INM

1
INMs contribution toELDAS project
J.A. Parodi, E. Rodriguez and B. Navascués INM
2
OUTLINE

• Three planned experiments REF,ELD,SVA, (TIL?)
• Definition of REF experiment
• Preliminary considerations
• Work on progress
• Evaluation strategy
• Time schedule

3
Definition of REF experiment

• Latest reference HIRLAM version (6.2.0)
• some of the latest improvements
• No tiles (acceptable degradation!)
• 3D-VAR assimilation, 6h cycling
• Lateral Boundaries ECMWF analysis
• 0.2ºx0.2ºx31L
• SL Time Step
• Domain
• 48h forecast from 00 (or 12) UTC analysis only

4
Three planned experiments

• REF
• ELD same as REF but with soil moisture generated
  by CNRM
• SVA same as REF but with SVA assimilation of
  soil moisture (Gianpaolo)

Possibly and depending on our resources, well
run a 4th experiment with tiling structure same
as REF, but including the same land tiles as in
the HIRLAM reference system
5
Preliminary considerations

• Both HIRLAM and the soil moisture generator
  model
• (CNRM) should use the same physiography
• Grid geometry? HIRLAM usually runs on a rotated
  grid.
• Minimize (or suppress) horizontal
  interpolations!

The CNRM has provided vegetation and soil fields
(from ECOCLIMAP) for both models
ARPEGE grid ? HIRLAM grid using closest grid
point for both climate files (ECOCLIMAP) and SWI
6
Work on progress
Implementation of the new physiographic
database ECOCLIMAP in the HIRLAM environment
Preliminary tests
Implementation of the SVA assimilation of soil
moisture (Balsamo) on the HIRLAM system
preliminary tests
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
Implementation of the SVA on the HIRLAM
environment
Preliminary tests Tiled sequential
assimilation (2 integrations) REF makes
bigger soil moisture corrections than SVA REF
corrects based also on T2m and RH2m errors
not related to incorrect SM! ... ...however
SVA performs in a more realistic way.
11
Evaluation strategy

• Soil moisture evolution (over the whole domain,
• over Europe, over specific regions)
• Study of soil moisture analysis increments.
• Near surface maps (T2m, RH2m) of monthly
  averaged bias,
• rms error.
• Verification scores against surface stations
  (T2m, RH2m)
• and radiosoundings.
• Surface heat fluxes
• Single grid point diagnostics
• Compare ELDAS products (precip., radiation and
  heating
• rates) with the corresponding HIRLAM fields

12
Computation requirements
Clock time needed for 6h cycle simulation 1
hour approx. For 1 month cycle simulation
1 week approx.
13
Time schedule

• First trial for 3 experiments (REF, ELD and SVA)
  (1-6 June 2000) (Dec. 2003)
• May ? Oct 2000 with REF exp. started (Jan. 2004)
• May ? Oct 2000 with SVA exp. started (Feb. 2004)
• May ? Oct 2000 with ELD exp. (?)

14
Acknowledgements
Thanks to Gianpaolo Balsamo (CNRM) Teresa
Fernández INMs Numerical Weather Dept....
15
ANEXOS
16
Tilling versus no tilling
17
ECOCLIMAP database(Masson et al. 2003)

• Global and high resolution dataset (1-km).
• Detailed information over Europe coming from
  CORINE and PELCOM projects.
• Use of full resolution maps of the vegetation
  index NDVI to provide the appropiate temporal and
  spatial scales.
• 215 ecosystems allowing a better assignement of
  vegetation parameter sets. (90 over Europe)
• Use of aggregation rules to derive surface
  parameters at the desired model resolution and
  for mixed ecosystem pixels.
• It allows the tiling approach, as used by the
  HIRLAM surface scheme.

18
ECOCLIMAP database II(topics)

• It is highly sophisticated as compared with the
  current HIRLAM physiographic description and it
  allows many possible choices
• Vegetation parameters (veg, lai, Zoh, Zom, alb,
  Rsmin, frac, emis, ...) with monthly or decennial
  (10 days) frequency.
• Compatible with tiling.
• Not yet enough tested in HIRLAM system.
• Straightforward usage for lat/lon coordinates.

19
(No Transcript)