Climate Dynamics

, Volume 42, Issue 11–12, pp 3015–3038 | Cite as

COSMO-CLM (CCLM) climate simulations over CORDEX-Africa domain: analysis of the ERA-Interim driven simulations at 0.44° and 0.22° resolution

  • Hans-Jürgen Panitz
  • Alessandro Dosio
  • Matthias Büchner
  • Daniel Lüthi
  • Klaus Keuler
Article

Abstract

We present the results of the application of the COSMO-CLM Regional Climate Model (CCLM) over the CORDEX-Africa domain. Two simulations were performed driven by the ERA-Interim reanalysis (1989–2008): the first one with the standard CORDEX spatial resolution (0.44°), and the second one with an unprecedented high resolution (0.22°). Low-level circulation and its vertical structure, the geographical and temporal evolution of temperature and precipitation are critically evaluated, together with the radiation budget and surface energy fluxes. CCLM is generally able to reproduce the overall features of the African climate, although some deficiencies are evident. Flow circulation is generally well simulated, but an excessive pressure gradient is present between the Gulf of Guinea and the Sahara, related to a marked warm bias over the Sahara and a cold bias over southern Sahel. CCLM underestimates the rainfall peak in the regions affected by the passage of the monsoon. This dry bias may be a consequence of two factors, the misplacement of the monsoon centre and the underestimation of its intensity. The former is related to the northern shift of the West African Heat Low. On the other hand, the underestimation of precipitation intensity may be related to the underestimation of the surface short-wave radiation and latent heat flux. The increase of the model resolution does not bring evident improvements to the results for monthly means statistics. As a result, it appears that 0.44° is a suitable compromise between model performances and computational constrains.

Keywords

COSMO-CLM Regional Climate Model CORDEX-Africa High resolution simulation ERA-Interim driven evaluation run 

1 Introduction

Africa is one of the most vulnerable continents to weather and climate variability (IPCC 2007). Over the past century West Africa has been affected by significant climate anomalies, which led, for instance, to the severe droughts in the 1970s and 1980s (e.g. Redelsperger et al. 2006). According to Li et al. (2012) the interdecadal variation of the West African Monsoon (WAM) has increased, with a time of change around 1998. Recently, other African regions such as the Horn of Africa have been affected by severe droughts. Projected climate change and low adaptive capacity may lead to even more severe impacts on many vital sectors such as agriculture, water management, and health (IPCC 2007).

Despite Global Climate Models (GCMs) have demonstrated the ability to generally replicate the main characteristics of the African climate over the second half of the twentieth century, at the time of the IPCC 4th Assessment Report (2007) they still presented significant deficiencies; for instance, the large majority of models overestimated precipitation by more than 20 % on average (and in some cases by as much as 80 %) over a wide area often extending into equatorial Africa. Regional climate Models (RCM) mostly focused either on Southern or West Africa (e.g. Jenkins et al. 2005; Afiesimama et al. 2006; Abiodun et al. 2008), where they generally improved the climate simulations by GCMs but also shared similar biases (IPCC 2007).

Since then, the scientific community has undertaken a comprehensive effort in data collecting and modeling activities focused primarily on the West Africa region. Such activities included the West African Monsoon Modelling and Evaluation (WAMME) initiative (Druyan et al. 2010; Xue et al. 2010) the African Multidisciplinary Monsoon Analysis (AMMA) (Redelsperger et al. 2006; Ruti et al. 2011) and the Ensembles-based prediction of Climate Changes and Their Impacts (ENSEMBLES) (Paeth et al. 2011).

The WAM is one of the key elements of the African climate; basically, it consists in a seasonal change in atmospheric circulation associated to high variability (both temporal and spatial) in precipitation. The monsoon system is driven by a complex interaction of atmosphere, ocean, and land-surface, initiated by differential heating of the ocean and land surface. The resulting low-level southwesterly flow advects relatively cool moist air from the Gulf of Guinea onto the hot dry continent (e.g. Steiner et al. 2009). A characteristic feature of the WAM is the African Easterly Jet (AEJ), a mid-tropospheric flow centered around 600–700 hPa at a latitude of around 15°N. The position and strength of the AEJ can influence the large-scale circulation in the region (Cook 1999).

It is very challenging for GCMs and RCMs to replicate the multitude of physical processes and the complexity of their feedbacks, which span multiple temporal and spatial scales, over such a large and heterogeneous continent. As pointed out by e.g. Xue et al. (2010), a systematic evaluation of the performances of climate models by fully exploiting the observational data, is still needed.

Recently, a new initiative endorsed by the World Climate Research Programme has emerged in order to produce an improved generation of regional climate change projections world-wide: the Coordinated Regional climate Downscaling Experiment (CORDEX) (Giorgi et al. 2009). Due to its vulnerability to climate variability, the significant impacts of projected climate change, and the general lack of climate projections based on Regional Climate Downscaling tools, Africa was selected as the first target region for the CORDEX activity.

Recent works by Hernández-Díaz et al. (2012) and Nikulin et al. (2012) presented the results of RCMs applied to the CORDEX-Africa domain; in particular, Nikulin et al. (2012) presented the first analysis of an ensemble of RCM simulations, focusing on the evaluation of the precipitation climatology; results indicate that RCMs are generally able to capture the main climatological features of precipitation, but substantial biases are still present depending on the model, region, and season.

In this work we present the results of the application of the COSMO-CLM (hereafter, CCLM) RCM to the CORDEX-Africa domain. We not only critically analyze the model’s ability in simulating the geographical and temporal distribution of temperature and precipitation, but we also analyze the radiation budget and surface fluxes, which have been proved to be crucial mechanisms in the formation, duration, and movement of the WAM (e.g. Steiner et al. 2009). Moreover, we compare the performances of the model in the standard CORDEX-Africa configuration (e.g. with a spatial resolution of 0.44°) to a new, and to our knowledge, unprecedented, high resolution (0.22°) simulation.

The paper is structured as follows: Sect. 2 describes the model set-up and the data used in the evaluation. In Sect. 3 results are shown and discussed. Concluding remarks are presented in Sect. 4.

2 Model and data

2.1 Model description and setup

The three-dimensional nonhydrostatic regional climate model CCLM (formerly CLM) is the climate version of the operational weather forecast model Consortium for Small-scale Modeling (COSMO, http://cosmo-model.org), (Baldauf et al. 2011) of the German Weather Service (DWD), the former Lokal Model LM (Steppeler et al. 2003). It prognostically solves the compressible equations for wind, temperature, pressure, specific humidity, cloud water and cloud ice content as well as rain, snow, and, optionally, graupel. The equations are solved on an Arakawa-C grid (Arakawa and Lambs 1977) on a rotated geographical coordinate system. In the vertical a hybrid coordinate is used: close to the surface, the numerical layers follow the terrain, whereas near the top of the model domain they are flat.

Numerical integration is performed with a Runge-Kutta scheme, using the time splitting method by Wicker and Skamarock (2002), which allows stable integration of high-frequency modes like acoustic and buoyancy modes.

The physical packages that have been utilized are: the radiative transfer scheme according to Ritter and Geleyn (1992), which is called only once per hour; the Tiedtke parameterization of convection (Tiedtke 1989) being modified by D. Mironow (DWD) related to the treatment of convective cloud condensate as a mixed water-ice phase and of detrained convective cloud condensate; a turbulence scheme (Raschendorfer 2001; Mironov and Raschendorfer 2001) based on prognostic turbulent kinetic energy closure at level 2.5 according to Mellor and Yamada (1982); a one-moment cloud microphysics scheme, a reduced version of the parameterization of Seifert and Beheng (2001), which predicts precipitation formation taking into account water vapour, cloud water, cloud ice, rain and snow (graupel and hail are not considered) with three-dimensional transport of all precipitating phases; a multi layer soil model (Schrodin and Heise 2001; Schrodin and Heise 2002; Heise et al. 2003); subgrid scale orography processes (Schulz 2008; Lott and Miller 1997). Detailed descriptions of the dynamics, numerics and physical parametrizations in the model can be found in the model documentation (e.g Doms 2011). A variety of CCLM applications, e.g. a simulation of East African precipitation patterns (Kaspar and Cubasch 2008), are gathered in a special issue of Meteorologische Zeitschrift (Rockel et al. 2008).

The numerical domain, common to all groups participating to the CORDEX-Africa initiative, is shown in Fig. 1, including a sponge zone of 10 grid-points at each side, where the Davies boundary relaxation scheme is used (Davies 1976, 1983). A horizontal grid resolution of 0.44° is used, as specified by the CORDEX protocol: thus, the model domain uses a grid of 214 points from West to East and 221 points from South to North, including the sponge zone. A time step of 240 s is used. For the high-resolution run (0.22°), the number of gridpoints increases to 427 × 441, and the time step is reduced to 120 s. The vertical coordinate with 35 levels remains identical for both cases, with the upper most layer at 30 km above sea level. Both simulations have been driven by the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis (Uppala et al. 2008; Dee et al. 2011), for the period 1989–2008.
Fig. 1

Surface height (m) and model domain for CORDEX Africa simulation. The domain includes a sponge zone of 10 grid points in each direction. Squares indicate the locations of the evaluation regions

After a series of sensitivity runs, the lower height of the damping layer was increased from its standard value, 11 km, to the approximate height of the tropical tropopause, 18 km, in order to avoid unphysical and unrealistic results. Also the soil albedo was replaced by a new dataset, derived from Moderate Resolution Imaging Spectroradiometer (MODIS) (Lawrence and Chase 2007), which gives more realistic results over the deserts.

2.2 Observational datasets

To evaluate the performance of the model, we employ a combination of available ground observations, satellite products, and reanalysis, summarized schematically in Table 1.
Table 1

Summary of available dataset used for the evaluation of the model’s results

Name

Type

Variable

Time period

Time res. (highest)

Spatial res. (deg.)

ERA-Interim

Reanalysis

T, Prec., fluxes

1989–2008

3-hourly

0.75

UDEL v2.01

Obs.

T, Prec.

1901–2008

Monthly

0.5

CRU v3.0

Obs.

T, Prec.

1901–2006

Monthly

0.5

GPCC v5

Obs.

Prec.

1951–2009

Monthly

0.5

GPCP v1.1

Satellite/obs.

Prec.

1998–2010

Daily

1

FEWS

Satellite/obs.

Prec.

2001–2011

Daily

0.1

TRMM 3B42v6

Satellite

Prec.

1998-2010

3-hourly

0.25

CMORPH

Satellite

Prec.

2003–2010

3-hourly

0.25

SRB v3.0

Reanalysis

Radiation

1989–2007

3-hourly

1

GSWP-2

Reanalysis

Radiation, fluxes

1986–1995

Monthly

1

Although high-quality observational datasets for Africa are scarce, in the last decade satellite-based products have become available, which deliver global precipitation analyses at very high spatial and temporal resolution. For instance, both the Tropical Rainfall Measuring Mission (TRMM) (Huffman et al. 2007), and the Climate Prediction Center Morphing Technique (CMORPH) (Joyce et al. 2004) datasets cover the African continent with 0.25° spatial and 3-hourly temporal resolution.

As pointed out by Nikulin et al. (2012), there may be discrepancies in the timing of peak rainfall between infrared sensor data and radar/microwave precipitation estimates. By analyzing several years of rain gauge observations in the sub-Sahel region, Pinker et al. (2006) showed that the CMORPH data correctly depict the diurnal cycle of rainfall. However, satellite estimates generally tend to overestimate precipitation over semi-arid regions, where rainfall evaporates significantly before reaching the surface. In addition, Nikulin et al. (2012) compared the TRMM-3B42 and TRMM-3G68 products, the latter including estimates from a precipitation radar, microwave imager and their combination, finding no significant difference in the daily phase of precipitation over Africa.

The GPCP dataset (Global Precipitation Climatology Project) (Adler et al. 2003) is based on a combination of satellite and gauge measurements, with a 1° spatial and daily temporal resolution (Huffman et al. 2001). Nikulin et al. (2012) found that over tropical Africa, TRMM shows a significant dry bias compared to GPCP, with differences up to 50 % in some regions, especially in winter. This discrepancy is due to the two dataset using different gauge analysis products (Huffman et al. 2009; Nikulin et al. 2012). The FEWS dataset (Famine Early Warning System) (Funk and Verdin 2010) is another satellite-gauge product at 0.1°, daily resolution. As pointed out by Sylla et al. (2012), who compared FEWS, GPCP, and TRMM, the number of observations used in these products varies over time and regions, and significant discrepancies in the observed precipitation may derive by differences in retrieval, merging and interpolation techniques.

Observational gridded datasets, statistically interpolated from station observations to a regular terrestrial grid, are also available, namely: the University of Delaware (UDEL) (Legates and Willmott 1990), the Climatic Research Unit (CRU, University of East Anglia) (Mitchell and Jones 2005), and the Global Precipitation Climatology Centre (CPCC) (Rudolf et al. 2010) datasets. Each data set covers different time periods, at 0.5° spatial and monthly temporal resolution. As pointed out by e.g. Zhang et al. (2012), over Africa the gauge network is very sparse, with areas where almost no data is available (e.g. central Africa). As a consequence of the interpolation process, therefore, in these areas station errors are particularly relevant because of their large spatial influence. However, Zhang et al. (2012) reports that the CPCC, GPCP, and CRU datasets generally agree well with each other, except for some discrepancies where the number of gauge stations is limited, such as Angola and equatorial Congo.

Reanalysis products such as ERA-Interim (Dee et al. 2011), which employs a 4D-variational approach, are also used for the evaluation of the RCM results: it is important to note that, e.g. precipitation and surface downwelling SW radiation are short-term forecast products, and their reliability depends therefore on the performance of the algorithm. For instance, ERA-Interim tropical precipitation tends to be overestimated compared to the other datasets, as shown by Zhang et al. (2012).

For the evaluation of the radiation budget the SRB database (NASA/GEWEX Surface Radiation Budget) (Stackhouse et al. 2011) is used. The SRB database contains global 3-hourly gridded estimates of shortwave (SW) and longwave (LW) surface parameters derived from a variety of input data sources with two sets of radiative transfer algorithms, namely the primary and Langley parameterized algorithms (Pinker and Laszlo 1992; Fu et al. 1997; Gupta et al. 1992). Input data include cloud parameters derived from the International Satellite Cloud Climatology Project (ISCCP) products, and meteorological inputs from the 4-D data assimilation System, level-4 (GEOS-4) from the NASA Global Modeling and Assimilation Office (GMAO). These satellite-derived parameters have a monthly uncertainty of around 5 Wm−2 for longwave radiation and between 5 and 20 Wm−2 for shortwave radiation.

Finally, the Global Soil Wetness Project (GSWP) (Dirmeyer et al. 2006) dataset was used for the evaluation of the land surface fluxes. GSWP is a modeling research activity of the Global Land-Atmosphere System Study (GLASS) and the International Satellite Land-Surface Climatology Project (ISLSCP). Its objective is to develop a land surface analog to the atmospheric reanalysis, by means of state-of-the-art land surface schemes. Models were forced by soil, vegetation and meteorological data from e.g, the USGS soil texture classification, the Global Gridded Surfaces of Selected Soil Characteristics (IGBP-DIS) or the NCEP-DOE Reanalysis. Model results were validated against in situ data from selected field campaigns and observational networks.

3 Results

In this section we critically analyze the ability of CCLM to reproduce the principal characteristics of the African climatology. We first discuss the seasonal general circulation by focusing on mean sea level pressure and low level winds. Subsequently, geographical and temporal distribution of temperature and precipitation are analyzed, with focus on the climatology of the WAM. Finally, the ability of CCLM to correctly simulate the surface radiation budget and the energy fluxes is discussed. Results are generally shown for the high resolution (0.22°) run, unless stated otherwise.

3.1 General circulation

The main characteristics of the African seasonal circulation are shown in Fig. 2, where modelled mean sea level pressure (MSLP) and low-level (925 hPa) wind vectors are shown, for austral and boreal summer (JFM and JAS, respectively). ERA-Interim data are also shown for comparison, together with the model bias. Wind pattern during JFM is characterized, over South-Eastern Africa, by the convergence of southeasterly and northeasterly winds over the Indian Ocean. As pointed out, e.g. by Sylla et al. (2010), this convergence of moist air is associated with the precipitation pattern over the Central and Southern continent. Over the South-West, the flow is characterized by the South Atlantic subtropical high with an anti-cyclonic circulation.
Fig. 2

Mean sea level pressure (MSLP) and low-level (925 hPa) wind vectors as modelled by CCLM, for austral and boreal summer (top and bottom row, respectively). ERA-Interim data are also shown for comparison, together with the model bias

On the continent, the pressure field is characterized by the West African Heat Low (WAHL), a region of high surface temperatures and low surface pressures, which plays a key role in the development of the WAM. According to e.g. Lavaysse et al (2009), the position and shape of the WAHL change considerably during the year; in JFM, the WAHL has the form of an elongated heat trough positioned south of the Darfur mountains. Between June and July, the WAHL migrates north-westward to a location over the Sahara, between the Hoggar and the Atlas mountains, where it remains quasi-stationary until September; the circulation is characterized by the convergence of humid and cooler southwesterly flow (the WAM) and the dry and hot northeasterly flow (Harmattan).

CCLM is generally able to reproduce the overall features of the seasonal circulation, compared to the reanalysis, and the bias in the pressure field remains generally low (<2 hPa). However, during the austral summer a negative bias is visible over central Africa, which may be related to the overestimation of surface temperature (discussed below). During JAS, CCLM slightly underestimates the MSLP over the northern continent and generally overestimates it along the south and east coasts. MSLP is also slightly overestimated over the Gulf of Guinea regions; this may be related to the underestimation of surface temperature in this area. The stronger simulated pressure gradient between the Gulf of Guinea and the Sahara is responsible for the overestimation of the monsoon flow, which, in turn, may result in the monsoon rainbelt being located too far north, as it will be discussed later on.

Figure 3 shows the vertical cross section of zonal wind, averaged between 10°W and 30°E, during boreal and austral summer (top and bottom row, respectively), for CCLM and ERA-Interim.
Fig. 3

Vertical cross section of zonal wind, averaged between 10°W and 30°E, during boreal and austral summer (top and bottom row, respectively), for CCLM and ERA-Interim

During the austral summer, a low level westerly flow is visible, centred around the Equator, whereas easterly flows are positioned around 18°N and 22°S, respectively; as explained e.g. by Sylla et al. (2010), these easterlies are associated to the precipitation pattern over Central and Southern Africa due to water vapor transport from the Indian Ocean (see also Fig. 2). Upper level circulation is characterized by the presence of high velocity westerly flows, at 25°N and 40°S. These Subtropical Jet Streams are originated from a temperature gradient in the upper troposphere. According to Reason and Rouault (2005) and Sylla et al. (2010), the location and the strength of these Streams may enhance subsidence over Southern Africa playing an important role in the rainfall generation.

CCLM is generally able to reproduce the vertical structure of the circulation in JFM, reproducing accurately both the location and the strength of the low level easterlies and the Subtropical Jet Streams. Also the position and the structure of the Equatorial westerly flow is well captured, although its strength is slightly overestimated.

The zonal wind profile during boreal summer is characterized, below 800 hPa, by the monsoon and the Harmattan flows, located, according to ERA-Interim reanalysis, between 0° and 15°N, and 20° and 30°N, respectively. Above this, at 600 hPa, the AEJ is centered at 13°N and, in the upper troposphere, at 200 hPa, the Tropical Easterly Jet (TEJ) is visible around 10°N. The AEJ is a key feature of the WAM, and for the growth of long-lived mesoscale convective systems (MCSs) (Parker et al. 2005), which are directly associated with the large majority of precipitation events in the Sahel (D’Amato and Lebel 1998). The AEJ is a consequence of the presence of a positive surface temperature gradient between the hot and dry Sahara and the cold and wet Guinea coast; according to the thermal wind relation, this gradient induces easterly shear above the surface monsoon flow (Cook 1999). Moreover, Cook (1999) showed that the presence of a negative soil moisture gradient between the Gulf of Guinea and the Sahara is important in the development of the AEJ. Finally, (Sylla et al. 2011) showed that deep convection is also important for the existence and development of the AEJ. The TEJ is stronger and more extensive than the AEJ; Nicholson (2008) suggests that the TEJ is associated to rainfall interannual variability, being markedly strong (weak) during wet (dry) years. In addition, the position and latitudinal extent of the tropical rainbelt is related to latitudinal position of the strong ascending motion of air, which is bounded by the axes of the AEJ and TEJ (Nicholson 2009). As a consequence, a northward shift of the AEJ latitude and a wider rainbelt may result in wetter summers over the Sahel and Guinea coast.

The vertical structure of the circulation in JAS is reproduced satisfactorily by CCLM. In particular, the shape, location and vertical extend of the monsoon flow is comparable to the reanalysis, although its strength is overestimated by around 3 m/s. Also the position of the TEJ core is generally well reproduced, although its vertical extension and intensity are slightly overestimated. The AEJ is visible, but not so well defined as in the reanalysis, and its core is shifted northward by around 2°. The overestimation of the monsoon flow is related to the overestimation of the surface pressure gradient discussed previously, and the consequence northern shift of the WAHL.

3.2 Temperature climatology

Figure 4 shows simulated seasonal mean 2-m temperature, for boreal and austral summer, along with the bias against ERA-Interim, the UDEL, and the CRU datasets. Although CCLM is able to capture the seasonal mean temperature consistently to other RCMs (Hernández-Díaz et al. 2012; Sylla et al. 2010) with a bias generally ranging between ±2 °C, some features of the bias’ spatial and seasonal distribution are worth to be discussed.
Fig. 4

Simulated seasonal mean 2-m temperature, for boreal and austral summer (top and bottom row, respectively), together with the model bias against ERA-Interim, the UDEL, and the CRU datasets

In JFM, CCLM generally underestimates the mean temperature over the Sahara and Northern Africa, whereas temperature over the Gulf of Guinea, Ethiopia, and Southern Africa is generally overestimated. Over Central Africa (DR Congo, Angola, Zambia) the simulated temperature is close to the observed one, whereas the bias compared to ERA-Interim reanalysis is larger. The difference between the ERA-Interim, UDEL, and CRU datasets is noteworthy and probably a consequence of the scarce density of measurement stations in this region.

In JAS, CCLM shows a marked warm bias over the Sahara and the Arabian Peninsula, and a cold bias over the Guinea region and southern Sahel, although the magnitude of the bias strongly depends on the reference dataset being used. Uncertainty in the observed temperature over the Sahara is related to the scarcity of measurements: Hernández-Díaz et al. (2012), for instance, claim that the CRU dataset may miss the hottest pool, therefore underestimating sensibly the temperature in the region. Cold bias in JAS over the Guinea area are also reported by Sylla et al. (2010) and Hernández-Díaz et al. (2012). In a study investigating the influence of lateral boundary condition on the performance of RCM over West Africa, Sylla et al. (2009) found that the cold bias over the Guinea region cannot be associated to the boundary forcing.

A recent study by Krähenmann et al. (2012) thoroughly investigated the capability of CCLM to simulate daily maximum and minimum temperature statistics over Africa. They found that CCLM overestimated both minimum and maximum temperature in the summer months in arid regions, probably related to a not correct representation of the cloud diurnal cycle. The underestimation of (maximum) summer temperature over the Guinea region can be related to an overestimation of convective activity. In addition, uncertainties in aerosol parameterization and soil thermal conductivity may be responsible for the misrepresentation of the diurnal temperature range.

3.3 Precipitation climatology

Figure 5 shows the spatial distribution of mean seasonal precipitation in JFM and JAS. During the austral summer, rainfall over the continent is mainly located over Central and South equatorial Africa. The two main sources of precipitation are the ITCZ and associated tropical rainbelt, located at its southernmost position, and the easterly flow from the Indian ocean discussed previously. Despite large differences exist in both intensity and spatial distribution amongst the observational datasets, CCLM is generally able to reproduce the main characteristics of the precipitation pattern. Particularly satisfactory is the simulation over Central Africa (Angola, DR Congo) where CCLM precipitation pattern is closer to the observation than ERA-Interim, which, on the contrary, shows a significant wet bias. However, some significant discrepancies are also evident. In particular, the position, shape and strength of the rainbelt over Indian Ocean are simulated quite reasonably (and even slightly overestimated); however, inland precipitation is usually underestimated, as, e.g. over Madagascar, the Guinea Coast and over the coasts of Tanzania and Mozambique.
Fig. 5

Mean seasonal precipitation in JFM (top rows) and JAS (bottom rows) as simulated by CCLM and reported by several observational datasets. Note that the observational datasets cover different periods

In JAS the rainbelt has moved to its northernmost position and rainfall associated to the WAM is at its maximum. Precipitation is generally confined between the Equator and 15°N; three precipitation peaks are also visible over the highlands of Guinea, Cameroon and Ethiopia. CCLM generally underestimates the strength of the WAM, locating the core of the rainbelt too far north, in accordance to the overestimation of the temperature and pressure gradient discussed earlier. As a consequence, precipitation over the Gulf of Guinea is clearly underestimated. A dry bias is also evident over DR Congo, although the FEWS database shows a pattern more in agreement to CCLM, compared to the other datasets. As for JFM, CCLM seems unable to fully transport inland the moisture from the ocean, as shown by the underestimation of the precipitation peak over Sierra Leone and Guinea. As the monsoonal rainbelt is generally well simulated by CCLM, this may be a consequence of the soil parameterization, as we will discuss later.

Annual cycles of mean precipitation, averaged over the CORDEX evaluation regions, are shown in Fig. 6. Here, CCLM results are reported for both simulations, at 0.44° and 0.22° to investigate the effect of spatial resolution on the results. First we note that CCLM agrees generally well with the observations in areas where precipitation is scarce, namely, AM, SA_WS, SA_WN, and SA_E. Also, satisfactory results are shown for the Ethiopian Highlands (EH) region, where CCLM is able to reproduce well the temporal evolution of precipitation. However, it has to be pointed out that large differences exist amongst the observational datasets. ERA-Interim overestimates the magnitude of the peak, and both TRMM and FEWS show the lowest intensities, resulting in an uncertainty of the averaged monthly precipitation intensity of up to 4 mm/day. In a recent work, Sylla et al. (2012) compared the performances of four gridded datasets (FEWS, GPCP, CRU, and TRMM) over Africa, showing that they exhibit substantial systematic differences in mean rainfall and in higher order daily precipitation statistics.
Fig. 6

Annual cycles of mean daily precipitation (mm/day) for CCLM and several observational datasets over 10 different sub-regions indicated in Fig. 1

It is noteworthy that, probably due to better representation of the complex topography on the region, the higher resolution run (0.22°) generally produces a more intense precipitation peak, except for Eastern Africa (EA) and South-Western Sahelian (WA_S).

Over Eastern Africa (EA), precipitation shows a peak during JFM due to the transport of moisture form the Indian Ocean. CCLM markedly underestimates the annual cycle of rainfall by around 2 mm/day, although the lower resolution run performs slightly better.

Over the equatorial area (WA_N, WA_S, CA_NH, and CA_SH), precipitation is strictly related to the evolution of the WAM and the passage of the tropical rainbelt. Over the Sahel region (WA_N) the precipitation pattern is unimodal, characterized by a single peak around August. CCLM reproduces the annual evolution of the rainfall, but the intensity of the peak is underestimated and delayed by one month. Similarly, over the regions characterized by a bimodal annual cycle, CCLM generally underestimates the precipitation during JAS, although the first peak of the precipitation is usually reproduced very well in all the regions (WA_S, CA_NH, and CA_SH). This underestimation is probably due to the rainbelt being shifted too far north (compare Fig. 5); in fact, after September, when the ITCZ migrates South, both the precipitation phase and intensity in CA_NH and WA_S are simulated correctly.

From this analysis, we can conclude that CCLM is generally able to reproduce the annual cycle of precipitation over various regions of Africa, although it usually underestimates the (boreal) summer rainfall peak in the regions affected by the passage of the monsoon. In particular, we note that CCLM produces results that are generally more satisfactory than the ERA-Interim reanalysis (except over EA and the summer peak in WA_N). In fact, Zhang et al. (2012) show that ERA-Interim tropical precipitation tends to be overestimated, although it must be noted that ERA-Interim precipitation is a pure simulated quantity (i.e. not assimilated). Figure 7 shows simulated and observed normalized interannual variabilities of seasonal mean precipitation, calculated as anomalies with respect to the precipitation mean, and normalized by the standard deviation. For each region, the analysis is performed only for the main rainfall season. First we note that the observational datasets (CRU, UDEL and CPCC) give similar results over most of the regions, although discrepancies exist especially over Central Africa (CA_NH, and CA_SH) where observations are very scarce (Nikulin et al. 2012). The ERA-Interim reanalysis generally reproduces the main characteristics of the precipitation anomaly, especially over South Africa, EA and AM, where correlation coefficient (compared to the UDEL dataset) are about 0.7 or higher. Over Central Africa, West Africa and EH, however, ERA-Interim is less satisfactory, with correlation coefficients ranging between 0.23 and 0.59. It has to be noted that over these areas ERA-Interim shows the largest discrepancies in the annual cycle of total precipitation compared to the observations (Fig. 6).
Fig. 7

Normalized interannual variability of seasonal mean precipitation. Also reported are the correlation coefficients for ERA-Interim and CCLM, calculated using the UDEL dataset as reference. Note that the analysis has been performed only for the main rainfall season of the respective region

Over the driest regions (South Africa and AM) CCLM generally reproduces the magnitude of the precipitation anomaly satisfactorily, but some discrepancies are evident (1996 peak in SA_WS; 1996, 2000, and 2006 peaks in SA_E; 2000 and 2006 peaks in SA_WN), which contribute to reduce the correlation coefficients. Over the other regions, CCLM usually performs similarly (e.g. EA and WA_S) or better than ERA-Interim, in particular over the period 1989–1997 in both EA and WA_N. Also, differences exist between the 0.22 and 0.44 runs; in particular, over SA_WN the two runs show very similar correlation coefficients, but the 0.22 run fails to reproduce the 2006 peak, whereas the 0.44 run underestimates the 1992 dry event. Over WA_N and EH the high-resolution run outperforms the lower-resolution one.

In summary, although failing to reproduce some specific events, CCLM is able to capture the main characteristics (magnitude and phase) of the observed interannual variability, especially over the most rainy regions, where the model performs similarly or better than ERA-Interim.

3.3.1 WAM climatology

In order to better quantify how CCLM simulates the WAM characteristics, Fig. 8 shows the time-latitude cross section (Hovmöller diagram) of observed and simulated annual cycle of precipitation over West Africa (averaged over the region 10°W–0°E). Both GPCP and TRMM show two main precipitation peaks, the first occurring around day 150 over the Guinea Gulf (0–7°N), and the second one around day 220 over the Sahel (centered around 11°N). This abrupt northward displacement is know as the Monsoon jump.
Fig. 8

Hovmöller diagram of mean annual cycle of precipitation (mm/day) over West Africa (averaged over the region 10°W–10°E). A 20-day moving average has been used to remove high-frequency variability

CCLM is somehow able to reproduce the main characteristics of the monsoon evolution, including the timing of the jump, although the first precipitation peak is overestimated and it occurs too early, and the second one is underestimated. However its geographical location is correct and better reproduced than the ERA-Interim reanalysis, which shows a general dry bias in the region above 10°N. CCLM shows a secondary maximum over the Guinea Gulf around day 320 that is not present in the observations (although somehow visible in the reanalysis). This ’southward jump’ as defined by Nikulin et al. (2012), is present in other RCM simulations, and it is related to the overestimation of precipitation over the ocean (see Fig. 5).

Figure 9 presents monsoon statistics proposed by D’Orgeval et al (2005) and Domínguez et al. (2010), namely, the monsoon precipitation intensity (above the threshold of 2 mm/day), the latitude of the monsoon centre, and the width (in latitude) of the monsoon band. These statistics are calculated considering only land points in the region 20°W–15°E, 0–20°N.
Fig. 9

Annual cycle of monsoon statistics as simulated by CCLM and accoring to observational datasets. Only land points in the region 20°W–15°E 0–20°N were used. For the intensity only values above a threshold of 2 mm/day have been used. Units: Precipitation intensity: mm/day; Monsoon centre latitude and width of monsoon band: degrees

According to D’Orgeval et al. (2005) the parameters of the monsoon statistics are defined as below. Let n be the number of grid points above 2 mm/day, i the index of one point, Pi the precipitation for this grid point and lati its latitude.
  • The (average precipitation) intensity above the threshold (Pt = 2 mm/day)
    $$ I = \frac{1}{n} \sum_{i=1}^{n}{(P_{i}-P_{t})} $$
  • The latitude of monsoon center:
    $$ C_{lat} = \frac{\sum\nolimits_{i=1}^{n}(P_i-P_t)lat_i}{\sum\nolimits_{i=1}^{n}(P_i-P_t)} $$
  • The width in latitude of the monsoon band:
    $$ W_{lat} = \sqrt{\frac{\sum\nolimits_{i=1}^{n}(P_i-P_t) (lat_i-C_{lat})^2}{\sum\nolimits_{i=1}^{n}(P_i-P_t)}} $$

CCLM generally reproduces the WAM characteristics satisfactorily in (boreal) winter and spring, but precipitation is underestimated in JAS and the monsoon is placed too far north, from July onwards, although the width of the rainbelt is consistent with the observations. It is therefore the combination of two factors, i.e. the misplacement of the monsoon centre and the underestimation of its intensity, that cause the significant dry bias in boreal summer in the WAM regions.

By comparing the results of 10 RCMs over the CORDEX-Africa regions, Nikulin et al. (2012) show that, over the regions affected by the WAM, individual RCMs perform very differently, showing either significant underestimation or overestimation of the mean precipitation. Although being usually drier than average, CCLM usually falls in the 25–75 percentile range of the RCMs’ ensemble.

3.4 Radiation budget

Figure 10 shows seasonal means of long- and short-wave surface net radiation. CCLM usually reproduces the long-wave net radiation (LWN) satisfactorily, especially in winter, over both land and oceans, in particular, if SRB is used as the reference data. The comparison with GSWP data show larger differences for the winter season. On the other hand, the short-wave net radiation (SWN) is clearly underestimated over the Atlantic ocean, over Central Africa in JFM, and over the areas corresponding to the monsoon rainbelt in JAS. A positive bias (up to 30 W/m2) is visible in JAS over sub-equatorial Africa.
Fig. 10

Modelled mean long-wave (LW) and short-wave (SW) surface net radiation, for boreal and austral summer, and biases compared to SRB and GSWP. Units in W/m2

Figure 11 shows the annual cycle of the different components of the radiation budget, spatially averaged over the evaluation regions. We note that both the phase and the magnitude of the two long-wave components (down- and up-wards) are well simulated, with biases generally remaining below 20 W/m2. Short-wave radiation is usually well simulated over South Africa and AM, whereas in the regions affected by the WAM (WA_N, WA_S, and, CA_NH), the short-wave downward flux (SWd) is clearly underestimated form May to October (and the upward flux, SWu, slightly overestimated). Notably, the SWd underestimation is larger than the uncertainty in the SRB database (20 W/m2). Due to a combination of further factors (e.g. the underestimation of the latent heat flux, discussed later) this is in accordance to the underestimation of precipitation in the same areas (see Fig. 6). SWd is also underestimated in JFM in CA_SH, and in JJAs in EH, where precipitation was modelled satisfactorily. It is also worth noting that the high resolution run is generally able to reproduce SWd more satisfactorily, especially in EA, SA_E, and SA_WN. Results are also improved in WA_N and EH, although large underestimation remains.
Fig. 11

Annual cycles of modelled and observed radiation components: long-wave up- (LWu) and downward (LWd), shortwave up- (SWu) and donwnward (SWd). Black curves represent the simulated quantities, accordingly components. Units in W/m2

Kothe and Ahrens (2010) investigated the uncertainties in the radiation budget simulated by several RCMs (including an earlier version of CCLM) for West Africa. They found that most models during boreal summer largely underestimated SWN over the equatorial oceans and in parts of the ITCZ, and they overestimated SWN in the Sahara and Sahel, with the largest differences, up to 80 W/m2 for the region offshore of Angola. They claim that uncertainties in the cloud fraction (\(\Updelta\)CFR) explains, on average, more than 40 % of the error variance over the oceans. Over land, on the other hand, the albedo parameterization is, on average, of similar importance as the \(\Updelta\)CFR, but its impact was much larger (more than 60 %) on areas characterized by very low cloud fractions, such as the Sahel and the Sahara. As discussed earlier, the version of CCLM used in this study utilizes a paramaterization of the albedo derived from MODIS, which improved the simulation of the radiation budget, especially over dry areas.

3.5 Surface fluxes

As stated earlier, the WAM extent and intensity is strongly related to the land-surface interaction, with soil moisture playing a key factor, as it controls the partitioning between the surface latent and sensible heat fluxes (Boone et al. 2009); in particular, a positive feedback exists between soil moisture and precipitation, as soil moisture can affect the precipitation recycling ratio, and, in addition, it increases the efficiency of precipitation processes (Steiner et al. 2009). Philippon and Fontaine (2002) suggested that wetter soils can reduce the Bowen ratio by cooling and adding moisture to the near-surface atmospheric layer, thus increasing the moist entropy flux and the amount of convective available potential energy in the relatively shallow boundary layer. The resulting vertical thermodynamic profile tends to become unstable, increasing the frequency and magnitude of convective rainfall events, as explained by Steiner et al. (2009).

Figure 12 shows the annual cycles of sensible (SH) and latent (LH) heat fluxes for the evaluation areas. It becomes evident that CCLM generally underestimated LH in all areas. However, the partitioning between LH and SH is different depending on the region: for instance, in CA_SH, CCLM reproduces fairly well both the fluxes until July. This is reflected in the reasonably good simulation of both precipitation and temperature in the austral summer. Even though the LH is too small, the reduction of precipitation, as a consequence of the underestimated evaporation, leads to drier soils, which is reflected in the overestimation of SH and, in turn, of the surface temperature (see Fig. 4). Similarly, in EA, the underestimation of LH and overestimation of SH are related to the underestimation of precipitation and overestimation of temperature, especially in JFM. It is noteworthy that in this area the coarse resolution run reproduces the fluxes more realistically; as a consequence, also the precipitation annual cycles is better simulated (Fig. 6). Over West Africa, CCLM behaves somehow differently: the evolution of the surface fluxes in the areas affected by the WAM has been explained by e.g. Boone et al. (2009). Over the Sahel (WA_N) sensible and latent heat flux peaks are approximately 5 months apart, with SH being dominant during the onset month of the WAM, and LH becoming larger at the end of each year, when water stored in the soil is evaporated, after precipitation has decreased, due to significant incoming solar radiation and dry atmospheric conditions. CCLM represents correctly the shape and phases of both fluxes, but constantly underestimates LH, whereas SH is more consistent with the observations/reanalysis.
Fig. 12

Annual cycles of modelled and observed latent (LH, lower panels) and sensible (SH, upper panels) heat fluxes. Units in W/m2

Over WA_S, the LH cycle shows a double peak with a relative minimum around July due to lower precipitation and incoming radiation and humid conditions (Boone et al. 2009). LH as simulated by CCLM is flatter than the observations, with only one peak around November–December.

From the analysis of the radiation budget and surface fluxes, it seems that the underestimation of precipitation, especially in the monsoon areas, is due to a combination of factors. Figure 11 shows that CCLM usually underestimates SWd, and slightly overestimates LWd, which can be related to an overestimation of the cloud fraction. The consequent reduction in SWd, together with a mis-reproduction of the soil water content, may be responsible for the underestimation in the evaporation, which is reflected by the underestimation of LH. Together with the misplacement of the monsoonal rainbelt, caused by the overestimation of the surface pressure/temperature gradient, this leads to the large underestimation of the precipitation described earlier.

4 Summary and concluding remarks

In this work we presented the results of the application of CCLM to the CORDEX-Africa domain, by critically analyzing the model’s ability in simulating not only the geographical and temporal distribution of temperature and precipitation, but also the radiation budget and surface fluxes, which are crucial mechanisms in the formation, duration, and evolution of the WAM. We compared the performances of the model in the standard CORDEX-Africa configuration (i.e. 0.44°) to a new, and to our knowledge, unprecedented high resolution (0.22°) simulation.

For the analysis of the results, we employed a combination of different observational datasets, satellite-based and reanalysis products. Due to the scarcity of the gauge stations over large areas of, e.g. central Africa, and different retrieval, merging, and interpolating technique used in the satellite algorithms, large discrepancies may and do exist amongst the different observational datasets (Nikulin et al. 2012; Sylla et al. 2012; Zhang et al. 2012).

Results show that CCLM is generally able to reproduce the overall features of seasonal MSLP and the related general circulation, although, a small negative bias <2 hPa is visible over central Africa during the austral summer, and over the Sahara during JAS. This excessive pressure gradient between the Gulf of Guinea and the Sahara is responsible for the overestimation of the monsoon flow, which, in turn, results in the monsoon rainbelt being located too far north. The vertical structure of the circulation is reproduced satisfactorily; in particular, the shape, location, and vertical extend of the monsoon flow is comparable to the reanalysis, although its strength is overestimated by around 3 m/s. Also the position of the TEJ core is generally well reproduced, with a weak overestimation of its intensity and vertical extension. The AEJ is visible, but not so well defined as in the reanalysis, and its core is shifted northward by around 2°.

CCLM is able to capture the seasonal mean temperature consistently to other RCMs (Hernández-Díaz et al. 2012; Sylla et al. 2010) with a bias generally ranging between ±2 °C; however, in JAS, CCLM results show a warm bias (between 2 °C and more than 5 °C) over the Sahara and the Arabian Peninsula, whereas a cold bias (around 3 °C is present over the Guinea region and southern Sahel. The magnitude of the bias can vary by more than 4 °C depending on the reference dataset. According to Krähenmann et al. (2012), the overestimation of temperature over arid regions in summer months is probably related to an incorrect representation of the cloud diurnal cycle, whereas the underestimation of (maximum) temperature over the Guinea region may be related to an overestimation of convective activity.

CCLM is generally able to reproduce the annual cycle and the interannual variability of seasonal precipitation, over most regions of Africa, although it usually underestimates the (boreal) summer rainfall peak in the regions affected by the passage of the monsoon. This significant dry bias seems to be caused by a combination of two factors, i.e. the misplacement of the monsoon centre and the underestimation of its intensity. The misplacement of the monsoon band is related to the overestimation of the surface pressure gradient and the consequence northern shift of the WAHL, which results in the overestimation of the monsoon flow. On the other hand, from the analysis of the radiation budget and surface fluxes, it seems that the underestimation of precipitation intensity, is due to a combination of different factors. In fact, CCLM usually underestimates SWd, and slightly overestimates LWd, which can be related to an overestimation of the cloud fraction. The reduction in SWd may be responsible, together with a mis-reproduction of the soil water content, for underestimation in the evaporation, which is reflected by the underestimation of LH. Together with the misplacement of the monsoonal rainbelt, caused by the overestimation of the surface pressure/temperature gradient, this leads to the large underestimation of the precipitation.

Some general concluding remarks can be drawn from this exercise:
  1. (1)

    When comparing CCLM results to the reanalysis, one has to remember that ERA-Interim horizontal wind and surface pressure are forcing data for the simulation. The capacity of reproducing the general features of e.g. seasonal low level circulation and vertical structure of the atmospheric circulation, confirms the ability of CCLM to maintain the large-scale information inherited by the forcing lateral conditions. In addition, CCLM results for mean annual precipitation cycles and interannual variability are usually better than the reanalysis. This makes CCLM an adequate tool for the second part of the CORDEX initiative, i.e. the dynamical downscaling of climate change scenario runs from the GCMs participating to the CMPI5 project.

     
  2. (2)

    The increase of the model resolution does not bring evident improvements to the model’s performances, at least for monthly mean statistics, as investigated in this study. For instance, precipitation is better simulated with the high-resolution run over the Ethiopian Highlands, whereas over Eastern Africa the standard 0.44° run gives the best results. Although the high resolution run is able to to simulate the radiation fluxes slightly better, a large underestimation of SWd is still present. Regarding the considerable amount of additional computational resources needed for a high resolution simulation over continental Africa, it appears that 0.44° is a suitable compromise between model performances and computational limits.

     
  3. (3)

    Our results show that the simulation of the African climate, with its complex system and its multiple scale interaction, is still not completely satisfactory, and further research is needed to thoroughly investigate the model deficiencies, in particular by analyzing the role of land surface interaction and parameterization schemes, which are responsible for the mis-representation of the surface fluxes, the radiation components, and, in turn, the precipitation intensity.

     

Notes

Acknowledgments

We would like to thank Grigory Nikulin (SMHI) for providing some of the observational dataset used in this study, and Diego Guizzardi (JRC) for preparing the FEWS database. The SRB data were obtained from the NASA Langley Research Center Atmospheric Sciences Data Center NASA/GEWEX SRB Project. GPCC Precipitation data is provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their Web site at http://www.esrl.noaa.gov/psd/ Computing resources have been provided by HLRS, Stuttgart, and the European Centre for Medium-Range Weather Forecast (ECMWF), Reading.

References

  1. Abiodun BJ, Pal JS, Afiesimama EA, Gutowski WJ, Adedoyin A (2008) Simulation of West African monsoon using RegCM3. Part II: impacts of deforestation and desertification. Theor Appl Climatol 93:245–261. doi:10.1007/s00704-007-0333-1 CrossRefGoogle Scholar
  2. Adler R, Huffman G, Chang A, Ferraro R, Xie P, Janowiak J, Rudolf B, Schneider U, Curtis S, Bolvin D, Gruber A, Susskind J, Arkin P, Elkin E (2003) The version 2 global precipitation climatology project (GPCP) monthly precipitation analysis (1979-present). J Hydrometeorol 4:1147–1167CrossRefGoogle Scholar
  3. Afiesimama EA, Pal JS, Abiodun BJ, Gutowski WJ, Adedoyin A (2006) Simulation of West African monsoon using the RegCM3. Part I: model validation and interannual variability. Theor Appl Climatol 86:23–37. doi:10.1007/s00704-005-0202-8 CrossRefGoogle Scholar
  4. Arakawa V, Lamb V (1977) Computational design of the basic dynamical processes in the UCLA general circulation model. In: Chang J (ed) Methods in computational physics: general circulation models of the atmosphere, vol 17. Academic Press, New York, pp 173–265. ISBN 0-12-460817-5Google Scholar
  5. Baldauf M, Seifert A, Förstner J, Majewski D, Raschendorfer M, Reinhardt T (2011) Operational convective-scale numerical weather prediction with the COSMO model: description and sensitivities. Mon Weather Rev 139(12):3887–3905. doi:10.1175/MWR-D-10-05013.1 CrossRefGoogle Scholar
  6. Boone AA, Poccard-Leclercq I, Xue Y, Feng J, Rosnay P (2009) Evaluation of the WAMME model surface fluxes using results from the MMA land-surface model intercomparison project. Clim Dyn 35(1):127–142. doi:10.1007/s00382-009-0653-1 CrossRefGoogle Scholar
  7. Cook K (1999) Generation of the African easterly jet and its role in determining West African precipitation. J Clim 12:1165–1184CrossRefGoogle Scholar
  8. D’ Amato N, Lebel T (1998) On the characteristics of the rainfall events in the Sahel with a view to the analysis of climatic variability. Int J Clim 18(9):955–974. doi:10.1002/(SICI)1097-0088(199807)18:9<955::AID-JOC236>3.0.CO;2-6 CrossRefGoogle Scholar
  9. Davies H (1983) Limitations of some common lateral boundary schemes used in regional NWP models. Mon Weather Rev 111:1002–1012. doi:10.1175/1520-0493(1983)111<1002:LOSCLB>2.0.CO;2 CrossRefGoogle Scholar
  10. Davies HC (1976) A lateral boundary formulation for multi-level prediction models. Q J R Meteorol Soc 102(432):405–418. doi:10.1002/qj.49710243210 Google Scholar
  11. Dee DP, Uppala SM, Simmons AJ, Berrisford P, Poli P, Kobayashi S, Andrae U, Balmaseda MA, Balsamo G, Bauer P, Bechtold P, Beljaars ACM, van de Berg L, Bidlot J, Bormann N, Delsol C, Dragani R, Fuentes M, Geer AJ, Haimberger L, Healy SB, Hersbach H, Hólm EV, Isaksen L, Kållberg P, Köhler M, Matricardi M, McNally AP, Monge-Sanz BM, Morcrette JJ, Park BK, Peubey C, de Rosnay P, Tavolato C, Thépaut JN, Vitart F (2011) The ERA-Interim reanalysis: configuration and performance of the data assimilation system. Q J R Meteorol Soc 137(656):553–597. doi:10.1002/qj.828 CrossRefGoogle Scholar
  12. Dirmeyer Pa, Gao X, Zhao M, Guo Z, Oki T, Hanasaki N (2006) GSWP-2: multimodel analysis and implications for our perception of the land surface. Bull Am Meteorol Soc 87(10):1381–1397. doi:10.1175/BAMS-87-10-1381 CrossRefGoogle Scholar
  13. Domínguez M, Gaertner Ma, Rosnay P, Losada T (2010) A regional climate model simulation over West Africa: parameterization tests and analysis of land-surface fields. Clim Dyn 35(1):249–265. doi:10.1007/s00382-010-0769-3 CrossRefGoogle Scholar
  14. Doms G (2011) A description of the nonhydrostatic regional COSMO model part 1: dynamics and numerics. DWD, Offenbach, Germany. http://www.cosmo-model.org/content/model/documentation/core/default.htm
  15. Druyan LM, Feng J, Cook KH, Xue Y, Fulakeza M, Hagos SM, Konaré A, Moufouma-Okia W, Rowell DP, Vizy EK, Ibrah SS (2010) The WAMME regional model intercomparison study. Clim Dyn 35(1):175–192. doi:10.1007/s00382-009-0676-7 CrossRefGoogle Scholar
  16. D’ Orgeval T, Polcher J, Li L (2005) Uncertainties in modelling future hydrological change over West Africa. Clim Dyn 26(1):93–108. doi:10.1007/s00382-005-0079-3 CrossRefGoogle Scholar
  17. Fu Q, Liou K, Cribb M, Charlock T, Grossman A (1997) Multiple scattering parameterization in thermal infrared radiative transfer. J Atmos Sci 54:2799–2812CrossRefGoogle Scholar
  18. Funk C, Verdin J (2010) Satellite rainfall applications for surface hydrology. Springer, Dordrecht. doi:10.1007/978-90-481-2915-7 Google Scholar
  19. Giorgi F, Jones C, Asrar G (2009) Addressing climate information needs at the regional level: the CORDEX framework. World Meteorol Organ (WMO) Bull 58(July):175–183Google Scholar
  20. Gupta SK, Darnell WL, Wilber AC (1992) A parameterization for longwave surface radiation from satellite data: recent improvements. J Appl Meteorol 31(12):1361–1367. doi:10.1175/1520-0450(1992)031<1361:APFLSR>2.0.CO;2 CrossRefGoogle Scholar
  21. Heise E, Lange M, Ritter B, Schrodin R (2003) Improvement and validation of the multilayer soil model. COSMO Newsl 3:198–203. http://www.cosmo-model.org/content/model/documentation/newsLetters/default.htm
  22. Hernández-Díaz L, Laprise R, Sushama L, Martynov A, Winger K, Dugas B (2012) Climate simulation over CORDEX Africa domain using the fifth-generation Canadian Regional Climate Model (CRCM5). Clim Dyn. doi:10.1007/s00382-012-1387-z
  23. Huffman G, Adler R, Morrissey M, Bolvin D, Curtis S, Joyce R, McGavock B, Susskind J (2001) Global precipitation at one-degree daily resolution from multisatellite observations. J Hydrometeorol 2:36–50CrossRefGoogle Scholar
  24. Huffman GJ, Adler RF, Bolvin DT, Gu G, Nelkin EJ, Bowman KP, Hong Y, Stocker EF, Wolff DB (2007) The TRMM multisatellite precipitation analysis (TMPA): quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J Hydrometeorol 8(1):38–55. doi:10.1175/JHM560.1 CrossRefGoogle Scholar
  25. Huffman GJ, Adler RF, Bolvin DT, Gu G (2009) Improving the global precipitation record: GPCP version 2.1. Geophys Res Lett 36(17):L17,808CrossRefGoogle Scholar
  26. IPCC (2007) 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, CambridgeGoogle Scholar
  27. Jenkins GS, Gaye AT, Sylla B (2005) Late 20th century attribution of drying trends in the Sahel from the Regional Climate Model (RegCM3). Geophys Res Lett 32. doi:10.1029/2005GL024225
  28. Joyce R, Janowiak J, Arkin P, Xie P (2004) CMORPH: a method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J Hydrometeorol 5:487–503CrossRefGoogle Scholar
  29. Kaspar F, Cubasch U (2008) Simulation of East African precipitation patterns with the regional climate model CLM. Meteorologische Zeitschrift 17(4):511–517. doi:10.1127/0941-2948/2008/0299 CrossRefGoogle Scholar
  30. Kothe S, Ahrens B (2010) On the radiation budget in regional climate simulations for West Africa. J Geophys Res 115(D23). doi:10.1029/2010JD014331
  31. Krähenmann S, Kothe S, Panitz H-J, Ahrens B (2012) Evaluation of daily maximum and minimum 2 m temperatures as simulated with the regional climate model COSMO-CLM over Africa. submitted to?Google Scholar
  32. Lavaysse C, Flamant C, Janicot S, Parker DJ, Lafore JP, Sultan B, Pelon J (2009) Seasonal evolution of the West African heat low: a climatological perspective. Clim Dyn 33(2–3):313–330. doi:10.1007/s00382-009-0553-4 CrossRefGoogle Scholar
  33. Lawrence PJ, Chase TN (2007) Representing a new MODIS consistent land surface in the community land model (CLM 3.0). J Geophys Res 112(G1):G01,023Google Scholar
  34. Legates DR, Willmott CJ (1990) Mean seasonal and spatial variability in gauge-corrected, global precipitation. Int J Climatol 10(2):111–127. doi:10.1002/joc.3370100202 CrossRefGoogle Scholar
  35. Li H, Wang H, Yin Y (2012) Interdecadal variation of the West African summer monsoon during 19792010 and associated variability. Clim Dyn. doi:10.1007/s00382-012-1426-9
  36. Lott F, Miller MJ (1997) A new subgrid-scale orographic drag parametrization: its formulation and testing. Q J R Meteorol Soc 123(537):101–127. doi:10.1002/qj.49712353704 CrossRefGoogle Scholar
  37. Mellor GL, Yamada T (1982) Development of a turbulence closure model for geophysical fluid problems. Rev Geophys 20(4):851–875. doi:10.1029/RG020i004p00851 CrossRefGoogle Scholar
  38. Mironov D, Raschendorfer M (2001) Evaluation of empirical parameters of the new LM surface-layer parameterization Scheme: results from numerical experiments including soil moisture analysis. Cosmo technical report 1, DWD, Offenbach, Germany. http://www.cosmo-model.org/content/model/documentation/techReports/default.htm
  39. Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int J Climatol 25(6):693–712. doi:10.1002/joc.1181 CrossRefGoogle Scholar
  40. Nicholson SE (2008) The intensity, location and structure of the tropical rainbelt over west Africa as factors in interannual variability. Int J Climatol 28(13):1775–1785. doi:10.1002/joc.1507 CrossRefGoogle Scholar
  41. Nicholson SE (2009) A revised picture of the structure of the monsoon and land ITCZ over West Africa. Clim Dyn 32(7–8):1155–1171. doi:10.1007/s00382-008-0514-3 CrossRefGoogle Scholar
  42. Nikulin G, Jones C, Samuelsson P, Giorgi F, Sylla MB, Asrar G, Büchner M, Cerezo-Mota R, Christensen OBs, Déqué M, Fernandez J, Hänsler A, van Meijgaard E, Sushama L (2012) Precipitation climatology in an ensemble of CORDEX-Africa regional climate simulations. J Clim. doi:10.1175/JCLI-D-11-00375.1
  43. Paeth H, Hall NM, Gaertner MA, Alonso MD, Moumouni S, Polcher J, Ruti PM, Fink AH, Gosset M, Lebel T, Gaye AT, Rowell DP, Moufouma-Okia W, Jacob D, Rockel B, Giorgi F, Rummukainen M (2011) Progress in regional downscaling of west African precipitation. Atmos Sci Lett 12(1):75–82. doi:10.1002/asl.306 CrossRefGoogle Scholar
  44. Parker DJ, Thorncroft CD, Burton RR, Diongue-Niang A (2005) Analysis of the African easterly jet, using aircraft observations from the JET2000 experiment. Q J R Meteorol Soc 131(608):1461–1482. doi:10.1256/qj.03.189 CrossRefGoogle Scholar
  45. Philippon N, Fontaine B (2002) The relationship between the Sahelian and previous 2nd Guinean rainy seasons: a monsoon regulation by soil wetness? Ann Geophys 20(4):575–582. doi:10.5194/angeo-20-575-2002 CrossRefGoogle Scholar
  46. Pinker RT, Laszlo I (1992) Modeling surface solar irradiance for satellite applications on a global scale. J Appl Meteorol 31(2):194–211. doi:10.1175/1520-0450(1992)031<0194:MSSIFS>2.0.CO;2 CrossRefGoogle Scholar
  47. Pinker RT, Zhao Y, Akoshile C, Janowiak J, Arkin P (2006) Diurnal and seasonal variability of rainfall in the sub-Sahel as seen from observations, satellites and a numerical model. Geophys Res Lett 33(7):2–5. doi:10.1029/2005GL025192 Google Scholar
  48. Raschendorfer M (2001) The new turbulence parameterization of LM. COSMO Newsl 1:90–98. http://www.cosmo-model.org/content/model/documentation/newsLetters/default.htm
  49. Reason CJC, Rouault M (2005) Links between the Antarctic Oscillation and winter rainfall over western South Africa. Geophys Res Lett 32(7):L07,705. doi:10.1029/2005GL022419 Google Scholar
  50. Redelsperger JL, Thorncroft CD, Diedhiou A, Lebel T, Parker DJ, Polcher J (2006) African monsoon multidisciplinary analysis: an international research project and field campaign. Bull Am Meteorol Soc 87(12):1739–1746. doi:10.1175/BAMS-87-12-1739 CrossRefGoogle Scholar
  51. Ritter B, Geleyn JF (1992) A comprehensive radiation scheme for numerical weather prediction models with potential applications in climate simulations. Mon Weather Rev 120(2):303–325. doi:10.1175/1520-0493(1992)120<0303:ACRSFN>2.0.CO;2 CrossRefGoogle Scholar
  52. Rockel B, Will A, Hense A (2008) The regional climate model COSMO-CLM (CCLM). Meteorologische Zeitschrift 17(4):347–348. doi:10.1127/0941-2948/2008/0309 CrossRefGoogle Scholar
  53. Rudolf B, Becker A, Schneider U, Meyer-christoffer A, Ziese M (2010) The new GPCC full data reanalysis Version 5 providing high-quality gridded monthly precipitation data for the global land-surface is public available since December 2010. GPCC status report (December):1–7Google Scholar
  54. Ruti PM, Williams JE, Hourdin F, Guichard F, Boone A, Van Velthoven P, Favot F, Musat I, Rummukainen M, Domínguez M, Gaertner MA, Lafore JP, Losada T, Rodriguez de Fonseca MB, Polcher J, Giorgi F, Xue Y, Bouarar I, Law K, Josse B, Barret B, Yang X, Mari C, Traore aK (2011) The West African climate system: a review of the AMMA model inter-comparison initiatives. Atmos Sci Lett 12(1):116–122. doi:10.1002/asl.305 CrossRefGoogle Scholar
  55. Schrodin R, Heise E (2001) The multi-mayer version of the DWD soil model TERRA-LM. Cosmo technical report 2, DWD, Offenbach, Germany. http://www.cosmo-model.org/content/model/documentation/techReports/default.htm
  56. Schrodin R, Heise E (2002) A new multi-layer soil-model. COSMO Newsl 2:139–151. http://www.cosmo-model.org/content/model/documentation/newsLetters/default.htm
  57. Schulz JP (2008) Introducing sub-grid scale orographic effects in the COSMO model. COSMO Newsl 9:29–36. http://www.cosmo-model.org/content/model/documentation/newsLetters/default.htm Google Scholar
  58. Seifert A, Beheng KD (2001) A double-moment parameterization for simulating autoconversion, accretion and selfcollection. Atmos Res 59–60:265–281CrossRefGoogle Scholar
  59. Stackhouse Jr. PW, Gupta SK, Cox SJ, Mikovitz C, Zhang T, Hinkelman LM (2011) He NASA/GEWEX surface radiation budget release 3.0: 24.5-year dataset. GEWEX News 21(1):10–12Google Scholar
  60. Steiner AL, Pal JS, Rauscher Sa, Bell JL, Diffenbaugh NS, Boone A, Sloan LC, Giorgi F (2009) Land surface coupling in regional climate simulations of the West African monsoon. Clim Dyn 33(6):869–892. doi:10.1007/s00382-009-0543-6 CrossRefGoogle Scholar
  61. Steppeler J, Doms G, Schättler U, Bitzer HW, Gassmann A, Damrath U, Gregoric G (2003) Meso-gamma scale forecasts using the nonhydrostatic model LM. Meteorol Atmos Phys 82(1–4):75–96. doi:10.1007/s00703-001-0592-9 CrossRefGoogle Scholar
  62. Sylla MB, Gaye AT, Pal JS, Jenkins GS, Bi XQ (2009) High-resolution simulations of West African climate using regional climate model (RegCM3) with different lateral boundary conditions. Theor Appl Climatol 98(3–4):293–314. doi:10.1007/s00704-009-0110-4 CrossRefGoogle Scholar
  63. Sylla MB, Coppola E, Mariotti L, Giorgi F, Ruti PM, DellAquila A, Bi X (2010) Multiyear simulation of the African climate using a regional climate model (RegCM3) with the high resolution ERA-Interim reanalysis. Clim Dyn 35(1):231–247. doi:10.1007/s00382-009-0613-9 CrossRefGoogle Scholar
  64. Sylla MB, Giorgi F, Ruti PM, Calmanti S, DellAquila A (2011) The impact of deep convection on theWest African summermonsoon climate: a regional climate model sensitivity study. Q J R Meteorol Soc 137:141–71430. doi:10.1002/qj.853 CrossRefGoogle Scholar
  65. Sylla MB, Giorgi F, Coppola E, Mariotti L (2012) Uncertainties in daily rainfall over Africa: assessment of gridded observation products and evaluation of a regional climate model simulation. Int J Climatol pp n/a–n/a, doi:10.1002/joc.3551
  66. Tiedtke M (1989) A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon Weather Rev 117(8):1779–1800. doi:10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2 CrossRefGoogle Scholar
  67. Uppala SM, Dee DP, Kobayashi S, Berrisford P, Simmons AJ (2008) Towards a climate adapt assimilation system: status update of ERA-Interim. ECMWF Newsl 115:12–18Google Scholar
  68. Wicker LJ, Skamarock WC (2002) Time-splitting methods for elastic models using forward time schemes. Mon Weather Rev 130(8):2088–2097. doi:10.1175/1520-0493(2002)130<2088:TSMFEM>2.0.CO;2 CrossRefGoogle Scholar
  69. Xue Y, Sales F, Lau WKM, Boone A, Feng J, Dirmeyer P, Guo Z, Kim KM, Kitoh A, Kumar V, Poccard-Leclercq L, Mahowald N, Moufouma-Okia W, Pegion P, Rowell DP, Schemm J, Schubert SD, Sealy A, Thiaw WM, Vintzileos A, Williams SF, Wu MLC (2010) Intercomparison and analyses of the climatology of the West African Monsoon in the West African Monsoon Modeling and Evaluation project (WAMME) first model intercomparison experiment. Clim Dyn 35(1):3–27. doi:10.1007/s00382-010-0778-2 CrossRefGoogle Scholar
  70. Zhang Q, Körnich H, Holmgren K (2012) How well do reanalyses represent the southern African precipitation? Clim Dyn. doi:10.1007/s00382-012-1423-z

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  • Hans-Jürgen Panitz
    • 1
  • Alessandro Dosio
    • 2
  • Matthias Büchner
    • 3
  • Daniel Lüthi
    • 4
  • Klaus Keuler
    • 5
  1. 1.Karlsruher Institut für TechnologieInstitut für Meteorologie und KlimaforschungEggenstein-LeopoldshafenGermany
  2. 2.European Commission Joint Research CentreIspraItaly
  3. 3.Potsdam Institute for Climate Impact Research (PIK)PotsdamGermany
  4. 4.Swiss Federal Institute of Technology (ETH)ZurichSwitzerland
  5. 5.Brandenburg University of Technology (BTU)CottbusGermany

Personalised recommendations