Search for: All records

Climatological rainfall across much of the Greater Horn of Africa has a bimodal annual cycle characterized by the short rains from October to December and the long rains from March to May. Previous generations of climate models from the Coupled Model Intercomparison Project (CMIP3 and CMIP5) generally misrepresented the bimodal rainfall distribution in this region by generating too much rainfall during the short rains and too little during the long rains. The peak of the long rains in these models also typically showed a pronounced 1-month lag relative to observations. Here, the ability of 21 CMIP6 models to properly simulate the observed, climatological annual cycle of Greater Horn rainfall is examined, comparing results with CMIP5 and CMIP3. As previous work has shown a connection between Greater Horn climatological rainfall biases and model biases in sea surface temperatures (SSTs), pattern correlations of climatological SST biases are also analysed. For the multi-model mean, it is found that the earlier biases in Greater Horn rainfall and associated SSTs persist in CMIP6. Examining only the three best performing models in each CMIP group reveals the CMIP6 models outperform those in CMIP3, with mixed results regarding improvements over CMIP5. For the best performing CMIP6 models,more »the SST and 850 hPa wind biases are reduced over the Indian Ocean relative to the other CMIP6 models examined. No statistically significant relationship was identified between CMIP6 model performance and the horizontal resolution of the model. Combined, these results indicate the importance of properly simulating the annual cycle of SSTs in order to successfully model the observed rainfall annual cycle in the Greater Horn.« less

In much of East Africa, climatological rainfall follows a bimodal distribution characterized by the long rains(March–May) and short rains (October–December). Most CMIP5 coupled models fail to properly simulatethis annual cycle, typically reversing the amplitudes of the short and long rains relative to observations. Thisstudy investigates how CMIP5 climatological sea surface temperature (SST) biases contribute to simulationerrors in the annual cycle of East African rainfall. Monthly biases in CMIP5 climatological SSTs (508S–508N)are first identified in historical runs (1979–2005) from 31 models and examined for consistency. An atmo-spheric general circulation model (AGCM) is then forced with observed SSTs (1979–2005) generating a set ofcontrol runs and observed SSTs plus the monthly, multimodel mean SST biases generating a set of ‘‘bias’’ runsfor the same period. The control runs generally capture the observed annual cycle of East African rainfallwhile the bias runs capture prominent CMIP5 annual cycle biases, including too little (much) precipitationduring the long rains (short rains) and a 1-month lag in the peak of the long rains relative to observations.Diagnostics reveal the annual cycle biases are associated with seasonally varying north–south- and east–west-oriented SST bias patterns in Indian Ocean and regional-scale atmospheric circulation and stability changes,the latter primarily associated with changes in low-levelmore »moist static energy. Overall, the results indicate thatCMIP5 climatological SST biases are the primary driver of the improper simulation of the annual cycle of EastAfrican rainfall. Some implications for climate change projections are discussed« less

Abstract The frequency of heat waves (defined as daily temperature exceeding the local 90th percentile for at least three consecutive days) during summer in the United States is examined for daily maximum and minimum temperature and maximum apparent temperature, in recent observations and in 10 CMIP5 models for recent past and future. The annual average percentage of days participating in a heat wave varied between approximately 2% and 10% in observations and in the model’s historical simulations during 1979–2005. Applying today’s temperature thresholds to future projections, heat-wave frequencies rise to more than 20% by 2035–40. However, given the models’ slight overestimation of frequencies and positive trend rates during 1979–2005, these projected heat-wave frequencies should be regarded cautiously. The models’ overestimations may be associated with their higher daily autocorrelation than is found in observations. Heat-wave frequencies defined using apparent temperature, reflecting both temperature and atmospheric moisture, are projected to increase at a slightly (and statistically significantly) faster rate than for temperature alone. Analyses show little or no changes in the day-to-day variability or persistence (autocorrelation) of extreme temperature between recent past and future, indicating that the future heat-wave frequency will be due predominantly to increases in standardized (using historical period statistics)more »mean temperature and moisture content, adjusted by the local climatological daily autocorrelation. Using nonparametric methods, the average level and spatial pattern of future heat-wave frequency is shown to be approximately predictable on the basis of only projected mean temperature increases and local autocorrelation. These model-projected changes, even if only approximate, would impact infrastructure, ecology, and human well-being.« less