Towards the Operational Implementation of the Canadian Land Data - - PowerPoint PPT Presentation

Bernard Bilodeau Marco L. Carrera Dorothée Charpentier Mario Lépine Bruce Brasnett Stéphane Bélair

Towards the Operational Implementation of the Canadian Land Data Assimilation System

6th WMO Symposium on Data Assimilation

Environment Canada

CaLD CaLDAS AS at En t Envir vironment

• nment Canada

Canada

Need to modernize the land surface data assimilation system used to initialize numerical prediction systems at EC Currently, assimilation of surface and near-surface

• bservations based on OI-

type approach CaLDAS  emphasis on space-based data and on more sophisticated assimilation techniques

http://collaboration.cmc.ec.gc.ca/science/rpn/SMAP

The he CaLD CaLDAS AS pr projects

• jects

CaLDAS-screen CaLDAS-SMOS/SMAP CaLDAS-SNOW CaLDAS-Veg

Assimilation of screen-level air temperature and dew point temperature to initialize soil moisture and surface temperature; ensemble OI for snow depth (surface obs) Assimilation of L-band TBs from SMOS (and eventually SMAP) to initialize soil moisture Screen-level data used as forcing Assimilation of GlobSnow SWE retrievals and of snow coverage area from MODIS (eventually VIIRS) In combination with CaLDAS-SMOS/SMAP Vegetation characteristics (LAI, veg) provided by CTEM Evaporation conductance obtained from photosynthesis in SVS Possible assimilation of MODIS / VIIRS for LAI

1 2 3

CaLD CaLDAS AS Sc Schema hematic tic

LAND MODEL (SPS) OBS ASSIMILATION EnKF + EnOI

xb y

EnKF xa = xb+ K { y – H(xb) } K = BHT ( HBHT+R)-1 with

CaLDAS IN OUT

Ancillary land surface data Atmospheric forcing Observations

Surface Temperature Soil moisture Snow depth or SWE Vegetation*

Screen-level (T, Td) Surface stations snow depth L-band passive (SMOS, SMAP) MW passive (AMSR-E) *Optical / IR (MODIS, VIIRS) Combined products (GlobSnow) T, q, U, V, Pr, SW, LW Orography, vegetation, soils, water fraction, ...

Analyses

• f…

*) not done yet…

CaLD CaLDAS AS-scr creen een Configur Configuration tion

Ensemble Kalman Filter for T2 and w2; Ensemble OI for snow depth. 24 members 3-h assimilation window On 25-kmYin-Yang grid (will be

• n 15-kmYin-Yang grid next

summer) First guess from the Surface Prediction System (SPS), also known as GEM-Surf Only minor changes to land surface scheme (still ISBA) No changes to geophysical fields Perturbed atmospheric forcing (precip, TT, SW, and LW) Ensemble of CaPA precipitation analyses Observations are OI analyses of T_2m and Td_2m (every 3h) and surface measurements of snow depth Observation error from dispersion of OIs (for T_2m and Td_2m); fixed values for snow depth No bias correction Perturbations to analyses and to

• bservations (except for snow)

Impr Improved ed Sno Snow w Depth Depth Anal Analysis ysis with with CaLD CaLDAS AS

CMC-OP Open Loop ENS-OI OBS

Mean snow depth Bias STDE

Bélair et al, to be submitted.

In Winter… Improved Near-Surf Surface ace For

• recas

ecasts ts (Gr (Grea eater Impact o ter Impact on T n Temp) emp)

Bias STDE

Control CaLDAS Control CaLDAS

60 cases, in February and March 2011

In Summer… Improved Near-Surf Surface ace For

• recas

ecasts ts (Gr (Grea eater Impact o ter Impact on Td) n Td)

Bias STDE

Control CaLDAS Control CaLDAS

60 cases, in July and August 2011

Near Near-Surf Surface ace and Upper and Upper-Air… Conflicting Conflicting Biases Biases for temper

• r temperatur

ture

120h, North America (against radiosondes)

Control CaLDAS

July and August 2011, 104 cases (against surface obs) GENERALLY TOO COLD at 2m BUT TOO WARM IN THE LOWER TROPOSPHERE

Plans Plans for Oper

• r Operational

tional Implementa Implementation tion

Currently being transferred to CMC-Operations as an experimental system (YY 25- km) Is expected to be part of YY 15-km implementation at CMC, together with EnVAR A 2.5-km version will soon be tested for regional forecasting system (Ops in Summer 2014) Version 2 (w/ SMOS, SMAP, GlobSnow) expected for 2015

Correlation between SMOS and CaLDAS first guess Tbs