Search

Variations in southern African precipitation have a major impact on local communities, increasing climate‐related risks and affecting water and food security, as well as natural ecosystems. However, future changes in southern African precipitation are uncertain, with climate models showing a wide range of responses from near‐term projections (2020–2040) to the end of the 21st century (2080–2100). Here, we assess the uncertainty in southern African precipitation change using five Ocean‐Atmosphere General Circulation single model initial‐condition large ensembles (30–50 ensemble members) and four emissions scenarios. We show that the main source of uncertainty in 21st Century projections of southern African precipitation is the internal climate variability. In addition, we find that differences between ensemble members in simulating future changes in the location of the Angola Low explain a large proportion (∼60%) of the uncertainty in precipitation change. Together, the internal variations in the large‐scale circulation over the Pacific Ocean and the Angola Low explain ∼64% of the uncertainty in southern African precipitation change. We suggest that a better understanding of the future evolutions of the southern African precipitation may be achieved by understanding better the model's ability to simulate the Angola Low and its effects on precipitation. Plain Language Summary: The variability of precipitation in southern Africa has a strong impact on local communities, rain‐fed agriculture, food security and water demand, hydropower production, lake levels, ecosystems, and wildlife. Above‐average rainfall increases the risk of flooding, while below‐average rainfall increases the risk of drought. However, future changes in precipitation in southern Africa are poorly understood. Here, we examine the potential sources of uncertainty in southern African precipitation change using five ocean‐atmosphere general circulation single‐model initial‐condition large ensembles and four emissions scenarios. We show that the main source of uncertainty is the simulation of internal climate variability throughout the 21st century. Among potential drivers, we show that the main driver of uncertainty in southern African precipitation change is the future change in the location of the Angola low. A future northward (southward) shift of the Angola Low is associated with a future decrease (increase) in southern African precipitation. We suggest that a better understanding of future changes in southern African precipitation could be achieved by better understanding the impact of internal climate variability on the Angola Low. Key Points: Future changes in southern African precipitation are uncertainThe main source of uncertainty in simulating southern African precipitation change is internal climate variabilityFuture changes in southern African precipitation depend on internal variations in the meridional location of the Angola Low [ABSTRACT FROM AUTHOR]

Future changes in Sahel precipitation are uncertain because of large differences among projections from various models. In order to explore this uncertainty, we use a storyline approach which seeks to identify alternative plausible evolutions of Sahel precipitation and their driving factors. By analyzing projections from the CMIP6 climate models, we show that changes in North Atlantic and in Euro‐Mediterranean temperatures explain up to 60% of the central Sahel precipitation change uncertainty. We then construct several storylines of Sahel precipitation change based on future plausible changes in North Atlantic and Euro‐Mediterranean temperatures. In one storyline, an amplified warming of both the North Atlantic and the Euro‐Mediterranean areas promotes a northward shift of the West African monsoon, increasing precipitation over the central Sahel, while, in the opposite storyline, a moderate warming in both regions is associated with a small change in precipitation over the central Sahel and a decrease in precipitation over the western Sahel, at the end of the 21st century. These results indicate that Sahel precipitation uncertainty will not be substantially reduced unless the uncertainty in the future warming of the North Atlantic and the Euro‐Mediterranean areas is constrained. Plain Language Summary: Variations in the strength of the West African Monsoon (WAM) circulation have societal impacts on around 80 million people from Senegal to Chad. However, future projections of the WAM precipitation are uncertain for the end of the 21st century because of strong differences from one climate model to another. We show here that sources of uncertainty in Sahel precipitation depend on changes in North Atlantic and Euro‐Mediterranean temperature, relative to changes in global mean surface air temperature. We show that differences in how climate models simulate the future warming over the North Atlantic and Euro‐Mediterranean area explain a large proportion of the uncertainty in precipitation change over the central Sahel. We provide a method that could be helpful at selecting models for impact studies, based on their sensitivity to climate change over the North Atlantic ocean and Mediterranean Sea. Key Points: Future changes in Sahel precipitation are uncertainChanges in North Atlantic and Euro‐Mediterranean temperatures explain up to 60% of the Sahel precipitation change uncertaintyUncertainty in changes in Sahel precipitation is associated with uncertainty in the future northward shift of the Saharan heat low [ABSTRACT FROM AUTHOR]

This study analyses the long‐term (1950–2100) observed and projected changes in springtime (March–May) heat waves (HWs) in West Africa under climate change. To that end, 28 climate models participating to the fifth Coupled Model Intercomparison Project (CMIP5) are considered, after a statistical post‐correction of their biases. A multi‐scale approach is proposed, covering the Sahel, Senegal, and three thermally‐coherent zones within Senegal. HWs are defined as a sequence of at least three consecutive days above a moving 95th percentile of current temperature distributions. Climate change over Senegal translates into a general shift of the whole statistical distribution towards higher temperature values, with a general stability in the shape of the distribution. Ongoing mean warming could reach +5°C in 2100 under RCP8.5 scenario, implying that coastal Senegal could experience then a mean climate comparable to the hinterland parts today. HWs have increased in intensity, frequency and duration across Sahel and Senegal over the past years, such intensification being higher on recent decades. Future HWs over all regions present intrinsic properties that radically differ from those observed so far. The severity and length of HWs displayed stationary conditions until the late 1990s, but started increasing since then. Projected changes show marked and rapid increase in these variables, the amplitude of which is primarily RCP‐dependent, and secondarily region‐dependent. For both metrics, the largest changes occur over hinterland Senegal and Sahel. There, under RCP8.5 and after the 2070s, the whole spring season could be considered as a permanent HW lasting 3 months. Along the coast, by contrast, average temperatures are both weaker and more variable, causing more frequent threshold crossings and limiting the duration of HWs. The multi‐scale approach used here highlights contrast within Senegal, which constitutes important information for public policy decision‐makers and its inhabitants in terms of adaptation to climate change. [ABSTRACT FROM AUTHOR]

In the second part of this study, possible impacts of climate change on Surface Solar Radiation (SSR) in Southern Africa (SA) are evaluated. We use outputs from 20 regional climate simulations from five Regional Climate Models (RCM) that participate in the Coordinated Regional Downscaling Experiment program over the African domain (CORDEX-Africa) along with their 10 driving Global Climate Models (GCM) from the Coupled Model Intercomparison Project Phase 5 (CMIP5). Multi-model mean projections of SSR trends are consistent between the GCMs and their nested RCMs. However, this consistency is not found for each GCM/RCM setup. Over the centre of SA, GCMs and RCMs project a statistically significant increase in SSR by 2099 of about + 1 W/m2 per decade in RCP4.5 (+ 1.5 W/m2 per decade in RCP8.5) during the DJF season in their multi-model means. Over Eastern Equatorial Africa (EA-E) a statistically significant decrease in SSR of about − 1.5 W/m2 per decade in RCP4.5 (− 2 W/m2 per decade in RCP8.5) is found in the ensemble means in DJF, whereas in JJA SSR is predicted to increase by about + 0.5 W/m2 per decade under RCP4.5 (+ 1 W/m2 per decade in RCP8.5). SSR projections are fairly similar between RCP8.5 and RCP4.5 before 2050 and then the differences between those two scenarios increase up to about 1 W/m2 per decade with larger changes in RCP8.5 than in RCP4.5 scenario. These SSR evolutions are generally consistent with projected changes in Cloud Cover Fraction over SA and may also related to the changes in atmosphere water vapor content. SSR change signals emerge earlier out of internal variability estimated from reanalyses (European Centre for Medium-Range Weather Forecasts Reanalysis ERA-Interim, ERAIN) in DJF in RCMs than in GCMs, which suggests a higher sensitivity of RCMs to the forcing RCP scenarios than their driving GCMs in simulating SSR changes. Uncertainty in SSR change projections over SA is dominated by the internal climate variability before 2050, and after that model and scenario uncertainties become as important as internal variability until the end of the 21st century. [ABSTRACT FROM AUTHOR]

This study evaluates the performance of climate models in reproducing surface solar radiation (SSR) over Southern Africa (SA) by validating five Regional Climate Models (RCM, including CCLM4, HIRHAM5, RACMO22T, RCA4 and REMO2009) that participated in the Coordinated Regional Downscaling Experiment program over Africa (CORDEX-Africa) along with their ten driving General Circulation Models (GCMs) from the Coupled Model Intercomparison Project Phase 5 over SA. The model simulated SSR was thereby compared to reference data from ground-based measurements, satellite-derived products and reanalyses over the period 1990-2005. Results show that (1) the references obtained from satellite retrievals and reanalyses overall overestimate SSR by up to 10 W/m2 on average when compared to ground-based measurements from the Global Energy Balance Archive, which are located mainly over the eastern part of the southern African continent. (2) Compared to one of the satellite products (Surface Solar Radiation Data Set—Heliosat Edition 2; SARAH-2): GCMs overestimate SSR over SA in terms of their multi-model mean by about 1 W/m2 (compensation of opposite biases over sub-regions) and 7.5 W/m2 in austral summer and winter respectively; RCMs driven by GCMs show in their multimodel mean underestimations of SSR in both seasons with Mean Bias Errors (MBEs) of about − 30 W/m2 in austral summer and about − 14 W/m2 in winter compared to SARAH-2. This multi-model mean low bias is dominated by the simulations of the CCLM4, with negative biases up to − 76 W/m2 in summer and − 32 W/m2 in winter. (3) The discrepancies in the simulated SSR over SA are larger in the RCMs than in the GCMs. (4) In terms of trend during the "brightening" period 1990-2005, both GCMs and RCMs (driven by European Centre for Medium-Range Weather Forecasts Reanalysis ERA-Interim, short as ERAINT and GCMs) simulate an SSR trend of less than 1 W/m2 per decade. However, variations of SSR trend exist among different references data. (5) For individual RCM models, their SSR bias fields seem rather insensitive with respect to the different lateral forcings provided by ERAINT and various GCMs, in line with previous findings over Europe. (6) Biases in SSR are overall qualitatively consistent with those in total cloud cover. The information obtained in present study is of crucial importance for understanding future climate projections of SSR and for relevant impact studies. [ABSTRACT FROM AUTHOR]

This paper documents the capability of the ARW/WRF regional climate model to regionalize near-surface atmospheric variables at high resolution (8 km) over Burgundy (northeastern France) from daily to interannual timescales. To that purpose, a 20-year continuous simulation (1989-2008) was carried out. The WRF model driven by ERA-Interim reanalyses was compared to in situ observations and a mesoscale atmospheric analyses system (SAFRAN) for five near-surface variables: precipitation, air temperature, wind speed, relative humidity and solar radiation, the last four variables being used for the calculation of potential evapotranspiration (ET). Results show a significant improvement upon ERA-Interim. This is due to a good skill of the model to reproduce the spatial distribution for all weather variables, in spite of a slight over-estimation of precipitation amounts mostly during the summer convective season, and wind speed during winter. As compared to the Météo-France observations, WRF also improves upon SAFRAN analyses, which partly fail at showing realistic spatial distributions for wind speed, relative humidity and solar radiation-the latter being strongly underestimated. The SAFRAN ET is thus highly under-estimated too. WRF ET is in better agreement with observations In order to evaluate WRF's capability to simulate a reliable ET, the water balance of thirty Douglas-fir stands was computed using a process-based model. Three soil water deficit indexes corresponding to the sum of the daily deviations between the relative extractible water and a critical value of 40 % below which the low soil water content affects tree growth, were calculated using the nearest weather station, SAFRAN analyses weather data, or by merging observation and WRF weather variables. Correlations between Douglas-fir growth and the three estimated soil water deficit indexes show similar results. These results showed through the ET estimation and the relation between mean annual SWDI and Douglas-fir growth index that the main difficulties of the WRF model to simulate soil water deficit is mainly attributable to its precipitation biases. In contrast, the low discrepancies between WRF and observations for air temperature, wind speed, relative humidity and solar radiation make then usable for the water balance and ET computation. [ABSTRACT FROM AUTHOR]

This study presents an evaluation of the ability of 10 regional climate models (RCMs) participating in the COordinated Regional climate Downscaling Experiment-Africa to reproduce the present-day spatial distribution of annual cycles of precipitation over the South African region and its borders. As found in previous studies, annual mean precipitation is quasi-systematically overestimated by the RCMs over a large part of southern Africa south of about 20°S and more strongly over South Africa. The spatial analysis of precipitation over the studied region shows that in most models the distribution of biases appears to be linked to orography. Wet biases are quasi-systematic in regions with higher elevation with inversely neutral to dry biases particularly in the coastal fringes. This spatial pattern of biases is particularly obvious during summer and specifically at the beginning of the rainy season (November and December) when the wet biases are found to be the strongest across all models. Applying a k-means algorithm, a classification of annual cycles is performed using observed precipitation data, and is compared with those derived from modeled data. It is found that the in-homogeneity of the spatial and temporal distribution of biases tends to impact the modeled seasonality of precipitation. Generally, the pattern of rainfall seasonality in the ensemble mean of the 10 RCMs tends to be shifted to the southwest. This spatial shift is mainly linked to a strong overestimation of convective precipitation at the beginning of the rainy season over the plateau inducing an early annual peak and to an underestimation of stratiform rainfall in winter and spring over southwestern South Africa. [ABSTRACT FROM AUTHOR]

The capability of a current state-of-the-art regional climate model for simulating the diurnal and annual cycles of rainfall over a complex subtropical region is documented here. Hourly rainfall is simulated over Southern Africa for 1998-2006 by the non-hydrostatic model weather research and forecasting (WRF), and compared to a network of 103 stations covering South Africa. We used five simulations, four of which consist of different parameterizations for atmospheric convection at a 0.5 × 0.5° resolution, performed to test the physic-dependency of the results. The fifth experiment uses explicit convection over tropical South Africa at a 1/30° resolution. WRF simulates realistic mean rainfall fields, albeit wet biases over tropical Africa. The model mean biases are strongly modulated by the convective scheme used for the simulations. The annual cycle of rainfall is well simulated over South Africa, mostly influenced by tropical summer rainfall except in the Western Cape region experiencing winter rainfall. The diurnal cycle shows a timing bias, with atmospheric convection occurring too early in the afternoon, and causing too abundant rainfall. This result, particularly true in summer over the northeastern part of the country, is weakly physic-dependent. Cloud-resolving simulations do not clearly reduce the diurnal cycle biases. In the end, the rainfall overestimations appear to be mostly imputable to the afternoon hours of the austral summer rainy season, i.e., the periods during which convective activity is intense over the region. [ABSTRACT FROM AUTHOR]

This study documents methodological issues arising when downscaling modes of large-scale atmospheric variability with a regional climate model, over a remote region that is yet under their influence. The retained case study is El Niño Southern Oscillation and its impacts on Southern Africa and the South West Indian Ocean. Regional simulations are performed with WRF model, driven laterally by ERA40 reanalyses over the 1971-1998 period. We document the sensitivity of simulated climate variability to the model physics, the constraint of relaxing the model solutions towards reanalyses, the size of the relaxation buffer zone towards the lateral forcings and the forcing fields through ERA-Interim driven simulations. The model's internal variability is quantified using 15-member ensemble simulations for seasons of interest, single 30-year integrations appearing as inappropriate to investigate the simulated interannual variability properly. The incidence of SST prescription is also assessed through additional integrations using a simple ocean mixed-layer model. Results show a limited skill of the model to reproduce the seasonal droughts associated with El Niño conditions. The model deficiencies are found to result from biased atmospheric forcings and/or biased response to these forcings, whatever the physical package retained. In contrast, regional SST forcing over adjacent oceans favor realistic rainfall anomalies over the continent, although their amplitude remains too weak. These results confirm the significant contribution of nearby ocean SST to the regional effects of ENSO, but also illustrate that regionalizing large-scale climate variability can be a demanding exercise. [ABSTRACT FROM AUTHOR]

This study aims at quantifying seasonal biases of regional climate model outputs during southern African summer, against a dense in situ measurement network (daily rain-gauge and surface air temperature records, and 12 h UTC radiosondes), and uncertainties associated with some physical parameterizations. Using the non-hydrostatic Advanced Research Weather Forecast (WRF) laterally forced by ERA40 reanalysis, twenty-seven experiments configured with three schemes of cumulus (CU), planetary boundary layer (PBL) and microphysics (MP), are performed at 35 km horizontal resolution during the core of a summer rainy season (December 1993 to February 1994 season) representative of the South African rainfall climatology. WRF simulates accurately seasonal large-scale rainfall patterns, as well as seasonal gradients of South African rainfall and 2-m temperature, and seasonal vertical profiles of the air temperature and humidity. However seasonal biases fluctuate strongly from an experiment to another, denoting considerable uncertainties generated by the physical package. Rainfall amounts are the most sensitive parameter to the tested schemes. Their geography, intensity, and intraseasonal characteristics are predominantly sensitive to CU schemes, and much less to PBL and MP schemes. Some CU-PBL combinations produce additive effects, which can dramatically either reduce or increase biases. Satisfactory configurations are found for South African climate, which would not have been possible without testing numerous physical parameterizations. [ABSTRACT FROM AUTHOR]

The authors analyze to what extent the internal variability simulated by a regional climate model is sensitive to its physical parameterizations. The influence of two convection schemes is quantified over southern Africa, where convective rainfall predominates. Internal variability is much larger with the Kain-Fritsch scheme than for the Grell-Dévényi scheme at the seasonal, intraseasonal, and daily time scales, and from the regional to the local (grid point) spatial scales. Phenomenological analyses reveal that the core (periphery) of the rain-bearing systems tends to be highly (weakly) reproducible, showing that it is their morphological features that induce the largest internal variability in the model. In addition to the domain settings and the lateral forcing conditions extensively analyzed in the literature, the physical package appears thus as a key factor that modulates the reproducible and irreproducible components of regional climate variability. [ABSTRACT FROM AUTHOR]