1. Rainfall data for validating satellite rainfall estimates - Precipitation network set up in Africa for AMMA
Henri Laurent
Marielle Gosset (Benin)
Christian Depraetere
Thierry Lebel (Niger)
IRD/LTHE, Grenoble, France
Abou Amani
Abdou Ali
Agrhymet, Niger

2. Ground validation issues
- time sampling effect
“ground truth”: estimation of areal rainfall and accuracy
Need for a good knowledge of rain fields
 raingauge networks
 radar
Observations and studies in the AMMA framework (African Monsoon Multidisciplinary Analyses)
regional scale – daily rainfall observations
meso scale – high resolution observations

3. Convective cloud cover – 1 month
Reference: all Meteosat images
Time sampling problem
Rainfall estimate error (%)
1°x1°
5°x5°
annual
Jul/Aug
Jun/Sep
SSM/I 40 73
101 29
TRMM 36 76 92 23 MT 25 65 18
Sampling: SSM/I
Sampling: TRMM
Using the dense raingauge network in Niamey (1°x1°), it has been shown that the time sampling error is reduced by % for MT compared to TRMM
Sampling: MT

4. Rainfall estimation at regional scale (from daily rainfall)
Raingauge network - Sahel • CILSS : CRA : O Synop : 85

5. Raingauge network - Sahel Summary on the kriging method used to create the areal rainfall for validation of satellite rainfall estimates
 Regression Kriging method used to estimate the mean areal rainfall
(grid: 0.5°x0.5°, 1°x1° or 2.5°x2.5°)
cumulated oved 10-day, monthly or annual periods
Anisotropy of rainfall fields: the drift has to be taken into account
 For details on the kriging method, see:
Ali et al., J. Appl. Meteo, 2005 (in press)

6. Comparison of different methods for areal rainfall estimate
3 krigging methods taking the drift into account
Theor. error
Observed error
Interpolated values are close,
but very different estimations of theoretical error

7. Intercomparison of different satellite products
Monthly rainfall, 2.5°x2.5° Sahel
CMAP (Sat+gauges)
GPCP (Sat+gauges)
GPCC  (gauges)
GPI (Sat)
SYN (gauges from synoptic network)
RMSE (%)

8. Rainfall at regional scale: daily estimation?
Tracking from METEOSAT infrared channel
19 July 1994
Kriging from CILSS daily rain gauges

9. African Monsoon Multidisciplinary Analysis
AMMA Long term experimental set up
3 meso-scale sites – 2 of which are already well equiped for precipitation measurements.
Upper OUEME valley
km2
Soudanian
rain : mm/year
specialized in water budget / hydrological processes.
data since 1997
Gourma
km2
Sahelian to Saharian
Rain : mm/year
site specialized in vegetation + satellite validation.
Few rain gages / Possibility to densify.
Niamey square degree
km2
Sahelian
rain : mm/year
Specialized in hydrology and the study of land / rain intreactions.
Data since 1990 (Hapex –Sahel, Epsat-Niger)

10. Niamey square degree Since 1989: between 30 and 109 raingauges
Many works, e.g.:
Lebel et al., Water Res. Res., 1992 – Amani et al., Water Res. Res., 1996 – Lebel et al., J. Hydro, 1997 – LeBarbé and Lebel, J. Hydro, 1997 – Amani and Lebel, J. Hydro, 1997 – Amani and Lebel, Sto. Hydro., 1998 – Lebel and Amani, J. Appl. Meteo, 1999 – Mathon et al., J. Appl. Meteo, 2002 – Ali et al., J. Hydromet., 2003
See
Very good knowledge of Sahelian rain fields
modelisation, downscaling issues
interpolation and estimation (kriging)
estimation error
Not simply transposable in another region
 First it is needed to study precipitation fields

11. Raingauge network – Mesoscale sites Available data sets for 2004
Niamey square degree (Niger)
33 recording rain gauges
Raw data 5 min, ~ from May to September (depends on the station)
Validated data:
10-day periods - station and grid (kriging 5 km x5 km)
rainfall events (i.e. >30% rainy stations) - station and grid (kriging 5 km x5 km)
Oueme (North Benin)
35 recording rain gauges
Raw data 5 min, all over the year (with some missing data)
Validated data: daily rainfall (stations)
Gridded estimates are not available (yet) by lack of climatological knowledge

12. Upper OUEME valley
Mesoscale Site:
Raingage network, river flow, ground water.
Surface, monitoring of vegetation dynamics.
Meteo/climate stations
“Super Site”: Donga watershed
Denser network (20 gages / 600 km2)
Meteorological radar Xport + optical disdrometer

13. Details: www.lthe.hmg.inpg.fr/catch/xport/
X port Radar developed at LTHE X band – 9.4 GHz diameter 1.8 m – 100 kW polarisation H and V doppler
Details:
Optical Spectrogranulometer, recording data every 1 min on the rain drop size distributions observed at ground level.

14. X-port : Data and objectives Donga / Bénin, 2004-2007 - EOP AMMA
Which measurement ?
-Reflectivity (power returned to the radar by the precipitation)  Amount of precipitation
-Polarimetric variables (difference between Horizontal and Vertical signal)
 median diameter of the drop size distribution
 attenuation correction
-Doppler  velocities of hydrometeors.
2D - Structure
3 D structure
multiparameters

15. Radar - vertical profiles analysis
Vertically pointing mode: vertical structure of precipitation
(images VPR McGill)
Derive statistics of vertical structure in a given climatic region
observe amount of convective vs stratiform rain
quantify occurrence of evaporation
 improve parameterization
 feed data base for satellite remote sensing algo.
(+ FFT analysis of Doppler spectrum -> evolution of DSD with height)

16. Radar + disdrometer data - Application High resolution 2D fields of precipitation – Homogeneity ? Propagation at ground ? - Down scaling issues - Observation of vertical profile of reflectivity within rain storm – gather profil types and assess variability, useful for inversion of satellite data - DSD analysis + radar polarimetric product : Analyze the time/space variability of Drop size distributions at the ground level.

17. Rainfall types during the rainy season (1999-2003)
Ongoing studies on the characterisation of the rainfall events in North Benin using rain gauge data
Development of a rainfall model valid for Sahel and sub-Sahel rain fields. Based on a modelisation of convective cells
Rainfall types during the rainy season ( )
Identification of Mesocale Convective Systems (rainfall) and determination of their propagation (speed and direction)
Depraetere et al., EGU Conference, April 2005, Vienna, Austria

18. Directional chronogram of the rainfall event
Computation of optimal direction and speed of the rainfall event
Directional chronogram of the rainfall event
Meso scale hyetogram derived from the pseudo-chronogram
Directional pseudo-chronogram of the rainfall event

19. Possible collaborations on ground validation issues
Summary
Possible collaborations on ground validation issues
rainfall ground truth – estimation and estimation error - regional scale - meso scale possibilities in the West African, sub-Sahelian zone : - Data from the Upper Ouémé meso-scale site - Rainfield modelling /Down scaling issues. - Use of a light X-Band, polarimetric radar for field observation of the precipitating systems