Extreme Drought Events Over the Amazon Basin: the Perspective from the Reconstruction of South American Hydroclimate

water

Article Extreme Drought Events over the Amazon Basin: The Perspective from the Reconstruction of South American Hydroclimate

Beatriz Nunes Garcia, Renata Libonati * and Ana M. B. Nunes

Departamento de Meteorologia, Instituto de Geociências, Universidade Federal do Rio de Janeiro, Rio de Janeiro 21941-916, Brazil; [email protected] (B.N.G.); [email protected] (A.M.B.N.) * Correspondence: [email protected]; Tel.: +55-21-3938-9470

Received: 27 September 2018; Accepted: 5 November 2018; Published: 7 November 2018

Abstract: The Amazon basin has experienced severe drought events for centuries, mainly associated with climate variability connected to tropical North Atlantic and Pacific sea surface temperature anomalous warming. Recently, these events are becoming more frequent, more intense and widespread. Because of the Amazon droughts environmental and socioeconomic impacts, there is an increased demand for understanding the characteristics of such extreme events in the region. In that regard, regional models instead of the general circulation models provide a promising strategy to generate more detailed climate information of extreme events, seeking better representation of physical processes. Due to uneven spatial distribution and gaps found in station data in tropical South America, and the need of more refined climate assessment in those regions, satellite-enhanced regional downscaling for applied studies (SRDAS) is used in the reconstruction of South American hydroclimate, with hourly to monthly outputs from January 1998. Accordingly, this research focuses on the analyses of recent extreme drought events in the years of 2005 and 2010 in the Amazon Basin, using the SRDAS monthly means of near-surface temperature and relative humidity, precipitation and vertically integrated soil moisture fields. Results from this analysis corroborate spatial and temporal patterns found in previous studies on extreme drought events in the region, displaying the distinctive features of the 2005 and 2010 drought events.

Keywords: extreme droughts; Amazon; downscaling; precipitation assimilation; hydroclimatology

1. Introduction Drought is a widespread phenomenon and affects most ecosystems, even those considered as humid biomes with high precipitation rates such as Amazonia [1,2]. Several studies have shown that anthropogenic activities associated with the conversion of natural vegetation into pasture and agriculture, fire practices, and climate variability may interact and enhance the occurrence of hydroclimate extremes such as droughts in the Amazon Basin (henceforward AB) [3–6]. Recent studies reveal that the Amazon rainforest is more vulnerable to future droughts than other tropical forests [7]. Accordingly, there is an increasing concern about loss of forest and ecosystem services together with climate impacts due to extreme droughts due to the large geographic extent of the AB, together with its influence in global biogeochemical cycles and the presence of high levels of biodiversity in the region. Several studies indicate that the region has been suffering severe drought since the end of the last century, as in 1997/1998, 2005, 2010 and 2015 [8–11]. The intensity and frequency of these extreme drought episodes in the AB during the last years, approximately one episode every five years with a significant increase in the coverage area, is remarkable [8]. However, each of these extreme episodes was dominated by different circulation regimes, and consequently, varied markedly in terms of

Water 2018, 10, 1594; doi:10.3390/w10111594 www.mdpi.com/journal/water Water 2018, 10, 1594 2 of 16 temporal intensity and spatial patterns [11–14]. Drought events in the Amazon basin are associated with teleconnections due to El-Niño Southern Oscillation and/or tropical North Atlantic sea surface temperature (SST) anomalous warming [15–17]. 1998’s drought was related to both teleconnections but especially to the strong El Niño of 1997–1998 and affected northern and eastern areas of AB [18]. Meanwhile, 2005’s drought was related to the elevated warming in the tropical North Atlantic SST, and its epicenter was in the west of the basin; 2010’s event was related to both teleconnections and occurred in a larger area of the basin having three different epicenters: west, northwest, and southeast areas [19,20]. 2015’s drought was related to the El-Niño Southern Oscillation and had precipitation deﬁcit through the whole basin but particularly in the eastern area [8,10]. Traditional methods of drought assessment and monitoring depend heavily on rainfall data as recorded in meteorological and hydrological networks. However, the availability of reliable modeling datasets covering wide regions over long periods of time has progressively strengthened the role of climate models outputs in environmental studies, in particular in those related to drought episodes [21]. Both general circulation and regional models have improved in recent years in terms of depiction of processes and phenomena to address impacts of global climate change [22]. Part of this breakthrough comes from increased spatial resolution and part comes from the inclusion of climate controls from new system components and their interaction. The use of regional downscaling of general circulation models provides a technique to generate more reﬁned climatic information on extreme events through better representation of physical processes [23,24]. In such context, the Satellite-Enhanced Regional Downscaling for Applied Studies (henceforth SRDAS) product has been used in the analysis of South American hydroclimate, with hourly to monthly outputs from January 1998 to the near present, with continuous assimilation of satellite-based precipitation estimates [25]. In the current study, SRDAS is used in the analyses of the drought episodes of 2005 and 2010 in the Amazon basin and corresponds to a 16-year numerical integration of a regional modeling system driven by the six hourly outputs a global reanalysis. Most of the features of larger synoptic to planetary waves from the global reanalysis are maintained in SRDAS through scale-selective bias correction, which is comparable to the spectral nudging technique from [26]. Accordingly, recent drought events (i.e., 2005 and 2010) provide a unique opportunity for assessing how SRDAS is able to reproduce the shifts in the amount and distribution of Amazonian rainfall and other drought-related variables, in turn quantifying the sensitivity of this regional reconstruction in replicate changes in oceanic modes such as ENSO and the Atlantic Multidecadal Oscillation (AMO) under the current trend of climate warming.

2. Materials and Methods

2.1. Data and Products

2.1.1. Station Data Observational monthly precipitation and temperature ﬁelds for this study are from National Institute of Meteorology (Instituto Nacional de Meteorologia—INMET) and Airspace Control Institute (Instituto de Controle do Espaço Aéreo—ICEA), both in Brazil. Table1 describes the meteorological stations used in the analysis, located near the epicenters of each one of the droughts in three different states in Brazil, namely Coari in Amazonas (AM), used in the characterization of the 2005 drought, and Cruzeiro do Sul, in Acre (AC), located in the western epicenter of the 2010 drought, within the time interval of 1981–2010. Both INMET’s stations, the ICEA station, Vilhena, in Rondônia (RO), was selected for the analysis of the southeastern epicenter of the 2010 drought. These meteorological station datasets have different available periods, for instance Coari and Cruzeiro do Sul are available from 1961 to the beginning of 2016, and Vilhena only from 1971 to October of 2015. In order to allow a comparison between them, we have selected the period from 1981 to 2010, which is deﬁned by the World Meteorological Organization as the base period of the current climate. The precipitation and temperature records were analyzed and the percentage of missing values was computed along the Water 2018, 10, 1594 3 of 16 entire time series. The obtained percentage of missing values was up to 7.30% for all stations, not hindering the climatological anomaly analysis.

Table 1. Station description and representativeness.

SID 1 Name (State Code) Altitude (m) Longitude Latitude Drought Epicenter 81770 Coari (Amazonas) 34 63.15◦ W 4.10◦ S 2005 81881 Cruzeiro do Sul (Acre) 220 72.68◦ W 7.61◦ S 2010 83208 Vilhena (Rondonia) 605 60.06◦ W 12.42◦ S 2010 1 Station Identiﬁcation Code (SID).

2.1.2. Regional Modeling System and Satellite-Based Products SRDAS is available over South American domain (93.6◦ W–25.8◦ W and 20.4◦ S–42.6◦ S), with atmospheric and surface initial conditions on 1 January 1998 at 0000 UTC, hourly outputs from 1000-hPa to 10-hPa levels and with horizontal resolution of approximately 25 km. The regional spectral model (RSM) [27] is the atmospheric model used in SRDAS, with the new version of scale-selective bias correction (NSSBC) already implemented in RSM [28], and driven by the six hourly outputs from the National Centers for Environmental Prediction–Department Of Energy (NCEP–DOE) R2 global reanalysis [29], except for the SST ﬁelds, which are updated daily from the product Optimum 1 Interpolation SST at 4 -degree resolution, version 2, based on satellite SST from only the advanced, very high-resolution radiometer (AVHRR) [30]. Most of the features of larger synoptic to planetary waves from NCEP–DOE R2 are maintained in SRDAS through NSSBC, which is mainly applied to the rotational part of the wind ﬁeld. The continuous assimilation of the three-hourly satellite-based precipitation estimates, obtained from the Climate Prediction Center Morphing technique (CMORPH) [31], is also included in RSM [25] to primary correct the representation of the deep convection in SRDAS, however atmosphere–land surface interactions are also expected to be enhanced. Ultimately, both NSSBC and the precipitation assimilation scheme (henceforward PA) are employed to improve the RSM long-term integrations. In SRDAS, the RSM version is that coupled to the four-layer Noah Land Surface Model (Noah LSM) [32]; and NCEP–R2 provides the atmospheric and land surface initial conditions to RSM and Noah LSM, respectively.

2.2. Methodology The study region corresponds to the Amazon Basin (Figure1) which has an area of about 6.3 million km2—approximately 5 million km2 in the Brazilian territory and the remainder divided among other South American countries, such as Peru (17%), Bolivia (11%), Colombia (5.8%), Ecuador (2.2%), Venezuela (0.7%), and Guyana (0.2%). In this study, monthly means of precipitation, 2-m above the ground air temperature and relative humidity, and 4-layer integrated soil moisture from January 1998 to December 2013 are computed over the region of 80◦ W–51◦ W and 6◦ N–21◦ S, using the SRDAS hourly outputs. The SRDAS monthly means are averaged over three distinctive regions located north (1), west (2) and southeast (3) in the Amazon basin, as displayed in the study area (Figure1), which also exhibits the elevation in meters of the entire study domain, as well as the station sites. The choice of each subregion from Figure1 was made based on each drought epicenters as described in previous studies [8,19,33]. Monthly values were used to calculate the long-term means and standard deviations, which were then used to obtain standardized anomalies for each month. To smooth short-term ﬂuctuations out, the ﬁve-month running mean is applied in the time series of standardized anomalies calculated over each of the three regions. Comparisons with selected station data located in each of the three regions are performed to establish SRDAS reliability to depict the drought-patterns along, taking into special attention the droughts of 2005 and 2010, with the spatial distribution of the anomalies computed from the monthly means over the Amazon basin. We only analyze the months with the lowest anomalies in precipitation, soil Water 2017, 9, x FOR PEER REVIEW 4 of 16 drought-patterns along, taking into special attention the droughts of 2005 and 2010, with the spatial Water 2018, 10, 1594 4 of 16 distribution of the anomalies computed from the monthly means over the Amazon basin. We only analyze the months with the lowest anomalies in precipitation, soil moisture, and relative humidity, andmoisture, with the and highest relative temperature humidity, and anomal withies the, considering highest temperature the lag between anomalies, the considering responses of the the lag atmospherebetween the and responses the surface of the conditions. atmosphere and the surface conditions.

Figure 1. Study area elevation (m) from the Satellite-Enhanced Regional Downscaling for Applied FigureStudies 1. (henceforthStudy area SRDAS).elevation Delimited(m) from withthe Satellite boxes (dashed-Enhanced lines) Regional are regions Downscaling north (1), for west Applied (2) and Studiessoutheast (hencefort (3) in theh SRDAS) Amazon. Delimited basin. Station with sites boxes are (dashed marked line withs) are “X”, regions namely: north Coari (1), (AM), west Cruzeiro(2) and southeastdo Sul (AC), (3) in and the Vilhena Amazon (RO). basin. Station sites are marked with “X”, namely: Coari (AM), Cruzeiro 3. Resultsdo Sul (AC), and Vilhena (RO).

3. ResultsHere, the 2005 and 2010 droughts in the Amazon basin are characterized using the interannual and intra-annual variability of precipitation and near-surface temperature from the three available Here, the 2005 and 2010 droughts in the Amazon basin are characterized using the interannual meteorological stations, namely, Coari, Cruzeiro do Sul, and Vilhena; and the SRDAS outputs for the and intra-annual variability of precipitation and near-surface temperature from the three available regions 1, 2, and 3, including relative humidity and soil moisture. The three stations are inside the meteorological stations, namely, Coari, Cruzeiro do Sul, and Vilhena; and the SRDAS outputs for the drought epicenters of the 2005 (Coari) and 2010 (Cruzeiro do Sul and Vilhena), as reported in previous regions 1, 2, and 3, including relative humidity and soil moisture. The three stations are inside the studies [19,20,33,34]. drought epicenters of the 2005 (Coari) and 2010 (Cruzeiro do Sul and Vilhena), as reported in Results from in situ measurements pointed to the existence of negative precipitation and positive previous studies [19,20,33,34]. temperature standardized anomalies from early 2005 and 2010, intensified during the winter (Figure2). Results from in situ measurements pointed to the existence of negative precipitation and Coari (green curve) had positive near-surface temperature anomaly throughout the year of 2005, positive temperature standardized anomalies from early 2005 and 2010, intensified during the reaching its maximum in June whereas its minimum in precipitation anomaly was in July. Precipitation winter (Figure 2). Coari (green curve) had positive near-surface temperature anomaly throughout (temperature) anomalies greatly oscillated during the year 2010 in Cruzeiro do Sul (blue curve), mostly the year of 2005, reaching its maximum in June whereas its minimum in precipitation anomaly was with negative (positive) values, with June (September) being the most extreme month associated in July. Precipitation (temperature) anomalies greatly oscillated during the year 2010 in Cruzeiro do with drought conditions. The lowest oscillation in the precipitation anomaly was registered during Sul (blue curve), mostly with negative (positive) values, with June (September) being the most the drought season in Vilhena (red curve), and September was the month showing the highest air extreme month associated with drought conditions. The lowest oscillation in the precipitation temperature anomaly value, as reported by [20] when analyzing the 2010 drought epicenters using anomaly was registered during the drought season in Vilhena (red curve), and September was the remote-sensed land surface temperatures. Although the two methods to determine the temperature month showing the highest air temperature anomaly value, as reported by [20] when analyzing the fields are different, there is still strong relationship between them [35]. The SRDAS product from the 2010 drought epicenters using remote-sensed land surface temperatures. Although the two methods 1999–2013 base period shows the same precipitation and temperature interannual variability for each to determine the temperature fields are different, there is still strong relationship between them [35]. drought epicenter (Figure3). The SRDAS product from the 1999–2013 base period shows the same precipitation and temperature In situ measurements of relative humidity and soil moisture are not available for none of the three interannual variability for each drought epicenter (Figure 3). ground stations considered here. Nevertheless, as expected, SRDAS shows negative relative humidity In situ measurements of relative humidity and soil moisture are not available for none of the and soil moisture anomalies during the two extreme drought events. Dryness of the soil is usually a three ground stations considered here. Nevertheless, as expected, SRDAS shows negative relative result of an extended period of severe rainfall deficit and prolonged low levels of relative humidity, and high incoming solar radiation [36,37]. Strong radiative anomalies due to clear sky conditions were Water 2017, 9, x FOR PEER REVIEW 5 of 16 humidity and soil moisture anomalies during the two extreme drought events. Dryness of the soil is Waterusually 2017 a, 9 result, x FOR ofPEER an REVIEWextended period of severe rainfall deficit and prolonged low levels of relative5 of 16 humidity, and high incoming solar radiation [36,37]. Strong radiative anomalies due to clear sky humidityconditions and were soil observedmoisture inanomalies the region during during the two the 2005extreme and drought 2010 drought events. eventsDryness [8] of, leadingthe soil is to usuallyextreme aly result dry soilof an conditions extended period such as of thesevere ones rainfall observed deficit in and SRDAS prolonged during low the levels 2005 of and relative 2010 humidity,droughts (Figureand high 3). incoming solar radiation [36,37]. Strong radiative anomalies due to clear sky conditionsWater 2018 were, 10, 1594 observed in the region during the 2005 and 2010 drought events [8], leading5 of 16 to extremely dry soil conditions such as the ones observed in SRDAS during the 2005 and 2010 droughtsobserved (Figure in the 3). region during the 2005 and 2010 drought events [8], leading to extremely dry soil conditions such as the ones observed in SRDAS during the 2005 and 2010 droughts (Figure3).

(a) (b)

Figure 2. Standardized anomaly of observed : (a) precipitation and (b) temperature at the following (a) (b) station sites: Coari in Amazonas (AM) (green curve, for 2005), Cruzeiro do Sul in Acre (AC) (blue curve,Figure for 2010) 2. Standardized and Vilhena anomaly in Rondônia of observed: (RO) (a )( precipitationred curve, for and 2010). (b) temperature Monthly mean at the followingand standard Figuredeviationstation 2. Standardized cal sites:culated Coari over in Amazonasanomaly the base of (AM) period observed (green 1981 curve,: (–a201) precipitation for0. 2005), Cruzeiro and ( dob) Sultemperature in Acre (AC) at (blue the curve,following stationfor sites 2010): andCoari Vilhena in Amazonas in Rondônia (AM) (RO) (green (red curve, curve, for for 2010). 2005), Monthly Cruzeiro mean do and Sul standard in Acre deviation (AC) (blue curve,calculated for 2010) over and the Vilhena base period in Rondônia 1981–2010. (RO) (red curve, for 2010). Monthly mean and standard deviation calculated over the base period 1981–2010.

(a) (b)

Figure 3. The standardized anomaly of: (a) precipitation; (b) temperature at 2-m above ground;

Figure(c) vertically3. The standardized integrated soil anomaly moisture; of (:d )(a relative) precipitation; humidity ( atb) 2-m temperature above the groundat 2-m asabove modeled ground by ; (c) SRDAS. Monthly( meansc) are averaged over the following regions: north (area 1,(d green) curve), west vertically integrated soil moisture; (d) relative humidity at 2-m above the ground as modeled by (area 2, blue curve), and southeast (area 3, red curve) in the Amazon basin, as delimited in Figure1. SRDAS. Monthly means are averaged over the following regions: north (area 1, green curve), west Monthly mean and standard deviation calculated over the base period 1999–2013. Figure(area 2, 3. blue The curve), standardized and southeast anomaly (area of: (3,a )red precipitation; curve) in the (b )Amazon temperature basin, at as 2 -delimitedm above groundin Figure; ( c1.) vMonthlyerticallyWe analyzed mean integrated and the standard soilintra-annual moisture deviation variability; (d) calculatedrelative of precipitation humidity over the atbase and2-m period temperature above 1999 the– ground2013. based on as the modeled historical by distributionSRDAS. Monthly in each means one of are the averaged three epicenters, over the considering following theregions 1981–2010: north period (area for1, green the meteorological curve), west station(areaWe analyzed2, datasetsblue curve), (Figures the and intr 4southeast–a6-)a andnnual the (area 1999–2013variability 3, red curve) period of precipitationin for the the Amazon SRDAS basin, and dataset temperatureas (Figuresdelimited7– in 9). based Figure For the on1. the historicalmeteorologicalMonthly distribution mean stationand standardin datasets each onedeviation (Figures of the4 calculated– 6three) each epicenters, monthly over the boxplot base considering period represents 1999 the– the2013. 1981 historical –2010 distribution period for the meteorologicalfor the 1981–2010 station period. dataset Fors the(Figures SRDAS 4, (Figures 5 and 76–)9 and) each the monthly 1999–2013 boxplot period represents for the the SRDAS historical dataset distributionWe analyzed for the the 1999–2013 intra-an period.nual variability Thus, each boxplot of precipitation represents the and minimum temperature value (the based whisker on the historicalbelow distribution the box), the in 1st each quartile one (the of the base three of the epicenters, box), the median considering or 2nd quartilethe 1981 (the–2010 line period inside thefor the meteorological station datasets (Figures 4, 5 and 6) and the 1999–2013 period for the SRDAS dataset

Water 2017, 9, x FOR PEER REVIEW 6 of 16

(Figures 7, 8 and 9). For the meteorological station datasets (Figures 4, 5 and 6) each monthly boxplot Waterrepresent 2017, 9s, xthe FOR historical PEER REVIEW distribution for the 1981–2010 period. For the SRDAS (Figures 7, 8 and6 of 169) each monthly boxplot represents the historical distribution for the 1999–2013 period. Thus, each (Figures 7, 8 and 9). For the meteorological station datasets (Figures 4, 5 and 6) each monthly boxplot boxplot represents the minimum value (the whisker below the box), the 1st quartile (the base of the representWater 2017, s9 , thex FOR historical PEER REVIEW distribution for the 1981–2010 period. For the SRDAS (Figures 7, 8 and6 of 9)16 box), the median or 2nd quartile (the line inside the box), the 3rd quartile (the cover of the box), the each monthly boxplot represents the historical distribution for the 1999–2013 period. Thus, each maximum (the whisker above the box), and the outliers (the values below or above the difference boxplot(Figures represents 7, 8 and 9). the For minimum the meteorological value (the station whisker datasets below (Figuresthe box), 4, the 5 and 1st 6quartile) each monthly (the base boxplot of the between the 3rd quartile and the 1st quartile multiplied by 1.5) of each historical distribution. The box),represent the medians the historical or 2nd quartiledistribution (the linefor theinside 1981 the–2010 box), period. the 3rd For quartile the SRDAS (the cover (Figur ofes the 7, box),8 and the 9) seasonal variability for the year of 2005, based on the 1981–2010 period of observed data, indicated maximumeach monthly (the boxplotwhisker representabove thes thebox) historical, and the distributionoutliers (the forvalues the below 1999–2013 or above period. the Thus,difference each monthly precipitation values above the median for almost half year. In particular, from February to betweenboxplot representsthe 3rd quartile the minimum and the value1st quartile (the whisker multiplied below by the 1.5 )box), of each the 1sthistorical quartile distribution. (the base of T thehe MayWater, monthly2018, 10 ,precipitation 1594 values were below the 1st quartile for almost the entire drought6 of season 16 seasonalbox), the variabilitymedian or for2nd the quartile year of (the 2005, line b insideased on the the box), 1981 the–2010 3rd period quartile of (the observed cover ofdata, the indicated box), the from June to September when most of the events happen, reaching the inferior limit in July (0.57 monthlymaximum precipitation (the whisker values above above the box) the ,median and the for outliers almost (the half values year. Ibelown particular, or above from the February difference to mm/d) (Figure 4a). Contrarily, the temperature was above the third quartile for the whole year and, Maybetween,box), monthly the 3rd3 rdprecipitation quartile quartile (the and cover values the of 1st thewere quartile box), below the multiplied maximum the 1st quartile (the by whisker1.5 for) of almost aboveeach historical the the box), entire and distribution. drought the outliers season The even though it decreased in July when compared to June, it reached its maximum (28.56 °C) in fromseasonal(the June valuesvariability to September below for or abovethe when year the most of difference 2005, of the b betweenased events on thethehappen 3rd1981 quartile,– 2010reaching andperiod thethe 1stof inferior observed quartile limit multiplied data, in Julyindicated by (0.57 August1.5) of(Figure each historical 4b). distribution. The seasonal variability for the year of 2005, based on the 1981–2010 mm/d)monthly (Figure precipitation 4a). Contrarily, values above the temperature the median was for almostabove thehalf third year .quartile In particular, for the from whole February year and, to period of observed data, indicated monthly precipitation values above the median for almost half evenMay, thoughmonthly it precipitation decreased in values July whenwere below compared the 1st to quartileJune, it forreached almost its the maximum entire drought (28.56 season °C) in fromyear. June In to particular, September from when February most to May,of the monthly events precipitation happen, reaching values were the belowinferior the limit 1st quartile in July for (0.57 Augustalmost (Figure the entire 4b). drought season from June to September when most of the events happen, reaching mm/d) (Figure 4a). Contrarily, the temperature was above the third quartile for the whole year and, the inferior limit in July (0.57 mm/d) (Figure4a). Contrarily, the temperature was above the third even though it decreased in July when compared to June, it reached its maximum (28.56 °C) in quartile for the whole year and, even though it decreased in July when compared to June, it reached its Augustmaximum (Figure (28.56 4b). ◦C) in August (Figure4b).

(a) (b)

Figure 4. Intra-annual(a )variability of observed: (a) precipitation (mm/d) and (b()b temperature) (°C, right panel) monthly values at the Coari station in Amazonas (AM). Each monthly boxplot represents the historical distribution for the 1981–2010. The green curve depicts the above variables for the year of Figure 4. Intra-annual variability of observed : (a) precipitation (mm/d) and (b) temperature (°C, right 2005. (a) (b) panel) monthly values at the Coari station in Amazonas (AM). Each monthly boxplot represents the historicalFigure distribution 4. Intra-annual for variability the 1981– of2010. observed: The green (a) precipitation curve depicts (mm/d) the andabove (b) temperaturevariables for (◦ C,the right year of 2005Figurepanel). 4. Intra monthly-annual values variability at the Coari of observed station in: Amazonas(a) precipitation (AM). Each(mm/d monthly) and ( boxplotb) temperature represents (°C, the right panel)historical monthly distribution values at forthe the Coari 1981–2010. station in The Amazonas green curve (AM) depicts. Each the monthly above variables boxplot for rep theresent years the historicalof 2005. distribution for the 1981–2010. The green curve depicts the above variables for the year of 2005.

(a) (b)

Figure 5. The same as(a )Figure 4 but for Cruzeiro do Sul station in Acre (AC) and(b )for 2010 (blue curve). Figure 5. The same as Figure4 but for Cruzeiro do Sul station in Acre (AC) and for 2010 (blue curve). Figure 5. The same as Figure 4 but for Cruzeiro do Sul station in Acre (AC) and for 2010 (blue curve) . (a) (b)

Figure 5. The same as Figure 4 but for Cruzeiro do Sul station in Acre (AC) and for 2010 (blue curve).

(a) (b)

Figure 6. The same as Figure4 but for Vilhena station in Rond ônia (RO) and for 2010 (red curve).

The SRDAS product(a) shows similar conﬁguration for precipitation and temperature(b) from observed data, but with lower (higher) precipitation (temperature) variability during the whole year of 2005. In agreement with Coari station, SRDAS had the lowest precipitation value in July (Figure5a). For the 2-m temperature, the(a month) with the highest value was September followed(b by) August in SRDAS,

Water 2017, 9, x FOR PEER REVIEW 7 of 16

Figure 6. The same as Figure 4 but for Vilhena station in Rondônia (RO) and for 2010 (red curve).

The SRDAS product shows similar configuration for precipitation and temperature from observed data, but with lower (higher) precipitation (temperature) variability during the whole year of 2005. In agreement with Coari station, SRDAS had the lowest precipitation value in July (Figure Water 2018, 10, 1594 7 of 16 5a). For the 2-m temperature, the month with the highest value was September followed by August in SRDAS, whereas at the station site, the maximum in temperature occurred in August (Figure 4b; and whereasFigure 7 atb the). The station 2-m site, soil the moisture maximum showed in temperature a lag from occurred both in variables, August (Figure reaching4b; and its Figure minimum7b). in OctoberThe 2-m (Figure soil moisture 7c). In showed 2005, from a lag from May both to November, variables, reaching 2-m soil its minimum moisture in values October were (Figure below7c). the In 2005, from May to November, 2-m soil moisture values were below the minimum limit of the month minimum limit of the month (Figure 7c). The relative humidity reached the lowest value in (Figure7c). The relative humidity reached the lowest value in September (Figure7d) in agreement September (Figure 7d) in agreement with the highest temperature value in the same month. with the highest temperature value in the same month.

(a) (b)

FigureFigure 7. Intra 7. Intra-annual-annual variabili variabilityty of of SRDAS SRDAS variables variables overover the the northern northern region region (area (area 1 from 1 from Figure Figure1): 1): ◦ (a) p(arecipitation) precipitation (mm/d) (mm/d);; (b ()b temperature) temperature at at 2 2-m-m aboveabove ground ground ( C);(°C) (c; )( 2-mc) 2- integratedm integrated soil moisturesoil moisture (mm);(mm); and and (d) (drelative) relative humi humiditydity at at 2 2-m-m above dede groundground (%). (%) Each. Each monthly monthly boxplot boxplot represents represent the s the historical distribution for the 1999–2013. The green curve depicts the above variables for the year historical distribution for the 1999–2013. The green curve depicts the above variables for the year of of 2005. 2005. In the 2010s, in the western epicenter, precipitation was below the ﬁrst quartile for almost the entire droughtIn the 2010 season,s, in except the western for August. epicenter, Although precipitation the lowest values was happened below the in Junefirst andquartile July, Septemberfor almost the entireexhibited drought precipitation season, except near thefor ﬁrst Aug quartileust. Although and was thethe month lowest with valu thees highest happened temperature in June valueand July, September(Figure5 ).exhibited Differently precipitation from the observed near data, the SRDASfirst quartile showed and decrease was in the precipitation month with from the March highest temperatureto August, value the latter (Figure being 5). the Differently month with from the lowest the observed value in data, 2010. SRDAS Although showed September decreas had e in precipitationslightly increased from March precipitation to August, in comparison the latter with being August, the month its value with was the still lowest below the value inferior in 2010. Althoughlimit (Figure September8a). SRDAS had slightly displayed increase similard temperature precipitation variability in comparison in comparison with toAugust the station,, its value with was the highest temperature value in September (Figure8b), and above the third quartile from February still below the inferior limit (Figure 8a). SRDAS displayed similar temperature variability in to October. The 2-m vertically integrated soil moisture had the lowest value in October, showing comparison to the station, with the highest temperature value in September (Figure 8b), and above two-month lag from the lowest value in precipitation and a month lag from the highest temperature the valuethird (Figurequartile8c). from The 2-m February relative to humidity October lowest. The value 2-m occurred vertically in Septemberintegrated as soil the moisturehighest value had the lowestin temperaturevalue in October (Figure, showing8d), however two- theirmonth intra-annual lag from the variability lowest showedvalue in better precipitation agreement and in thea month lag fromnorthern the area.highest temperature value (Figure 8c). The 2-m relative humidity lowest value occurred in September as the highest value in temperature (Figure 8d), however their intra-annual variability showed better agreement in the northern area.

Water 2017, 9, x FOR PEER REVIEW 8 of 16

Water 2017, 9, x FOR PEER REVIEW 8 of 16 Water 2018, 10, 1594 8 of 16

(a) (b)

Figure 8. The same as Figure 5 but for the western region (area 2 in Figure 1) and for 2010 (blue curve). (c) (d)

In theFigure southeast 8. The same area, as Figure precipitation5 but for the western exhibited region large (area 2 invariability Figure1) and throughout for 2010 (blue curve).the year 2010, howeverFigure was 8. closeThe same to zero as Figureduring 5 thebut droughtfor the western season regionin both (area station 2 in and Figure SRDAS 1) and datasets for 2010 (Figure (blue 6a; In the southeast area, precipitation exhibited large variability throughout the year 2010, however and curve) Figure. 9a). In contrast to the other two areas, the intra-annual variability of the SRDAS temperaturewas close showed to zero duringless agreement the drought with season the observed in both station data, but and reached SRDAS datasetsits peak (Figurein September6a; and as in Figure9a). In contrast to the other two areas, the intra-annual variability of the SRDAS temperature the observationIn the southeast, surpassing area, the precipitation maximum limitexhibited and thelarge third variability quartile, respectivelythroughout (Figurethe year 6b ; 2010, and showed less agreement with the observed data, but reached its peak in September as in the observation, Figurehowever 9b was). For close the 2to-m zero soil during moisture the, adrought three-month season lag in from both thestation precipitation and SRDAS minimum datasets was (Figure seen 6 ina; surpassing the maximum limit and the third quartile, respectively (Figure6b; and Figure9b). For the this area, with the lowest value occurring in September (Figure 9c), which was also the month with and 2-m Figure soil moisture,9a). In contrast a three-month to the lag other from two the precipitation areas, the intra minimum-annual was variability seen in this of area, the with SRDAS the highest temperature value. Similarly, the 2-m relative humidity also had the lowest value in temperaturethe lowest showed value occurring less agreement in September with the (Figure observed9c), which data, was but alsoreached the month its peak with in theSeptember highest as in Septemberthe observationtemperature (Figure, value.surpassing 9d). Similarly, the maximum the 2-m relative limit humidityand the third also hadquartile, the lowest respectively value in (Figure September 6b; and Figure(Figure 9b).9 Ford). the 2-m soil moisture, a three-month lag from the precipitation minimum was seen in this area, with the lowest value occurring in September (Figure 9c), which was also the month with the highest temperature value. Similarly, the 2-m relative humidity also had the lowest value in September (Figure 9d).

(a) (b)

Figure 9. Cont.

(a) (b)

Water 2017, 9, x FOR PEER REVIEW 9 of 16 Water 2017, 9, x FOR PEER REVIEW 9 of 16 Water 2018, 10, 1594 9 of 16

FigureFigure 9. The 9.sameThe as same Figure as 7 Figure but for7 thebut south for theeastern southeastern region (area region 3 in (areaFigure 3 1) in and Figure for 12010) and (red for curve). 2010 Figure(red 9. The curve). same as Figure 7 but for the southeastern region (area 3 in Figure 1) and for 2010 (red curve). The SRDAS precipitation (2-m temperature above ground) values are averaged over an entire regionTheT he what SRDASSRDAS might precipitationprecipitation explain the (2-m(2 - lowerm temperaturetemperature (higher) abovevalues ground) of precipitation values are (2 averagedaveraged-m temperature over anan above entireentire groundregionregion what) whatin com might mightparison explain explain with the the the lower observed lower (higher) (higher) values values at values station of precipitation of sites precipitation. (2-m temperature (2-m temperature above ground) above inground comparisonThe) followingin com withparison analysis the observedwith of the spatial observed values distribution at values station at sites.of station the SRDAS sites. anomalies was performed only for the monthTheThe followingfollowings exhibiting analysisanalysis the most of spatial spatialextreme distributiondistribution drought conditions of thethe SRDASSRDAS for each anomaliesanomal stationies wasw(Figureas performedperformed 4, 5 and onlyonly 6) and forfor SRDASthethe monthsmonth (Figus exhibitingexhibitingre 7, 8 and thethe 9). most Themost 2005’s extreme extreme drought drought drought had conditions conditions its driest for month for each each in station stationJuly for (Figures (Figure precipitation4– 46,) 5 and and, and SRDAS 6) theand lowest(FiguresSRDAS relative 7(Figu–9). There humidity 7, 2005’s 8 and drought9 and). The the 2005’s had highest its drought driest temperature month had its in driestanomal July for monthy precipitation,values in Julyin September for and precipitation the, lowest as previously, relativeand the discussedhumiditylowest relative in and this thehumidity section; highest and and temperature soil the moisture highest anomaly temperatureshowing values a month anomal in September,lagy fromvalues temperature in as September previously and, asthree discussed previously-month in lagthisdiscussed from section; precipitation in andthis section; soil moisture, with and the soil showing lowest moisture anomaly a monthshowing value lag a month from in October temperature lag from (Figure temperature and 10 three-month). The and area three 1 lag(seen-month from in Figureprecipitation,lag from 1) showprecipitation withed the the anomaly, lowestwith the lowest/highest anomaly lowest valueanomaly values in Octobervalue for thein (Figure Octoberfour analyzed 10 (Figure). The variables area10). 1The (seen along area in with1 Figure (seen part 1in) ofshowedFigure area 2.1) the Inshow anomalycontrasted the to lowest/highestanomaly the 2005’s lowest/highest drought, values the for values the2010’s four f ordrought analyzed the four occurred analyzed variables in variables alonga larger with areaalong part of with ofthe area ABpart 2., startingInof contrastarea in2. Inthe to contrast thenorth 2005’s and to the drought,ending 2005’s in the drought,the 2010’s southeast droughtthe 2010’s, but occurredthe drought deficits in occurred ain larger precipitation in area a larger of theand area AB, soil startingof moisture the AB in , happenedthestarting north in andin the the endingnorth west and in(Figure the ending southeast, 11a,c in) thewhereas butsoutheast the higher deﬁcits, but temperature the in precipitationdeficits valuesin precipitation and and soil the moisture rel andat ivesoil happenedhumidity moisture deficitinhappened the westwere in (Figureobserved the west 11 ina,c) (Figure the whereas basin’s 11a,c highersoutheast) whereas temperature region higher (Figure temperature values 11 andb,d). thevalues relative and humiditythe relative deﬁcit humidity were observeddeficit were in theobserved basin’s in southeast the basin’s region southeast (Figure region 11b,d). (Figure 11b,d).

(a) (b) (a) (b)

(c) (d) (c) (d) Figure 10. The 2005 drought anomaly ﬁelds of: (a) precipitation in July (mm/d); (b) temperature at ◦ Figure2-m above 10. The ground 2005 indrought September anomaly ( C); fields (c) 2-m of integrated: (a) precipitation soil moisture in July in (mm/d); October (b (mm);) temperature (d) relative at 2humidityFigure-m above 10 at. groundThe 2-m 2005 above in droughtSeptember ground anomaly in (°C) September; ( cfields) 2-m (%). ofin:tegrated Boxes(a) precipitation (dashed soil moisture lines) in July delimit in (mm/d);October areas: (mm)( northb) temperature; (1),(d) westrelative (2), at humidityand2-m southeastabove at ground2-m (3). above Solid in September ground black contour in (°C)September;encompasses (c) 2-m (%) in.tegrated Boxes the Amazon (dashed soil moisture basin.lines) delimitin October areas (mm): north; (d (1),) relative west (2humidity), and southeast at 2-m above(3). Solid ground black in contour September encompasses (%). Boxes the (dashed Amazon lines) basin. delimit areas: north (1), west (2), and southeast (3). Solid black contour encompasses the Amazon basin.

WaterWater 20172018,, 910, x, 1594FOR PEER REVIEW 1010 of of 16 16

(a) (b)

Figure 11. The 2010 drought anomaly fields of: (a) precipitation in August (mm/d); (b) temperature at ◦ Fi2-mgure above 11. The ground 2010 indrought September anomaly ( C); fields (c) 2-m of: integrated (a) precipitation soil moisture in August in October (mm/d) (mm);; (b) temperature (d) relative athumidity 2-m above at 2-mground above in groundSeptember in September (°C); (c) 2- (%).m integrated Boxes (dashed soil moisture lines) delimit in October areas (mm) north; (1),(d) westrelative (2), humidityand southeast at 2-m (3). above Solid ground black contour in September encompasses (%). Boxes the Amazon (dashed basin. lines) delimit areas north (1), west (2), and southeast (3). Solid black contour encompasses the Amazon basin. 4. Discussion 4. DiscussionClimate projections point out to an intensification of El Niño–Southern Oscillation- and AMO-related anomalies impacting the Amazon, including an increased frequency of extreme droughts Climate projections point out to an intensification of El Niño–Southern Oscillation- and due to the global warming [38,39]. The tropical forests sustain the hydrological cycle through AMO-related anomalies impacting the Amazon, including an increased frequency of extreme evapotranspiration, contributing to physical, chemical and biological processes and feedbacks, which droughts due to the global warming [38,39]. The tropical forests sustain the hydrological cycle non-linearity may buffer or amplify climate changes [40]. Furthermore, recent studies have shown through evapotranspiration, contributing to physical, chemical and biological processes and that the Amazon Basin in the beginning of the 21-century have had an unprecedented number of feedbacks, which non-linearity may buffer or amplify climate changes [40]. Furthermore, recent extreme drought events [8,11,16,19] and undergone a large-scale conversion of forests into pasture and studies have shown that the Amazon Basin in the beginning of the 21-century have had an cropland over the last decades [41–43], thus altering the land–atmosphere interface and contributing unprecedented number of extreme drought events [8,11,16,19] and undergone a large-scale to changes in the regional and local hydrological cycle [44–46]. Deforestation and land use process conversion of forests into pasture and cropland over the last decades [41–43], thus altering the over the region are associated with fire practices, increasing atmospheric aerosol concentrations, which land–atmosphere interface and contributing to changes in the regional and local hydrological cycle in turn promote feedback mechanisms between clouds and precipitation [47,48], thus contributing to [44–46]. Deforestation and land use process over the region are associated with fire practices, changes in the water cycle. The interaction between temperature trends, deforestation and extreme increasing atmospheric aerosol concentrations, which in turn promote feedback mechanisms events in the last centuries reveals an intensification of the Amazon hydrological cycle since 1990 [49], between clouds and precipitation [47,48], thus contributing to changes in the water cycle. The beyond the natural variability [50]. Evidence of external ocean forcing in the Amazon’s hydroclimate interaction between temperature trends, deforestation and extreme events in the last centuries shows that both equatorial Pacific and tropical Atlantic SST anomalies are correlated with many reveals an intensification of the Amazon hydrological cycle since 1990 [49], beyond the natural variables that influence the Amazon basin [51,52]. As stated by Panisset et al. [8], different physical variability [50]. Evidence of external ocean forcing in the Amazon’s hydroclimate shows that both climate mechanisms were responsible for the contrasting spatial patterns of the 2005 and 2010 droughts. equatorial Pacific and tropical Atlantic SST anomalies are correlated with many variables that Because the 2005 drought has been associated with the tropical North Atlantic SST anomalous warming, influence the Amazon basin [51,52]. As stated by Panisset et al. [8], different physical climate its core of the precipitation, humidity, and soil moisture deficit was mainly over the western part mechanisms were responsible for the contrasting spatial patterns of the 2005 and 2010 droughts. of the AB whereas the eastern sector was less impacted exhibiting normal conditions and positive Because the 2005 drought has been associated with the tropical North Atlantic SST anomalous anomalies. The warming of the North Atlantic Ocean induces changes in the north-south divergent warming, its core of the precipitation, humidity, and soil moisture deficit was mainly over the circulation, the weakening of the northeast trade winds and the moisture fluxes due to the northward western part of the AB whereas the eastern sector was less impacted exhibiting normal conditions shift of the Intertropical Convergence Zone (ITCZ) [11]. Consequently, resulting in less precipitation and positive anomalies. The warming of the North Atlantic Ocean induces changes in the over the western AB as shown in previous studies [8,14,19,53] as the most severe of the last 100 years, north-south divergent circulation, the weakening of the northeast trade winds and the moisture the 2005 drought occurred in the western and southern area of AB, but also in the northern area as fluxes due to the northward shift of the Intertropical Convergence Zone (ITCZ) [11]. Consequently, resulting in less precipitation over the western AB as shown in previous studies [8,14,19,53] as the

Water 2018, 10, 1594 11 of 16 corroborated by the SRDAS product. Previous studies detected the highest precipitation deficit in the southwestern area of the basin [19], SRDAS also indicates drought occurrence in this region, albeit the minimum in precipitation during the dry season appears mainly in the northern area of AB, decrease in precipitation can be seen in the western area as well. Even though the detected drought core by SRDAS differs slightly from previous studies, the product indicated that the winter season was abnormally hot and dry with rainfall reaching the climatological first quartile. Departures from the month presenting the extremes in precipitation and temperature from the SRDAS results when compared to in situ data are due to the spatial differences because the observed data are from stations located in the delimited drought area, and the SRDAS product corresponds to an average over the whole regions 1, 2, and 3 (Figure1). The drought event of 2010 was more spatially extensive than the former, presenting another epicenter in the southeastern region besides the northwestern/western epicenter [15,20,33,54]. The northwestern/western epicenter of the drought of 2010 are attributed to AMO, whereas the southeastern one is related to a moderate to strong warming of the Pacific SST during a positive ENSO event [20]. This warming of Pacific SST induces changes in the west–east Walker circulation, reflecting dry conditions over Eastern Amazonia due to the subsidence that inhibits rainfall over the region [10,55]. According to our results, the 2010 drought was associated with reduced precipitation, humidity, and soil moisture and extreme temperatures over the west and southwest AB but also over the southeast sector. SRDAS showed that the precipitation deficit occurred mainly in the western region where a larger area of negative anomaly of soil moisture was found, with maximum detected in almost the whole area 3. Changes in soil moisture rely on modification not only of relative humidity (shown in Figures 10 and 11) in which the maximum deficit occurred in the same area, especially in the 2005 drought, with 1-month lag, but as well as precipitation anomalies, which precede the soil moisture deficits [4–6,33,56–59]. SRDAS shows that precipitation returns back to normal or above normal conditions on the same or following month of the soil moisture minimum, and temperature shows a distinct positive anomaly at the drought peak. SRDAS also detects the relation between higher temperature values and lower relative humidity in both drought events that favors the occurrence and propagation of fire [60,61]. Although SRDAS and the Gravity Recovery and Climate Experiment (GRACE) observe different parameters (SRDAS uses only the soil moisture, and GRACE uses the integral of surface water, soil moisture, and groundwater), it is interesting to compare both results. When compared SRDAS integrated soil moisture with GRACE terrestrial water storage shown in [53], for the 2005 drought, GRACE shows its minimum in August/September and the month that represents it in SRDAS is October but, in 2010, both of them show a minimum in October, during the drought peak in different regions though [62]. The SRDAS precipitation showed consistency to previous studies [11] that exhibited daily minimum rainfall (GPCP) values during the dry season to both of the droughts. The 2005 drought had the minimal influence of temperature differently from the 2010 drought, which experienced extreme temperature anomalies in the western area of the basin, and it was also proved by SRDAS, but in area 3 [8].

5. Conclusions Traditional methods of drought assessment and monitoring depend heavily on rainfall data as recorded in meteorological and hydrological networks. However the availability of reliable satellite imagery covering wide regions over long periods of time has progressively strengthen the role of remote sensing in environmental studies, in particular in those related to drought episodes [8,63]. Accordingly, the aim of SRDAS is to provide long-term downscaled fields from global reanalyses over South America taking advantage of the spatial and temporal coverage of the remote sensing products. Here we show that the newly released reconstruction of South American hydroclimate, corroborates previous studies on the intensity and location of the 2005 and 2010 droughts in the Amazon Basin [8,11,16], therefore contributing to the analysis of environmental disasters and risk Water 2018, 10, 1594 12 of 16 assessment in areas with sparse gauge data. Indeed, the SRDAS temperature and precipitation anomalies showed consistency with in situ observations along each drought epicenter. The results show that the product can identify similar and divergent spatiotemporal patterns of the two different extremes from 2005 and 2010 [8]. The SRDAS product makes use of the data assimilation improvements brought from the coarse-resolution R2, through the use of a new scale-selective bias correction, mainly applied to the rotational component of the horizontal wind [25]. Thus, the SRDAS concept of combining regional models and the assimilation of remote sensing products shows to be useful to the analyses of extreme climate events, such as drought is Amazon region. Moreover, the product showed the capacity of recognizes the N-SE propagating pattern of the 2010 drought. NSSBC maintains synoptic to planetary waves from the NCEP-R2 boundary conditions nearly unchanged in the inner domain of SRDAS. Therefore, planetary modes, such as ENSO and AMO, are expected to influence the SRDAS hydrometeorological variables. Thus, our results highlight the ability of this product to reproduce the effects of the two main teleconnections mechanisms responsible for drought occurrence over the region, namely the ENSO and AMO. Albeit in situ measurements of soil moisture are unavailable for the ground station analyzed here, the one-month lag in soil moisture anomalies detected by SRDAS over the drought episodes is in accordance to the literature. Precipitation assimilation of satellite-based products is an essential feature of SRDAS that provides more accurate high-resolution gridded variables to better reproduce the Amazon extreme droughts, especially the soil moisture, which isn’t represented well in global hydrology and land surface models [64]. There is an increasing concern about loss of forest and ecosystem services due to extreme drought occurrence in the Amazon basin. Deforestation, fire and climate change have been reported as primary agents altering hydrological cycle in the region [44,45,65,66]. Such interactions among anthropogenic pressures, climate changes, and forest responses present potential positive feedbacks that may increase forest degradation and loss [67]. Despite this, drought should be recognized as having an important impact—patterns of drought incidence need to be further studied to allow for resources for innovative ideas for drought monitoring and environmental policies. Such concerns support the need to provide reliable drought information. Here we show that the newly released reconstruction of South American hydroclimate provides baseline and refined information for most of actors involved in the drought management process, such as precipitation, temperature, soil moisture, and relative humidity. This is a preliminary analysis that requires further research aiming to evaluate the differences between the SRDAS product and global reanalysis datasets, as well to include other in situ and remote sensing-based indicators (e.g. soil moisture, cloud cover) and to better understand the advantages and caveats of the newly released hydroclimate reconstruction in reproducing Amazon drought patterns. Finally, in 2015, a new record-breaking drought related to a strong El-Niño event also took place in the Amazon Basin and thus further requires more research in the context of the SRDAS response to these extreme events.

Author Contributions: B.N.G. wrote the first draft of the paper; B.N.G. and R.L. analyzed the data and/or products; B.N.G., R.L. and A.M.B.N. elaborated the paper; and A.M.B.N. developed the product, performed the numerical experiment, and contributed to the analyses. Funding: Serrapilheira Institute (grant number Serra-1708-15159) and FAPERJ award 111.381/2014. Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

References

1. González, J.; Valdés, J.B. New drought frequency index: Deﬁnition and comparative performance analysis. Water Resour. Res. 2006, 42.[CrossRef] 2. Mishra, A.K.; Singh, V.P. Review paper A review of drought concepts. J. Hydrol. 2010, 391, 202–216. [CrossRef] Water 2018, 10, 1594 13 of 16

3. Hoffmann, W.A.; Schroeder, W.; Jackson, R.B. Positive feedbacks of ﬁre, climate, and vegetation and the conversion of tropical savanna. Geophys. Res. Lett. 2002, 29.[CrossRef] 4. Choat, B.; Jansen, S.; Brodribb, T.J.; Cochard, H.; Delzon, S.; Bhaskar, R.; Bucci, S.J.; Feild, T.S.; Gleason, S.M.; Hacke, U.G.; et al. Global convergence in the vulnerability of forests to drought. Nature 2012, 491, 752–755. [CrossRef][PubMed] 5. Rowland, L.; Da Costa, A.C.L.; Galbraith, D.R.; Oliveira, R.S.; Binks, O.J.; Oliveira, A.A.R.; Pullen, A.M.; Doughty, C.E.; Metcalfe, D.B.; Vasconcelos, S.S. Death from drought in tropical forests is triggered by hydraulics not carbon starvation. Nature 2015, 528, 119–122. [CrossRef][PubMed] 6. Salazar, L.F.; Nobre, C.A.; Oyama, M.D. Climate change consequences on the biome distribution in tropical South America. Geophys. Res. Lett. 2007, 34.[CrossRef] 7. Kooperman, G.J.; Chen, Y.; Hoffman, F.M.; Koven, C.D.; Lindsay, K.; Pritchard, M.S.; Swann, A.L.S.;

Randerson, J.T. Forest response to rising CO2 drives zonally asymmetric rainfall change over tropical land. Nat. Clim. Chang. 2018, 8, 434–440. [CrossRef] 8. Panisset, J.S.; Libonati, R.; Gouveia, C.M.P.; Machado-Silva, F.; França, D.A.; França, J.R.A.; Peres, L.F. Contrasting patterns of the extreme drought episodes of 2005, 2010 and 2015 in the Amazon Basin. Int. J. Climatol. 2018, 38, 1096–1104. [CrossRef] 9. Cai, W.; Borlace, S.; Lengaigne, M.; Van Rensch, P.; Collins, M.; Vecchi, G.; Timmermann, A.; Santoso, A.; Mcphaden, M.J.; Wu, L.; et al. Increasing frequency of extreme El Niño events due to greenhouse warming. Nat. Clim. Chang. 2014, 4, 111–116. [CrossRef] 10. Jiménez-Muñoz, J.C.; Mattar, C.; Barichivich, J.; Santamaría-Artigas, A.; Takahashi, K.; Malhi, Y.; Sobrino, J.A.; van der Schrier, G. Record-breaking warming and extreme drought in the Amazon rainforest during the course of El Niño 2015–2016. Sci. Rep. 2016, 6, 33130. [CrossRef][PubMed] 11. Marengo, J.A.; Espinoza, J.C. Extreme seasonal droughts and floods in Amazonia: Causes, trends and impacts. Int. J. Climatol. 2016, 36, 1033–1050. [CrossRef] 12. Fu, R.; Dickinson, R.; Chen, M.; Wang, H. How do tropical sea surface temperatures influence the seasonal distribution of precipitation in the equatorial Amazon? J. Clim. 2001, 4003–4026. [CrossRef] 13. Liebmann, B.; Marengo, J.A. Interannual variability of the rainy season and rainfall in the Brazilian Amazon Basin. J. Clim. 2001, 14, 4308–4318. [CrossRef] 14. Yoon, J.H.; Zeng, N. An Atlantic influence on Amazon rainfall. Clim. Dyn. 2010, 34, 249–264. [CrossRef] 15. Andreoli, R.V.; De Oliveira, S.; Kayano, T.; Viegas, J.; Augusto, R.; De Souza, F.; Antonio, L. The influence of different El Niño types on the South American rainfall. Int. J. Climatol. 2017, 1390, 1374–1390. [CrossRef] 16. Coelho, C.A.S.; Cavalcanti, I.A.F.; Costa, S.; Freitas, S.R.; Ito, E.R.; Luz, G.; Santos, A.F.; Nobre, C.A.; Marengo, J.A.; Pezza, A.B. Climate diagnostics of three major drought events in the Amazon and illustrations of their seasonal precipitation predictions. Meteorol. Appl. 2012, 19, 237–255. [CrossRef] 17. Tedeschi, R.G.; Collins, M. The influence of ENSO on South American precipitation during austral summer and autumn in observations and models. Int. J. Climatol. 2016, 36, 618–635. [CrossRef] 18. Nepstad, D.C.; Veríssimo, A.; Alencar, A.; Nobre, C.; Lima, E.; Lefebvre, P.; Schlesinger, P.; Potter, C.; Moutinho, P.; Mendoza, E.; et al. Large-scale impoverishment of amazonian forests by logging and fire. Nature 1999, 398, 505–508. [CrossRef] 19. Marengo, J.A.; Nobre, C.A.; Tomasella, J.; Oyama, M.D.; Sampaio de Oliveira, G.; De Oliveira, R.; Camargo, H.; Alves, L.M.; Brown, I.F.; de Oliveira, G.S.; et al. The drought of Amazonia in 2005. J. Clim. 2008, 21, 495–516. [CrossRef] 20. Marengo, J.A.; Tomasella, J.; Alves, L.M.; Soares, W.R.; Rodriguez, D.A. The drought of 2010 in the context of historical droughts in the Amazon region. Geophys. Res. Lett. 2011, 38.[CrossRef] 21. Huntingford, C.; Fisher, R.A.; Mercado, L.; Booth, B.B.; Sitch, S.; Harris, P.P.; Cox, P.M.; Jones, C.D.; Betts, R.A.; Malhi, Y.; et al. Towards quantifying uncertainty in predictions of Amazon “dieback”. Philos. Trans. R. Soc. B Biol. Sci. 2008, 363, 1857–1864. [CrossRef][PubMed] 22. Barros, V.R.; Field, C.B.; Dokke, D.J.; Mastrandrea, M.D.; Mach, K.J.; Bilir, T.E.; Chatterjee, M.; Ebi, K.L.; Estrada, Y.O.; Genova, R.C.; et al. Climate Change 2014: Impacts, Adaptation, and Vulnerability; Part B: Regional Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. In Proceedings of the Intergovernmental Panel on Climate Change (IPCC), New York, NY, USA, 1 January 2014. Water 2018, 10, 1594 14 of 16

23. Caya, D.; Biner, S.; Diaconescu, E. Challenging some tenets of Regional Climate Modelling. Meteorol. Atmos. Phys. 2008, 22, 3–22. [CrossRef] 24. Nunes, A.M.B.; Roads, J.O. Improving regional model simulations with precipitation assimilation. Earth Interact. 2005, 9.[CrossRef] 25. Nunes, A.M.B. Satellite-enhanced dynamical downscaling for the analysis of extreme events. J. Geophys. Res. Atmos. 2016, 121, 617–636. [CrossRef] 26. Von Storch, H.; Langenberg, H.; Feser, F. A Spectral Nudging Technique for Dynamical Downscaling Purposes. Mon. Weather Rev. 2000, 128, 3664–3673. [CrossRef] 27. Juang, H.-M.H.; Hong, S.-Y.; Kanamitsu, M. The NCEP Regional Spectral Model: An Update. Bull. Am. Meteorol. Soc. 1997, 78, 2125–2143. [CrossRef] 28. Kanamitsu, M.; Yoshimura, K.; Yhang, Y.-B.; Hong, S.-Y. Errors of Interannual Variability and Trend in Dynamical Downscaling of Reanalysis. J. Geophys. Res. 2010, 115, D17115. [CrossRef] 29. Kanamitsu, M.; Ebisuzaki, W.; Woollen, J.; Yang, S.-K.; Hnilo, J.J.; Fiorino, M.; Potter, G.L. NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Am. Meteorol. Soc. 2002, 83, 1631–1644. [CrossRef] 30. Reynolds, R.W.; Smith, T.M.; Liu, C.; Chelton, D.B.; Casey, K.S.; Schlax, M.G. Daily High-Resolution-Blended Analyses for Sea Surface Temperature. J. Clim. 2007, 20, 5473–5496. [CrossRef] 31. Joyce, R.J.; Janowiak, J.E.; Arkin, P.A.; Xie, P. CMORPH: A Method that Produces Global Precipitation Estimates from Passive Microwave and Infrared Data at High Spatial and Temporal Resolution. J. Hydrometeorol. 2004, 5, 487–503. [CrossRef] 32. Ek, M.; Mitchell, K.; Lin, Y.; Rogers, E.; Grunmann, P.; Koren, V.; Gayno, G.; Tarpley, J.D. Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res. 2003, 108, 1–16. [CrossRef] 33. Lewis, S.L.; Brando, P.M.; Phillips, O.L.; Van Der Heijden, G.M.F.; Nepstad, D. The 2010 Amazon Drought. Science 2011, 331.[CrossRef][PubMed] 34. Zeng, N.; Yoon, J.-H.; Marengo, J.A.; Subramaniam, A.; Nobre, C.A.; Mariotti, A.; Neelin, J.D. Causes and impacts of the 2005 Amazon drought. Environ. Res. Lett. 2008, 3, 014002. [CrossRef] 35. Mutiibwa, D.; Strachan, S.; Albright, T. Land Surface Temperature and Surface Air Temperature in Complex Terrain. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2013, 8, 4762–4774. [CrossRef] 36. Fischer, E.; Seneviratne, S.; Vidale, P.; Lüthi, D.; Schär, C. Soil Moisture–Atmosphere Interactions during the 2003 European Summer Heat Wave. J. Clim. 2007, 20, 5081–5099. [CrossRef] 37. Seneviratne, S.I.; Wilhelm, M.; Stanelle, T.; Van Den Hurk, B.; Hagemann, S.; Berg, A.; Cheruy, F.; Higgins, M.E.; Meier, A.; Brovkin, V.; et al. Impact of soil moisture-climate feedbacks on CMIP5 projections: First results from the GLACE-CMIP5 experiment. Geophys. Res. Lett. 2013, 40, 5212–5217. [CrossRef] 38. Duffy, P.B.; Brando, P.; Asner, G.P.; Field, C.B. Projections of future meteorological drought and wet periods in the Amazon. Proc. Natl. Acad. Sci. USA 2015, 112, 13172–13177. [CrossRef][PubMed] 39. Guimberteau, M.; Ronchail, J.; Espinoza, J.C.; Lengaigne, M.; Sultan, B.; Polcher, J.; Drapeau, G.; Guyot, J.-L.; Ducharne, A.; Ciais, P. Future changes in precipitation and impacts on extreme streamﬂow over Amazonian sub-basins. Environ. Res. Lett. 2013, 8, 014035. [CrossRef] 40. Bonan, G.B. Forests and Climate Change: Forcings, Feedbacks, and the Climate Beneﬁts of Fiorests. Science 2008, 320, 1444–1449. [CrossRef][PubMed] 41. Ometto, J.P.; Sousa-Neto, E.R.; Tejada, G. Land Use, Land Cover and Land Use Change in the Brazilian Amazon (1960–2013); Springer: Berlin/Heidelberg, Germany, 2016. 42. Venter, O.; Sanderson, E.W.; Magrach, A.; Allan, J.R.; Beher, J.; Jones, K.R.; Possingham, H.P.; Laurance, W.F.; Wood, P.; Fekete, B.M.; et al. Sixteen years of change in the global terrestrial human footprint and implications for biodiversity conservation. Nat. Commun. 2016, 7, 12558. [CrossRef][PubMed] 43. Lapola, D.M.; Martinelli, L.A.; Peres, C.A.; Ometto, J.P.H.B.; Ferreira, M.E.; Nobre, C.A.; Aguiar, A.P.D.; Bustamante, M.M.C.; Cardoso, M.F.; Costa, M.H.; et al. Pervasive transition of the Brazilian land-use system. Nat. Clim. Chang. 2014, 4, 27–35. [CrossRef] 44. Zemp, D.C.; Schleussner, C.F.; Barbosa, H.M.J.; Rammig, A. Deforestation effects on Amazon forest resilience. Geophys. Res. Lett. 2017, 44, 6182–6190. [CrossRef] 45. Staal, A.; Tuinenburg, O.A.; Bosmans, J.H.C.; Holmgren, M.; Van Nes, E.H.; Scheffer, M.; Zemp, D.C.; Dekker, S.C. Forest-rainfall cascades buffer against drought across the Amazon. Nat. Clim. Chang. 2018, 8, 539–543. [CrossRef] Water 2018, 10, 1594 15 of 16

46. Lejeune, Q.; Davin, E.L.; Guillod, B.P.; Seneviratne, S.I. Influence of Amazonian deforestation on the future evolution of regional surface fluxes, circulation, surface temperature and precipitation. Clim. Dyn. 2015, 44, 2769–2786. [CrossRef] 47. Cox, P.M.; Harris, P.P.; Huntingford, C.; Betts, R.A.; Collins, M.; Jones, C.D.; Jupp, T.E.; Marengo, J.A.; Nobre, C.A. Increasing risk of Amazonian drought due to decreasing aerosol pollution. Nature 2008, 453, 212–215. [CrossRef][PubMed] 48. Bevan, S.L.; North, P.R.J.; Grey, W.M.F.; Los, S.O.; Plummer, S.E. The impact of atmospheric aerosol from biomass burning on Amazon dry-season drought. J. Geophys. Res. Atmos. 2009, 114, 1–11. [CrossRef] 49. Gloor, M.; Brienen, R.J.W.; Galbraith, D.; Feldpausch, T.R.; Schöngart, J.; Guyot, J.L.; Espinoza, J.C.; Lloyd, J.; Phillips, O.L. Intensification of the Amazon hydrological cycle over the last two decades. Geophys. Res. Lett. 2013, 40, 1729–1733. [CrossRef] 50. Davidson, E.A.; de Araújo, A.C.; Artaxo, P.; Balch, J.K.; Brown, I.F.; Bustamante, M.M.; Coe, M.T.; De Fries, R.S.; Keller, M.; Longo, M.; et al. The Amazon basin in transition. Nature 2012, 481, 321–328. [CrossRef] [PubMed] 51. Chen, Y.; Randerson, J.T.; Morton, D.C.; DeFries, R.S.; Collatz, G.J.; Kasibhatla, P.S.; Giglio, L.; Jin, Y.; Marlier, M.E. Forecasting Fire Season Severity in South America Using Sea Surface Temperature Anomalies. Science 2011, 334, 787–791. [CrossRef][PubMed] 52. Chen, Y.; Velicogna, I.; Famiglietti, J.S.; Randerson, J.T. Satellite observations of terrestrial water storage provide early warning information about drought and fi re season severity in the Amazon. J. Geophys. Res. Biogeosci. 2013, 118, 495–504. [CrossRef] 53. Chen, J.L.; Wilson, C.R.; Tapley, B.D.; Yang, Z.L.; Niu, G.Y. 2005 drought event in the Amazon River basin as measured by GRACE and estimated by climate models. J. Geophys. Res. Solid Earth 2009, 114, 1–9. [CrossRef] 54. Espinoza, J.C.; Ronchail, J.; Guyot, J.L.; Junquas, C.; Vauchel, P.; Lavado, W.; Drapeau, G.; Pombosa, R. Climate variability and extreme drought in the upper Solimões River (western Amazon Basin): Understanding the exceptional 2010 drought. Geophys. Res. Lett. 2011, 38.[CrossRef] 55. Jiménez-Muñoz, J.C.; Barichivich, J.; Mattar, C.; Takahashi, K.; Sobrino, A.; Malhi, Y. Spatio-temporal patterns of thermal anomalies and drought over tropical forests driven by recent extreme climatic anomalies. Philos. Trans. R. Soc. B 2018, 373, 20170300. [CrossRef][PubMed] 56. Seneviratne, S.I.; Corti, T.; Davin, E.L.; Hirschi, M.; Jaeger, E.B.; Lehner, I.; Orlowsky, B.; Teuling, A.J. Investigating soil moisture-climate interactions in a changing climate: A review. Earth-Sci. Rev. 2010, 99, 125–161. [CrossRef] 57. Poveda, G.; Jaramillo, A.; Gil, M.M.; Quiceno, N.; Mantilla, R. Seasonality in ENSO related precipitation, river discharges, soil moisture, and vegetation index (NDVI) in Colombia. Water Resour. Res. 2001, 37, 2169–2178. [CrossRef] 58. Fischer, E.M.; Seneviratne, S.I.; Lüthi, D.; Schär, C. Contribution of land-atmosphere coupling to recent European summer heat waves. Geophys. Res. Lett. 2007, 34, 1–6. [CrossRef] 59. Nicolai-Shaw, N.; Zscheischler, J.; Hirschi, M.; Gudmundsson, L.; Seneviratne, S.I. Remote Sensing of Environment A drought event composite analysis using satellite remote-sensing based soil moisture. Remote Sens. Environ. 2017, 203, 216–225. [CrossRef] 60. Brando, P.M.; Balch, J.K.; Nepstad, D.C.; Morton, D.C.; Putz, F.E.; Coe, M.T.; Silverio, D.; Macedo, M.N.; Davidson, E.A.; Nobrega, C.C.; et al. Abrupt increases in Amazonian tree mortality due to drought-fire interactions. Proc. Natl. Acad. Sci. USA 2014, 111, 6347–6352. [CrossRef][PubMed] 61. Nepstad, D.; Lefebvre, P.; Da Silva, U.L.; Tomasella, J.; Schlesinger, P.; Solórzano, L.; Moutinho, P.; Ray, D.; Benito, J.G. Amazon drought and its implications for forest flammability and tree growth: A basin-wide analysis. Glob. Chang. Biol. 2004, 10, 704–717. [CrossRef] 62. Frappart, F.; Ramillien, G.; Ronchail, J. Changes in terrestrial water storage versus rainfall and discharges in the Amazon basin. Int. J. Climatol. 2013, 33, 3029–3046. [CrossRef] 63. Vicente-Serrano, S.M.; Gouveia, C.; Camarero, J.J.; Beguería, S.; Trigo, R.; López-Moreno, J.I.; Azorín-Molina, C.; Pasho, E.; Lorenzo-Lacruz, J.; Revuelto, J.; et al. Response of vegetation to drought time-scales across global land biomes. Proc. Natl. Acad. Sci. USA 2013, 110, 52–57. [CrossRef][PubMed] 64. Zeng, N. Seasonal cycle and interannual variability in the Amazon hydrologic cycle. J. Geophys. Res. Atmos. 1999, 104, 9097–9106. [CrossRef] Water 2018, 10, 1594 16 of 16

65. Erfanian, A.; Wang, G.; Fomenko, L. Unprecedented drought over tropical South America in 2016: Signiﬁcantly under-predicted by tropical SST. Sci. Rep. 2017, 7, 22–24. [CrossRef][PubMed] 66. Nowell, H.; Homes, C.; Robertson, K.; Teske, C.; Hiers, J. A New Picture of Fire Extent, Variability, and Drought Interaction in Prescribed Fire Landscapes: Insights From Florida Government Records. Geophys. Res. Lett. 2018, 7874–7884. [CrossRef] 67. Yang, J.; Tian, H.; Pan, S.; Chen, G.; Zhang, B.; Dangal, S. Amazon drought and forest response: Largely reduced forest photosynthesis but slightly increased canopy greenness during the extreme drought of 2015/2016. Glob. Chang. Biol. 2018, 24, 1919–1934. [CrossRef][PubMed]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).