Downscaling of South America present climate driven by 4
Transcrição
Downscaling of South America present climate driven by 4
Clim Dyn (2012) 38:635–653 DOI 10.1007/s00382-011-1002-8 Downscaling of South America present climate driven by 4-member HadCM3 runs Sin Chan Chou • José A. Marengo • André A. Lyra • Gustavo Sueiro • José F. Pesquero • Lincoln M. Alves • Gillian Kay • Richard Betts • Diego J. Chagas • Jorge L. Gomes • Josiane F. Bustamante • Priscila Tavares Received: 4 August 2010 / Accepted: 17 January 2011 / Published online: 12 February 2011 Ó Springer-Verlag 2011 Abstract The objective of this work is to evaluate climate simulations over South America using the regional Eta Model driven by four members of an ensemble of the UK Met Office Hadley Centre HadCM3 global model. The Eta Model has been modified with the purpose of performing long-term decadal integrations and has shown to reproduce ‘‘present climate’’—the period 1961–1990— reasonably well when forced by HadCM3. The global model lateral conditions with a resolution of 2.5° latitude 9 3.75° longitude were provided at a frequency of 6 h. Each member of the global model ensemble has a different climate sensitivity, and the four members were selected to span the range of uncertainty encompassed by the ensemble. The Eta Model nested in the HadCM3 global model was configured with 40-km horizontal resolution and 38 layers in the vertical. No large-scale internal nudging was applied. Results are shown for austral summer and winter at present climate defined as 1961–90. The upper and low-level circulation patterns produced by the Eta-CPTEC/HadCM3 experiment set-up show good agreement with reanalysis data and the mean precipitation and temperature with CRU observation data. The spread in the downscaled mean precipitation and temperature is small when compared against model errors. On the other hand, the benefits in using an ensemble is clear in the S. C. Chou (&) J. A. Marengo A. A. Lyra G. Sueiro J. F. Pesquero L. M. Alves D. J. Chagas J. L. Gomes J. F. Bustamante P. Tavares National Institute for Space Research (INPE), Cachoeira Paulista, São Paulo 12630-000, Brazil e-mail: [email protected] G. Kay R. Betts UK Met Office Hadley Centre, FitzRoy Road, Exeter, Devon EX1 3PB, UK improved representation of the seasonal cycle by the ensemble mean over any one realization. El Niño and La Niña years were identified in the HadCM3 member runs based on the NOAA Climate Prediction Center criterion of sea surface temperature anomalies in the Niño 3.4 area. The frequency of the El Niño and La Niña events in the studied period is underestimated by HadCM3. The precipitation and temperature anomalies typical of these events are reproduced by most of the Eta-CPTEC/HadCM3 ensemble, although small displacements of the positions of the anomalies occur. This experiment configuration is the first step on the implementation of Eta-CPTEC/HadCM3 upcoming experiments on climate change studies that are discussed in a companion paper. Keywords South America Regional climate model Eta model Present climate Climate simulation uncertainty 1 Introduction Global Climate Models (GCMs) are the main tools for studying long-term climate variability and change. However, these models have rather coarse resolution, which poses limitations to the explicit simulation of mesoscale processes and to the representation of topography, land use, and land-sea distribution. Increases in spatial resolution can be achieved by downscaling techniques: refining part of the grid of a variable-resolution global model; nesting high resolution Regional Climate Models (RCMs) in GCMs conditions; and statistical downscaling of GCM simulations. Climate change projections derived from RCMs may be considered useful for impact studies because of the subcontinental nature of the patterns and magnitude of 123 636 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs changes. The improvement of the description of orography and other land surface characteristics helps to detail the structure of weather systems. However, it should be noted that the climate generated from a regional model is strongly dependent upon the lateral boundary conditions. Seth and Rojas (2003) and Rojas and Seth (2003) performed 5-month RegCM3 regional model for the South American domain for two extreme wet and dry years, using both reanalysis and atmospheric general circulation model (AGCM) data as lateral boundary forcing with observed sea surface temperatures (SST). They found that errors in the low level circulation and moisture fields of the GCM significantly degraded the nested model simulation, but that the interannual signal was well reproduced. The Eta version used at National Centers for Environmental Prediction has been used for process studies by Berbery et al. (2003), Silva and Berbery (2006) and Collini et al. (2008), among others. The Brazilian Center for Weather Forecasts and Climate Studies (CPTEC) has used the Eta regional model (hereafter referred as Eta-CPTEC) operationally since 1996 to provide weather forecasts over South America. Due to its vertical coordinate (Mesinger 1984), the Eta model is able to produce satisfactory results in the regions containing steep orography such as the Andes Cordillera. On prediction and predictability issues, Chou et al. (2000) did one of the first experiments with a regional model for South America, with 1 month of continuous integrations. The Eta-CPTEC was used to investigate the precipitation predictability at different time scales–seasonal, monthly and weekly–over South America (Chou et al. 2005). Comparisons of the CPTEC GCM seasonal climate forecasts with those from Eta-CPTEC against observations show that this regional model provides considerable improvement over the driver model. Various papers have assessed simulations of present time climatology (1961–90) of South America from different regional models using the HadAM3P global model as lateral boundary conditions (Alves and Marengo 2009 for the HadRM3P, da Rocha et al. 2009 for RegCM3; Pisnichenko and Tarasova 2009 for Eta CCS, and Pesquero et al. 2009 for Eta-CPTEC). When compared to observations, the regional simulations exhibit systematic errors which might be related to the physics of the RCMs (e.g., convective schemes, topography and land surface processes) and the lateral boundary conditions and possible biases inherited from the global model. Dynamical downscaling experiments for the purpose of constructing climate change scenarios in South America have started to become available for various emission scenarios and time-slices until the end of the 21st Century. These examples have used various regional models forced with the global future climate change scenarios as 123 boundary conditions from various global climate models (Marengo et al. 2009a, b; Núñez et al. 2006; Alves and Marengo 2009, Solman et al. 2007, Garreaud and Falvey 2008, Cabré et al. 2010, Urrutia and Vuille 2009, Pesquero et al. 2009, and Chou et al. 2009). Pisnichenko and Tarasova (2009) describe two time-slice integrations of the Eta model modified for climate change studies (the so-called Eta-CCS) driven by the HadAM3P global model. Their results show precipitation fields with strong negative bias over a large part of South America during summer for present-climate simulations and weak precipitation activity along the Atlantic Intertropical Convergence Zone (ITCZ). Another version of the Eta model for decadal runs was developed by Pesquero et al. (2009) and shows a considerable improvement in the representation of the patterns of precipitation in austral summer and winter. These results indicate that the Eta model provides added value in comparison to the HadAM3P global model. Other articles on South American regional simulation of present climate include Rauscher et al. (2006, 2007), Seth et al. (2007), Solman et al. (2007), Alves and Marengo (2009), Chou et al. (2009), Menéndez et al. (2010) and references quoted therein. The studies listed above generally use a single model. In this study, we run a four-member regional model ensemble in order to address some of the uncertainties inherent in any model simulation. The boundary conditions are taken from four members of the HadCM3 ‘‘Perturbed Physics Ensemble’’ (PPE) in which the standard model structure is used and perturbations are introduced to the physical parameterization schemes to produce variants of the same model (Murphy et al. 2004; Stainforth et al. 2005, Collins et al. 2006a, Collins et al. 2010). The objective of this work is to evaluate this regional climate ensemble model system, Eta-HadCM3 members, and give some measure of the uncertainty of the results provided by this ensemble over South America. This Eta-CPTEC regional model was run as part of the impact analyses and vulnerability assessments needed for the preparation of the Second National Communication of Climate Change of Brazil to the United National Framework Climate Change Convention UNFCCC. A companion paper analyzes regional future climate change projections over South America for the A1B scenario derived from the Eta-CPTEC nested in HadCM3 runs, referred hereafter as Eta-CPTEC/HadCM3. 2 The models 2.1 The HadCM3 global model The lateral boundary conditions used to drive the EtaCPTEC regional model are supplied by the UK Met Office S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Hadley Centre coupled climate model HadCM3 (Gordon et al. 2000; Collins et al. 2001). The HadCM3 atmosphere has a resolution of 2.5° 9 3.75° latitude-longitude, with 19 levels in the vertical (Pope et al. 2000). The HadCM3 model has shown to perform favorably amongst current climate models in the simulation of the climate in the Amazon, as well as important teleconnections with largescale modes of variability in the tropical Pacific and Atlantic (e.g., Gedney et al. 2000; Li et al. 2006; Good et al. 2008), although some biases may still exist (Harris et al. 2008). The uncertainties present in any model simulation can be explored through ensemble experiment design. One way in which this can be achieved is through the multi-model ensemble method, in which a coordinated effort such as the Coupled Model Intercomparison Projects (CMIP, e.g. Covey et al. 2003; Meehl et al. 2007), which brings together GCM output from different contributing climate modeling centers. An advantage of this method is that a wide variety of model designs and configurations form the ensemble, often described as an ‘‘ensemble of opportunity’’. However, as such, the ensemble has not been designed to completely encompass the range of model uncertainty. The other method follows the perturbed physics ensemble (PPE) approach (Murphy et al. 2004; Stainforth et al. 2005; Collins et al. 2006b, 2010), which is designed to quantify the modeling uncertainty in the simulation or projections of climate that depends on the way processes are represented in the model, i.e. in their physics parameters. The HadCM3 ensemble was designed to quantify uncertainty in projections of climate change derived from uncertainty in model physics as per the second method described above. Through expert elicitation, key uncertain parameters were identified, primarily in the atmosphere but also in the land surface, and their plausible ranges were defined. These parameters were modified within these plausible ranges to form a large (300 ? member) ensemble, run with a computationally-efficient slab ocean. From this ensemble, a subset of 16 model variants, each with a different combination of parameter settings, was selected according to their performance in simulating current climate while still letting parameter sampling space widely (Murphy et al. 2007). Together with the standard HadCM3 model, the 16 model variants were run in fully coupled transient mode, forced with SRES A1B emissions scenariogenerated CO2 concentrations (Nakicenovic et al. 2000) to the end of the 21st century. Even though each member of the ensemble is forced with the same CO2 concentrations, the effect of the different combinations of parameter settings alters the degree and to some extent the patterns of climate change. The range in global mean temperature rise by the end of the 21st century provided by variants of this 637 one model is of a similar magnitude to the range given by the AR4 multi-model ensemble (Collins et al. 2006b). 2.2 The Eta-CPTEC regional model The Eta model (Mesinger et al. 1988; Black 1994; Janjić 1994) is a grid-point model based on the eta coordinate, g (Mesinger 1984), which is defined as. ð p pt Þ ðpref ðZsfc Þ pt Þ g¼ ðpsfc pt Þ ðpref ð0Þ pt Þ where p is the air pressure and Z is the height. The indices t and sfc indicate model top and model surface, respectively. The index ref refers to values from a reference atmosphere, therefore, pref(0) is the air pressure at the height 0, and pref(Zsfc) is the air pressure at the surface, both are taken from a reference atmosphere. The surfaces of the g coordinate are approximately horizontal everywhere which is particularly adequate for regions with steep orography such as the Andes Cordillera in South America. The time scheme is the forward–backward scheme modified by Janjić (Janjić 1979) for the adjustment terms and a modified Euler-Backward scheme for the advection terms. The space difference scheme prevents the two-grid internal gravity wave separation. The prognostic variables are temperature, specific humidity, horizontal wind, surface pressure, the turbulent kinetic energy and cloud liquid water/ice. These variables are distributed on the Arakawa type E-grid. Large scale forcing is defined in one row along the lateral boundaries. The second row is a merge between the outer and the inner third rows. Integration domain starts from the third row. The lateral boundary formulation is described by Mesinger (1977). This scheme does not adopt the relaxation zones as the commonly used Davies (1976) scheme. One-way nesting was used. No large-scale nudging is applied in the integration domain. Veljović et al. (2010) have shown that the Eta Model maintains the large scale features even using large domain without the need of internal large scale nudging. Model precipitation is produced by Betts-Miller-Janjić cumulus parameterization scheme (Janjić 1994) and by the Zhao cloud microphysics scheme (Zhao et al. 1997). The land surface scheme (Chen et al. 1997; Ek et al. 2003) has 4 soil layers for temperature and moisture of depths, from top to bottom: 10, 30, 60 and 100 cm.The scheme distinguishes 12 types of vegetation and 7 types of soil texture. The vegetation map includes revised Amazon deforestation arc (Sestini et al. 2002). The radiation scheme package was developed by the Geophysical Fluid Dynamics Laboratory. The scheme includes short-wave (Lacis and Hansen 1974) and long-wave radiation (Fels and Schwarzkopf 1975). The radiation tendencies are recalculated every 1 h and are 123 638 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs applied every time step. The atmospheric turbulence scheme has the turbulent kinetic energy as prognostic variable. Monin–Obukhov similarity theory combined with Paulson stability functions (Paulson 1970) are applied at the surface layer. 2.2.1 Model setup The Eta model has been used as the operational weather and seasonal climate forecast model at INPE (Chou et al. 2005) for a few years. The seasonal climate version of the Eta model was adapted to run decadal time range integrations, with focus on the study of climate change scenarios related to different levels of atmospheric CO2 concentration. For the present climate studies, the CO2 concentration was set to a constant value of 330 ppm. The Eta model was set to have a calendar of 360 days in a year in order to follow the HadCM3 models’ calendar. Boundary conditions for driving the Eta model were taken from the HadCM3 model ensemble, chosen to span the range of uncertainty described by the model variants. The HadCM3 conditions were input at every 6 h, and linear update of the Fig. 1 The top row refers to mean 200-hPa DJF streamlines for 1961–1990 from the Eta Model members a Low, b Medium, c High, and d Unperturbed; and e HadCM3 ensemble mean. The bottom row refers to mean 200-hPa JJA streamlines for 1961–1990 from the Eta 123 values along the boundaries was applied at every time step. Accordingly, three members were selected, which displayed high, medium and low sensitivity in global mean temperature response. Together with the standard, unperturbed model, these provide the boundary conditions for driving multiple realizations of the Eta-CPTEC regional model, hereafter referred to as the high, medium, low, and unperturbed experiments. Sea surface temperature values were taken from the coupled ocean–atmosphere model monthly means, sea surface temperature is updated daily and linearly to the monthly values on the 15th day of each month. Similar linear interpolation is applied to the vegetation greenness which is prescribed in fixed monthly mean values. The results for 1961–90 present climate are described here, and for 2010–2100 in the companion paper. The model integration started on the 1st of January of 1960 and a one-year spin-up time was adopted. Therefore the results are based on the period from 1st January 1961 until 30th December 1990. The integrations were continuous for 31 years. Initial soil moisture started from a January climatology and albedo started from a seasonal climatology. The model was setup to run with 40-km horizontal resolution, 38 vertical layers and 90 s time step. Model members f Low, g Medium, h High, and i Unperturbed; and j HadCM3 ensemble mean. Wind speeds greater than 25 and 30 m/s are shaded S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 3 Results The results focus on the climate features of the seasons December–January–February (DJF) and June–July–August (JJA), which are the austral summer and winter seasons, respectively. For verification purposes, circulation fields from the ERA40 reanalyses (Uppala et al. 2005) and rainfall and temperature fields from the Climate Research Unit CRU-University of East Anglia (New et al. 2000) are used. 3.1 Large scale circulation The downscaled large-scale flow is compared against the driver global model and the ECMWF reanalysis data. This comparison will show whether the nested model can maintain the large-scale conditions input by the driver model and whether the downscaled flow is accurate enough for further use in impact studies. The average circulation from the driver model is taken over the four HadCM3 members. Figure 1a–d show the mean upper-level mean circulation simulated by the Eta-CPTEC for the austral summer (DJF) season. The comparison with HadCM3 upper level circulation (Fig. 1e) shows that the regional model follows closely the conditions provided by the lateral boundaries by 639 centering the upper level anticyclone over the Bolivian Plateau and spanning it over most of the tropical part of the continent. The downstream cyclonic vortex has its axis along the coast of Northeast Brazil. These features were transferred from the GCM. The correct description of the large scale convection observed over the South American continent during summer is strongly dependent on the position and intensity of the circulation associated with the anticyclone. During JJA (Fig. 1f–i), the upper level anticyclone is weak and has its centre positioned over northern part of South America. This is in agreement with the absence of deep convection in the central part of the continent during the JJA season. The subtropical upper level jet is a clear feature with its mean magnitude exhibiting values over 30 m/s. Figure 2 shows the mean low level circulation, averaged over the 30 years during the austral summer season. Important features of the low level circulation are the tropical Northeast trade winds, and the deflection of these winds by the Andes. These winds to the east of the Andes are important in the moisture transport from tropical into subtropical latitudes, and on some occasions these winds reach strong intensity and become the so called South American Low Level Jet east of the Andes (SALLJ), particularly during summer (Saulo et al. 2000; Marengo et al. 2004, and references quoted therein). The intense Fig. 2 Same as Fig. 1, but for 850-hPa streamlines 123 640 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Northeast trades contribute to maintaining the moisture transport to the central and southeastern part of the continent where the maximum total precipitation occurs and forms the South Atlantic Convergence Zone (SACZ). Furthermore, the intense trades also contribute to the rainfall and convective activity at the northern side of the La Plata Basin. During winter, JJA, the Northeast trades are weaker than in summer, but the SALLJ is still present, exhibiting a strong northerly component, and contributing to the frontogenesis over Argentina (Satyamurty et al. 1998). In a manner different from summer, this northerly flow brings air masses from the subtropical South Atlantic, compared to the summertime Amazonian air masses. The Eta-CPTEC model shows an improved representation of the maximum jet over the Andes, which is too wide in the global model, through the increased spatial resolution and consequent better representation of the complex Andes orography. Figure 3 shows the seasonal large scale circulation averaged over the 30 year from ERA40 reanalyses, as well as the CRU climatology of rainfall and temperature. At upper levels, the anticyclones simulated by Eta are centered in about the same position in summer as the reanalyses; in winter, JJA, the intensity and position of the subtropical jets in the Eta runs are also close to the reanalysis data. The low level circulation features in the regional model are comparable to the reanalyses, with a slight model overestimation of the speed in the Northeast trades. The Eta-CPTEC runs are able to add small scale features that are absent in the global driving field or in the ECMWF reanalyses (Figs. 2, 3), such as the details of the low-level flow blocked in the surroundings of the Andes Cordillera in Ecuador and Colombia or the flow convergence in the Serra do Mar mountains that are located along the coast of southeast Brazil. Besides these small scale features that which are mostly attributed to the difference in resolution of mountain description between the EtaCPTEC (40 km), HadCM3 (2.5° 9 3.75° latitude-longitude) and the ERA40 reanalyses (2.5° 9 2.5° latitudelongitude), there are no significant differences in behavior and spatial pattern of large-scale circulation of the EtaCPTEC runs as compared to the HadCM3 and ECMWF Fig. 3 ERA40 wind reanalyses at 200 hPa for a DJF and e JJA; at 850 hPa for b DJF and f JJA. Wind speeds greater than 25 and 30 m/s are shaded; CRU precipitation for c DJF and g JJA; CRU air temperature for d DJF and e JJA. Color scale is shown on the lower side of each panel 123 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 641 Although the Eta CPTEC model reproduces very closely the driver model large scale circulation patterns, the precipitation patterns of Eta and HadCM3 exhibit larger differences which may have several sources, for example the different resolutions, the different cumulus and cloud microphysics schemes, land-surface scheme, etc. Therefore, these differences can produce different precipitation biases between the regional and global model. The differences among the four Eta CPTEC simulations are related to the different boundary conditions provided by the driver model PPE. Here, each member of the EtaCPTEC regional model seasonal precipitation is shown and compared against the ensemble mean HadCM3 seasonal rainfall (Fig. 4), and should be looked together with the CRU seasonal mean observations from Fig. 3. At first glance, the precipitation patterns are very similar among the 4 members and the HadCM3 model (Fig. 4). The maxima of precipitation observed over the central part of the continent, the SACZ, and the ITCZ over the Equatorial Atlantic, are all features present in the downscaling runs, and a common underestimation of precipitation can be observed along the northern coast of the continent, where the SACZ precipitation band would connect with ITCZ band as seen in the CRU data (Fig. 3c). However, a closer look reveals differences among the members in the quantity of precipitation. For example, over northern Argentina and the Uruguay region, as well as over central Amazonia, along the SACZ and Northeast Brazil, some members underestimate precipitation more than the others. The ITCZ also shows different precipitation amounts among the members. However, verification of precipitation over the oceans is difficult as satellite estimates are also questionable for such a long period. The dry bias in northern Amazonia could have various sources, the stronger Eta CPTEC/HadCM3 simulated Northeast trades may also have a contribution to the bias, by pushing moisture farther south and inland, which could displace the moisture saturation region. In JJA, the ITCZ is displaced to the north as shown by the maximum of precipitation. Within the four members, the maxima of rainfall varies in location from nearer to the coast to further inland, while the shape of the rainfall maxima can be either narrow or wide, depending on the member considered. All members show an area of no precipitation in the central part of the continent, comparing reasonably well with CRU data. Large overestimates of precipitation are found over the Andes between southern Fig. 4 The top row refers to mean DJF precipitation (mm/d) for 1961–1990 from the Eta Model members a aeyjb, b aeyjj, c aeyjo, d aenwl, and e HadCM3 ensemble mean. The bottom row refers to mean JJA precipitation (mm/d) for 1961–1990 from the Eta Model members f aeyjb, g aeyjj, h aeyjo, i aenwl and j HadCM3 ensemble mean. Shading intervals are 1, 3, 6, 9, and 12 mm/d reanalyses. This shows the benefits of downscaling the HadCM3 model over South America. 3.2 Precipitation 123 642 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Chile and Argentina. Regions such as the Andes Mountains and the Amazon frequently lack observations, so verification in those areas is difficult given the uncertainties introduced by the interpolation techniques used by the CRU in the processing of their rainfall data set. The large scale patterns simulated by HadCM3 in both summer and winter are also simulated by Eta CPTEC, with perhaps the main difference being in the rainfall over the Andes region, where in the global model it is organized in a wide band extending to the east of the Andes, while the regional models display rainfall more concentrated over the Andes themselves. 3.3 2-m temperature The 2-m temperature is strongly dependent on the landsurface schemes and the surface layer scheme. These schemes are different in the driver and nested models which produce differences in the 2-m temperature pattern and therefore different temperature bias between the regional and driver models. Figure 5 shows the simulation of seasonal air temperatures from the HadCM3 (ensemble mean) and from each member from the Eta-CPTEC/ HadCM3 runs (observations are shown in Fig. 3). The RCM simulations show the details of low temperatures over the mountain areas, especially over the Andes Cordillera and the Brazilian Plateau, where both the observations and the Eta simulations show detail that the global models do not capture. In DJF, all members show underestimate of temperature over the Amazon region of about 2°C. On the other hand, over the northern part of Argentina, where precipitation is underestimated, the temperatures are overestimated by all members by about 2°C. In JJA, although major temperature patterns are similar to CRU observations, in the Amazon region the members show larger dispersion, with some members underestimating and others overestimating temperatures, especially over western Amazon. Near the Central Brazilian Plateau, temperatures are underestimated by the model members, with the magnitude of the errors varying between 2 and 3°C. Over northern Argentina, where temperatures are overestimated by the members during austral summer season, in the winter, temperatures are underestimated by almost 4°C. This is a region of frontal passage and may suggest that the model generates much colder air masses behind the cold fronts. Temperatures over mountain areas are also much colder than observations during winter. Fig. 5 Same as Fig. 4, but for 2-m temperature (°C). Shading interval is 2°C, and starts at 2°C 123 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 3.4 Annual cycle The model output annual cycle is verified for four regions (Fig. 6): the Amazon, Northeast Brazil, the La Plata Basin and a region centered in Brazil that includes portions of Uruguay, Paraguay and Bolivia, and parts of Colombia, Venezuela, Peru, Argentina and Chile. However, we will refer to this region as ‘‘Brazil’’, for simplicity. These regions were chosen particularly due to differences in their precipitation regimes. Monthly mean precipitation values are taken over the period 1961–1990 for the four regions (Fig. 7). The annual cycle of the Amazon region shows a large underestimation of precipitation by the members during the rainy period from December until May, but good agreement during the drier period, between June and November. There is small inter-member dispersion. In northern part of Northeast of Brazil, where the rainy season occurs between February and May, some model members overestimate precipitation during that season but underestimate in the transition period, from October to December. Members show larger 643 spread over Northeast Brazil than over Amazon region, but this may also be affected in part by the difference in sizes of the two regions. The benefits of running even a small ensemble instead of relying on a single realization are apparent, as some of those members lie closer to the observations than the standard member, hence reducing the mean error. In the La Plata Basin, the precipitation annual cycle has smaller amplitude and varies from about 5 mm/d in rainy season to 2 mm/d in drier season. Except in the rainy months of January, February and March, the model members generally follow closely the annual cycle of precipitation in this region. The Brazil region summarizes the evaluation of the other three regions. The maximum and minimum of the annual cycle show agreement in time with the observed CRU cycle. The underestimation in the rainy period is largely due to negative errors produced in the Amazon region and partly in the La Plata Basin. The annual cycle of temperature for the four regions is shown in Fig. 8. In the Amazon region, model temperature tends to follow the observations, except in September and October when mean temperatures are overestimated. In Fig. 6 Selected regions of South America where the annual cycle is assessed 123 644 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Northeast Brazil, although the model reproduces reasonably the timing of the warm and cold season, the temperatures are underestimated throughout the year. In the La Plata Basin, the model members reproduce the larger amplitude of the annual cycle, although most of the members overestimate the temperatures between September and November. Overall, the model members are successful in reproducing the temperature annual cycles, with an spread among members of less than 2°C. For rainfall, the observed and simulated annual cycles are comparable, with perhaps the caveat that there is an underestimation of observed rainfall in Amazonia by at least 2.5 mm/d, with a spread among members of less than 1 mm/day in this region. 3.5 Errors and spread Figure 9a, d show the seasonal mean RMSE of precipitation from the Eta CPTEC. The error of each ensemble member is Fig. 7 Annual cycle of precipitation (mm/d) mean of 1961–1990 period for the a Amazon region, b Northeast Brazil, c La Plata Basin, d and the entire Brazil. Color curves refer to: Eta Model members: 123 calculated by subtracting the CRU observed climatology of 1961–1990 from the 30-year mean ensemble member precipitation. Observations are only available over land areas. In DJF, the largest precipitation errors are found in southeastern Brazil, the tropical Andes, southern Chile, and along the northern coast of Brazil and Guyana. Except for the latter region, the errors are related to model overestimation of precipitation near mountain areas. More observations over these mountain areas could help this verification. Errors along the northern coast are due to model systematic underestimation of rainfall. Some errors occur in the Amazon, Paraguay and northern Argentina regions, and they are related to underestimation of precipitation. In JJA, largest errors are found in the northern Amazon near 2–3°N where the ITCZ is positioned at this time of year. In this season the errors in Chile increase and spread to the central part of the country. Figure 9b, e show the ensemble mean of precipitation and the spread among the members in DJF and JJA from aeyjo (red), aeyjj (blue), aeyjb (green) and aenwl (orange); and CRU observations (black) S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Eta. The spread is calculated as the standard deviation of the 30-year mean of the four members. This spread of the four members can be considered as some sort of measurement of the Eta model’s internal variability. While the maximum RMSE range about 2 and 6 mm/d, in both DJF and JJA, the largest ensemble spread ranges from 0.5 to 1.0 mm/d over land, and reaches 2.0 mm/d over the ocean. Large precipitation spread is also found in southern Chile. In JJA, in southern Brazil, which forms part of the La Plata Basin, although there is no significant increase in precipitation mean or RMSE, a small increase in the ensemble spread occurs. In the construction of the ensemble members, it is desirable that the RMSE be of comparable magnitude of the ensemble spread. However, here we have found ensemble precipitation spread to be smaller than model RMSE, particularly over the continent where observations are available. Figure 9c, f show the ensemble mean of precipitation and the spread among the members in DJF and JJA from the HadCM3 driver model. The magnitude of the spread as shown by the STDV is larger 645 over the Andes in the global than in the regional model. It’s greater over the Andes in the GCM, and it’s greater along the SACZ, Chile, ITCZ too In general the magnitude of the ensemble spread in precipitation from both regional and global models is smaller than the model rainfall RMSE errors over the continent. Figures 10a, d show the mean RMSE of 2-m temperature in DJF and JJA. The largest temperature errors occur over the Andes Mountains, and are generally underestimates. As observations are scarce over mountain areas, CRU observations there contain large uncertainty. In DJF a major error area is found over northern Argentina and Paraguay, where the model overestimates the temperatures. These errors reach about 4°C. In JJA, errors in this area in northern Argentina fall to a minimum, and large errors are found in the Northeast Brazil region. Figure 10c, e show the ensemble mean temperature and the spread among the four members from the Eta CPTEC, in DJF and JJA, respectively. The spread varies by about 0.5°C over most of the continent, and the smallest spread is found over the Fig. 8 Same as Fig. 7, but for 2-m temperature (°C) 123 646 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs ocean, opposite to the pattern of spread in precipitation, and which is probably a consequence of the small variability of sea surface temperatures. In summer, areas over Bolivia, Paraguay and northern Argentina, northern Amazon and Guyana, and coastal areas in Peru exhibit largest spread in temperature, whereas in winter, the largest temperature spread occurs in the central part of the continent and over the Andes Mountains. The spread among members from HadCM3 (Fig 10c, f) is comparable in magnitude and distribution with that from Eta- CPTEC for both summer and winter. In general the magnitude of the ensemble spread in temperature is smaller than the model temperature errors over the continent. Evaluation carried out here show an under-dispersive character of the ensemble as the spread is smaller than the RMSE. In the calculation of the errors, we must also consider the uncertainties in the CRU data set over regions such as Amazonia and the Andes, due to the poor data coverage in those regions, as compared to the well covered Northeast Brazil of Southeastern South America regions. Fig. 9 Root mean square errors of Eta Model precipitation of 1961–1990 for a DJF and d JJA. Ensemble mean (shaded) and spread (contours) of Eta Model precipitation for b DJF and e JJA; Ensemble mean (shaded) and spread (contours) of the HadCM3 model precipitation for c DJF and f JJA; Mean precipitation shaded intervals are 1, 3, 6, 9 and 12 mm/d. Spread contour interval is 0.5 mm/d 123 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 647 Fig. 10 Same as Fig. 9, but for 2-m temperature (°C). Shading interval is 2°C, and starts at 2°C. Contour interval is 0.2°C Table 1 Number of El Niño and La Niña events observed and simulated by the 4 EtaCPTEC/HadCM3 runs La Niña El Niño Observed 7 9 Unperturbed 8 4 High 4 5 Mid 4 5 Low 7 6 3.6 El Niño and La Niña events A crucial question frequently posed by users of model output for impact studies is how reliable are the HadCM3 model and Eta CPTEC/HadCM3 downscaled simulations of El Niño and La Niña events, which are major climate signals and whose impacts on different socio-economic sectors have been studied around the world (e.g. see Trenberth 2007 and references therein). To answer the question, El Niño and La Niña events were identified according to the criterion adopted by the US Climate Prediction Centre. This criterion identifies El Niño (La Niña) when the 3-month average of sea surface temperature anomalies along the equatorial Pacific in the region tagged as Niño 3.4 (120 W–170 W and 5 N–5 S) is larger/smaller than (0.5°C/- 0.5°C) in 5 consecutive overlapping seasons (Trenberth 1997). Based on this 123 648 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs criterion, between 1961 and 1990, about 9 El Niño and about 7 La Niña events were identified from observed SST (Reynolds et al. 2002). The HadCM3 sea surface temperature anomalies tend to underestimate the frequency of occurrence of these events. Table 1 shows the number of events for each member in the present climate simulation. This table is based on Fig. 11, which shows the time series of sea surface temperature anomalies from Niño 3.4. The magnitude of the SST anomalies varies considerably among the members. The unperturbed member has some El Niño events with anomalies reaching almost 4°C, and although La Niña events are more frequent then El Niño, the negative anomalies are weaker for this member. The low sensitivity member produces similar numbers of El Niño and La Niña events, and the magnitude of the positive and negative anomalies are more comparable, reaching about ±2°C. The medium and high sensitivity members have similarities as both exhibit weak anomalies, reaching about 1°C, and both have similar numbers of El Niño and La Niña events. The strength of these anomalies and also the number of events will affect the temperature and precipitation anomalies of the downscaling runs. It should be noted, however, that coupled model simulated ENSO events can vary in magnitude and frequency through natural internal climate variability, and the particular characteristics of the events in the 1961–1990 period may be representative of these variations rather than the parameter modifications. It does, however, reinforce the benefits of running an ensemble, such that the anomalies connected with a range of ENSO types can be assessed. The DJF periods during established El Niño or La Niña events were selected based on Fig. 11 to show the impacts of these events on temperature and precipitation. Figure 12a–d shows the mean precipitation anomalies for each member of the Eta-CPTEC ensemble El Niño events. Negative anomalies are found over the Atlantic region of ITCZ, whereas positive anomalies are found in the region of SACZ. These anomalies often cause droughts in Northeast Brazil and the Brazilian Amazon, and floods in Southern Brazil (Ropelewski and Halpert 1987, Villar et al. 2008, Grimm and Tedeschi 2009), which are typical features during El Niño years. However, not all droughts in those regions are related to El Niño, especially in tropical South America east of the Andes (Marengo et al. 2008). The precipitation anomalies of El Niño events from the unperturbed member are more consistent as the negative anomalies reach the Amazon and Northeast Brazil regions and positive anomalies are present in southeast Brazil. The positive precipitation anomalies in the eastern part of the continent in the perturbed members is positioned too far north for an accurate reproduction of El Niño impacts. During La Niña years (Fig. 12e–h), the precipitation 123 anomaly pattern is roughly opposite to El Niño years, that is, rainier along the ITCZ region and drier in the SACZ region. The temperature anomalies during El Niño DJF (Figs. 13a–d) show warm anomalies over northern and central part of the continent and cold anomalies around northern parts of Argentina and the south of Brazil. This is consistent with anomalies observed during El Niño events in South America. The unperturbed member also shows the largest amplitude of anomalies. In La Niña years (Figs. 13e–h) the temperature anomalies are opposite, with colder conditions in the northern and central parts of Brazil. The members show large disagreement in the temperature anomalies over Argentina. 4 Conclusions The regional Eta Model was configured over South America and applied to downscale HadCM3 members of the PPE experiment for the present climate, considered here the period between 1961 and 1990. As a step prerequisite for employing these downscaling conditions for impact studies, the ability of the regional model to reproduce present climate, the estimate of the errors and the ensemble members spread, was evaluated here. The evaluation focused on DJF and JJA seasons, which are the austral summer and rainy season and the austral winter and dry season, respectively. The results show that the upper and low level large-scale circulations reproduce closely the circulation from the global model without the need of internal large-scale nudging and using a single row at the lateral boundaries, which shows the efficiency of the lateral boundary scheme. The continuous long–term integrations show robustness and suitability of the model for climate studies. The low level flow shows that Eta-CPTEC could reproduce the driver model patterns and was able to add small scale features that were absent in the HadCM3 fields due to its coarse resolution. The downscaled climatology of precipitation and temperature is close to CRU observations. Major continental precipitation and temperature errors have been identified in areas of Amazonia and northern Argentina and Paraguay. Mountain area verification would require more in situ data as CRU observations do not show high precipitation values over orography. In austral summer in the Amazon region there is an underestimation of rainfall, which is a characteristic common to various simulation experiments (Marengo et al. 1994, 2003, Stern and Miyakoda 1995, Cox et al. 2000, among others), and which has been linked to the convection, land-surface and planetary boundary layer schemes. In the subtropical latitudes of northern Argentina, the S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 649 Fig. 11 Times series of 3-month running mean HadCM3 sea surface temperature anomalies (°C) from 1960 to 1990 for the Niño 3.4 region, for the a aenwl, b aeyjb, c aeyjj and d aeyjo members. Red shading refers to El Niño events, blue refers to La Niña events and black are neutral periods 123 650 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Fig. 12 Precipitation anomalies (mm/d) for austral summer (DJF) of El Niño and La Niña years. Anomalies are calculated subtracting the 1961–1990 climatology of the respective member. Maps are for the laeyjo a, e, aeyjj b, f, aeyjo c, g and aenwl d, h members. Blue shading refers to positive anomalies and red to negative anomalies underestimation of precipitation is accompanied by model overestimation of temperature. The ensemble members from the Eta-CPTEC/HadCM3 simulations exhibit a small spread when compared against model rmse. However, the annual cycle of precipitation and temperature in some areas, such as Northeast Brazil, shows some dispersion among the members. Verification of model ability to reproduce the regional signatures of El Niño and La Niña events shows precipitation and temperature anomaly signs typical of those events for summer, although some anomaly patterns appear displaced, such as during El Niño events in summer in Northeast Brazil. The HadCM3 SST generally underestimates the number of El Niño and La Niña events, which may have caused some weaker signal in the regional Eta CPTEC/HadCM3 simulations. The strengths and weaknesses identified in the EtaCPTEC should not be regarded as permanent defects, since the model is undergoing continuous improvement. Besides some regional systematic biases, especially in the convectively active equatorial regions, it is clear that some areas exhibit systematic biases, such as the underestimation of rainfall in northern Amazonia. Model physics require further investigation in order to identify and reduce the model errors. The results showed an under-dispersive character of these downscaling runs when compared with model errors. The regional model climatology results from the combination of the long series of atmospheric conditions coming from the lateral boundaries and regional model internal dynamics and physics. The small spread in the regional model climatology suggested that for this time period, the final climate has stronger dependence on the regional model internal characteristics given similar large-scale conditions provided by the driver model. Tests with another global model long-term integration outputs are ongoing. In addition, perturbations of regional model physics parameters are also being tested. These tests can help to give a measure of the magnitude of the regional model uncertainty. The evaluation carried out here shows that these regional climate outputs can be employed for climate change 123 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 651 Fig. 13 Same as Fig. 12, but for 2-m temperature anomalies (°C). Orange to red shadings refers to positive anomalies and blue shading refers to negative anomalies studies, and errors should be considered when these outputs are used to drive further impact studies. The Eta-CPTEC/ HadCM3 experiments on climate change projections are discussed in a companion paper. Acknowledgments The authors thank the UNDP Project BRA/05/ G31 and the FCO GOF-Dangerous Climate Change DCC project from the UK. SC and JM were funded by the Brazilian National Research Council CNPq. Additional funds came from the Brazilian programs Rede-CLIMA, the National Institute of Science and Technology for Climate Change (INCT-CC), and from the European Community’s Seventh Framework Programme (FP7/2007–2013) under Grant Agreement no. 212492 (CLARIS LPB—A Europe-South America Network for Climate Change Assessment and Impact Studies in La Plata Basin). References Alves L, Marengo JA (2009) Assessment of regional seasonal predictability using the PRECIS regional climate modeling system over South America. Theor App Climat. doi:10.1007/ s00704-009-0165-2 Berbery EH, Luo Y, Mitchell K, Betts A (2003) Eta model estimated land surface processes and the hydrological cycle of the Mississippi Basin. J Geophys Res 108:8852. doi:10.1029/2002 JD003192 Black TL (1994) NMC notes. the new NMC mesoscale Eta model: description and forecast examples. Weather Anal Forecast 9:256–278 Cabré MF, Solman S, Nuñez M (2010) Creating regional climate change scenarios over southern South America for the 2020 and 2050s using the pattern scaling technique: validity and limitations. Climat Chan 98:449–469. doi:10.1007/s10584-009-9737-5 Chen F, Janjić ZI, Mitchell K (1997) Impact of atmospheric surfacelayer parameterization in the new land-surface scheme of the NCEP mesoscale Eta model. Bound-Layer Meteor 85:391–421 Chou SC, Nunes AMB, Cavalcanti IFA (2000) Extended range forecasts over South America using the regional Eta model. J Geophys Res 105:10147–10160 Chou SC, Bustamante JF, Gomes JL (2005) Evaluation of Eta model seasonal precipitation forecasts over South America. Nonlinear Process Geophys 12(4):537–555 Chou SC, Lyra A, Pesquero F, Alves LM, Sueiro G, Chagas DJ, Marengo JA, Djurdjevic V (2009) Improvement of Long-term integrations by increasing RCM domain size. In: Challenges In regional-scale climate modelling, twenty-first century, Lund. Proceedings. ISBN 16816471 123 652 S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs Collini EA, Berbery EH, Barros V, Pyle M (2008) How does soil moisture influence the early stages of the South American monsoon? J Climat 21:195–213 Collins M, Tett SFB, Cooper C (2001) The internal climate variability of a HadCM3, a version of the Hadley centre coupled model without flux adjustments. Clim Dyn 17:61–81. doi:10.1007/ s003820000094 Collins M, Booth BBB, Harris GR, Murphy JM, Sexton DMH, Webb MJ (2006) Towards quantifying uncertainty in transient climate change. Clim Dyn. doi:10.1007/s00382-006-0121-0 Collins WV et al (2006b) Radiative forcing by well-mixed greenhouse gases: estimates from climate models in the IPCC AR4. J Geophys Res 111:D14317. doi:10.1029/2005JD006713 Collins M, Booth BBB, Bhaskaran B, Harris GR, Murphy JM, Sexton DMH, Webb MJ (2010) Climate model errors, feedbacks and forcings: a comparison of perturbed physics and multi-model ensembles. Clim Dyn. doi:10.1007/s00382-010-0808-0 Covey C, AchutaRao KM, Cubasch U, Jones P, Lambert SJ, Mann ME, Phillips TJ, Taylor KE (2003) An overview of results from the coupled model intercomparison project. Global Planet Change 37:103–133 Cox PM, Betts RA, Jones CD, Spall SA, Totterdell IJ (2000) Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature 408:184–187. doi:10.1038/ 35041539 da Rocha RP, Morales CA, Cuadra SV, Ambrizzi T (2009) Precipitation diurnal cycle and summer climatology assessment over South America: an evaluation of regional climate model version 3 simulations. J Geophys Res 114:D10108. doi:10.1029/ 2008JD010212 Davies HC (1976) A lateral boundary formulation for multi-level prediction models. QJR Meteor Soc 102:405–418 Ek MB, Mitchell KE, Lin Y, Rogers E, Grummen P, Koren V, Gayno G, Tarpley JD (2003) Implementation of NOAH land surface advances in the National centers for environmental prediction operational mesoscale Eta model. J Geophys Res 108:8851. doi: 10.1029/2002JD003246 Fels SB, Schwarzkopf MD (1975) The simplified exchange approximation: a new method for radiative transfer calculations. J Atmos Sci 32:1475–1488 Garreaud R, Falvey M (2008) The coastal winds off western subtropical South America in future climate scenarios. Int J Climatol 29:543–554. doi:10.1002/joc.1716 Gedney N, Cox P, Douville H, Polcher J, Valdes P (2000) Characterizing land surface schemes to understand their responses to climate change. J Clim 13:3066–3079. doi:10.1175/ 1520-0442(2000)013\3066:CGLSST[2.0.CO;2 Good P, Lowe J, Collins M, Moufouma-Okia W (2008) An objective tropical Atlantic sea surface temperature gradient index for studies of South Amazon dry-season climate variability and change. Philos Trans R Soc Ser B 363:1761–1766 Gordon CC et al (2000) The simulation of SST, sea ice extents and ocean heat transport in a version of the Hadley centre coupled model without flux adjustments. Clim Dyn 16:147–168 Grimm AM, Tedeschi RG (2009) ENSO and extreme rainfall events in South America. J Clim 22:1589–1609 Harris P, Huntingford C, Cox PM (2008) Amazon basin climate under global warming: the role of the sea surface temperature. Philos Trans R Soc Ser B 363:1753–1759 Janjić ZI (1979) Forward-backward scheme modified to prevent two grid-interval noise and its application in sigma coordinate models. Contrib Atmos Phys 52:69–84 Janjić ZI (1994) The step-mountain Eta coordinate model: further developments of the convection, Viscous sub layer and turbulence closure schemes. Mon Wea Rev 122:927–945 123 Lacis AA, Hansen JE (1974) A parameterization of the absorption of solar radiation in earth’s atmosphere. J Atmos Sci 31:118–133 Li W, Fu R, Dickinson RE (2006) Rainfall and its seasonality over the Amazon in the twenty-first century as assessed by the coupled models for the IPCC AR4. J Geophys Res 111:D02111. doi: 10.1029/2005JD006355 Marengo JA, Miller J, Russell G, Rosenzweig C, Abramopoulos F (1994) Calculations of river-runoff in the GISS GCM: impact of a new land-surface parameterization and runoff routing model on the hydrology of the Amazon river. Clim Dyn 10:349–361 Marengo JA, Cavalcanti IFA, Satyamurty P, Trosnikov I, Nobre CA, Bonatti JP, Camargo H, Sampaio G, Sanches MB, Manzi AO, Castro CAC, D’Almeida C, Pezzi LP, Candido L (2003) Assessment of regional seasonal rainfall predictability using the CPTEC/COLA atmospheric GCM. Clim Dyn 21:459–475 Marengo JA, Soares W, Saulo C, Nicolini M (2004) Climatology of the LLJ east of the Andes as derived from the NCEP reanalyses. J Clim 17:2261–2280 Marengo JA, Nobre C, Tomasella J, Oyama M, Sampaio G, Camargo H, Alves L, Oliveira R (2008) The drought of Amazonia in 2005. J Clim 21:495–516 Marengo JA, Jones R, Alves LM, Valverde MC (2009a) Future change of temperature and precipitation extremes in South America as derived from the PRECIS regional climate modeling system. Int J Climatol 15:2241–2255 Marengo JA, Ambrizzi T, Rocha RP, Alves LM, Cuadra SV, Valverde MC, Ferraz SET, Torres RR, Santos DC (2009b) Future change of climate in South America in the late XXI century: intercomparison of scenarios from three regional climate models. Clim Dyn. doi:10.1007/s00382-009-0721-6 Meehl GA, Covey C, Taylor KE, Delworth T, Stouffer RJ, Latif M, McAvaney B, Mitchell JFB (2007) THE WCRP CMIP3 multimodel dataset: a new era in climate change research. Bull Amer Meteor Soc 88:1383–1394 Menéndez C, de Castro M, Boulanger J-P, D’Onofrio A, Sanchez E, Sörensson AA, Blazquez J, Elizalde A, Jacob D, Le Treut H, Li ZX, Núñez MN, Pessacg N, Pfeiffer S, Rojas M, Rolla A, Samuelsson P, Solman SA, Teichmann C (2010) Downscaling extreme month-long anomalies in southern South America. Climat Chang 98:379–403. doi:10.1007/s10584-009-9739-3 Mesinger F (1977) Forward-backward scheme, and its use in a limited area model. Contrib Atmos Phys 50:200–210 Mesinger F (1984) A blocking technique for representation of mountains in atmospheric models. Rivista di Meteorologia Aeronautica 44(1–4):195–202 Mesinger F, Janjić ZI, Ničković S, Gavrilov D, Deaven DG (1988) The step-mountain coordinate: model description and performance for cases of Alpine lee cyclogenesis and for a case of Appalachian redevelopment. Mon Wea Rev 116:1493–1518 Murphy JM, Sexton DMH, Barnett DN, Jones GS, Webb MJ, Collins M, Stainforth DA (2004) Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature 430:768–772 Murphy JMB, Booth BBB, Collins M, Harris GR, Sexton DMH, Webb MJ (2007) A methodology for probabilistic predictions of regional climate change from perturbed physics ensembles. Philos Trans Soc R Ser A 365:1993–2028 Nakicenovic N, Alcamo J, Davis G, de Vries B, Fenhann J, Gaffin S, Gregory K, Grubler A, Jung TY, Kram T, La Rovere EL, Michaelis L, Mori S, Morita T, Pepper W, Pitcher H, Price L, Riahi K, Roehrl A, Rogner H-H, Sankovski A, Schlesinger M, Shukla P, Smith S, Swart R, van Rooijen S, Victor N, Dadi Z (2000) Special report on emissions scenarios. Cambridge University Press, United Kingdom, p 599 New M, Hulme M, Jones P (2000) Representing twentieth-century space time climate variability. part II: development of S. C. Chou et al.: Downscaling of South America present climate driven by 4-member HadCM3 runs 1901–1996 monthly grids of terrestrial surface climate. J Clim 13:2217–2238 Núñez M, Solman S, Cabré M (2006) Mean climate and annual cycle in a regional climate change experiment over Southern South America. II: climate change scenarios (2081–2090). In: Proceedings of 8 ICSHMO, 24–28 April 2006. Foz do Iguacu, Brazil, pp 325–331 Paulson CA (1970) The mathematical representation of wind speed and temperature profiles in the unstable atmospheric surface layer. J App Meteorol 9:857–861 Pesquero JF, Chou SC, Nobre CA, Marengo JA (2009) Climate downscaling over South America for 1961–1970 using the Eta model. Theor Appl Climatol. doi:10.1007/s00704-009-0123-z Pisnichenko IA, Tarasova TA (2009) Climate version of the ETA regional forecast model. Evaluating the consistency between the ETA model and HadAM3P global model. Theor Appl Climatol. doi:10.1007/s00704-009-0139-4 Pope V, Gallani M, Rowtree P, Stratton R (2000) The impact of new physical parameterizations in the Hadley centre climate model. Clim Dyn 16:123–146 Rauscher SA, Seth A, Qian J-H, Camargo SJ (2006) Domain choice in an experimental nested modeling prediction system for South America. Theor Appl Climatol 86:229–246. doi:10.1007/ s00704-006-0206-z Rauscher SA, Seth A, Liebmann B, Qian J-H, Camargo SJ (2007) Regional climate model simulated timing and character of seasonal rains in South America. Mon Wea Rev 135:2642–2657 Reynolds RW, Rayner NA, Smith TM, Stokes DC, Wang W (2002) An improved in situ and satellite SST analysis for climate. J Climate 15:1609–1625 Rojas M, Seth A (2003) Simulation and sensitivity in a nested modeling system for South America, part II: GCM boundary forcing. J Climat 16:2454–2471 Ropelewski CF, Halpert MS (1987) Global and regional scale precipitation patterns associated with El Niño/Southern Oscillation. Mon Wea Rev 115:1606–1626 Satyamurty P, Nobre CA, Dias PLS (1998) South America. In: Karoly DJ, Vincent DG (eds) Meteorology of the southern hemisphere. American Meteorology Society, Boston, pp 243–282 Saulo C, Nicolini M, Chou SC (2000) Model characterization of the South American low-level flow during the 1997–1998 springsummer seasons. Clim Dyn 16:867–881 Sestini MF, Alvalá RCS, Mello EMK, Valeriano DM, Chou SC, Nobre CA, Paiva JAC, Reimer ÉS (2002) Vegetation map elaboration for use in numerical models (‘‘Elaboração de mapas de vegetação para utilização em modelos meteorológicos e hidrológicos’’) internal report. Instituto Nacional de Pesquisas Espaciais, São José dos Campos, SP. Brazil 653 Seth A, Rojas M (2003) Simulation and sensitibity in a nested modeling system for South America. part I: reanalysis boundary forcing. J Clim 16:2437–2453 Seth A, Rauscher SA, Camargo SJ, Qian J-H, Pal JS (2007) RegCM3 regional climatologies for South America using reanalysis and ECHAM global model driving fields. Clim Dyn 28:461–480. doi:10.1007/s00382-006-0191-z Silva VBS, Berbery EH (2006) Intense rainfall events affecting the La Plata basin. J Hydrometeor 7:769–787 Solman S, Nuñez M, Cabré MF (2007) Regional climate change experiments over southern South America. I: present climate. Clim Dyn 30:533–552. doi:10.1007/s00382-007-0304-3 Stainforth DA et al (2005) Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature 433:403–406 Stern WF, Miyakoda K (1995) Feasibility of seasonal forecasts inferred from multiple GCM simulations. J Clim 8:1071–1085 Trenberth KE (1997) The definition of El Ninõ. Bull Amer Meteor Soc 78:2771–2777 Trenberth KE, Jones PD, Ambenje P, Bojariu R, Easterling D, Klein Tank A, Parker D, Rahimzadeh F, Renwick JA, Rusticucci M, Soden B, Zhai P (2007) Observations: surface and atmospheric climate change. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp 235-336 Uppala SM et al (2005) The ERA-40 re-analysis. Q J R Meteorol Soc 131:2961–3012 Urrutia R, Vuille M (2009) Climate change projections for the tropical andes using a regional climate model: temperature and precipitation simulations for the end of the twenty-first century. J Geophys Res 114:D2. D02108 Veljović K, Rajković B, Fennessy MJ, Altshuler EL, Mesinger F (2010) Regional climate modeling: should one attempt improving on the large scales? Lateral boundary condition scheme: any impact? Meteor Zeitschrift 19:237–246. doi:10.1127/0941-2948/ 2010/0460 Villar JCE, Ronchail J, Guyot JL, Cochonneau G, Filizola N, Waldo L, De Oliveira E, Pombosa R, Vauchel P (2008) Spatio-temporal rainfall variability in the Amazon basin countries (Brazil, Peru, Bolivia, Colombia, and Ecuador). Int J Climatol Published online in Wiley InterScience Zhao Q, Black TL, Baldwin ME (1997) Implementation of the cloud prediction scheme in the Eta model at NCEP. Weather Forecast 12:697–712 123
Documentos relacionados
Vera et al. (2006)
Figure 1. Seasonal mean precipitation computed for 1970 – 1999 period from observations (UDEL), and IPCC-AR4 models (see list in the text). Contour level is 1 mm day 1, values larger than 2 mm day ...
Leia mais