Françoise GUICHARD 1*, Laurent KERGOAT 2 , Eric MOUGIN 2 and Frédéric HOURDIN 3
(*) [email protected] 1: CNRM-GAME (CNRS & Météo-France) 2: GET (CNRS, IRD, UPS, CNES) 3: LMD, IPSL (CNRS & UPMC)
References Guichard et al. 2009: Surface thermodynamics and radiative budget in the Sahelian Gourma: seasonal and diurnal
cycles, J. Hydrology, 375, 161-177. doi : 10.1016/j.jhydrol.2008.09.007Guichard et al. 2012 and Hourdin et al. 2012: deliverables ESCAPE (D1.1 and D3.2)Hourdin et al., 2010 : AMMA-Model Intercomparison Project, Bull. Amer. Meteor. Soc., doi : 10.1175/2009BAMS2791.1Roehrig et al. 2012: The present and future of the West African monsoon: a process-oriented assessment of CMIP5
simulations along the AMMA transect. J. Climate, submitted.
The ANNUAL CYCLE OF TEMPERATURE in the SAHEL and its CLIMATIC SENSITIVITYGC33A1004
▶ local meteorological data: SYNOP, Hombori historical data, also GISS & BEST [> 60 years] ▶ CRU [> 60 years] and MSU [~ 30 years] gridded datasets
▶ high-frequency (15 min) AMMA-Catch ground station data [~ 10 years]
▶ reanalyses : ERA40 [ ~ 50 years ] and ERA Interim, NCEP CFSR, MERRA [~ 60 years]▶ CMIP5 climate simulations: amip [~ 30 years] , (nat) historical [> 60 years] …
2. OBSERVATIONS, DATASETS, MODELS
▶ In the subtropical semi-arid Sahel, a major climatic feature is the bimodal structure of the annual cycle of temperature at the surface, with two maxima surrounding the rainy season (e.g. Guichard et al. 2009). The first one, in Spring, is particularly large ( monthly mean temperature = 35°C at 15°N). [Fig 1]
▶ We analyse this specific structure, explore how it has been affected by the multi-decadal warming observed in this region, and evaluate whether reanalyses and climate models (CMIP5) can reproduce observations.
Figure 1 : an example of time series of temperature and specific humidity at 2m in the Sahel (function of day of year).
Observations from a few years are overplotted, using automa-tic weather station data from Agoufou (Mali, 1.5°W, 15.3°N) – 10-day mean values.
Temperature
Specific humidityrainfall
annual maximum
1. CLIMATOLOGY and QUESTIONS▶ Observations: a strong warming in 60 years, > 1 K in Spring [Fig 2] - larger increase at night. ▶ In Spring: weaker short-term inter-annual variability (> 10 years), stronger multidecennal trend. ▶ Spring warming: not the same processes at play than during the rainy season (rain-T2m couplings) ▶ A strong warming in Spring (already very hot) is likely to have strong societal implications.
3. WARMING TREND and SEASONS
Figure 2 : Trends in the annual cycle of temperature (T2m): SYNOP data of Hombori (central Sahel, 15°N) from 1950 to 2009. For each month, the series of dots correspond to monthly-mean values from 1950 to 2009. A linear fit is added (for comparison, a 2 nd order fit is displayed below). The trends over the period is also indicated for each month on the top-right (coloured bars). Dots are coloured red for correlations > 0.4.
dominated by short interannual variability
multi-decacal trend dominates
coupling with rainfall(70's 80's severe droughts)
TRENDEACH DOT
= 1 MONTH
FROM1950 to 2009
5. CMIP5 CLIMATE MODELS: CLIMATOLOGIES & TRENDS
6. SUMMARY , PERSPECTIVES ▶ Observations show a strong warming in the Sahel during the past 60 years, with:
- no clear warming during the dry cool season ~ JFM, ND“dry” meaning “very low moisture”, not “no rain”
- stronger warming trend during warmer moist months ~ AMJ- weaker warming during the full monsoon ~ JAS
▶ a large radiative impact of water vapour in Spring ? (monsoon flow), explore coupling of temperature trends with water vapour changes? data & models
▶ an impact on the monsoon onset? on rainfall intensity? daily data needed
societal repercussions? (agriculture, health.. . )
0 K
1 K
TEMPERATURE TREND 1950-2010, MONTH per MONTH
J F M A M J J A S O N D
▶ CRU gridded dataset consistent with local SYNOP data▶ ERA40 & ERA-Interim consistent with CRU [Fig 3] ▶ MERRA & NCEP-CFSR: data assimilation pb? [Fig 3b]
4. WARMING TREND in REANALYSES
Figure 3
(a) same as Fig. 2 except average over 10°W-10°E, 10°N-20°N] from 1958 to 2009 for CRU (top) and ERA-40 (bottom) [note ERA-Interim is used after 2002 when ERA-40 stops, with adjustement.
(b) as (a) except 1979 to 2009, for CRU (top) and reanalyses (3 last plots).
(a)
(b)
also, not shown: (i) NCEP-CFSR cold bias in climatology, (ii) ∃ variables affected by unphysical trends & discontinuities (e.g. MERRA rainfall)
▶ very large spread in annual cycles + smooth annual max [Fig 4a] ▶ large differences in Tmin (pb with nocturnal boundary layers)▶ no consistency in magnitude & seasonality of warming trends [Fig 4b] , ≠ temperature-humidity couplings (suggests ≠ feedbacks)
from weaker to stronger
warming trend
bcc-csm1-1
CNRM-CM5
HadGEM2-ES
MRI-CGCM3
CanESM2
CSIRO-MK3-6-0
IPSL-CM5A-LR
NorESM1-M
Temperature trend Relative humidity trendmodels
OBSCMIP5 models
JJAS: daytime warm bias
JFMA: nighttime cold bias
Figure 4 (a) monthly-mean diurnal cycles of temperature, observations (grey) and models (pink) in the Sahel (cfSites files, Agoufou pt) (b) temperature and relative humidity trends in CMIP5 climate models as a function of month, period ~[1950,2010], models are ordered by increasing warm trends.
(a)
(b)