Abstract

This study investigates the variability of the seasonal cycle of convection in the Brazilian Amazon basin during the last decades, and examines physical mechanisms that potentially trigger these modifications. A new methodology to evaluate the onset and length of the rainy season using long-term cloud fraction observations from geostationary satellites is proposed and the connection between cloud cycle variability, surface properties, and thermodynamic and dynamic conditions is explored. The results show that cloud cover has significantly decreased over the last decades. The decline in cloudiness is steeper at 1200 UTC (0800 LT), when a trend of up to −6% decade−1 is observed over the central and eastern Amazon. High-cloud-cover reduction is the major contributor to the observed decline in total cloud fraction. Delayed onsets and a reduction of up to 4 days yr−1 in the northern and central Amazon wet-season length are observed. Correlation analyses indicate that the El Niño phenomenon affects the interannual variability of cloudiness in the Amazon, leading to delayed onset and early demise of the rainy season. The tropical South Atlantic, the Pacific warm pool, and the North Atlantic tripole also play a small, but significant, role in the Amazon’s cloudiness variability. The decrease in cloudiness over the Amazon basin reduces the amount of solar radiation reflected back to space while increasing irradiance at the surface. This local warming alters surface heat fluxes and the atmospheric thermodynamic profile, further affecting cloud development. The strong tendencies reported here indicate a significant shift in the Amazonian hydroclimate during the last few decades.

1. Introduction

The Amazon rain forest plays an important role in regulating the global climate system. It is in the rain forest that the water vapor transported from the Atlantic Ocean is recycled, contributing to the transport of heat and moisture from the tropics to the subtropics (Arraut and Satyamurty 2009). In the last decades, this region has undergone significant environmental changes caused by deforestation and biomass burning (e.g., Davidson et al. 2012). The response of this delicate environment to natural climate variability and human-induced environmental changes is largely unknown. Climate modeling studies indicate that a warmer climate could possibly induce a dieback of the forest and an incoming process of savanization of the Amazon basin (Cox et al. 2004; Malhi et al. 2009; Boers et al. 2017). Should this happen, there will be dramatic consequences for the South American climate, with socioeconomic and environmental impacts. Therefore, understanding mechanisms associated with variations and changes in convection is of paramount importance to evaluate how the environment will respond to natural and anthropogenic forcings in the coming years.

Observational studies already indicate pronounced alterations in the Amazonian rainfall regime in recent decades. An increase in dry-season length in the southern Amazon has been reported (Marengo et al. 2011; Fu et al. 2013), evidencing variations in the seasonal cycle of convection in the region. Longer dry seasons may increase the number of forest fire outbreaks (Running 2006; Fernandes et al. 2011; Arias et al. 2015), aggravating deforestation and the emission of greenhouse gases and aerosol to the atmosphere. Increased dry-season lengths were also linked to a delay in the onset of the rainy season (Fu and Li 2004). Long-term analysis of cloudiness and outgoing longwave radiation (OLR) also suggest drier conditions in the Amazon (Arias et al. 2011; Butt et al. 2009). Butt et al. (2009) analyzed geostationary satellite retrievals over five regions in the Amazon and observed a decreasing trend in cloud cover during the dry season and a modest increase during the rainy season from 1984 to 2007. These trends were not spatially homogeneous: in central-eastern and southeastern regions the reduction in cloud cover along the years was higher, while positive cloud-cover trends were more significant over northwest portion of the Amazon. Using monthly retrievals, Arias et al. (2011) showed a reduction in high cloud cover associated with an increase in shortwave radiation at the surface over the Amazon during the same period.

Besides altering local surface energy partition, global warming can also modify large-scale coupled modes that impact the Amazon rainfall regime. Most Amazonian droughts have been associated either with the occurrence of El Niño events or with warmer sea surface temperature (SST) anomalies in the tropical North Atlantic, or with a combination of both processes (e.g., Marengo and Espinoza 2016). According to most studies, the same processes are also related to delayed onsets of the wet season (e.g., Fu and Li 2004; Marengo et al. 2001; Liebmann and Marengo 2001). Recently, the Amazon experienced two severe droughts (2005 and 2010) within a period of only 5 years (Marengo et al. 2008; Zeng et al. 2008; Lewis et al. 2011; Marengo et al. 2011). Furthermore, both the 2005 and 2010 droughts were classified as once-in-a-century events (Marengo and Espinoza 2016). Global coupled climate models project potential increase in the frequency of occurrence of extreme El Niño events in response to global warming (Cai et al. 2014). The combination of higher than normal SST and stronger El Niño events will likely lead to an increase in the frequency of severe droughts in the Amazon.

Previous studies have proposed different approaches to assess the spatial variability of the onset and duration of the wet and dry seasons over tropical South America. Most methods rely on ground-based measurements and gridded daily precipitation data from rain gauges (Marengo et al. 2001; Butt et al. 2011; Liebmann and Marengo 2001; Fu et al. 2013); others use OLR as a proxy to determine the occurrence of convection (Kousky 1988; Horel et al. 1989). These methods have been applied to general circulation model outputs to investigate their ability to simulate the South American monsoon (e.g., Liebmann et al. 2007; Bombardi and Carvalho 2009). Although these approaches may provide useful information on rainfall regimes they also have some disadvantages. First, OLR may not be the most appropriate proxy for convection, since it relies only on the infrared channel of sensors aboard satellites to identify cloudiness, disregarding most of the low clouds that can provide an important contribution for the seasonal variability of cloudiness and precipitation in the Amazon. Second, rain gauge measurements are sparse, restricting the ability to accurately observe the spatial variability of the onset and duration of wet season. This study aims to advance understanding of recent changes in the development of convection and cloud life cycle in the Amazon. Here we examine changes in the diurnal and seasonal cycle of clouds over the Amazon in recent decades and investigate their relation with the onset and length of the rainy season. We also identify possible mechanisms explaining the observed changes in cloudiness and in the rainfall regime.

To address these questions we propose a new method to estimate the onset and duration of Amazon wet season based on cloud fraction retrievals from the International Satellite Cloud Climatology Project (ISCCP) (Schiffer and Rossow 1983). The new methodology to estimate the rainy-season onset is described in section 2. Section 3 discusses the spatial distribution of the wet-season onset and length obtained according to this new methodology (section 3a) and examines how the diurnal and seasonal cycles of convection have changed along the years (section 3b). Trends in cloudiness at different times of the day, onset, demise, and length of the wet season are also examined (sections 3b and 3c). The contributions of potential drivers and possible outcomes of these changes are also explored (sections 3d and 3e). Section 4 summarizes the main conclusions of this work.

2. Methodology

a. Observational datasets

1) Cloud fraction from ISCCP

This study proposes a new methodology to estimate the onset and end of the Amazon’s wet season based on cloud fraction. To identify clouds, ISCCP applies a sophisticated algorithm to visible and infrared radiances from geostationary satellites (Rossow and Schiffer 1999), providing reliable estimates of cloud properties. Daily cloud fraction retrievals from the International Satellite Cloud Climatology Project (the ISCCP-D1 product) on a 280-km equal-area (2.5° × 2.5°) grid with 3-h resolution from July 1983 to December 2009 were used. During that period, most of the cloud properties over the study region were retrieved using Geostationary Operational Environmental Satellites (GOES)-6, -7, -8, and -12. ISCCP follows a strict and careful procedure to retrieve cloud properties. In particular, during daytime the cloud mask algorithm takes into account analysis of the spatial and temporal variations of visible (0.6 μm) and infrared (11 μm) radiances to classify sets of pixels. Nighttime cloud retrievals are similar, but only the infrared radiances are used. Differences in surface properties, as well as particular geographic and climatological features, are also considered. Cloud fraction is defined as the amount of pixels classified as cloudy over the total number of pixels within a given area. The data are divided according to cloud-top pressure (PC) and classified according to pressure level as low cloud if PC > 680 hPa, midlevel cloud if 400 < PC ≤ 680 hPa, or high cloud if PC ≤ 400 hPa.

2) Climate modes and ERA-Interim

Variations in the Amazonian rainfall regime have been linked to the occurrence of different coupled climate modes in the Pacific and Atlantic Oceans (e.g., Marengo and Espinoza 2016). To evaluate the influence of these modes on the interannual variability of the wet season in the Amazon according to the metric proposed here, we calculated the Pearson correlation coefficient between the time series of the onset, end, and duration of the wet season, and maximum and minimum cloud fraction (fcmax and fcmin, respectively) observed in each year and several climate indices. The indices investigated in this study are the Atlantic multidecadal oscillation (AMO; Enfield et al. 2001), El Niño–Southern Oscillation (ENSO), the Pacific decadal oscillation (PDO; Mantua et al. 1997), the tropical southern Atlantic index (TSA; Enfield et al. 1999), the Pacific–North America index (PNA), the North Atlantic Oscillation index (NAO), the North Atlantic tripole (Deser and Timlin 1997), and the Pacific warm pool. These climate indices were provided by the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL) Climate Prediction Center (CPC) (https://www.esrl.noaa.gov/psd/data/climateindices/list/).

To better understand the processes associated with changes in the convection cycle over the Amazon, we examined connections between cloud variability and changes in thermodynamics and dynamics during the study period. Monthly averages of variables from the European Centre for Medium Range Weather and Forecast (ECMWF) Re-Analysis, ERA-Interim/Land and ERA-Interim (Dee et al. 2011), at different atmospheric levels were analyzed. The variables used in this study were surface sensible flux (SHF), latent heat flux (LHF), convective available potential energy (CAPE), specific humidity (q), temperature (T), relative humidity (RH), and zonal (u) and meridional (υ) components of the wind.

b. Method to estimate the onset and end of the wet season

The study area extends from 12.5°S to 2.5°N in latitude and 50° to 74°W in longitude. According to the Brazilian National Institute of Space Research [Instituto Nacional de Pesquisas Espaciais (INPE)] definition, this area encompasses the Brazilian Amazon rain forest, a crucial environment for heat and moisture transport across South America (Arraut and Satyamurty 2009). To obtain the average behavior of the seasonal cycle of cloudiness for each grid cell and 3-h period, the daily time series of cloud fraction is smoothed using penalized B-splines. In this study the mean cloud fraction (fcmed) observed over the whole area during the study period (1983–2009) at a given time of the day is used as a threshold to characterize the transition between dry and wet seasons.

In this case, the onset is defined as the date after which the smoothed cloud fraction systematically exceeds the mean annual threshold for a particular time of the day. Conversely, the end of the wet season is defined when cloud fraction becomes systematically below the same threshold. Duration is defined as the period between the onset and end of the wet season based on the discussed criterion. Figure 1 shows an example of the smoothed time series of cloud fraction at 1200 UTC for the grid cell centered at 6.25°S, 56.64°W. The pronounced seasonality observed in this figure indicates that this region experiences a large amplitude of variation in cloud fraction, from about 20% in the dry season to up to 100% in the wet season. The average cloud fraction observed during the study period within the whole study area was 68.3% (dashed red line). On average, higher cloudiness and lower differences between summer and winter are observed in earlier years compared to the entire period, indicating that changes in the processes that affect local convection may have occurred over time.

Fig. 1.

Smoothed time series of cloud fraction at 1200 UTC for the grid cell centered at 6.25°S, 56.64°W. The red line represents the average cloud fraction observed over the whole study area at 1200 UTC between 1983 and 2009. This value was used to define the onset and demise of the rainy season at each cell.

Fig. 1.

Smoothed time series of cloud fraction at 1200 UTC for the grid cell centered at 6.25°S, 56.64°W. The red line represents the average cloud fraction observed over the whole study area at 1200 UTC between 1983 and 2009. This value was used to define the onset and demise of the rainy season at each cell.

Notice that a given grid cell may exhibit different mean cloud fraction at different times of the day. Therefore, it is necessary to identify the time of the day that shows the largest and statistically significant differences in cloud fraction before and after the onset of the wet season. This was accomplished by applying the t test of the difference between the mean cloud fraction 1 month before and 1 month after the onset date at different times.

The most statistically significant results were found at 0600, 1200, and 1500 UTC (around 0200, 0800, and 1100 LT) when about 96% of the grid cells presented p values < 0.05. Cloud fraction at 1800 UTC exhibited less seasonal variability than at other times of the day and therefore larger p values (not shown). Besides the significance analysis, other potential caveats were considered for the choice of the time: 1) as previously described, diurnal ISCCP retrievals of cloud fraction are more reliable than nighttime retrievals, since they rely on radiance from two spectral channels (the visible and infrared); 2) the period that precedes the wet season is characterized by an increase in convective processes during the afternoon (Saraiva et al. 2016), reducing the ability to distinguish variations in cloud regimes from dry to wet season at that time of the day; and 3) as the land is cooler in the early morning, variations in cloudiness due to turbulent fluxes are reduced at that time. Based on the previous analyses and considerations, the cloud fraction thresholds observed at 1200 UTC (around 0800 LT) were chosen to define the dry-to-wet transition dates in each cell. The time series of the mean cloud fraction at 1200 UTC in the study area averaged over different time intervals (1.0, 1.5, 2.0, and 2.5 months) before and after the onset of wet season are shown in Fig. 2. All the time intervals analyzed show a clear difference between the mean cloud fraction observed before and after the wet-season onset, indicating different cloud regimes before and after the dry-to-wet transition date. The longer the time interval considered for averaging cloud fraction pre- and post-transition date, the larger the differences between cloud fractions. In the first years of study, cloud fraction was generally higher than in the present, and therefore the differences between cloudiness before and after the transition were less pronounced in the beginning of the time series, more specifically between 1984 and 1986.

Fig. 2.

Time series of the mean cloud fraction at 1200 UTC in the Amazon basin (12.5°S–2.5°N, 50°–74°W) averaged over different time intervals before and after the onset of the rainy season: (a) 1, (b) 1.5, (c) 2, and (d) 2.5 months.

Fig. 2.

Time series of the mean cloud fraction at 1200 UTC in the Amazon basin (12.5°S–2.5°N, 50°–74°W) averaged over different time intervals before and after the onset of the rainy season: (a) 1, (b) 1.5, (c) 2, and (d) 2.5 months.

Different thresholds, other than the mean cloud fraction (e.g., fcmed − 5%, fcmed − 10%), have also been tested to define the transition from dry to wet season. However, the results indicate that the choice of fcmed maximizes the difference between cloudiness before and after the wet season. Furthermore, as discussed in section 3, the results obtained using the mean cloud fraction to define the onset and demise of the wet season show good agreement with previous studies that applied different techniques to identify the wet season (e.g., Kousky 1988; Marengo et al. 2001).

3. Results

a. Spatial distribution of the rainfall regime in the Amazon basin

In this section the spatial variability of the rainfall regime in the Amazon basin is analyzed. The seasonal cycle of rainfall over the Brazilian Amazon is largely influenced by the South American monsoon system (SAMS). SAMS is primarily driven by temperature gradients between the South America and the Atlantic Ocean (e.g., Marengo et al. 2012; Silva Dias and Carvalho 2017; Silva and Kousky 2012; Vera et al. 2006). Unlike other monsoon systems, SAMS is best characterized by a seasonal reversal of wind anomaly over the continent (Zhou and Lau 1998). Although during the wet-season rain rates are typically lower than most monsoon systems (Petersen and Rutledge 2001), the total annual precipitation in the Amazon basin is similar to other monsoon regions (e.g., Xie and Arkin 1998; Liebmann and Mechoso 2011).

The spatial distributions of the average onset, demise, and duration of the wet season between 1983 and 2009, evaluated using the methodology proposed in section 2, are shown in Fig. 3. Because of the large extent of the Amazon, which lies in both the Northern and the Southern Hemispheres, rainfall spatial patterns are quite heterogeneous across the basin. Therefore, precipitation exhibits opposite rainfall regimes between the northern and southern Amazon (e.g., Espinoza et al. 2009a; Liebmann and Mechoso 2011). These opposing rainfall phases are also evident in the onset and demise dates of the wet season (Fig. 3). There is a clear spatial variability of the wet-season demise dates, with earlier demise dates being observed from the northwest toward the southeast. According to Fig. 3b, the wet season ends between 22 March and 10 April in the southeastern Amazon and after mid-July in the northwestern Amazon. The beginning of the wet season exhibits a more complex spatial pattern, with earlier onset dates following a diagonal pattern, from northwest to southeast, mostly between the equator and 7.5°S in the western Amazon and between 5.0° and 12.5°S in the eastern Amazon (lighter colors of Fig. 3a). For most of this area, these earlier onsets occur between 12 and 22 October. Progressively later onset dates are usually observed in adjacent diagonal patterns following these earlier onsets. The latest onset dates are located over the northern Amazon. Over the northeastern Amazon the onset of the wet season occurs in December, while over the northwestern Amazon the average onset is observed in the beginning of the following year, up until February. In fact, over the northwestern Amazon the wet season prevails during most of the year, lasting more than 9 months each year, on average, and therefore there is no meaning in defining onset and demise dates for this region. In general, the eastern and southeastern Amazon present shorter wet seasons, lasting from 140 up to 200 days.

Fig. 3.

Spatial distribution of the average (a) onset [day of year (DoY)], (b) demise (DoY), and (c) duration (days) of the Amazon’s wet season between 1983 and 2009. The onset and end of the wet season are defined according to the methodology proposed in section 2.

Fig. 3.

Spatial distribution of the average (a) onset [day of year (DoY)], (b) demise (DoY), and (c) duration (days) of the Amazon’s wet season between 1983 and 2009. The onset and end of the wet season are defined according to the methodology proposed in section 2.

The similarity between the onset and demise spatial patterns obtained in this work with previous studies that relied on completely independent approaches to determine the beginning and withdrawal dates of the wet season in the Amazon is remarkable. Kousky (1988) defined the onset date of the wet season (from 1979 to 1987) as the date when pentad-averaged OLR was below 240 W m−2. It was also required that the OLR in 10 out of 12 pentads before the onset were above this threshold and that in 10 out of 12 pentads after the onset OLR were below this threshold. Using rain gauge observations (from 1979 to 1996), Marengo et al. (2001) defined the onset as the pentad in which rainfall exceeds 4 mm day−1, provided that in 6 out of 8 preceding pentads precipitation was below 3.5 mm day−1 and that in 6 out of 8 pentads after the onset precipitation was larger than 4.5 mm day−1. Demise dates were defined analogously. As northwestern Amazon (west of 60°W and north of 5°S) presents a less defined annual cycle, the required persistence and threshold criteria to identify onset and demise dates are not met by these rainfall and OLR-based methodologies. Additionally, because of the scarce rain gauge data Marengo et al. (2001) do not provide estimates east of 52°W. The southeast–northwest evolution of rainy-season demise dates observed in the present work is also consistent with these previous studies. Therefore, our results show good agreement with estimates based on other datasets, methods, and periods. For example, the end of the wet season over the southeast Amazon occurred from 16 to 20 April according to Kousky (1988), from 20 April to 10 May according to Marengo et al. (2001), and between 21 March and 10 April in this work. The spatial patterns of the onset dates obtained by these two previous works also closely resemble the patterns shown in Fig. 3a. It is interesting to notice the evolution of the mean onset dates in the study region as the time series analyzed progressively include more recent time periods. For example, in general, the earliest onset dates reported by Kousky (1988) from 1979 to 1987 occurred between 19 August and 12 September; analyzing a longer period (1979–96), Marengo et al. (2001) found that the earliest onset dates occurred later, between 8 and 27 September on average; here we show that the earliest onset dates from 1983 to 2009 occurred even later, between 12 and 22 October. These results indicate a delay in the beginning of the wet season along the years, a topic that is further explored in section 3c.

To evaluate regional differences in the rainfall regime statistical measures of central tendency and dispersion of the wet-season onset, demise, and length were assessed in four different sectors of the basin: the northwest (NW): 5°S–2.5°N, 62°–74°W; northeast (NE): 5°S–2.5°N, 50°–62°W; southwest (SW): 12.5°–5°S, 62°–74°W; and southeast (SE): 12.5°–5°S, 50°–62°W). Tables 1 and 2 show the mean, median, and 10th and 90th percentiles (P10 and P90) of these dates averaged over these four different regions. According to Table 1, on average the rainy season lasts from 19 October to 15 April in southeastern Amazon, from 23 October to 21 April in southwestern Amazon, from 24 November to 8 June in northeastern Amazon, and from 13 November to 15 July in northwestern Amazon. The differences between mean and median onset and demise dates were small for all regions, except for the northwestern Amazon, where median values indicate that the wet season lasts from 26 October to 2 July. The largest differences between median and mean values are observed in the northern Amazon and indicate that the distribution of the onset, end, and length of the wet season is more asymmetric in this region than in the southern Amazon. The difference of 83 days between the mean and median wet-season length in the northwestern Amazon indicates that only a few grid cells, where wet conditions persist throughout the year, contribute to the elevated mean wet-season length in this region. The smallest difference between P10 and P90 observed in southern Amazon indicates that this region presents less variability in the wet-season onset and end dates compared to the northern Amazon. The frequency distributions of the onset, demise, and length are broader over the northwestern Amazon than in other regions. This is likely due to the higher cloudiness observed over the area throughout the year, which restricts the ability of the method to accurately determine the wet season. The same issues are observed when rain gauges or OLR are used. On the other hand, the rainy season is very well defined over the southwestern Amazon, where the differences between P10 and P90 are only 36 and 47 days for the onset and demise, respectively.

Table 1.

Mean and median of the onset and demise dates and length (days) of the wet season averaged over four different regions in the Amazon. The values in parentheses represent the dates in terms of day of the year.

Mean and median of the onset and demise dates and length (days) of the wet season averaged over four different regions in the Amazon. The values in parentheses represent the dates in terms of day of the year.
Mean and median of the onset and demise dates and length (days) of the wet season averaged over four different regions in the Amazon. The values in parentheses represent the dates in terms of day of the year.
Table 2.

The 10th and 90th percentiles (P10 and P90) of the onset (day of the year), demise (day of the year), and length (days) of the wet season averaged over four different regions in the Amazon. An asterisk indicates a day in the following year.

The 10th and 90th percentiles (P10 and P90) of the onset (day of the year), demise (day of the year), and length (days) of the wet season averaged over four different regions in the Amazon. An asterisk indicates a day in the following year.
The 10th and 90th percentiles (P10 and P90) of the onset (day of the year), demise (day of the year), and length (days) of the wet season averaged over four different regions in the Amazon. An asterisk indicates a day in the following year.

b. Observed tendencies for cloud fraction

To verify how cloudiness has changed over the last decades, linear regressions of pentad-averaged cloud fraction as a function of time were evaluated (Fig. 4). As ISCCP provides cloud fraction every 3 h, each time of the day was analyzed separately to verify possible variations in the diurnal cycle of convection. To avoid the influence of outliers in the calculation of trends, the Sen’s slope estimator was applied to the time series. The Mann–Kendall test was used to verify the significance of the trends. Only statistically significant results (p value < 0.05) were considered. The negative trends shown in Fig. 4 evidence an overall decrease in cloud fraction over the Amazon basin at all times analyzed. The largest reduction in cloudiness is observed at 1200 UTC (~6% decade−1) over the central and eastern Amazon. Notice that at that time the entire study area experiences significant decreases in cloud fraction. The most significant trends in cloudiness are observed in the eastern Amazon (east of 62.5°W), regardless of the time. No significant cloudiness trends were observed in the western and extreme eastern Amazon at 1500 UTC (1100 LT). This is likely due to the large variability in cloud fraction observed at that time (not shown) resulting in large dispersion of the data and leading to less significant trends.

Fig. 4.

Spatial distribution of the linear trend of pentad-averaged cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 4.

Spatial distribution of the linear trend of pentad-averaged cloud fraction during the years 1983–2009 for each 3-h period of the day.

The apparent east–west discontinuity around 62.5°W was further investigated. Time series of cloud fraction east and west of the transition do not show discontinuities before and after the dates when GOES-East was replaced (not shown). However, between around mid-1984 and the end of 1985, GOES-East was not operating. Although GOES-West was moved to help coverage, there was still a remaining gap between GOES-West and Meteosat at the time. Two La Niña events occurred in 1984 and 1985 (http://ggweather.com/enso/oni.htm) that are linked to higher than average cloud fraction in Amazonia (e.g., Espinoza et al. 2013). Therefore, the lack of data in the region immediately west of 62.5°W during this period may have caused a slight underestimation of cloud fraction trends in this area. Cloud fraction trends calculated removing the first 4 years of the study period still present an east–west transition (not shown), although it is less sharp, indicating that the missing cloud fraction data have a small, but significant, contribution to the east–west discontinuity at 62.5°W.

Modeling and observational studies indicate that in the morning shallow clouds predominate in the Amazon basin, reducing the amount of incoming solar radiation at the surface and therefore delaying deep convection (e.g., Betts and Jakob 2002; Machado et al. 2002, 2004; Wu et al. 2009; Schiro et al. 2016; Zhuang et al. 2017). Convection builds up quickly in the beginning of the afternoon, resulting in deep, high clouds and increased cloud fraction (e.g., Machado et al. 2002, 2004). To further explore changes in the cloudiness regime over the years, we evaluated the contribution of different cloud types to the decreasing cloudiness trends (Fig. 4). Figures 57 show cloud fraction trends at different times in the Amazon basin separated according to cloud type. The results indicate that negative trends in high cloudiness were the major contributors to the decline of total cloud fraction. Figure 5 shows a decline of up to 4% decade−1 in high-cloud fraction at all times analyzed, especially in the central and eastern Amazon (east of 62.5°W). This decreasing trend is more pronounced at 2100, 0000, and 0300 UTC and less pronounced at 0600 UTC, indicating that high-cloud fraction decreases in the afternoon and throughout nighttime. Therefore the significant decrease in total cloud fraction observed in the morning at 1200 UTC is due to a reduction in the remaining cirrus clouds from anvils of deep convective clouds formed during the previous afternoon.

Fig. 5.

Spatial distribution of the linear trend of pentad-averaged high-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 5.

Spatial distribution of the linear trend of pentad-averaged high-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 6.

Spatial distribution of the linear trend of pentad-averaged midlevel-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 6.

Spatial distribution of the linear trend of pentad-averaged midlevel-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 7.

Spatial distribution of the linear trend of pentad-averaged low-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Fig. 7.

Spatial distribution of the linear trend of pentad-averaged low-cloud fraction during the years 1983–2009 for each 3-h period of the day.

Conversely, a mild but statistically significant reduction of less than 2% decade−1 in low-cloud fraction (Fig. 7) is observed over western Amazon (west of 62.5°W) during nighttime and early morning (0000–1200 UTC). During the same period of the day, midcloud fraction show an increasing trend of up to 2.5% decade−1 over the whole basin, except in the northeast Amazon (Fig. 6). This may indicate that cloud vertical development has changed over time, usually favoring the formation of midlevel clouds over low and high clouds. During the afternoon (1500–1800 UTC), the amount of both high and midlevel clouds is reduced over the years in eastern Amazon, while low-cloud fraction increased over the same region. It is important to notice, however, that as sensors aboard geostationary satellites only detect the highest cloud layer, underlying layers may be obscured by higher clouds. Therefore, part of the increase in midlevel and low clouds may be due to the detection of lower layers of clouds that were not detected previously when high cloud cover was larger. The apparent transition between east and west cloudiness trends (Fig. 4) likely results from the significant reduction of high clouds east of 62.5°W (Fig. 5), whereas west of this longitude the reduction in cloud fraction is mainly explained by low-level clouds, which show a much lower decreasing trend (Fig. 7). That is, over the more humid western Amazon, where on average cloud fraction is higher and cloudiness variability is lower than in other regions (Fig. 8), there has been no significant change in the frequency of occurrence of deep clouds.

Fig. 8.

Spatial distribution of the (a) mean and (b) standard deviation of cloud fraction at 1200 UTC from 1983 to 2009 over the Brazilian Amazon.

Fig. 8.

Spatial distribution of the (a) mean and (b) standard deviation of cloud fraction at 1200 UTC from 1983 to 2009 over the Brazilian Amazon.

ISCCP cloud fraction variability over the Amazon basin has been previously analyzed, but the diurnal cycle of clouds was not taken into account (Butt et al. 2009; Arias et al. 2011). Arias et al. (2011) showed a reduction in cloud fraction over the southern Amazon throughout the year, reaching a maximum negative trend of up to −6% decade−1 during austral summer. The same study also showed reduced cloudiness in the northern Amazon, more prominently during austral spring. By analyzing five different regions in Amazon, Butt et al. (2009) reported a decrease in cloudiness of −0.3% yr−1 during the dry season and a modest increase of 0.1% yr−1 during the wet season. They identified a decrease of up to 5% decade−1 in the east-central Amazon in November and up to 7% decade−1 in the southeastern Amazon in August. According to their study decreasing trends of cloud fraction were observed during all months in both of these regions. Previous studies that analyzed rainfall and hydrological trends are also compared with our findings. Espinoza et al. (2009a) estimated a mean rainfall decrease of −0.32% yr−1 over the entire Amazon basin from 1975 to 2003, emphasizing that this decreasing trend was more pronounced after 1982. Long-term measurements of the annual minimum river discharges in southern Amazon show a decrease of −0.61% yr−1 from 1967 to 2013 (Molina-Carpio et al. 2017). These results are consistent with the decreasing cloud fraction trends found in the present work at different times of the day (Fig. 4). Regional-scale observations and a hydrological model forced with reanalysis (Espinoza et al. 2009b; Wongchuig-Correa et al. 2017) also identified a downward trend in minimum annual river discharge, especially over the southern and eastern Amazon. Mean and maximum annual discharges show opposite trends, increasing in the northwestern Amazon and decreasing in the south and southwestern Amazon (Espinoza et al. 2009b; Wongchuig-Correa et al. 2017; Gloor et al. 2013). Long-term observations of rainfall patterns show decadal-time-scale variations in precipitation with opposing phases in the southern and northern Amazon (Marengo 2004). The smaller rate of decrease in cloud fraction over the western Amazon, shown in Fig. 4, is also consistent with these previous studies.

To further illustrate the temporal variation of cloudiness we show examples of time series of cloud fraction with their respective trends at 1200 UTC averaged over the four different Amazonian regions as defined in the previous section (Fig. 9). Figure 9 indicates that the decreasing trends in cloudiness are more pronounced in the eastern Amazon, confirming the findings discussed previously. It is also clear that the seasonal variability of precipitation is more pronounced over the southern than over the northern Amazon, a result that also agrees with previous studies (e.g., Liebmann and Mechoso 2011).

Fig. 9.

Smoothed time series of the cloud fraction at 1200 UTC and its trends from 1983 to 2009 averaged over four regions in the Amazon.

Fig. 9.

Smoothed time series of the cloud fraction at 1200 UTC and its trends from 1983 to 2009 averaged over four regions in the Amazon.

To verify whether the reduction in cloudiness is seasonally dependent, the spatial distributions of the trends of the mean cloud fraction at 1200 UTC during the wet and dry seasons were analyzed (Figs. 10a,b). The results indicate that more pronounced trends are observed in southeastern Amazon during the wet season, whereas in the northeastern Amazon larger trends are observed during the dry season. More pronounced trends were observed for the minimum rather than for the maximum cloud fraction (Figs. 10c,d).

Fig. 10.

Spatial distribution of the trends of the mean cloud fraction at 1200 UTC during the (a) wet and (b) dry seasons from 1983 to 2009. Also shown are trends of the (c) maximum and (d) minimum cloud fraction at 1200 UTC during the same study period. The Sen’s slope estimator was applied to the time series and the Mann–Kendall test was used to verify the significance of the trends. Only statistically significant results (p value < 0.05) were considered.

Fig. 10.

Spatial distribution of the trends of the mean cloud fraction at 1200 UTC during the (a) wet and (b) dry seasons from 1983 to 2009. Also shown are trends of the (c) maximum and (d) minimum cloud fraction at 1200 UTC during the same study period. The Sen’s slope estimator was applied to the time series and the Mann–Kendall test was used to verify the significance of the trends. Only statistically significant results (p value < 0.05) were considered.

Although it has been suggested that long-term global trends of cloud amount may be due to viewing geometry artifacts (Evan et al. 2007), several independent analyses have shown that cloudiness is, indeed, decreasing. Ground-based measurements of cloud cover indicate that global cloudiness has decreased since the 1970s (Norris 2005). Using observations from meteorological stations from 1971 to 1996, Warren et al. (2007) observed a reduction of −2.2% decade−1 in cloudiness over South America as well as a decline of −0.7% decade−1 in global cloudiness. A comparison between monthly averaged high-cloud-cover retrievals from ISCCP and OLR showed a consistent reduction of both variables over the Amazon basin along the years (Arias et al. 2011). Arias et al. (2011) also pointed out that satellite viewing angle does not seem to systematically reduce total and high-cloud amount over the Amazon, since cloudiness changes mostly occurred farther from the edges of the satellite field of view. They also show that estimates of low-level clouds over oceans, along the edge of the satellite field of view, are affected by satellite viewing angle change. However, since this work only focuses on cloud retrievals over land, low-level-cloud trends are not expected to be significantly affected by satellite geometry in the study region. Besides, as will be shown in section 3d, thermodynamic conditions and surface fluxes, obtained using completely independent data sources, are consistent with a decrease in cloud fraction over the area. The use of reanalysis data and other physical parameters to corroborate satellite observations of cloud variability (section 3d) is, in fact, recommended (Norris 2001; Norris and Slingo 2009) and provides us confidence on the overall results obtained in this work.

c. Wet-season onset and length trends

Figure 11 shows the spatial distributions of the yearly trend of the wet-season onset, end, and duration, calculated using the time series of daily cloud fraction at 1200 UTC according to the methodology explained in section 2. Once again, only statistically significant trends (p values < 0.05) are considered. A delay of the onset of the wet season is observed (Fig. 11a), especially east of 62.5°W. The northern and central Amazon present the largest positive onset trends, with delays of up to 4 days yr−1 in the extreme central-northern Amazon. In the southern Amazon the onset of the wet season is also delayed, at a smaller but statistically significant rate of about 1 day yr−1. For most of this area the demise of the wet season did not change significantly, except over a small area in the southwestern Amazon and Rondonia, where a weak negative trend indicates that the end of the wet season is being anticipated at a rate of about −1 day yr−1. The combination of late onset and earlier demise dates lead to a significant and generalized decrease in the duration of the wet season especially over the eastern Amazon. In the northeastern Amazon the duration of the rainy season decreased at high rates, from −4 to −3 days yr−1. These results indicate that a high-paced drying of the Amazon has been occurring since the last two to three decades. Although several studies have observed variations in the Amazonian rainfall regime, those which report estimates of precipitation or cloudiness trends have focused on the southwestern Amazon (Butt et al. 2011; Fu et al. 2013). Using rainfall measurements from meteorological stations, Butt et al. (2011) observed a trend of delay of −0.6 days yr−1 for the wet-season onset in the state of Rondonia in the southwestern Amazon. This value is in good agreement with the trend observed in Fig. 11a, in the surroundings of the state, within 8°–12°S, 60°–65°W. Debortoli et al. (2015) analyzed rainfall patterns from rain gauges located in the southern Amazon from 1971 to 2010. According to this study 88% of the rain gauges indicated a shortening of the wet season, due to a combination of delayed onset and premature demise dates. Fu et al. (2013) observed an increase in the dry-season length of 6.5 ± 2.5 days decade−1 in the southern Amazon. Although the present study has focused on the wet-season (instead of the dry-season) length, this value seems to be compatible with the ones shown in Fig. 11c for the same area. Figure 11 shows that, despite the massive deforestation observed in Rondonia in the last decades, the trend of the onset and length of the rainy season is more significant in other areas, such as the northeastern Amazon. The strong tendencies observed here suggest a shift in the seasonal hydrological cycle in the Amazon, with unknown impacts on climate, the ecosystem, and agriculture. As previously pointed out in section 3b, several recent studies indicate a decrease in rainfall and minimum annual river discharge trends in the Amazon over the years, in accordance with the results obtained in this work (Espinoza et al. 2009a; Marengo and Espinoza 2016, Espinoza et al. 2009b; Wongchuig-Correa et al. 2017; Molina-Carpio et al. 2017).

Fig. 11.

Spatial distribution of the trends of the onset, end, and duration of the wet season during the years 1983–2009. A significant reduction in the wet-season length is observed east of 62.5°W, mainly caused by a delay of the onset in this period.

Fig. 11.

Spatial distribution of the trends of the onset, end, and duration of the wet season during the years 1983–2009. A significant reduction in the wet-season length is observed east of 62.5°W, mainly caused by a delay of the onset in this period.

d. Association between cloudiness and other atmospheric variables

Previous studies have shown a consistent trend of increasing sea surface temperature in large ocean basins (Parker et al. 2007; Grimm and Saboia 2015). The combination of reduced cloudiness and warmer SST may alter thermodynamic and dynamic conditions, modifying atmospheric properties and surface heat partitioning (e.g., Bony et al. 2015). To explore these issues we calculated trends of several atmospheric variables and surface fluxes, using the monthly averaged ERA-Interim (Dee et al. 2011) from 1983 to 2009 in the study region (Figs. 12 and 13).

Fig. 12.

Linear trends, computed using monthly averages, of (a) sensible heat flux, (b) latent heat flux, (c) temperature at the surface, and (d) CAPE.

Fig. 12.

Linear trends, computed using monthly averages, of (a) sensible heat flux, (b) latent heat flux, (c) temperature at the surface, and (d) CAPE.

Fig. 13.

Linear trends, computed using monthly averages, of (a) zonal wind at 700 hPa, (b) meridional wind at 700 hPa, (c) specific humidity at the surface, and (d) relative humidity at the surface.

Fig. 13.

Linear trends, computed using monthly averages, of (a) zonal wind at 700 hPa, (b) meridional wind at 700 hPa, (c) specific humidity at the surface, and (d) relative humidity at the surface.

As cloud fraction decreases, the amount of solar radiation reaching the surface increases, modifying the energy heat flux partition. Figures 12a and 12b show that the surface sensible heat flux has increased over the Amazon basin, whereas the surface latent heat flux has decreased. Over eastern Amazon SHF has increased at a rate of up to 0.3 W m−2 yr−1. A similar magnitude of decrease in LHF is observed in the northeastern and south-central Amazon. These findings are consistent with previous studies that showed that decreases in rainfall during the transition period lower surface latent heat flux, delaying the beginning of the wet season (Li and Fu 2004). Changes in sensible and latent heat flux due to deforestation have also been linked to the delay of the wet-season onset in the southwestern Amazon (Butt et al. 2011).

The atmospheric temperature profile is also altered as a response to reduced cloudiness (Figs. 12c). Surface temperature increased, especially in the eastern and southern Amazon, an area severely deforested over the last decades (e.g., Laurance et al. 2004; Fearnside 2008), where a positive trend of 0.05°C yr−1 is observed for that variable. The increase in surface temperature and sensible heat flux are likely associated to a widespread increase of CAPE over the Amazon basin, possibly explaining the positive trend of up to 10 J kg−1 yr−1 in the central Amazon (Fig. 12d).

Dynamic features, such as wind patterns, also varied during the analyzed period. Figures 13a and 13b show that the zonal component of the wind velocity decreased at rates of up to −0.1 m s−1 yr at 700 hPa, particularly in the eastern Amazon. As suggested by Gloor et al. (2013), this intensification of trade winds may be related to an increased water vapor inflow from the Atlantic Ocean to the Amazon basin, modifying circulation patterns and increasing the specific humidity over the region (Fig. 13c). The trends of the meridional component of the wind at 700 hPa showed opposite signs depending on the region analyzed. More intense northerly winds were observed in the northwestern Amazon whereas the southeastern Amazon has experienced a weakening of northerly winds along the years. These changes in circulation patterns are likely due to modifications in the position and intensity of the Hadley cell, which have also been linked to the shortening of the American monsoons (Yoon and Zeng 2010; Arias et al. 2015; Espinoza et al. 2016). Although the specific humidity has increased over the years (Fig. 13c), RH at the surface decreased at a rate of up to −0.25% yr−1 (Fig. 13d). A reduction in RH may be associated to the cloudiness reduction, previously observed in section 3b, as more lifting is needed to induce cloud formation in a lower-RH environment.

In the previous sections we showed evidence of a persistent reduction in cloudiness over the Amazon basin that appeared related to delayed onsets and a shortening of the rainy season. Although the reduction in deep clouds reported in this study (Fig. 5) may lead to less precipitation over the area, cloudiness reduction itself could be the response to variations in other variables responsible for the delay of the wet-season onset. Understanding the interplay between changes in land and atmospheric processes, the development of convection, and rainfall regime is challenging. Simulations using a coupled atmosphere–land model show that an increase in surface albedo may lead to more or less rainfall, depending on other vegetation properties such as roughness and canopy height, which could modify the amount of solar radiation reaching the ground (Doughty et al. 2012). During the study period from 1983 to 2009, an immense area of approximately 484 000 km2 was deforested in the so-called legal Amazonia (http://www.obt.inpe.br/prodes/index.php). It is estimated that the increase in surface albedo due to deforestation leads to an annual radiative forcing of −7.3 W m−2 over a deforested area in the southern Amazon (Sena et al. 2013). The seasonal impact of biomass burning aerosols also affects the radiative balance over the study area significantly. Regional climate models show that high emission of biomass burning aerosols may increase thermodynamic stability, affecting the onset of the South American monsoon system (Zhang et al. 2009). During the months of August and September the average direct effect of aerosols on shortwave radiation over the Brazilian Amazon basin was −5.3 W m−2 (Sena and Artaxo 2015). The combined effects of deforestation and aerosol on the radiation balance of the study area is high, leading to cooling near the surface and heating in the top of the boundary layer due to solar radiation absorption by aerosol (Davidi et al. 2009), modifying the energy available for cloud formation. On the other hand, the decrease in cloudiness itself also modifies the surface heat balance, triggering an increase in surface temperature, as shown in Fig. 12. The shortening of the wet season, reported in this study, can lead to drier vegetation in the long term, which in turn may impact cloud formation and development through land–atmosphere interactions (e.g., Werth and Avissar 2002; Durieux et al. 2003; Heiblum et al. 2014; Wright et al. 2017). Observational studies show that, in the absence of anthropogenic aerosols, deep clouds are observed mainly over preserved forest areas (Heiblum et al. 2014; Wang et al. 2009). Reduced rain forest evapotranspiration could inhibit the development of shallow clouds, acting as a moisture pump for deep convection and destabilizing the atmosphere during the initial stages of the dry-to-wet-season transition period in this region (Wright et al. 2017), leading to delayed onset and a shortening of the wet season. Although the thermodynamic variations shown in Figs. 12 and 13 may be associated with a reduction in cloudiness, future work exploring the influence of dynamics in cloud fraction trends is suggested.

e. Relationships with coupled climate modes

Correlations between the detrended time series of climatic coupled mode indices and the yearly minimum and maximum cloud fraction (fcmin and fcmax) are shown in Fig. 14. Only variables that best correlated with fcmin (fcmax) and cells where the correlation was statistically significant (p values < 0.05) are shown. Similar analyses were performed for the onset, end, and duration of the wet season (Figs. 15 and 16). The aim of these analyses is to verify climatic coupled modes associations with the wet-season onset and demise date (and fcmin and fcmax). Since detrended annual averages of the climatic indices are used, both the phase and intensity of the climatic modes are taken into account in these correlations.

Fig. 14.

Spatial distribution of the correlation R between (a) the minimum cloud fraction (fcmin) and ENSO-3.4, (b) the maximum cloud fraction (fcmax) and ENSO-3.4, and (c) the minimum cloud fraction (fcmin) and the North Atlantic tripole of SST. The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Fig. 14.

Spatial distribution of the correlation R between (a) the minimum cloud fraction (fcmin) and ENSO-3.4, (b) the maximum cloud fraction (fcmax) and ENSO-3.4, and (c) the minimum cloud fraction (fcmin) and the North Atlantic tripole of SST. The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Fig. 15.

Spatial distribution of the correlation R between the wet-season onset and (a) ENSO-3.4, (b) the area of the Pacific warm pool (PWP), and (c) the anomaly of SST of the North Atlantic tripole. The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Fig. 15.

Spatial distribution of the correlation R between the wet-season onset and (a) ENSO-3.4, (b) the area of the Pacific warm pool (PWP), and (c) the anomaly of SST of the North Atlantic tripole. The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Fig. 16.

Spatial distribution of the correlation R between the wet-season demise and (a) ENSO-3.4, and (b) the anomaly of SST of the tropical South Atlantic (TSA). The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown. The Pacific warm pool did not show significant correlation with the wet-season demise.

Fig. 16.

Spatial distribution of the correlation R between the wet-season demise and (a) ENSO-3.4, and (b) the anomaly of SST of the tropical South Atlantic (TSA). The correlation was computed using detrended time series of the climatic coupled modes. Only cells where the correlation was statistically significant (p values < 0.05) are shown. The Pacific warm pool did not show significant correlation with the wet-season demise.

Figure 14 indicates that positive phases of ENSO-3.4 are associated with a decrease in the maximum and minimum cloud fraction observed over the northern Amazon along the years. Over the state of Rondonia in the southeastern Amazon, ENSO-3.4 correlates positively with fcmin, showing a possible link between the occurrence of El Niño and a decrease in the minimum cloud fraction over that region. The North Atlantic tripole of SST also influences fcmin, especially in the central and southwestern Amazon, where negative correlations between these variables are observed.

ENSO-3.4, the Pacific warm pool, and the North Atlantic tripole also seem to influence the onset date of the rainy season, as indicated by the positive correlations between these variables (Figs. 15a–c). Therefore, delayed onsets of the wet season are likely related to the occurrence of El Niño events, larger areas of the Pacific warm pool, and positive anomalies of SST in the North Atlantic tripole. The influence of the ENSO-3.4 on the wet-season onset is observed over the eastern Amazon, whereas the Pacific warm pool influences the northern Amazon and the North Atlantic tripole impacts a small area of the western Amazon.

ENSO appears as the dominant factor modulating the characteristics of the wet season. Negative correlations between ENSO-3.4 and the withdrawal of the wet season are observed over a large area of the basin, except the northern and southeastern Amazon (Fig. 16a). These results indicate that, over these areas, the wet season ends earlier during El Niño years. TSA, on the other hand, correlates positively with wet-season demise date over some regions of the western Amazon (Fig. 16b), suggesting that higher SST in the tropical South Atlantic may be related to delays in wet-season demise. As the TSA and ENSO have concurring impacts on the end of the Amazon’s wet season, the resulting trend in the demise date is very weak, as shown previously in Fig. 11b. The Pacific warm pool did not show significant correlation with the wet-season demise.

Because ENSO influences both the onset and the demise date of the wet season, ENSO’s most significant impact on the wet-season length is observed particularly in the central and eastern parts of the basin (Fig. 17). Previous studies have also proposed a possible link between wet-season onset/length and variations associated with these and other climate coupled modes. Positive (negative) phases of ENSO are commonly regarded as possible drivers of Amazonian droughts (floods) (e.g., Marengo et al. 2008; Ronchail et al. 2006; Espinoza et al. 2013). The results shown in these previous studies are consistent with the findings of Figs. 15a, 16a, and 17, where the occurrence of El Niño (La Niña) is related to delayed (earlier) onsets, earlier (delayed) demises, and shorter (longer) wet seasons, therefore indicating drought (flood) scenarios. It is worth emphasizing that the signals shown in Figs. 1517 are observed because ENSO modulates the interannual rainfall variability in the Amazon. However, the long-term variability that is evident from the linear trends of cloudiness and wet-season onset/demise (Figs. 4 and 11) cannot be totally explained by ENSO and is likely due to a combination of several other factors of natural and anthropogenic origin. To objectively address this question, modeling and observational studies that comprise much longer periods (ideally to cover more AMO and PDO phases) are needed.

Fig. 17.

Spatial distribution of the correlation between wet-season length and detrended time series of ENSO-3.4. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Fig. 17.

Spatial distribution of the correlation between wet-season length and detrended time series of ENSO-3.4. Only cells where the correlation was statistically significant (p values < 0.05) are shown.

Temperature anomalies in the Atlantic Ocean also have a significant influence on precipitation in South America. A warmer tropical North Atlantic is associated with subsidence over the Amazon basin that leads to decreased moisture convergence flux, reducing rainfall over the area (Yoon and Zeng 2010). This mechanism can lead to severe droughts, such as the one observed in 2005 (e.g., Marengo et al. 2008; Zeng et al. 2008; Cox et al. 2008; Espinoza et al. 2011). Yoon and Zeng (2010) state that tropical North Atlantic SST influences on rainfall in Amazon are comparable to that of Pacific SST, whereas the tropical South Atlantic SST shows a weaker but significant influence on rainfall during wet-to-dry-season transition. Recent flood events in Amazon were attributed to warmer tropical South Atlantic in the absence of El Niño (Marengo et al. 2010; Satyamurty et al. 2013; Espinoza et al. 2014). These previous results agree with the positive correlation between TSA and wet-season end found in this study (Fig. 16b), which may imply longer wet seasons and possible floods. An increase in the area of the Pacific warm pool has been quoted as a possible cause of reduced cloudiness over Amazon (Arias et al. 2011) and therefore is likely associated to delayed onsets of the wet season as suggested by Fig. 15b. It is important to emphasize that coupled modes and land-use change are not the only factors influencing the interannual variability of the wet season. Recent studies show that the wet-season onset results from complex competing processes that are nonlinearly related to SST anomalies (Fu et al. 2013; Yin et al. 2014; Arias et al. 2015). In southern Amazon delays of the dry-to-wet-season transition have been also related to 1) a smaller CAPE, 2) a poleward shift of the Southern Hemisphere subtropical jet, and 3) an enhanced South American low-level jet (Yin et al. 2014). The equatorial contraction of the intertropical convergence zone (ITCZ) over the east Pacific–American–Atlantic sector and the westward expansion of the North Atlantic subtropical high (NASH) also contribute to delayed wet-season onsets over the Amazon basin (Arias et al. 2015; Fu et al. 2013). These processes are modulated by SST warming and climatic modes (Yin et al. 2014; Arias et al. 2015).

4. Summary and conclusions

A new methodology to assess the rainy-season onset and length in the Brazilian Amazon, based on geostationary satellite retrievals of cloud fraction, is proposed. The advantage of this method is that it does not rely on rain gauges (which are very sparse in the region). Previous studies that reported trends on wet- and dry-season lengths have focused on the analysis of climatological variations of precipitation over the southwestern Amazon (Butt et al. 2011; Fu et al. 2013), a region that has been severely deforested over the last decades. However this study shows that the trend of the shortening of the rainy season is more significant over the northeast Amazon, where a reduction of up to 4 days yr−1 in the wet season is observed. This result suggests that large-scale influences in rainfall regimes in Amazon are likely more important than local influences, such as the rate of deforestation. Nevertheless, modeling studies are necessary to further investigate these issues.

Perhaps more relevant, this study shows a consistent and widespread decline in cloudiness over Amazon from 1983 to 2009. Cloud fraction has reduced regardless of the time of the day or location, with more prominent trends of up to −6% decade−1 at 1200 UTC (0800 LT) over central and eastern Amazon. We show that most of the reduction in total cloud fraction is caused by a decrease in the amount of high clouds, especially in the central and eastern Amazon. In particular, high cloud occurrence decreases during the afternoon and throughout the night. In the morning, total cloud fraction is still reduced likely due to the decrease in cirrus debris from anvils of deep convective clouds formed during the previous day. The frequency of occurrence of deep clouds over the forested and wetter western Amazon has not varied significantly; however, a small, but significant, reduction in low-level clouds is observed in this region. The results also suggest a modification in cloud vertical development in the last decades, favoring the formation of midlevel clouds against low and high clouds. To confirm this hypothesis, however, further observational studies that use different techniques to detect cloud type are recommended, since high clouds observed by passive sensors aboard satellites may obscure underlying cloud layers. The significant reduction in wet-season length and cloudiness over the Amazon basin found in this study shows good agreement with recent studies that reveal a decrease in rainfall and river discharge trends in the region over the years (e.g., Espinoza et al. 2009a,b; Marengo and Espinoza 2016).

Less cloud cover over the Amazon basin lead to higher incoming solar radiation at the surface, warming the surface and modifying surface heat fluxes and the atmospheric thermodynamic profile (Figs. 12 and 13). However, additional studies are necessary to infer the feedbacks between decreased cloudiness and shortening of the wet season. The reduction in cloudiness and precipitation and in the wet-season length may enhance dry environmental conditions, ultimately contributing to delayed onsets and shorter wet seasons (Wright et al. 2017). High SST anomalies in the equatorial Atlantic Ocean enhance subsidence over the Amazon, decreasing cloudiness and modifying the rainfall regime (Yoon and Zeng 2010). We show that trade winds have intensified in central and eastern Amazon and northerly winds have intensified over the northwestern Amazon (Fig. 13), likely associated with changes in the position and intensity of the Hadley cell (Yoon and Zeng 2010; Arias et al. 2015; Espinoza et al. 2016). These changes in circulation could lead to an increase in the moisture flow over the Amazon basin. The results indicate a reduction in surface RH along the years, which may be associated to the cloudiness reduction, since in a lower RH environment more lifting is needed to cloud formation. Although these thermodynamic conditions may influence cloud development over the region future work exploring the influence of dynamics in cloud fraction trends is suggested.

Correlation analyses indicate that positive phases of ENSO influence the interannual variability of the rainfall regime in Amazon, leading to delayed onsets and earlier demises of the wet season. Larger areas of the Pacific warm pool are associated to delayed onsets over the northern Amazon. The North Atlantic tripole is also linked to delayed onsets in a small region of the western Amazon, whereas TSA is associated to later rainy-season demise dates over the western and central parts of the basin. However, the interannual variability of the rainy season in Amazonia is the result of complex competing processes that are nonlinearly related to SST anomalies (Fu et al. 2013; Yin et al. 2014; Arias et al. 2015). Therefore, the trends in cloudiness and wet-season onset/demise (Figs. 4 and 11) cannot be completely attributed to ENSO and other climatic modes and are likely caused by a combination of several other physical mechanisms of natural and anthropogenic origin, such as land-use change, global warming, and impacts on the hydrological cycle in South America. Modeling and observational studies comprising much longer periods (ideally to cover more AMO and PDO phases) are needed. The correct simulation of these coupled modes is fundamental to properly predict the future of the Amazon. The critical changes in cloud life cycle evidenced in this study may have considerable impacts in South America’s hydrological cycle and climate, with important implications for the ecosystem and agriculture.

Acknowledgments

The ISCCP D1 data were obtained from the International Satellite Cloud Climatology Project web site https://isccp.giss.nasa.gov maintained by the ISCCP research group at the NASA Goddard Institute for Space Studies, New York, NY. This work was funded by FAPESP Grants 2016/12342-1 and 2017/17047-0 and by CNPq Grant 168160/2017-0.

REFERENCES

REFERENCES
Arias
,
P. A.
,
R.
Fu
,
C. D.
Hoyos
,
W.
Li
, and
L.
Zhou
,
2011
:
Changes in cloudiness over the Amazon rainforests during the last two decades: Diagnostic and potential causes
.
Climate Dyn.
,
37
,
1151
1164
, https://doi.org/10.1007/s00382-010-0903-2.
Arias
,
P. A.
,
R.
Fu
,
C.
Vera
, and
M.
Rojas
,
2015
:
A correlated shortening of the North and South American monsoon seasons in the past few decades
.
Climate Dyn.
,
45
,
3183
3203
, https://doi.org/10.1007/s00382-015-2533-1.
Arraut
,
J. R.
, and
P.
Satyamurty
,
2009
:
Precipitation and water vapor transport in the Southern Hemisphere with emphasis on the South American region
.
J. Appl. Meteor. Climatol.
,
48
,
1902
1912
, https://doi.org/10.1175/2009JAMC2030.1.
Betts
,
A. K.
, and
C.
Jakob
,
2002
:
Study of diurnal cycle of convective precipitation over Amazonia using a single column model
.
J. Geophys. Res.
,
107
,
4732
, https://doi.org/10.1029/2002JD002264.
Boers
,
N.
,
N.
Marwan
,
H. M.
Barbosa
, and
J.
Kurths
,
2017
:
A deforestation-induced tipping point for the South American monsoon system
.
Sci. Rep.
,
7
,
41489
, https://doi.org/10.1038/srep41489.
Bombardi
,
R. J.
, and
L. M. V.
Carvalho
,
2009
:
IPCC global coupled climate model simulations of the South America Monsoon System
.
Climate Dyn.
,
33
,
893
916
, https://doi.org/10.1007/s00382-008-0488-1.
Bony
,
S.
, and Coauthors
,
2015
:
Clouds, circulation and climate sensitivity
.
Nat. Geosci.
,
8
,
261
268
, https://doi.org/10.1038/ngeo2398.
Butt
,
N.
,
M.
New
,
G.
Lizcano
, and
Y.
Malhi
,
2009
:
Spatial patterns and recent trends in cloud fraction and cloud-related diffuse radiation in Amazonia
.
J. Geophys. Res.
,
114
,
D21104
, https://doi.org/10.1029/2009JD012217.
Butt
,
N.
,
P. A.
de Oliveira
, and
M. H.
Costa
,
2011
:
Evidence that deforestation affects the onset of the rainy season in Rondonia, Brazil
.
J. Geophys. Res.
,
116
,
D11120
, https://doi.org/10.1029/2010JD015174.
Cai
,
W.
, and Coauthors
,
2014
:
Increasing frequency of extreme El Niño events due to greenhouse warming
.
Nat. Climate Change
,
4
,
111
116
, https://doi.org/10.1038/nclimate2100.
Cox
,
P. M.
,
R. A.
Betts
,
M.
Collins
,
P. P.
Harris
,
C.
Huntingford
, and
C. D.
Jones
,
2004
:
Amazonian forest dieback under climate-carbon cycle projections for the 21st century
.
Theor. Appl. Climatol.
,
78
,
137
156
.
Cox
,
P. M.
, and Coauthors
,
2008
:
Increasing risk of Amazonian drought due to decreasing aerosol pollution
.
Nature
,
453
,
212
215
, https://doi.org/10.1038/nature06960.
Davidi
,
A.
,
I.
Koren
, and
L.
Remer
,
2009
:
Direct measurements of the effect of biomass burning over the Amazon on the atmospheric temperature profile
.
Atmos. Chem. Phys.
,
9
,
8211
8221
, https://doi.org/10.5194/acp-9-8211-2009.
Davidson
,
E. A.
, and Coauthors
,
2012
:
The Amazon basin in transition
.
Nature
,
481
,
321
328
, https://doi.org/10.1038/nature10717.
Debortoli
,
N. S.
,
V.
Dubreuil
,
B.
Funatsu
,
F.
Delahaye
,
C.
Henke de Oliveira
,
S.
Rodrigues-Filho
,
C.
Hiroo Saito
, and
R.
Fetter
,
2015
:
Rainfall patterns in the Southern Amazon: A chronological perspective (1971–2010)
.
Climatic Change
,
132
,
251
264
, https://doi.org/10.1007/s10584-015-1415-1.
Dee
,
D. P.
, and Coauthors
,
2011
:
The ERA-Interim reanalysis: Configuration and performance of the data assimilation system
.
Quart. J. Roy. Meteor. Soc.
,
137
,
553
597
, https://doi.org/10.1002/qj.828.
Deser
,
C.
, and
M. S.
Timlin
,
1997
:
Atmosphere–ocean interaction on weekly timescales in the North Atlantic and Pacific
.
J. Climate
,
10
,
393
408
, https://doi.org/10.1175/1520-0442(1997)010<0393:AOIOWT>2.0.CO;2.
Doughty
,
C. E.
,
S. R.
Loarie
, and
C. B.
Field
,
2012
:
Theoretical impact of changing albedo on precipitation at the southernmost boundary of the ITCZ in South America
.
Earth Interact.
,
16
, https://doi.org/10.1175/2012EI422.1.
Durieux
,
L.
,
L. A. T.
Machado
, and
H.
Laurent
,
2003
:
The impact of deforestation on cloud cover over the Amazon arc of deforestation
.
Remote Sens. Environ.
,
86
,
132
140
, https://doi.org/10.1016/S0034-4257(03)00095-6.
Enfield
,
D. B.
,
A. M.
Mestas-Nuñez
,
D. A.
Mayer
, and
L.
Cid-Serrano
,
1999
:
How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures?
J. Geophys. Res.
,
104
,
7841
7848
, https://doi.org/10.1029/1998JC900109.
Enfield
,
D. B.
,
A. M.
Mestas-Nuñez
, and
P. J.
Trimble
,
2001
:
The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S
.
Geophys. Res. Lett.
,
28
,
2077
2080
, https://doi.org/10.1029/2000GL012745.
Espinoza
,
J. C.
, and Coauthors
,
2009a
:
Spatiotemporal rainfall variability in the Amazon Basin Countries (Brazil, Peru, Bolivia, Colombia and Ecuador)
.
Int. J. Climatol.
,
29
,
1574
1594
, https://doi.org/10.1002/joc.1791.
Espinoza
,
J. C.
, and Coauthors
,
2009b
:
Contrasting regional discharge evolutions in the Amazon basin (1974–2004)
.
J. Hydrol.
,
375
,
297
311
, https://doi.org/10.1016/j.jhydrol.2009.03.004.
Espinoza
,
J. C.
,
J.
Ronchail
,
J. L.
Guyot
,
C.
Junquas
,
P.
Vauchel
,
W.
Lavado
,
G.
Drapeau
, and
R.
Pombosa
,
2011
:
Climate variability and extreme drought in the upper Solimões River (western Amazon Basin): Understanding the exceptional 2010 drought
.
Geophys. Res. Lett.
,
38
,
L13406
, https://doi.org/10.1029/2011GL047862.
Espinoza
,
J. C.
,
J.
Ronchail
,
F.
Frappart
,
W.
Lavado
,
W.
Santini
, and
J. L.
Guyot
,
2013
:
The major floods in the Amazonas River and tributaries (western Amazon basin) during the 1970–2012 period: A focus on the 2012 flood
.
J. Hydrometeor.
,
14
,
1000
1008
, https://doi.org/10.1175/JHM-D-12-0100.1.
Espinoza
,
J. C.
,
J. A.
Marengo
,
J.
Ronchail
,
J. Molina
Carpio
,
L. Noriega
Flores
, and
J. L.
Guyot
,
2014
:
The extreme 2014 flood in south-western Amazon basin: The role of tropical-subtropical South Atlantic SST gradient
.
Environ. Res. Lett.
,
9
,
124007
, https://doi.org/10.1088/1748-9326/9/12/124007.
Espinoza
,
J. C.
,
H.
Segura
,
J.
Ronchail
,
G.
Drapeau
, and
O.
Gutierrez-Cori
,
2016
:
Evolution of wet-day and dry-day frequency in the western Amazon basin: Relationship with atmospheric circulation and impacts on vegetation
.
Water Resour. Res.
,
52
,
8546
8560
, https://doi.org/10.1002/2016WR019305.
Evan
,
A. T.
,
A. K.
Heidinger
, and
D. J.
Vimont
,
2007
:
Arguments against a physical long-term trend in global ISCCP cloud amounts
.
Geophys. Res. Lett.
,
34
,
L04701
, https://doi.org/10.1029/2006GL028083.
Fearnside
,
P. M.
,
2008
:
The roles and movements of actors in the deforestation of Brazilian Amazonia
.
Ecol. Soc.
,
13
,
23
, https://doi.org/10.5751/ES-02451-130123.
Fernandes
,
K.
, and Coauthors
,
2011
:
North Tropical Atlantic influence on western Amazon fire season variability
.
Geophys. Res. Lett.
,
38
,
L12701
, https://doi.org/10.1029/2011GL047392.
Fu
,
R.
, and
W.
Li
,
2004
:
The influence of the land surface on the transition from dry to wet season in Amazonia
.
Theor. Appl. Climatol.
,
78
,
97
110
, https://doi.org/10.1007/s00704-004-0046-7.
Fu
,
R.
, and Coauthors
,
2013
:
Increased dry-season length over southern Amazonia in recent decades and its implication for future climate projection
.
Proc. Natl. Acad. Sci. USA
,
110
,
18 110
18 115
, https://doi.org/10.1073/pnas.1302584110.
Gloor
,
M.
, and Coauthors
,
2013
:
Intensification of the Amazon hydrological cycle over the last two decades
.
Geophys. Res. Lett.
,
40
,
1729
1733
, https://doi.org/10.1002/grl.50377.
Grimm
,
A. M.
, and
J. P.
Saboia
,
2015
:
Interdecadal variability of the South American precipitation in the monsoon season
.
J. Climate
,
28
,
755
775
, https://doi.org/10.1175/JCLI-D-14-00046.1.
Heiblum
,
R. H.
,
I.
Koren
, and
G.
Feingold
,
2014
:
On the link between Amazonian forest properties and shallow cumulus cloud fields
.
Atmos. Chem. Phys.
,
14
,
6063
6074
, https://doi.org/10.5194/acp-14-6063-2014.
Horel
,
J. D.
,
A. N.
Hahmann
, and
J. E.
Geisler
,
1989
:
An investigation of the annual cycle of convective activity over the tropical Americas
.
J. Climate
,
2
,
1388
1403
, https://doi.org/10.1175/1520-0442(1989)002<1388:AIOTAC>2.0.CO;2.
Kousky
,
V. E.
,
1988
:
Pentad outgoing longwave radiation climatology for the South American sector
.
Rev. Bras. Meteor.
,
3
,
217
231
.
Laurance
,
W. F.
,
A. K. M.
Albernaz
,
P. M.
Fearnside
,
H. L.
Vasconcelos
, and
L. V.
Ferreira
,
2004
:
Deforestation in Amazonia
.
Science
,
304
,
1109
1111
, https://doi.org/10.1126/science.304.5674.1109b.
Lewis
,
S. L.
,
P. M.
Brando
,
O. L.
Phillips
,
G. M.
van der Heijden
, and
D.
Nepstad
,
2011
:
The 2010 Amazon drought
.
Science
,
331
,
554
, https://doi.org/10.1126/science.1200807.
Li
,
W.
, and
R.
Fu
,
2004
:
Transition of the large-scale atmospheric and land surface conditions from the dry to the wet season over Amazonia as diagnosed by the ECMWF Re-Analysis
.
J. Climate
,
17
,
2637
2651
, https://doi.org/10.1175/1520-0442(2004)017<2637:TOTLAA>2.0.CO;2.
Liebmann
,
B.
, and
J.
Marengo
,
2001
:
Interannual variability of the rainy season and rainfall in the Brazilian Amazon basin
.
J. Climate
,
14
,
4308
4318
, https://doi.org/10.1175/1520-0442(2001)014<4308:IVOTRS>2.0.CO;2.
Liebmann
,
B.
, and
C. R.
Mechoso
,
2011
: The South American monsoon system. The Global Monsoon System: Research and Forecast, 2nd ed. C. Chang, Y. Ding, and R. H. Johnson, Eds., World Scientific, 137–157.
Liebmann
,
B.
,
S. J.
Camargo
,
A.
Seth
,
J. A.
Marengo
,
L. M.
Carvalho
,
D.
Allured
,
R.
Fu
, and
C. S.
Vera
,
2007
:
Onset and end of the rainy season in South America in observations and the ECHAM 4.5 atmospheric general circulation model
.
J. Climate
,
20
,
2037
2050
, https://doi.org/10.1175/JCLI4122.1.
Machado
,
L. A.
,
H.
Laurent
, and
A. A.
Lima
,
2002
:
Diurnal march of the convection observed during TRMM-WETAMC/LBA
.
J. Geophys. Res.
,
107
,
8064
, https://doi.org/10.1029/2001JD000338.
Machado
,
L. A.
,
H.
Laurent
,
N.
Dessay
, and
I.
Miranda
,
2004
:
Seasonal and diurnal variability of convection over the Amazonia: A comparison of different vegetation types and large scale forcing
.
Theor. Appl. Climatol.
,
78
,
61
77
, https://doi.org/10.1007/s00704-004-0044-9.
Malhi
,
Y.
, and Coauthors
,
2009
:
Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest
.
Proc. Natl. Acad. Sci. USA
,
106
,
20 610
20 615
, https://doi.org/10.1073/pnas.0804619106.
Mantua
,
N. J.
,
S. R.
Hare
,
Y.
Zhang
,
J. M.
Wallace
, and
R. C.
Francis
,
1997
:
A Pacific interdecadal climate oscillation with impacts on salmon production
.
Bull. Amer. Meteor. Soc.
,
78
,
1069
1080
, https://doi.org/10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2.
Marengo
,
J. A.
,
2004
:
Interdecadal variability and trends of rainfall across the Amazon basin
.
Theor. Appl. Climatol.
,
78
,
79
96
, https://doi.org/10.1007/s00704-004-0045-8.
Marengo
,
J. A.
, and
J. C.
Espinoza
,
2016
:
Extreme seasonal droughts and floods in Amazonia: Causes, trends and impacts
.
Int. J. Climatol.
,
36
,
1033
1050
, doi:.
Marengo
,
J. A.
,
B.
Liebmann
,
V. E.
Kousky
,
N. P.
Filizola
, and
I. C.
Wainer
,
2001
:
Onset and end of the rainy season in the Brazilian Amazon basin
.
J. Climate
,
14
,
833
852
, https://doi.org/10.1175/1520-0442(2001)014<0833:OAEOTR>2.0.CO;2.
Marengo
,
J. A.
, and Coauthors
,
2008
:
The drought of Amazonia in 2005
.
J. Climate
,
21
,
495
516
, https://doi.org/10.1175/2007JCLI1600.1.
Marengo
,
J. A.
,
J.
Tomasella
,
W.
Soares
,
L. M.
Alves
, and
C. A.
Nobre
,
2010
:
Extreme climatic events in the Amazon basin: Climatological and hydrological context of previous floods
.
Theor. Appl. Climatol.
,
107
,
73
85
, https://doi.org/10.1007/s00704-011-0465-1.
Marengo
,
J. A.
,
J.
Tomasella
,
L. M.
Alves
,
W. R.
Soares
, and
D. A.
Rodriguez
,
2011
:
The drought of 2010 in the context of historical droughts in the Amazon region
.
Geophys. Res. Lett.
,
38
,
L12703
, https://doi.org/10.1029/2011GL047436.
Marengo
,
J. A.
, and Coauthors
,
2012
:
Recent developments on the South American monsoon system
.
Int. J. Climatol.
, 32,
1
21
, https://doi.org/10.1002/joc.2254.
Molina-Carpio
,
J.
,
J. C.
Espinoza
,
P.
Vauchel
,
J.
Ronchail
,
B.
Gutierrez
,
J. L.
Guyot
, and
L.
Noriega
,
2017
:
The hydroclimatology of the upper Madeira River basin: Spatio-temporal variability and trends (1967–2013)
.
Hydrol. Sci. J.
,
62
,
911
927
, https://doi.org/10.1080/02626667.2016.1267861.
Norris
,
J. R.
,
2001
:
What can cloud observations tell us about climate variability?
Space Sci. Rev.
,
94
,
375
380
, https://doi.org/10.1023/A:1026704314326.
Norris
,
J. R.
,
2005
:
Multidecadal changes in near-global cloud cover and estimated cloud cover radiative forcing
.
J. Geophys. Res.
,
110
,
D08206
, https://doi.org/10.1029/2004JD005600.
Norris
,
J. R.
and
A.
Slingo
,
2009
: Trends in observed cloudiness and Earth’s radiation budget: What do we not know and what do we need to know? Clouds in the Perturbed Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics, and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., MIT Press, 17–36.
Parker
,
D.
,
C.
Folland
,
A.
Scaife
,
J.
Knight
,
A.
Colman
,
P.
Baines
, and
B.
Dong
,
2007
:
Decadal to multidecadal variability and the climate change background
.
J. Geophys. Res.
,
112
,
D18115
, https://doi.org/10.1029/2007JD008411.
Petersen
,
W. A.
, and
S. A.
Rutledge
,
2001
:
Regional variability in tropical convection: Observations from TRMM
.
J. Climate
,
14
,
3566
3586
, https://doi.org/10.1175/1520-0442(2001)014<3566:RVITCO>2.0.CO;2.
Ronchail
,
J.
,
J.
Guyot
,
J.
Villar
,
P
.
Fraizy
,
G.
Cochonneau
,
E.
De Oliveira
,
N.
Filizola
, and
J. J.
Ordenez
,
2006
. Impact of the Amazon tributaries on major floods at Óbidos. IAHS Publ., 308, 220–225.
Rossow
,
W. B.
, and
R. A.
Schiffer
,
1999
:
Advances in understanding clouds from ISCCP
.
Bull. Amer. Meteor. Soc.
,
80
,
2261
2288
, https://doi.org/10.1175/1520-0477(1999)080<2261:AIUCFI>2.0.CO;2.
Running
,
S. W.
,
2006
:
Is global warming causing more, larger wildfires?
Science
,
313
,
927
928
, https://doi.org/10.1126/science.1130370.
Saraiva
,
I.
,
M. A. F.
Silva Dias
,
C. A. R.
Morales
, and
J. M. B.
Saraiva
,
2016
:
Regional variability of rain clouds in the Amazon basin as seen by a network of weather radars
.
J. Appl. Meteor. Climatol.
,
55
,
2657
2675
, https://doi.org/10.1175/JAMC-D-15-0183.1.
Satyamurty
,
P.
,
C. P.
Wanzeler da Costa
, and
A. O.
Manzi
,
2013
:
Moisture sources for the Amazon basin: A study of contrasting years
.
Theor. Appl. Climatol.
,
111
,
195
209
, https://doi.org/10.1007/s00704-012-0637-7.
Schiffer
,
R. A.
, and
W. B.
Rossow
,
1983
:
The International Satellite Cloud Climatology Project (ISCCP): The first project of the World Climate Research Programme
.
Bull. Amer. Meteor. Soc.
,
64
,
779
784
, https://doi.org/10.1175/1520-0477-64.7.779.
Schiro
,
K. A.
,
J. D.
Neelin
,
D. K.
Adams
, and
B. R.
Lintner
,
2016
:
Deep convection and column water vapor over tropical land versus tropical ocean: A comparison between the Amazon and the tropical western Pacific
.
J. Atmos. Sci.
,
73
,
4043
4063
, https://doi.org/10.1175/JAS-D-16-0119.1.
Sena
,
E. T.
, and
P.
Artaxo
,
2015
:
A novel methodology for large-scale daily assessment of the direct radiative forcing of smoke aerosols
.
Atmos. Chem. Phys.
,
15
,
5471
5483
, https://doi.org/10.5194/acp-15-5471-2015.
Sena
,
E. T.
,
P.
Artaxo
, and
A. L.
Correia
,
2013
:
Spatial variability of the direct radiative forcing of biomass burning aerosols and the effects of land use change in Amazonia
.
Atmos. Chem. Phys.
,
13
,
1261
1275
, https://doi.org/10.5194/acp-13-1261-2013.
Silva
,
V.
, and
V. E.
Kousky
,
2012
: The South American monsoon system: Climatology and variability. Modern Climatology, S.-Y. Wang, Ed., InTech, 123–152.
Silva Dias
,
M. A.
, and
L. M.
Carvalho
,
2017
. The South American monsoon system. The Global Monsoon System: Research and Forecast, 3rd ed. World Scientific, 25–33.
Vera
,
C.
, and Coauthors
,
2006
:
Towards a unified view of the American monsoon system
.
J. Climate
,
19
,
4977
5000
, https://doi.org/10.1175/JCLI3896.1.
Wang
,
J.
, and Coauthors
,
2009
:
Impact of deforestation in the Amazon basin on cloud climatology
.
Proc. Natl. Acad. Sci. USA
,
106
,
3670
3674
, https://doi.org/10.1073/pnas.0810156106.
Warren
,
S. G.
,
R. M.
Eastman
, and
C. J.
Hahn
,
2007
:
A survey of changes in cloud cover and cloud types over land from surface observations, 1971–96
.
J. Climate
,
20
,
717
738
, https://doi.org/10.1175/JCLI4031.1.
Werth
,
D.
, and
R.
Avissar
,
2002
:
The local and global effects of Amazon deforestation
.
J. Geophys. Res.
,
107
,
8087
, https://doi.org/10.1029/2001JD000717.
Wongchuig-Correa
,
S.
,
R. C. D.
de Paiva
,
J. C.
Espinoza
, and
W.
Collischonn
,
2017
:
Multi-decadal hydrological retrospective: Case study of Amazon floods and droughts
.
J. Hydrol.
,
549
,
667
684
, https://doi.org/10.1016/j.jhydrol.2017.04.019.
Wright
,
J. S.
,
R.
Fu
,
J. R.
Worden
,
S.
Chakraborty
,
N. E.
Clinton
,
C.
Risi
,
Y.
Sun
, and
L.
Yin
,
2017
:
Rainforest-initiated wet season onset over the southern Amazon
.
Proc. Natl. Acad. Sci. USA
,
114
,
8481
8486
, https://doi.org/10.1073/pnas.1621516114.
Wu
,
C.-M.
,
B.
Stevens
, and
A.
Arakawa
,
2009
:
What controls the transition from shallow to deep convection?
J. Atmos. Sci.
,
66
,
1793
1806
, https://doi.org/10.1175/2008JAS2945.1.
Xie
,
P.
, and
P. A.
Arkin
,
1998
:
Global monthly precipitation estimates from satellite-observed outgoing longwave radiation
.
J. Climate
,
11
,
137
164
, https://doi.org/10.1175/1520-0442(1998)011<0137:GMPEFS>2.0.CO;2.
Yin
,
L.
,
R.
Fu
,
Y. F.
Zhang
,
P. A.
Arias
,
D. N.
Fernando
,
W.
Li
,
K.
Fernandes
, and
A. R.
Bowerman
,
2014
:
What controls the interannual variation of the wet season onsets over the Amazon?
J. Geophys. Res. Atmos.
,
119
,
2314
2328
, https://doi.org/10.1002/2013JD021349.
Yoon
,
J.-H.
, and
N.
Zeng
,
2010
:
An Atlantic influence on Amazon rainfall
.
Climate Dyn.
,
34
,
249
264
, https://doi.org/10.1007/s00382-009-0551-6.
Zeng
,
N.
,
J.-H.
Yoon
,
J. A.
Marengo
,
A.
Subramaniam
,
C. A.
Nobre
,
A.
Mariotti
, and
J. D.
Neelin
,
2008
:
Causes and impacts of the 2005 Amazon drought
.
Environ. Res. Lett.
,
3
,
014002
, https://doi.org/10.1088/1748-9326/3/1/014002.
Zhang
,
Y.
,
R.
Fu
,
H.
Yu
,
Y.
Qian
,
R.
Dickinson
,
M. A. F.
Silva Dias
,
P. L.
da Silva Dias
, and
K.
Fernandes
,
2009
:
Impact of biomass burning aerosol on the monsoon circulation transition over Amazonia
.
Geophys. Res. Lett.
,
36
,
L10814
, https://doi.org/10.1029/2009GL037180.
Zhou
,
J.
, and
K. M.
Lau
,
1998
:
Does a monsoon climate exist over South America?
J. Climate
,
11
,
1020
1040
, https://doi.org/10.1175/1520-0442(1998)011<1020:DAMCEO>2.0.CO;2.
Zhuang
,
Y.
,
R.
Fu
,
J. A.
Marengo
, and
H.
Wang
,
2017
:
Seasonal variation of shallow- to-deep convection transition and its link to the environmental conditions over the central Amazon
.
J. Geophys. Res.
,
122
,
2649
2666
, https://doi.org/10.1002/2016JD025993.

Footnotes

© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).