USE OF CRU data within AMMA-CROSS

In this annexe, we provide some additional information regarding the rainfall estimations that we prepared for AMMA-CROSS with CRU data. (http://www.cru.uea.ac.uk/cru/data/hrg.htm )

More precisely, we made use of CRU TS 2.1 dataset: http://www.cru.uea.ac.uk/~timm/grid/CRU_TS_2_1.html

This figure shows the [10W°,10°E] monthly mean rainfall climatology as obtained for a 30-year average (1961-1990). It clearly displays specific characteristics of the West African monsoon with an abrupt monsoon onset (rainfall maximum jump from towards 5°N in June to around 11°N in July-August) and a more continuous monsoon retreat in September-October.

As explained on the CRU web page, "CRU TS 2.0 is our best estimate of the spatial pattern of climate at each moment in time; it is complete in space and time. In this sense, CRU TS 2.0 is 'space-optimised' rather than 'time-optimised'. Because the emphasis is placed on obtaining best estimates of the instantaneous spatial patterns, inhomogeneities may be present in the time-series an individual grid-box." (http://www.cru.uea.ac.uk/~timm/grid/ts-advice.html).

This may apply to CRU TS 2.1 dataset as well. Although we focus on a somewhat wider domain, it corresponds nevertheless to a region where data are sparse. Furthermore, we specifically focus on a few particular years within AMMA-CROSS, not on a climatology.

Over the 2000-2002 period for instance there were not as much data available on the AMMA [10°W,10°E] cross-section as found in other years, especially in the south-western and eastern borders of the domain, both areas which are climatologically more rainy that the center of the domain [5°W,5°E], due to orography. These areas are expected to contribute the most to the rainfall amount between 5°N and 10°N.

These features may be a cause for the absence of a real monsoon shift in 2000 (the August maximum remains centered along the coast after the onset even though it extends further North into the Sahel), and for a relatively weak latitudinal rainfall gradient between 5°N and 10-12°N in 2001 and 2002 when compared with the other rainfall estimates. Indeed, the gradient sharpens when rainfall values are retain only when the station counts within radius is positive.

It does not mean that such a procedure is the more appropriate. In the present case, according to climatological values, it acts to decrease the [10W°,10°E] mean rainfall amount over [5°N,10°N]. This points however again to the range of uncertainties affecting quantitative estimates of rainfall at large scale, over the ITCZ as well as its northern and southern sides.

For AMMA-CROSS datafiles - [10W°,10°E] monthly mean values - we decided to perform a simple average from the CRU datafiles, without particular treatment for areas where actual data are missing. This must be kept in mind for further analysis and comparison.