Fig 1.
Predicted number of El Niño and La Niña events during the 21st century compared to historical simulations.
Data are based on 27 CMIP5 models and using the RCP8.5 scenario. The overall sum of predicted El Niño events is +10, while that of La Niñas is -19. Overall, 70% of the models indicate either no change or an increase in the number of El Niño events for the coming century; this figure goes down to 52% for La Niñas (data from 21).
Fig 2.
Unified ENSO proxy (UEP) variability compared to projected PDSI variability across the American Southwest in the coming century.
The UEP (in black) combines ENSO and PDO/IPO Pacific climate variability and is based on 10 commonly used ENSO proxies that were consolidated via Principal Component Analysis to capture the joint features of these reconstructions (data from 23). The multimodel mean summer (JJA) PDSI variability over the American Southwest for 1850–2100 (in red) is based on 17 CMIP5 model projections and using the RCP 8.5 emissions scenario (data from 24).
Fig 3.
North American Drought Atlas (NADA) gridded dataset (http://www.ncdc.noaa.gov/paleo/pdsi.html; https://iridl.ldeo.columbia.edu/SOURCES/.LDEO/.TRL/.NADAv2a-2008/PDSI/datafiles.html) used in this study. The bold black line delineates the western portion of the American Southwest.
Fig 4.
Palmer Drought Severity Index (PDSI) variability in the American Southwest over the past millennia.
The annual time series (a) displays greater variability during the LIA than the MCA, as indicated by the number of years during which variability is above the 90th percentile (b). The time series was filtered using a 10-year high-pass to compute a time series of PDSI variance (c) on the basis of 10-year and 50-year running standard deviations (SD). A wavelet analysis shows the evolution of the power spectrum of tree-ring derived PDSI over the past millennia (d). The black contours are the 10% significance regions, using a red-noise background spectrum [46].
Table 1.
Moments of the distribution of filtered paleoecological records and climate simulations.
Fig 5.
Factors affecting and reflecting hydrologic variability in the American Southwest over the past millennia.
Global solar irradiance reconstruction [48–50] and ice-core based sulfate (SO4) influx in the Northern Hemisphere [51] from volcanic activity (a); mean annual temperature (MAT) reconstructions for the Northern Hemisphere [52], North America [29], and the American Southwest* expressed as anomalies based on 1961–1990 temperature averages (b); changes in ENSO-related variability based on El Junco diatom record [41], oxygen isotopes records from Palmyra [42], and the unified ENSO proxy [UEP; 23] (c); changes in PDSI variability for the American Southwest (d), and changes in winter precipitation variability as simulated by CESM model ensembles 2 to 5 [43]. *Data for the American Southwest temperature reconstruction is from PRISM temperature data from CMIP5 [43].