AMMA icon
  Executive Summary
  Input Data
  Output Data
  AMMA Related:
  AMMA International
ALMIP title banner

Input Data

(last updated: Sep. 30, 2008)

Page Contents:

  1. Atmospheric Forcing
  2. Land Surface Parameter data
  3. Forcing data download

1. Atmospheric Forcing

The atmospheric forcing data which will be provided to the participants is listed below. It will be provided in NetCDF format, using the conventions outlined on the Assistance for Land Modeling Activities (ALMA) web site. Further details can be found on that web site.

Variable Name Description Units (MKS)
Tair Near surface air temperature K
Qair Near surface specific humidity kg/kg
PSurf Surface pressure Pa
SWdown Surface incident shortwave radiation W/m2
LWdown Surface incident longwave radiation W/m2
Rainf (Total) Rainfall rate kg/(m2 s)
CRainf Convective rainfall rate kg/(m2 s)
Snowf Snowfall rate kg/(m2 s)
Wind_N Near surface northward wind component>
Wind_E Near surface eastward wind component>

* It is assumed that most land surface schemes will use the total rainfall rate (Rainf) and the mean wind speed (magnitude computed from Wind_N and Wind_E), but additional information has been added for generality (e.g. CRainf: the total rain rate is the sum of the convective and the large scale/stratiform rain rates).

Experiment 1
data is provided at a 0.5 degree resolution at a 3 hour time step over the regional window defined as -20 to 30 east longitude, and -5 to 20 north latitude. This corresponds to a 101x51 grid which uses a cylindrical equidistant or platte carre projection. Each forcing file corresponds to a single year. There are 4 files available, from 2001 through 2004. Each file begins on Jan. 1 at 0Z, and ends at Dec. 31, 24Z (note that this implies that the forcing values for the last time step for a given year are identical to the those for the initial time step for the following year). This dataset is based purely on numerical weather prediction (NWP) forecast model output diagnostics (from ECMWF).

For convenience (and possible use by other projects), the atmospheric forcing data have been provided on a rectangular grid. Note that while there are 5151 grid points, only 3650 correspond to land points (i.e. those containing a non-negligible fraction of land using the provided ECOCLIMAP land mask). Results only need be reported for these points. However, one may report results on grids with all 5151 points with the non-land points flagged. A sample NetCDF header for Experiment 1 is given here. As an example, the total Exp.1 rainrate averaged over the year 2004, is shown below in mm day-1.

Exp.1 (NWP-based) precipitation averaged over the entire year 2004.

Experiment 2
data is provided at a 0.5 degree resolution at a 3 hour time step over the regional window defined as -20 to 30 east longitude, and -5 to 20 north latitude (as in Exp.1). There are atmospheric forcing data for the year 2004 (and soon, 2005). This dataset is based on the merging of numerical weather prediction (NWP) forecast model output diagnostics (from ECMWF) and remote sensing-based products. The downwelling longwave and shortwave radiative fluxes for 2004 are from the
OSI-SAF (Ocean and Sea Ice Satellite Application Facility) and the precipitation is from EPSAT) (Estimation of Precipitation by SATellite) product available within AMMA-SAT (AMMA-PRECIP). The use of remote sensing-based products based on MSG has resulted in fairly consistent forcing: an example is shown below. In 2005, the downwelling radiative fluxes are from LAND-SAF (LAND Satellite Application Facility). Documentation describing the Shortwave and Longwave flux algorithms are available. Note that the satellite data is only incorporated during the monsoon months (since these are the only months for which EPSAT data is currently available).

An example of the EPSAT 0.10 degree precipitation estimate (used for Exp.s 2-3: lower panel) and the aggregated values (used for Exp .2: using a 0.5 degree resolution: upper panel) for Aug. 7, 2005.

The total Exp.2 rainrate averaged over the year 2004, is shown below in mm day-1. The difference bewteen the Exp.2 rain rates less the Exp.1 values for the same period is also shown (lower panel).

Exp.2 merged precipitation averaged over 2004.

Difference in annually averaged rain rate for 2004 : Exp.2 less Exp.1 values. There is a general northward shift in the active zone of precipitation in Exp.2.

An example of a time series from the Exp. 2 forcing dataset at a single grid point. This point was selected because the NWP-based data placed the zone of active precipitation well to the south, as is reflected in the low rainfall (Rainf) plot (lowest panel: black curve) and the high solar radiation (uppermost panel: thick black curve). The precipitation given by the AMMA-SAT product (lowest panel: blue curve) is significantly larger, showing a farther northward penetration of the active precipitation zone associated with the WAM. The OSI-SAF downwelling shortwave (uppermost plot: red curve) and longwave fluxes are also shown. Note that the OSI-SAF shortwave radiation is significantly reduced on DoY 185 and 189, corresponding to two rain events seen in the AMMA-SAT precipitation. The latter two rain events on DoY 192 and 196 occur at night, therefore the OSI-SAF longwave radiation has a strong response. Note that the NWP longwave radiative flux is strongly correlated with the solar cycle, while the SAF based product is better correlated with the near-surface atmospheric air temperature (it is computed with a significant contribution from this variable in addition to cloudiness information).

Experiment 3
data is provided at a 0.5 degree resolution and 3 hourly time step (as in Exp.s 1-2). This merged 0.5 degree dataset for 2002-2007 is currently being tested and consists in ECMWF atmospheric state variables, and
TRMM-3B42 3-hourly data for 2002-2007 for the precipitation. OSI-SAF radiation flux data is used for the summer 2004 as in Exp.2, but in contrast with Exp.2, LAND-SAF for all available months from 07-2005 to 2007 are used for Exp.3 (in Exp.2, this product was only used in summer months). Plots of the Exp.1, 2, and 3 radiation and precipitation monthly averages are shown here. We would like to acknowledge Remi Meynadier of IPSL/Université Pierre et Marie Curie (UPMC) for providing us the TRMM-3B42 data on the ALMIP 0.5 grid for 2002-2007.

Precipitation Input Check

In order to ensure that you are reading the input precipitation data correctly and using the grid mask (compressed land-only vector) information correctly, here is a simple table which gives precipitation averages (mm and mm day-1 in parenthesis) over the land vector (LAND). In addition, averages for the sub-period June-July-Aug-Sept are also given. NOTE that these are the pixel averages, not the actual area-weighted averages. This is provided as an aid to to those running the experiments. Note that an updated version of the EPSAT precipitation product has been used to create a new forcing dataset as of May, 2007 (here referred to as Exp2b).

1 2004 753 (2.06) 357 (3.07)
2 2004 813 (2.22) 435 (3.57)
2b 2004 725 (1.98) 347 (2.84)
1 2005 824 (2.26) 398 (3.26)
2 2005 775 (2.12) 348 (2.85)
2b 2005 777 (2.13) 356 (2.91)

The monthly average precipitation (mm day-1) for Exp2(b) is shown in the Figures below (the data in the forcing files). Note that the zone east of 25 degrees E is masked since the satellite-based forcings do not cover this region (the forcings default to NWP data over this region, and are therefore the same as in Exp1).

Exp.2(b) monthly precipitation averages for 2004.

Exp.2(b) monthly precipitation averages for 2005.

Radiative and Precipitation Forcing Monthly AVG plots

Plots of the monthly average fluxes can be viewed here. The experiment numbers corresponding to each Exp. are shown below. Exp. numbers in brackets imply that the forcing are the same as those for the Exp. not in brackets.

Year Rainf SWdown LWdown
2002 1[2], 3 1[2, 3] 1[2, 3]
2003 1[2], 3 1[2, 3] 1[2, 3]
2004 1, 2, 3 1, 2[3] 1, 2[3]
2005 1, 2, 3 1, 2, 3 1, 2, 3
2006 1, 2, 3 1, 2, 3 1, 2, 3
2007 1, 3 1, 3 1, 3

2. Land Surface Parameter data

The land surface parameter data is based on values from ECOCLIMAP used at Meteo-France. Soil parameters and monthly varying (for a single climatological annual cycle) vegetation parameters are provided for both aggregated or tile-based land surface schemes.

ECOCLIMAP monthly average LAI (Leaf Area Index)

    The monthly averaged LAI from ECOCLIMAP aggregated to a 0.5 degree spatial resolution. This data is used for ALMIP experiments 1 and 2. Note that the LAI values in ECOCLIMAP actually use a decadal time step. A climatological annual cycle is used (so that the year in the date above is irrelevant).

    NOTE, that there are 2 land masks in the forcings. The ECMWF land mask is given in the forcing just as general information. We ask that modelers use the land mask in the land forcing parameter files (ecoclimap*.nc) when reporting results to us. As we are only examining the land surface in ALMIP, users have 2 options (which ever is better for them). They can either:

  • Report results on the full 2D grids as in the forcing files, but with all non-land points FLAGGED
  • Report results on a compressed land vector (which will defined in early April)

Obviously the latter option results in considerably smaller files, but again, this will be up to the participants (whichever option is easier for them).

3. ALMIP Forcing data downloads

Server Access

    The ALMIP input data server is password protected. In order to obtain the password and login information, please contact us. The ftp server hostname is The directory with data for downloading is almip. The current list of files to be used in ALMIP is shown below:


    The Forcing*gz files are gzipped and in NetCDF format. The yearly forcing data is at a three-hour time step on a 0.5 degree cylindrical equidistant (platte carre) grid (101x51 grid points). Each Forcing* file begins at 00UTC Jan. 1, and ends on 24UTC Dec. 31 of the same year. The files with the mrg extension indicate the combined NWP forcing and satelite product forcings.

    The ecoclimap*z files contain the land surface data. Several fields vary approximately every 10 days (36 time values). The file contains land surface data for multiple tiles or patches at each grid box. Note that the ECOCLIMAP files with the 2 extension are to be used.

Listing of Current (and Pending) Data on the server

    The data which can currently be downloaded are shown below. Each cell corresponds to gridded data covering a year-long period. Last update on Sep. 30, 2008.

File Type (Input data) 2001200220032004200520062007
ECMWF 0.5o
Merged (ECMWF+EPSAT+SAF) 0.5o
Merged (ECMWF+TRMM+SAF) 0.5o
Merged 0.1o
ECOCLIMAP Tiles 0.5o
ECOCLIMAP Tiles 0.1o

Currently available data files (on the ALMIP server) are indicated using green shading, and files which will soon be available (are under preparation) are indicated using yellow shading. Note that ECOCLIMAP files contain climatological data for a single annual cycle, so that a single year-long file can be used for all years.

Sample NetCDF header for Experiment 1

netcdf Forcing_AMMA_0.5_2003 {
	x = 101 ;
	y = 51 ;
	level = 1 ;
	timestep = UNLIMITED ; // (2921 currently)
	int timestep(timestep) ;
		timestep:units = "timesteps since 2003-01-01 00:00:00" ;
		timestep:title = "Time steps" ;
		timestep:tstep_sec = 10800.f ;
		timestep:long_name = "Time step axis" ;
		timestep:time_origin = " 2003-JAN-01 00:00:00" ;
	int x(x) ;
		x:valid_range = 1, 101 ;
		x:axis = "1D" ;
		x:units = "-" ;
		x:long_name = "Navigation in x direction" ;
		x:associate = "x" ;
	int y(y) ;
		y:valid_range = 1, 51 ;
		y:axis = "1D" ;
		y:units = "-" ;
		y:long_name = "Navigation in y direction" ;
		y:associate = "y" ;
	int level(level) ;
		level:valid_range = 1, 1 ;
		level:axis = "1D" ;
		level:units = "-" ;
		level:long_name = "Level Index (above surface)" ;
		level:associate = "level" ;
	float time(timestep) ;
		time:units = "seconds since 2003-01-01 00:00:00" ;
		time:calendar = "gregorian" ;
		time:title = "Time" ;
		time:long_name = "Time axis" ;
		time:time_origin = " 2003-JAN-01 00:00:00" ;
	float lon(y, x) ;
		lon:valid_range = -20.f, 30.f ;
		lon:axis = "2D" ;
		lon:units = "Degrees East" ;
		lon:long_name = "Latitude" ;
		lon:associate = "y x" ;
	float lat(y, x) ;
		lat:valid_range = -5.f, 20.f ;
		lat:axis = "2D" ;
		lat:units = "Degrees North" ;
		lat:long_name = "Longitude" ;
		lat:associate = "y x" ;
	float land_mask(y, x) ;
		land_mask:valid_range = 0.f, 1.f ;
		land_mask:axis = "2D" ;
		land_mask:units = "0 (water) or 1 (for land point)" ;
		land_mask:long_name = "Land Mask" ;
		land_mask:associate = "y x" ;
	float height(level) ;
		height:valid_range = 2.f, 100.f ;
		height:axis = "1D" ;
		height:units = "m" ;
		height:long_name = "Distance in the atmosphere above the surface" ;
		height:associate = "level" ;
	float Tair(timestep, level, y, x) ;
		Tair:valid_range = 213.f, 333.f ;
		Tair:axis = "4D" ;
		Tair:units = "K" ;
		Tair:long_name = "Near surface air temperature" ;
		Tair:associate = "time level y x" ;
		Tair:missing_value = 1.e+20f ;
	float Qair(timestep, level, y, x) ;
		Qair:valid_range = 0.f, 0.03f ;
		Qair:axis = "4D" ;
		Qair:units = "kg/kg" ;
		Qair:long_name = "Near surface specific humidity" ;
		Qair:associate = "time level y x" ;
		Qair:missing_value = 1.e+20f ;
	float Wind_E(timestep, level, y, x) ;
		Wind_E:valid_range = 0.f, 75.f ;
		Wind_E:axis = "4D" ;
		Wind_E:units = "m/s^1" ;
		Wind_E:long_name = "Near surface eastward wind component" ;
		Wind_E:associate = "time level y x" ;
		Wind_E:missing_value = 1.e+20f ;
	float Wind_N(timestep, level, y, x) ;
		Wind_N:valid_range = 0.f, 75.f ;
		Wind_N:axis = "4D" ;
		Wind_N:units = "m/s^1" ;
		Wind_N:long_name = "Near surface northward wind component" ;
		Wind_N:associate = "time level y x" ;
		Wind_N:missing_value = 1.e+20f ;
	float PSurf(timestep, y, x) ;
		PSurf:valid_range = 5000.f, 110000.f ;
		PSurf:axis = "3D" ;
		PSurf:units = "Pa" ;
		PSurf:long_name = "Surface atmospheric pressure" ;
		PSurf:associate = "time y x" ;
		PSurf:missing_value = 1.e+20f ;
	float Rainf(timestep, y, x) ;
		Rainf:valid_range = 0.f, 0.02f ;
		Rainf:axis = "3D" ;
		Rainf:units = "kg/m^2s" ;
		Rainf:long_name = "Rainfall rate" ;
		Rainf:associate = "time y x" ;
		Rainf:missing_value = 1.e+20f ;
	float Snowf(timestep, y, x) ;
		Snowf:valid_range = 0.f, 0.02f ;
		Snowf:axis = "3D" ;
		Snowf:units = "kg/m^2s" ;
		Snowf:long_name = "Snowfall rate" ;
		Snowf:associate = "time y x" ;
		Snowf:missing_value = 1.e+20f ;
	float SWdown(timestep, y, x) ;
		SWdown:valid_range = 0.f, 1360.f ;
		SWdown:axis = "3D" ;
		SWdown:units = "W/m^2" ;
		SWdown:long_name = "Surface incident shortwave radiation" ;
		SWdown:associate = "time y x" ;
		SWdown:missing_value = 1.e+20f ;
	float LWdown(timestep, y, x) ;
		LWdown:valid_range = 0.f, 750.f ;
		LWdown:axis = "3D" ;
		LWdown:units = "W/m^2" ;
		LWdown:long_name = "Surface incident longwave radiation" ;
		LWdown:associate = "time y x" ;
		LWdown:missing_value = 1.e+20f ;
	float CRainf(timestep, y, x) ;
		CRainf:valid_range = 0.f, 0.02f ;
		CRainf:axis = "3D" ;
		CRainf:units = "kg/m^2s" ;
		CRainf:long_name = "Convective Rain rate" ;
		CRainf:associate = "time y x" ;
		CRainf:missing_value = 1.e+20f ;

// global attributes:
		:file_name = "" ;
		:CreatedBy = "Aaron Boone at CNRM Meteo-France" ;
		:title = "Atmospheric Forcing for AMMA from ECMWF FC data interpolated to 0.5 deg grid" ;
		:production = "Transfered to the single file on 20060208 at 120007.070" ;