
Springer, New York. https ://doi.org/10.1007/978-1-4419-9863-7
Hewitson B, Crane R (2002) Self-organizing maps: applications to
synoptic climatology. Clim Res 22:13–26
Hong Y, Hsu K, Sorooshian S, Gao X (2005) Self-organizing nonlin-
ear output (SONO): a neural network suitable for cloud patch-
based rainfall estimation at small scales. Water Resour Res.
https ://doi.org/10.1029/2004W R0031 42
Huffman G (2015) GPM IMERG final precipitation L3 1 month
0.1 degree × 0.1 degree V04, Greenbelt, MD, Goddard Earth
Sciences Data and Information Services Center (GES DISC).
Accessed 15 Feb 2018. https ://doi.org/10.5067/GPM/IMERG
/3B-MONTH /04
Huffman GJ, Bolvin DT, Braithwaite D, Hsu K, Joyce R, Kidd C, Nel-
kin EJ, Sorooshian S, Tan J, Xie P (2017) Algorithm theoretical
basis document (ATBD) version 5.1 NASA global precipitation
measurement (GPM) integrated multi-satellite retrievals for GPM
(IMERG).https ://pmm.nasa.gov/sites /def au lt/files /docum ent_files
/IMERG _ATBD_V5.1.pdf. Accessed 15 Feb 2018
IGP (2005) Atlas climatológico de precipitaciones y temperaturas en
la Cuenca del Río Mantaro. CONAM, Lima, Perú. http://www.
met.igp.gob.pe/publi cacio nes/2000_2007/Atlas _Clima tico.pdf
INDECI (2017) Resúmen ejecutivo—Temporada de lluvias Diciem-
bre 2016—Marzo 2017. http://www.indec i.gob.pe/objet os/alert a/
MjYxN g==//20170 33016 5025.pdf
Johnson NC, Feldstein SB, Tremblay B (2008) The continuum of
Northern Hemisphere teleconnection patterns and a description
of the NAO shift with the use of self-organizing maps. J Clim
21:6354–6371
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L,
Iredell M, Saha S, White G, Woollen J (1996) The NCEP/NCAR
40-year reanalysis project. Bull Am Meteorol Soc 77:437–471
Kao HY, Yu JY (2009) Contrasting Eastern-Pacific and Central-Pacific
types of ENSO. J Clim 22:615–632
Kohonen T (2001) Self-organizing maps. Springer, New York
Kug JS, Jin FF, An SI (2009) Two types of El Niño events: cold tongue
El Niño and warm pool El Niño. J Clim 22:1499–1515
L’Heureux M etal (2017) Observing and predicting the 2015/16
El Niño. Bull Amer Meteor Soc. https ://doi.org/10.1175/
BAMS-D-16-0009.1
Lagos P, Silva Y, Nickl E, Mosquera K (2008) El Niño—related pre-
cipitation variability in Peru. Adv Geosci 14:231–237
Larkin NK, Harrison DE (2005) On the definition of El Niño and asso-
ciated seasonal average US weather anomalies. Geophys Res Lett
32:L13705. https ://doi.org/10.1029/2005G L0227 38
Lavado-Casimiro W, Espinoza J (2014) Impactos de El Niño y La
Niña en las lluvias del Perú (1965–2007). Rev Bras Meteorol
29:171–182
Lavado-Casimiro WS, Labat D, Ronchail J, Espinoza JC, Guyot JL
(2013) Trends in rainfall and temperature in the Peruvian Ama-
zon–Andes basin over the last 40 years (1965–2007). Hydrol
Process. https ://doi.org/10.1002/hyp.9418
Lenters JD, Cook KH (1997) On the origin of the Bolivian High and
related circulation features of the South American climate. J
Atmos Sci 54:656–677
Lenters JD, Cook KH (1999) Summertime precipitation variability
over South America: role of the large-scale circulation. Mon
Weather Rev 127:409–431
Lynch A, Uotila P, Cassano JJ (2006) Changes in synoptic weather
patterns in the polar regions in the twentieth and twenty-first
centuries (Part 2). Antarctic. Int J Climatol 26:1181–1199
Morata A, Martin ML, Luna MY, Valero F (2006) Self-similarity
patterns of precipitation in the Iberian Peninsula. Theor Appl
Climatol 85:41–59
NCAR (2014) Outgoing longwave radiation (OLR): AVHRR. The cli-
mate data guide. Boulder, CO.https ://clima tedat aguid e.ucar .edu/
clima te-data/outgo ing-longw ave-radia tion-olr-avhrr . Accessed
15 Jun 2018
NOAA (2017) ENSO diagnostic discussion (climate prediction
center, Nov–Dec 2017).http://origi n.cpc.ncep.noaa.gov/produ
cts/analy sis_monit oring /enso_disc_jun20 14/. Accessed 25 Feb
2018
Nogues-Paegle J, Mo KC (1997) Alternating wet and dry condi-
tions over South America during summer. Mon Weather Rev
125:279–291
Paccini L, Espinoza JC, Ronchail J, Segura H (2017) Intraseasonal
rainfall variability in the Amazon basin related to large-scale
circulation patterns: a focus on western Amazon-Andes transi-
tion region. Int J Clim 38(5):2386–2399
PAHO (2017) Emergencia por impacto del Fenómeno “El Niño Cos-
tero”—Perú, 2017. Pan American Health Organization, Wash-
ington, DC
Rasmusson EM, Carpenter TH (1982) Variations in tropical sea sur-
face temperature and surface wind fields associated with the
Southern oscillation/El Niño. Mon Weather Rev 110:354–384
Rau P etal (2017) Regionalization of rainfall over the Peruvian
Pacific slope and coast. Int J Clim. https ://doi.org/10.1002/
joc.4693
Rayner NA, Parker DE, Horton EB, Folland CK, Alexander LV, Rowell
DP, Kent EC, Kaplan A (2003) Global analyses of sea surface
temperature, sea ice, and night marine air temperature since the
late nineteenth century. J Geophys Res 108(D14):4407. https ://
doi.org/10.1029/2002J D0026 70
Reusch DB, Alley RB, Hewitson BC (2005a) Relative performance
of self-organizing maps and principal component analysis in pat-
tern extraction from synthetic climatological data. Polar Geogr
29:227–251
Reusch DB, Hewitson BC, Alley RB (2005b) Towards ice core-based
synoptic reconstructions of West Antarctic climate with artificial
neural networks. Int J Climatol 25:581–610

C.Rodríguez-Morata et al.
1 3
Reusch DB, Alley RB, Hewitson BC (2007) North Atlantic climate
variability from a self-organizing map perspective. J Geophys Res.
https ://doi.org/10.1029/2006J D0074 60
Rodrigues-Chaves R, Nobre P (2004) Interactions between sea surface
temperature over the South Atlantic Ocean and the South Atlan-
tic Convergence Zone. Geophys Res Lett 31:L03204. https ://doi.
org/10.1029/2003G L0186 47
Rodriguez-Morata C, Ballesteros-Canovas JA, Rohrer M, Espinoza JC,
Beniston M, Stoffel M (2018) Linking atmospheric patterns with
hydro-geomorphic disasters in Peru. Int J Climatol. https ://doi.
org/10.1002/joc.5507
Rodwell MJ, Hoskins BJ (2001) Subtropical anticyclones and summer
monsoons. J Clim 14:3192–3211
Sanabria J, Bourrel L, Dewitte B, Frappart F, Rau P, Olimpio S, Labat
D (2018) Rainfall along the coast of Peru during strong El Niño
events. J Int Climatol. https ://doi.org/10.1002/joc.5292
Schneider U, Becker A, Finger P, Meyer-Christoffer A, Ziese M
(2015) GPCC monitoring product: near real-time monthly land-
surface precipitation from rain-gauges based on SYNOP and
CLIMAT data. Deutscher Wetterdienst. https ://doi.org/10.5676/
DWD_GPCC/MP_M_V5_100
Schuenemann KC, Cassano JJ, Finnis J (2009) Synoptic forcing of
precipitation over Greenland: climatology for 1961–99. J Hydro-
meteorol 10:60–78
Schwarb M, Acuña D, Konzelmann T, Rohrer M, Salzmann N, Serpa
Lopez B, Silvestre E (2011) A data portal for regional climatic
trend analysis in a Peruvian High Andes region. Adv Sci Res
6:219–226. https ://doi.org/10.5194/asr-6-219-2011
Seefeldt MW, Cassano JJ (2008) An analysis of low-level jets in the
Greater Ross Ice Shelf region based on numerical simulations.
Mon Weather Rev 136:4188–4205
Skific N, Francis JA, Cassano JJ (2009a) Attribution of projected
changes in atmospheric moisture transport in the Arctic: a self-
organizing map perspective. J Clim 22:4135–4153
Skific N, Francis JA, Cassano JJ (2009b) Attribution of seasonal
and regional changes in Arctic moisture convergence. J Clim
22:5115–5134
Sulca J, Vuille M, Silva Y, Takahashi K (2016) Teleconnections
between the Peruvian central Andes and northeast Brazil dur-
ing extreme rainfall events in austral summer. J Hydrometeorol
17:499–515
Sulca J, Takahashi K, Espinoza JC, Vuille M, Lavado-Casimiro W
(2017) Impacts of different ENSO flavors and tropical Pacific con-
vection variability (ITCZ, SPCZ) on austral summer rainfall in
South America, with a focus on Peru. Int J Climatol 38:420–435.
https ://doi.org/10.1002/joc.5185
Takahashi K, Martínez AG (2017) The very strong El Niño in 1925 in
the far-eastern Pacific. Clim Dyn. https ://doi.org/10.1007/s0038
2-017-3702-1
Takahashi K, Montecinos A, Goubanova K, Dewitte B (2011) ENSO
regimes: reinterpreting the canonical and Modoki El Niño. Geo-
phys Res Lett 38:L10704. https ://doi.or g/10.1029/2011G L0473 64
Trenberth K (2016) The climate data guide: Nino SST indices (Nino
1 + 2, 3, 3.4, 4; ONI and TNI). https ://clima tedat aguid e.ucar.
edu/clima te-data/nino-sst-indic es-nino-12-3-34-4-oni-and-tni.
Accessed 22 Jun 2018
Trenberth K, Stepaniak DP (2001) Indices of El Niño evolution. J Clim
14:1697–1701
Uotila P, Lynch AH, Cassano JJ, Cullather RI (2007) Changes in
Antarctic net precipitation in the 21st century based on Intergov-
ernmental Panel on Climate Change (IPCC) model scenarios. J
Geophys Res. https ://doi.org/10.1029/2006J D0074 82
Waring S, Brown B (2005) The threat of communicable diseases fol-
lowing natural disasters: a public health response. Disaster Manag
Response 3:41–47
Wise E, Dannenberg M (2014) Persistence of pressure patterns over
North America and the North Pacific since AD 1500. Nat Com-
mun. https ://doi.org/10.1038/ncomm s5912
Yeh SW, Kug JS, Dewitte B, Kwon MH, Kirtman B, Jin FF (2009)
El Niño in a changing climate. Nature. https ://doi.org/10.1038/
natur e0831 6
Yu JY, Kao HY, Lee T, Kim ST (2011) Subsurface ocean temperature
indices for central-Pacific and eastern-Pacific types of El Niño
and La Niña events. Theor Appl Climatol. https ://doi.org/10.1007/
s0070 4-010-0307-6
Zhang R, Wang Y, Liu W etal (2006) Cloud classification based on
self-organizing feature map and probabilistic neural network. In:
Proceedings of the 6th World Congress on Intelligent Control and
Automation, 21–23 June 2006, Dalian, China, pp41–45