Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Water availability is a major constraint to rural agriculture in the Bolivian Altiplano. The ability of medium and high-resolution CMIP3 models to simulate observed relationships between interannual Altiplano summer precipitation anomalies and large-scale patterns of precipitation, 200-hPa winds and 200-hPa geopotential height is examined. Altiplano precipitation anomalies are known to be related to upper-level wind anomalies, with westerly (easterly) anomalies associated with deficit (excess) rainfall. A majority of models are able to simulate an easterly/wet-westerly/dry relationship in response to changes in the zonal flow produced by fluctuations in tropical tropospheric temperatures that affect the meridional temperature gradient, consistent with observations.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The Bolivian Altiplano is a semi-arid region, located on a high plateau in the central Andes of South America. Water availability is a major constraint for agriculture, where approximately 50% of the rural population employs traditional farming methods, depending entirely on water supplied by spring and summer rainfall (Garcia et al., 2007). Over 60% of annual precipitation falls during the summer months (DJF) in association with the South American Monsoon (SAM) and interannual variations are large (e.g. Garreaud, 1999; Lenters and Cook, 1999; Garreaud and Aceituno, 2001; Garreaud et al., 2003). This research examines the ability of the medium and high-resolution coupled models from the World Climate Research Program (WCRP) Coupled Model Intercomparison Project version 3 (CMIP3) (Meehl et al., 2007) to simulate the observed relationships between Altiplano interannual precipitation anomalies and large-scale atmospheric variables (i.e. precipitation, 200-hPa winds, and 200-hPa geopotential height) for summer (DJF). Because precipitation variability is so important to agriculture and water resources in the Altiplano, this research also examines the evolution of these relationships through the 21st century under conditions of greenhouse warming.
A number of previous studies have investigated Altiplano summertime precipitation variability and its relationship to the atmospheric circulation (e.g. Garreaud, 1999; Lenters and Cook, 1999; Vuille, 1999; Garreaud and Aceituno, 2001; Falvey and Garreaud, 2005). On interannual timescales, precipitation variability is related to changes in the mid- and upper-level wind patterns; easterly flow favours moisture transport from the lowlands east of the Altiplano, while westerly flow inhibits moisture transport from the east. These changes in the zonal flow are associated with changes in the meridional temperature gradient between the tropics and mid-latitudes (Garreaud and Aceituno, 2001; Garreaud et al., 2003). Previous studies have shown a relationship between Altiplano precipitation anomalies and El Niño Southern Oscillation (ENSO) with wet (dry) summers related to cooling (warming) of the tropical troposphere and associated changes in the upper-level zonal flow during La Niña (El Niño) events (Garreaud and Aceituno, 2001; Garreaud et al., 2003). The strength and position of the Bolivian High, which develops over the central Andes in summer, also affects the direction of low-level flow east of the Altiplano, influencing precipitation variability on various timescales (Lenters and Cook, 1999). Northward (southward) displacement of the Bolivian High is associated with low-level southerly (northwesterly) flow and dry, stable (humid, unstable) conditions, which are unfavourable (favourable) for convection on the Altiplano. In this study, our focus is on the upper-level flow.
Vera and Silvestri (2009) have examined precipitation interannual variability for South America using a subset of CMIP3 models, and show deficiencies in the representation of precipitation variability in the South Atlantic convergence zone (SACZ), located over southeastern South America. However, the models are able to reproduce interannual precipitation variability for the SAM region, capturing the northward migration of the variability peak during fall and winter, and its southward return in spring and summer. Our main objective in this study is to examine the ability of the coupled climate models to simulate relationships between Altiplano precipitation anomalies and large-scale precipitation and atmospheric circulation patterns using regression analysis methods employed in previous observational studies (e.g. Garreaud, 1999; Garreaud and Aceituno, 2001; Garreaud et al., 2003). We also examine future variability in the context of mean changes in the circulation. It is our firm opinion that a first-order study of the ability of the global models to represent the present-day relationships and their evolution into the future is a necessary precursor to dynamical downscaling.
Studies of the precipitation annual cycle also provide useful context for this analysis. The CMIP3 models are able to simulate the main features of the precipitation annual cycle for South America, but they tend to overestimate precipitation amounts near the Andes (Vera et al., 2006) due to the coarse resolution of the models. While there is some disagreement regarding future changes in annual precipitation for the Altiplano, results suggest a wetter summer (Seth et al., 2010). Projections of climate extremes for the Altiplano suggest that summer precipitation will be less frequent but more intense (Thibeault et al., 2010).
The Altiplano experiences water shortages due to high evapotranspiration rates and poor soils, as well as normally low amounts of precipitation, which can lead to food shortages for humans and livestock (Garcia et al., 2007). In semi-arid regions, the projected large increases in temperature are likely to result in higher evapotranspiration rates, reducing soil moisture even where precipitation experiences a small increase (Wang, 2005). Temperatures in the Altiplano are expected to increase by approximately 4 °C throughout the annual cycle in the high emissions A2 scenario (Seth et al., 2010). The increased precipitation intensity projected for summer (Thibeault et al., 2010) will likely result in an increase in the runoff ratio, resulting in soil moisture reductions. Furthermore, the small increase projected for total summertime precipitation (Seth et al., 2010; Thibeault et al., 2010) may be insufficient to offset the effects of higher temperatures on soil moisture, exacerbating reductions related to increased precipitation intensity. To test this hypothesis, this research also examines summer soil moisture projections for the Altiplano.
The remainder of the paper is structured as follows. The datasets and methods used in the analysis are described in Section 2. Section 3 presents the results, beginning with the observed relationships between Altiplano summer precipitation anomalies and large-scale atmospheric variables. Results for the 20th century simulations and the 21st century projections follow, including soil moisture projections. The results are followed by a summary and conclusions in Section 4.
2. Data and methods
This research focuses on precipitation variability for the northern Altiplano; specifically the relationships between Altiplano precipitation anomalies and anomalies of large-scale precipitation, 200-hPa zonal (u) and meridional (v) winds, hereafter designated as (u,v), and 200-hPa geopotential height. Simulations from eleven CMIP3 global coupled climate models are analysed for 20th century historical simulations and the 21st century A2 emissions scenario using a single realisation from each model. Because the Altiplano region is characterized by complex topography across small spatial scales, models selected are limited to those with medium to high resolution and include those employed in our earlier analysis of climate extremes (Thibeault et al., 2010). Two additional models (ECHAM5 and HadCM3), for which soil moisture data are available, are included in this analysis, however soil moisture data are not available for all models (Table I). All the models selected have grid-point elevations in the Altiplano above 2000 m, with the exception of one grid point in CCSM3 (1828 m). The sensitivity of area-averaged temperature to this grid point is negligible compared to the warm bias in the model (Thibeault et al., 2010), so it is retained in this study. A2 scenario data are not available for the MIROC-HiRes model. Altiplano precipitation anomalies and Altiplano zonal wind anomalies are averaged for the area 16–19S by 67–70 W. Extending the region beyond this area increases the number of lower-elevation grid points, increasing wet biases, especially in higher-resolution models (Thibeault et al., 2010). Relationships between Altiplano precipitation anomalies and large-scale atmospheric variables are calculated for the region 120E–10E and 30N–45S, depicting the large-scale patterns of precipitation and upper-level circulation that are related to Altiplano rainfall anomalies. This region covers much of the tropical Pacific and Atlantic oceans as well as the South American continent and extends into the Southern Hemisphere mid-latitudes.
Table I. CMIP3 coupled ocean-atmosphere models used in this analysis. Atmospheric resolution is shown in longitude by latitude degrees, respectively. Ocean resolution is defined as the number of grids in longitude and latitude, respectively
Data not provided for A2 scenario.
Soil moisture data not available.
National Center for Atmospheric Research
1.4 × 1.4
320 × 395
Meteo-France, Centre National de Recherches Meteorologiques
Twentieth century model simulations are compared to observed results from Climate Prediction Center Merged Analysis of Precipitation (CMAP) (Xie and Arkin, 1996) data for the period of 1979–2004 and the University of East Anglia Climatic Research Unit Global 0.5° Monthly Time Series, Version 2.1 (CRU-TS2.1) (Mitchell and Jones, 2005) covering 1950–2002. Twentieth century model simulations of 200-hPa winds and geopotential height are compared to observed results for monthly NCEP/NCAR Reanalysis 1 (Kalnay et al., 1996).
Regression maps are used to show the relationships between Altiplano precipitation anomalies and the large-scale patterns of precipitation, 200-hPa winds, and 200-hPa geopotential height for summer (DJF). Prior to performing the regression, all variables are detrended. The regression coefficients are tested for significance at the 90% confidence level using a Student's t-test. Regressions computed for observations cover the period of 1979–2004 and employ CMAP precipitation data. Regressions for twentieth century simulations cover the period of 1970–1999.
The relationships between area-averaged precipitation and 200-hPa zonal wind anomalies for the Altiplano (16S–19S by 67 W–70 W) are also examined by calculating Spearman's rho. Anomalies are calculated with respect to a base period of 1970–1999 and are detrended before analysis. Relationships are tested for their significance at the 90% confidence level. Observed relationships are examined for 1950–2002, and employ CRU-TS2.1 precipitation data. Simulated relationships are examined for 1950–1999 and 2050–2099 for each model.
Time evolution of Altiplano area-averaged (16S–19S by 67 W–70 W) summer (DJF) soil moisture is examined by calculating multi-model average time series for the period of 1970–2099. Time series of normalized soil moisture anomalies are first calculated for each model in each scenario with respect to a base period of 1970–1999. The normalized anomalies are then averaged to produce a multi-model time series of soil moisture anomalies for each scenario. Each time series is smoothed by an 11-year running average, indicating the direction of change. The use of an 11-year running average allows the mean to be calculated using 5 years on either side of the filter's centre. The standard deviation around the ensemble mean is used as a representation of inter-model variability. MIROC-HiRes is excluded from the soil moisture analysis because the A2 scenario is not available.
3.1. Observed Altiplano precipitation variability
Regression maps between observed Altiplano precipitation anomalies and precipitation, and 200-hPa winds and geopotential height (Figure 1) indicate a statistically significant relationship between wet periods on the Altiplano and easterly wind anomalies over the central Andes. This result is similar to that of Garreaud and Aceituno (2001), shown in their Figure 7a. The pattern of precipitation anomalies shown in Figure 1(a) resembles the spatial pattern of precipitation anomalies typical for the cold phase of ENSO: dry anomalies in the central Pacific and wet anomalies in the western Pacific. Zonal wind anomalies over the central Andes are related to changes in the temperature gradient between the tropics and mid-latitudes (Garreaud, 1999; Garreaud and Aceituno, 2001). Figure 1(b) is similar to Figure 7b in Garreaud and Aceituno (2001), showing the relationship between zonal wind anomalies over the central Andes and tropospheric temperatures. Spearman's rho for summer area-averaged Altiplano precipitation and 200-hPa zonal wind (1950–2002) (Table II) is statistically significant, consistent with the easterly/wet-westerly/dry relationship between upper-level zonal wind anomalies and Altiplano precipitation previously documented in other studies (e.g. Garreaud and Aceituno, 2001).
Table II. Spearman's rho for detrended Altiplano precipitation versus detrended 200-hPa zonal wind for summer (DJF). Observed values are shown for 1950–2002. Simulated values are shown for 1950–1999 and 2050–2099. Bold (italic) indicates values that are statistically significant at the 95% (90%) confidence level
3.2. Twentieth century simulated Altiplano precipitation variability
Most of the models are able to simulate 200-hPa easterly wind anomalies over the central Andes during anomalously wet periods on the Altiplano (Figure 2, left column). Results shown in Figure 2 are limited to those models that better simulate the patterns of the present relationships and demonstrate statistically significant relationships between area-averaged Altiplano precipitation and zonal winds (Table II, i.e. CCSM3, CNRM-CM3, ECHAM5, GFDL2.0, HadCM3, MRI, and MIROC-HiRes, hereafter selected models). These easterly anomalies are associated with a large-scale pattern of dry precipitation anomalies in the central Pacific and wet anomalies in the western Pacific, resembling that typically observed in a cold ENSO phase, but the models vary in how they simulate the spatial extent of these anomalies in comparison to the observed pattern (Figure 1(a)). The tropospheric cooling (represented by 200-hPa geopotential height anomalies, Figure 2, right column) observed during wet periods is simulated by the selected models, but the patterns vary among the models as well as in comparison to observations (Figure 1(b)).
The Spearman's rho correlations for 1950–1999 (Table II) show that a majority of models are able to simulate a negative relationship between Altiplano summer rainfall and 200-hPa zonal winds, consistent with the observed easterly/wet-westerly/dry relationship. The relationship is significant in CCSM3, CNRM-CM3, ECHAM5, GFDL2.0, HadCM3, MIROC-HiRes, and MRI, but is somewhat weaker than observed in GFDL2.0, MIROC-HiRes, and MRI.
3.3. Twentyfirst century projections
By 2050–2099, the selected models having A2 scenario data continue to simulate a statistically significant negative relationship between Altiplano summer precipitation and zonal wind anomalies (Table II), indicating that the fundamental requirement that easterly winds transport moisture to the region does not change in a warming climate, but models do not clearly demonstrate whether it will strengthen or weaken.
Garreaud and Aceituno (2001) documented that warming in the tropical troposphere, such as that which occurs during the warm phase of ENSO, is associated with a strengthening of westerly winds over the central Andes, resulting in dry summer conditions for the Altiplano. The projected changes in the mean circulation (increase in tropical tropospheric warming described by Vecchi and Soden (2007) and its influence on upper-level wind anomalies) may explain the projected decrease in the frequency of summertime precipitation events in the Altiplano.
Probability density functions (PDFs) of Altiplano (16S–19S by 67 W–70 W) summer (DJF) zonal winds in the selected models having A2 scenario data are examined for 1950–1999 and 2050–2099 (Figure 3) to evaluate whether the models project an increase in the frequency of westerly winds, possibly explaining the projected reduction in the frequency of summer rainfall events (Thibeault et al., 2010). PDFs show that a majority of the selected models project an increase in upper-level westerly winds for 2050–2099 in comparison to 1950–1999. Kolmogorov-Smirnov tests indicate that this change in distribution is statistically significant in all of the selected models at the 95% confidence level, except for CNRM-CM3 (p-value of 0.549). These results suggest that a decrease in the frequency of easterly wind anomalies may explain projections for less frequent Altiplano summer rainfall.
Seth et al. (2010) and Thibeault et al. (2010) have previously examined changes in the Altiplano annual cycle of precipitation. Both studies (using different subsets of the CMIP3 models) suggest a decrease in total springtime precipitation and increase in total summer precipitation. Here, we further explore precipitation projections for the Altiplano by comparing two sets of models: all the models employed in this analysis, and the selected models with A2 scenario data available. Changes in the annual cycle are examined for the difference between 2070–2099 and 1970–1999 monthly precipitation. The multi-model statistics for present day and future climate are represented by boxplots, providing information on variability among the models (Figure 4). The multi-model averages for all of the models and the selected models project decreased precipitation in November and December. Results based on the selected models suggest that Altiplano precipitation will decrease during the late spring and early summer by the late 21st century. These new results differ from our previous conclusion for January–March precipitation (Thibeault et al., 2010). Multi-model median results from our previous work (Thibeault et al., 2010, Figure 9f) suggest decreased rainfall in December and January, with little change in February precipitation, which is qualitatively similar to our new results using the selected models. In the selected models, there is a larger reduction in December and January precipitation in both the multi-model means and medians. February multi-model mean and median results for the selected models show little change in precipitation by 2070–2099.
These results suggest that the easterly/wet-westerly/dry relationship will likely remain important to Altiplano precipitation variability in the future. The physical constraint that moisture transport is from the east will not change. Correlation analysis of the models suggests that the relationship between precipitation and zonal wind anomalies will continue to exist in the future. PDFs of summer 200-hPa zonal winds suggest that changes in the mean circulation will result in an increase in the frequency of westerly wind anomalies (compared to the present day climatology) by the latter half of the 21st century. This result may explain the projections for less frequent rainfall events in the Altiplano (Thibeault et al., 2010). A reduction in easterly wind anomalies is also consistent with projections for reduced summer precipitation in the selected models, suggesting that dry summer conditions will become more frequent in the future.
3.4. Soil moisture projections
Multi-model soil moisture projections (Figure 5) show that Altiplano summer soil moisture starts to decrease by approximately 2020 and is reduced about 1.0 standard deviation (SD) by the end of the century. Models were limited to the selected models that have A2 scenario soil moisture data available (i.e. CCSM3, ECHAM5, GFDL2.0, and HadCM3). The shaded area in Figure 5, showing the width of 1.0 SD of the ensemble mean, is below the zero line for much of the latter 21st century, indicating agreement among the models for summer drying. This result is consistent with projections for reduced December and January precipitation in the selected models. Increased summertime rainfall intensity will likely lead to further reductions in soil moisture due to an increase in the runoff ratio as the rainfall rate exceeds the infiltration capacity of the soil. Higher temperatures will also increase evapotranspiration rates in the Altiplano, further reducing soil moisture. Increased summer drying will likely have serious implications for traditional rural agriculture and water resources in the region.
4. Summary and conclusions
This research has examined the ability of a subset of medium and high-resolution CMIP3 models to simulate the large-scale patterns of precipitation, 200-hPa winds (u,v), and 200-hPa geopotential height related to interannual summer (DJF) precipitation anomalies in the Bolivian Altiplano. Model simulations were compared to observed estimates of CMAP and CRU precipitation and NCEP/NCAR reanalysis 200-hPa winds (u,v) and geopotential height. 21st century projections were also examined to explore whether and how the observed relationships might change with greenhouse warming. This research examined precipitation projections for the annual cycle, comparing results for all the models to those that better simulate the observed relationships important to precipitation variability. Soil moisture projections for the Altiplano were also examined.
Our results indicate that despite the models' inadequacies in resolving the topography of the Altiplano, the precipitation response to large-scale forcing is captured in the selected models, but to varying degrees. Results for the 20th century are summarized below.
A majority of models are able to simulate: (1) an easterly/wet-westerly/dry relationship in DJF, (2) the observed relationship between wet (dry) summers on the Altiplano and cooling (warming) of the tropical troposphere that affects the meridional temperature gradient and associated upper-level zonal flow anomalies, consistent with observations, and (3) easterly (westerly) anomalies associated with the large-scale pattern of precipitation anomalies in the central Pacific and western Pacific that resemble those typically observed in a cold (warm) ENSO phase.
Correlation analysis shows that a majority of models simulate a statistically significant negative relationship between Altiplano precipitation and 200-hPa zonal winds during summer, but they vary in how they simulate the strength of the relationship.
Modeled results for the 21st century are summarized below.
The negative relationship between summer precipitation and zonal wind anomalies will continue to be important in the future, but models do not clearly demonstrate a change in the strength of the relationship.
PDFs of upper-level zonal winds in models that simulate a statistically significant negative relationship between Altiplano precipitation and 200-hPa zonal winds suggest an increase in the frequency of westerly winds (due to changes in the mean circulation), which are unfavourable for precipitation in the Altiplano.
Higher temperatures projected for the tropical troposphere may be responsible for the increase in the frequency of westerly wind anomalies during summer, possibly explaining projections for less frequent summer precipitation (Thibeault et al., 2010), as well as the summer precipitation reductions projected by models that better capture observed Altiplano precipitation variability.
Despite the uncertainty in precipitation projections, there is good agreement among the selected models for Altiplano summer soil moisture projections. Soil moisture is expected to decrease in summer (DJF) from approximately 2020 onward due to: (1) reduced late spring and early summer precipitation, (2) an increase in the runoff ratio with more intense rainfall, (3) increased evapotranspiration, and (4) a decrease in precipitation frequency, allowing more time for drying and hardening of the soil between rainy episodes, which will further increase runoff. Soil moisture reductions are likely to have serious implications for those who practice traditional methods of agriculture and raising livestock in the Altiplano.
Our results suggest that the known relationships between Altiplano precipitation anomalies and large-scale atmospheric variables will remain important in the future. The requirement that moisture transport is from the east will not change. The expected increase in tropical tropospheric warming and its influence on the atmospheric circulation may lead to a reduction in upper-level easterly wind anomalies during summer, resulting in reduced precipitation in late spring and early summer as well as less frequent summertime rainfall events in the Altiplano. Consistencies between the results presented here and expected large-scale changes in atmospheric circulation patterns provide a measure of confidence in the projections. These results have serious implications for water resources and food security in the Altiplano, but they should be taken with caution. Further testing of these results with improved higher-resolution models will be required in order to better understand the future of precipitation variability in the Altiplano.
The authors thank two anonymous reviewers for their constructive comments that have greatly improved the quality of this research. The authors also thank the international modeling groups for providing their data for analysis, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) for collecting and archiving the model data, the JSC/CLIVAR Working Group on Coupled Modelling (WGCM) and their Coupled Model Intercomparison Project (CMIP3) and Climate Simulation Panel for organising the model data analysis activity, and the IPCC WG1 TSU for technical support. The IPCC Data Archive at Lawrence Livermore National Laboratory is supported by the Office of Science, US Department of Energy. This research was supported by a Long-Term Research award (LTR-4) in the Sustainable Agriculture and Natural Resource Management (SANREM) Collaborative Research Program (CRSP) with funding from USAID. Portions of this research were also supported by a University of Connecticut Center for Environmental Sciences and Engineering (CESE) Multidisciplinary Environmental Research Award.