Influence of initial soil moisture in a Regional 1 Climate Model study over West Africa . Part 1 : 2 Impact on the climate mean 3

12 The impact of soil moisture initial conditions on the mean climate over West Africa was 13 examined using the latest version of the Regional Climate Model of the International Centre for 14 Theoretical Physics (RegCM4) at a horizontal resolution of 25 km × 25 km. The soil moisture 15 reanalysis of the European Centre Meteorological Weather Forecast’s reanalysis of the 20th 16 century ERA20C is used to initialize the control experiment, while its minimum and maximum 17 values over the entire domain are used to establish the initial dry and wet soil moisture 18 conditions respectively (hereafter dry and wet experiments). For the control, the wet and dry 19 experiments, an ensemble of five runs from June to September are performed. In each 20 experiment, we analyzed the two idealized simulations most sensitive to the dry and wet soil 21 moisture initial conditions. The impact of soil moisture initial conditions on precipitation in West 22 Africa is linear over the Central and West Sahel where dry (wet) experiments lead to rainfall 23 decrease (increase). The strongest precipitation increase is found over the West Sahel for wet 24 experiments with a maximum change value of approximately 40%, while the strongest 25 precipitation decrease is found for dry experiments over Central Sahel with a peak of change of 26 approximately −4%. The sensitivity of soil moisture initial condition can persist for three to four 27 months (90-120 days) depending on the region. However, the influence on precipitation is no 28 longer than one month (between 15 and 30 days). The strongest temperature decrease is located 29 over the Central and West Sahel with a maximum change of approximately −1.5 °C in wet 30 experiments, while the strongest temperature increase is found over the Guinea Coast and 31 Central Sahel for the dry experiments, with a maximum of change around 0.6°C. A significant 32

The European 20th Century Weather Prediction Center ERA20C soil moisture reanalysis is used 132 to initialize the control experiment, while its domain-wide minimum and maximum values are 133 used to establish the initial dry and wet soil moisture conditions respectively (hereafter dry and 134 wet experiments). We initialized the dry and wet soil moisture initial conditions (in volumetric 135 fraction m 3 .m -3 ) respectively at the minimum value (=0.117*10 -4 ) and the maximum value 136 (=0.489). 137 We designed three experiments (reference, wet, and dry), each with an ensemble of five (5) 138 simulations starting from June 1st to September 30th. The difference between these three 139 experiments is the change in the initial soil moisture condition (reference initial soil moisture 140 condition, wet initial soil moisture condition, and dry initial soil moisture condition) during the

167
For the two years most sensitive to soil moisture initial conditions, the Student t-test is used to 168 compare the significance of the difference between a wet or dry sensitivity test (sample 1) and 169 the control (sample 2) in assuming that our two samples are independent and in considering that Where X ( X ) are the sample means, n ( n ) are the sample sizes and S ( S ) are the sample 179 variances. In this study, the t-test at the 95% confidence level is used to consider statistically     (Table 2). We noted that the impacts on precipitation of the wet experiments are greater 236 than those from dry experiments (Table 2). These results are consistent with previous studies that 237 supported a strong relationship between precipitation and soil moisture in particular over the  in magnitude than that of dry experiments over most studied domains (Fig. 7). For dry 257 experiments, the strongest daily precipitation response (about −4mm.day -1 ), is found over the 258 Guinea Coast in the run JJAS 2003 (Fig. 7c). While for the wet experiments, the strongest impact 259 on the daily precipitation is more than 8mm.day -1 and it is found over the West Sahel and the 260 Guinea Coast (Fig. 7b, c, respectively). It is worth to note that the impact of initial soil moisture 261 conditions on daily precipitation is much shorter than the duration of the impact on daily soil 262 moisture. The significant impact on daily precipitation is found only for wet experiments, and 263 did not last more than 15 days in large parts of the study domain, excepted over wetter sub-region of Guinea Coast where it lasts approximately one month. We noted that the precipitation 265 peaks over West Sahel and Guinea Coast (Fig. 7b and c, respectively) during August and 266 September coincide with fluctuation in the daily soil moisture impact (Fig.6b and c). This  The impacts on relative humidity and air temperature (Fig.8 and Fig.9, respectively) (Fig.8a and Fig. 9a).

279
For the upper troposphere, the significant impact on relative humidity and temperature is found 280 only for wet experiments, and exhibited a drying and a warming over most of studied domains 281 ( Fig.8 and Fig.9). This impact for the wet experiments was also reported by Hong and Pal  (Fig. 10).

286
For the dry experiments (Fig. 10a, c), we found that the moistening of the lower atmosphere 287 decreases over most of the study domain. However, the strong wind magnitude changes over the 288 Atlantic Ocean bring the moistening from the ocean to the Guinea Coast and West Sahel. This 289 can explain the precipitation increase over these sub-regions in the dry experiments. Over 290 Central Sahel, the strong decrease in precipitation seems to be associated with the decrease of 291 specific humidity which is particularly notable in the run JJAS 2003 (Fig.4a). Conversely, for the 292 wet experiments (Fig.10b, d), an increase in the moistening of the atmosphere is found mainly 293 over the Sahel band while further South, a decrease of the specific humidity is simulated over 294 Guinea Coast. The strong change in wind magnitude shifts the moistening from the North to the 295 South, leading to precipitation increase over most part of study domain (Fig.4 b and d). These   (Table 3).   Figure 13. The impact on temperature is linear 333 over the Central Sahel, Guinea Coast and the whole West African domain (Fig.13a, c and d). The 334 strongest mean temperature decrease is observed over the Central and West Sahel in wet 335 experiences with the maximum change approximately −1.5 °C (Table 2)  to temperature increase (Fig.13, Table2). 342 We now analyze the influence of soil moisture initial conditions anomalies on land energy 343 balance, particularly on the surface fluxes sensible and latent heat. Figure 14 shows changes in respect to the control exhibits significant increase (decrease) of the sensible heat (Fig.14).  (Fig. 15). The impact in wet experiments is strong over Central and West Sahel 352 compared to the dry experiments, but not for Guinea Coast (Fig. 15, Table 2). In the dry 353 experiments, the strongest sensible heat flux increase is found over Guinea Coast, with 354 maximum change about 9.18 W.m -2 during JJAS 2004 (see Table 2). In the wet experiments, the  Table 2).  (Table 2).

379
For a quantitative evaluation, the PDF distribution of the Bowen ratio is showed on Figure 19. In Pal J. S., Small E. E. and Elthair E. A.: Simulation of regional scale water and energy budgets: