JCSDA: Ocean Data Assimilation

JCSDA: Ocean Data Assimilation • JCSDA partners - ocean analyses: • NCEP/EMC and NASA/GMAO - seasonal climate forecasts • Navy - ocean weather forecasts • SST - Key intersection with Meteorology colleagues • Key contributions sought in JCSDA collaborations: • improved SST • improved surface forcing • improved error covariances for surface forcing (particularly winds) • improved observational error covariances (particularly altimetry), including representation error • improved ocean data assimilation methodology JCSDA 3rd Workshop on Satellite Data Assimilation

Seasonal to Interannual Forecasting at NCEP Ocean Initial Conditions Altimeter Argo Seasonal Forecasts for North America with Climate Atmosphere GCM SST XBT TAO Coupled Ocean Atmosphere Forecast System (CFS03) CCA, OCN MR, ENSO Global Ocean Data Assimilation System (GODAS) IRI SST Anomaly Surface Temperature & Rainfall Anomalies Stress Heat Fluxes CCA, CA Markov IRI Forecasters E-P Official Probabilistic Surface Temperature & Rainfall Forecasts Scatterometer Official SST Forecast JCSDA 3rd Workshop on Satellite Data Assimilation

JCSDA: Ocean Data Assimilation • Improved SST analysis (Xu Li and John Derber, NCEP/EMC) • Detection and correction of aerosol contamination in infrared SST retrievals (Jim Cummings, NRL/Monterey; Andrew Harris, NESDIS) • Ocean emissivity/reflectance model (N. Nalli, NESDIS, JSDI) • Improving analysis of tropical upper ocean conditions for forecasting (Jim Carton) • Errors in sea level height analyses (Alexey Kaplan, LDEO) • Salinity impacts in the GMAO ocean data assimilation (Chaojiao Sun, GMAO) • Ensemble Kalman filter and altimetry assimilation (Christian Keppenne, GMAO) JCSDA 3rd Workshop on Satellite Data Assimilation

Improved NCEP SST AnalysisXu Li, John DerberEMC/NCEP Project Objective:To Improve SSTAnalysis Use satellite data more effectively Resolve diurnal variation Improve first guess JCSDA 3rd Workshop on Satellite Data Assimilation

Improved NCEP SST AnalysisXu Li, John DerberEMC/NCEP • Progress • SST physical retrieval code has been merged into GSI and provided to NCEP marine branch for test in operational mode • An extensive diagnostic study on the diurnal variation signals in in situ and satellite observations, SST retrievals, SST analysis and associated air-sea fluxes (NCEP GFS product) shows the SST diurnal variation needs to be addressed to improve the SST analysis product. • 7-day 6-hourly SST analysis has been produced with GSI, after a new analysis variable, in situ and AVHRR data were introduced into GSI. • Plan • Analyze SST by assimilating satellite radiances directly with GSI • Active ocean in the GFS • Aerosol effects JCSDA 3rd Workshop on Satellite Data Assimilation

Progress (1):Use satellite data more effectively • SST retrieval (with AVHRR Data) • Navy Retrieval  Physical Retrieval • Improved Analysis (Exp. done 2003-2004) • Physical retrieval code has been merged into GSI • Physical retrieval algorithm is running operationally since March 2005 • SST analysisby assimilating satellite radiances directly with GSI • Use more satellite data • Add a new analysis variable in GSI: skin temperature of ocean • Errors of observation and first guess • Add SST In Situ and AVHRR observations to GSI • Experiments on SST or Skin Temperature analysis with GSI • Control: No In Situ & AVHRR, daily first guess (weekly analysis) • EXP1: With In Situ & AVHRR, daily first guess (weekly analysis) • EXP2: In Situ & AVHRR, 6-hourly first guess (previous 6-hourly analysis) JCSDA 3rd Workshop on Satellite Data Assimilation

Progress (2):Resolve Diurnal Variation • Problems caused by the lack of SST diurnal cycle • Radiance bias correction • SST Analysis bias • Others: DV is an essential weather variation • Boundary condition: flux calculation precision • Evaluation of cost function in data assimilation • Feasibility to resolve diurnal variation (Diagnostics) • Observation • Buoy • Satellite retrieval • Flux (from GFS) • SST prediction in hourly time scale • 6-hourly SST analysis by GSI JCSDA 3rd Workshop on Satellite Data Assimilation

JCSDA 3rd Workshop on Satellite Data Assimilation

Detection and correction of aerosol contamination in infrared seas surface temperature retrievals Contributors: NRL: Jim Cummings (PI), Doug Westphal (PI), Jeff Hawkins; NAVO: Doug May; UMD: Andy Harris Project Objectives: • Detection of aerosol contamination in infrared satellite sea surface temperature (SST) retrievals using Navy Aerosol Analysis Prediction System (NAAPS) aerosol distributions. • Correction of satellite SSTs for aerosol contamination using NAAPS aerosol products. JCSDA 3rd Workshop on Satellite Data Assimilation

JCSDA Workshop on Satellite Data Assimilation Project Tasks • Collocate NAAPS optical depth forecast fields valid for the time SST retrievals are generated (Doug May, NAVOCEANO). • Estimate SST retrieval reliability relationship to AOD content (Doug May, NAVOCEANO - Jim Cummings, NRL) • Develop SST quality control schemes to recognize aerosol contamination (Jim Cummings, NRL). • Correct satellite SSTs for aerosol contamination (Andy Harris, NESDIS). • Validate NAAPS aerosol products using using independent data - improve NAAPS model (Jeff Hawkins, Doug Westphal, NRL).

Detection and correction of aerosol contamination in infrared seas surface temperature retrievals Contributors: NRL: Jim Cummings (PI), Doug Westphal (PI), Jeff Hawkins; NAVO: Doug May; UMD: Andy Harris MODIS NAAPS • Summary of Accomplishments • Collocation of NAVO AVHRR SSTs and NAAPS aerosol AODs for each aerosol type 4 times daily • Development of discriminant QC function to identify possible aerosol contamination • Validation of NAAPS AODs for select Saharan dust events using MODIS AOD retrievals • Development of reduced predictor scheme which improves correction of satellite BTs by at least a factor 3 over AOD alone 0.1 0.4 1.6 6.4 Point-for-point comparison in case study shows bi-modal distribution of bias Addition of total clear-sky transmittance and air-sea T predictors improves BT correction 3.7 µm 11 µm 12 µm Future: - Test correction scheme against aerosol-independent (e.g. MW) SST. Use independent data to improve NAAPS product JCSDA 3rd Workshop on Satellite Data Assimilation

Suggested form of k-estimation k-coefficients will be different for different aerosol types JCSDA 3rd Workshop on Satellite Data Assimilation

Ocean Surface Reflection / Emissivity Model Nicholas R. Nalli and Chris Barnet, Walter Wolf, Mitch Goldberg (Co-Is) NOAA/NESDIS/ORA Paul van Delst, John Derber (EMC POC) NOAA/NCEP/EMC JSDI - FY05 JCSDA 3rd Workshop on Satellite Data Assimilation

Contributors ORA: Nicholas Nalli (PI), C. Barnet, W. Wolf, M. Goldberg EMC: P. van Delst, J. Derber M-AERI Brightness Temperatures, AEROSE 06-Mar-04 Enhancement of Ocean Reflection/Emissivity Model Summary of Accomplishments • Theoretical development of quasi-specular reflection model in earlier work (Nalli et al., 2001) • Sample M-AERI spectra collected from AEROSE 2004 to be published in JGR Special Issue on AIRS validation (Nalli et al., 2005) – these data will be used to verify and validate models • UMBC Kcarta installed and compiled on Linux workstation From Nalli et al. (2005) Future • Conduct statistical analyses of M-AERI field data to verify/quantify model error. • Characterize window channel spectral response functions and centroid Planck approximations. • Run forward radiance calculations for window channels over range of wind speeds, zenith angles and atmospheric conditions. • Derive lookup tables and/or parametric fit of reflection diffusivity angle designed to be used in conjunction with the Wu-Smith emissivity model currently implemented by NCEP/EMC. • Validate model calculations against M-AERI spectra acquired during oceanographic field campaigns, including AEROSE 2004. • Tech transfer to NCEP/EMC for operational implementation into CRTM. • Implementation within AIRS forward model.

Ocean Surface IR Emissivity and Reflectance • Radiance emissivity models (e.g., Masuda et al., 1988; Watts et al., 1996; Wu and Smith, 1997) have been derived from Cox-Munk wave slope statistics. • Lookup tables (LUT) of model emissivity are used in radiative transfer modeling: • Reflectance of atmospheric radiance is a more challenging problem: • Surface is neither specular nor Lambertian, but quasi-specular • Depends upon the hemispherical radiance distribution • Using 1  may lead to systematic errors of forward radiance in window channels • These errors may be significant for applications requiring high accuracy (e.g., SST) JCSDA 3rd Workshop on Satellite Data Assimilation

Marine Atmospheric Emitted Radiance Interferometer (M-AERI) • Ship-based FTS designed to sample downwelling and upwelling calibrated IR spectra near the surface (Minnett et al., 2001) • High accuracy calibration is achievedusing 2NIST-traceable blackbodies • Derived products include • High accuracy radiometric skin SST derived from semi-opaque spectral region (~7.7 µm) (Smith et al., 1996) • Essential for accurate cal/val of advanced instruments and algorithms • 0.1 K absoluteaccuracy– this simply is the “gold-standard” for satellite “ground truth” • Continuous retrievals of lower tropospheric profiles at turbulent time scales (e.g., Feltz et al., 1998) • Retrieval of ocean surface spectral emissivity JCSDA 3rd Workshop on Satellite Data Assimilation

Improved Analysis of Tropical Upper Ocean Conditions for Seasonal to Interannual Forecasting Contributors: UMD: Jim Carton (PI), Gennady Chepurin; EMC: David Behringer Collaborators: Hailong Liu, Ching-Yee Chang, Sumant Nigam Project Objectives: Identify the forecast bias in the current NCEP ocean assimilation system (GODAS) and explore methods to control its impact on forecasts. Observations have spatially correlated errors while ocean models have slowly varying biases. Most current data assimilation schemes assume zero bias.

Improved Analysis of Tropical Upper Ocean Conditions for Seasonal to Interannual Forecasting 95m temp. b ias Contributors: UMD: Jim Carton (PI), Gennady Chepurin; EMC: David Behringer Collaborators: Hailong Liu, Ching-Yee Chang, Sumant Nigam Bias stationarity, seasonality suggests a simple forecast bias model • Summary of Accomplishments • Identification of time mean and seasonal bias in ocean data assimilation • Development and coding of reduced space two-stage bias correction algorithm • Examination of systematic errors in seasonal cycle of CFS Reduced-space bias correction Future: - implementation and extension to multivariate. Exploration of seasonal bias in CFS Where the bias is projected on our model EOFs :

Bias in temperature annual cycle JCSDA 3rd Workshop on Satellite Data Assimilation

95m Temperature bias EOF decomposition JCSDA 3rd Workshop on Satellite Data Assimilation

Bias model • Based on EOF analysis we propose the following bias model: , and Gi is i-th EOF. where JCSDA 3rd Workshop on Satellite Data Assimilation

Test with climatology Bias added to Levitus climatology. Then the climatology is sampled at 5% of grid points and the bias correction algorithm applied. JCSDA 3rd Workshop on Satellite Data Assimilation

Sea level height errors in models and data:Monte Carlo simulations and spectral distributions Contributors, LDEO: A.Kaplan (PI), M.Cane (co-PI), T.Merlis; Collaborators, GMAO: A.Borovikov, M.Rienecker, C.Keppenne Project goal: to provide realistic initializations for ocean components of seasonal-to-interannual climate forecast systems by assimilating satellite observations. To achieve this goal we have to investigate the errors of observational and model data sets and the ways to model them statistically. Dynamical simulations of the ocean model error are still quite rare (Cane,Miller,Kaplan et al.; Fukumori et al.; Borovikov et al. 2005)

Sea level height errors in models and data:Monte Carlo simulations and spectral distributions Latitudinal dependence of time evolution of normalized spread in sea surface height for the GMAO ensemble Theoretical estimates for tide gauge error show good consistency with the actual errors for all tropical Pacific tide gauge stations but four (known to manifest local effects or phase shifts of Rossby waves) Contributors, LDEO: A.Kaplan (PI), M.Cane (co-PI), T.Merlis; Collaborators, GMAO: A.Borovikov, M.Rienecker, C.Keppenne • Summary of Accomplishments • The AMIP-based ocean ensemble from GMAO (Borovikov et al. 2005) was analyzed and evaluated vs ensembles driven by small-scale noise. • Ratios of temporal and spatial contributions to the small-scale sea surface height variability were analyzed and used for modeling observational error in monthly tide gauge sea level values; wavenumber spectra are used for satellite altimetry error estimates with regards to the model grid size. Future: Finalizing error estimates; ensembles of ocean simulations perturbed by small-scale variability; data assimilation implementations

Connection between surface geostrophic kinetic energy and small-scale variability in sea surface height: • <s2>=C(f/g)2 <K> where C=a (Lx2+Ly2)/6, and • depends on the wavenumber power spectrum of the ocean sea surface height. Parameter adescribes how small differences in sea surface height scale to the LxXLy box JCSDA 3rd Workshop on Satellite Data Assimilation

Summary • A lot of progress • Some good collaborations - more will evolve from this workshop • From this workshop: • Collaborations on SST • use of aerosol data - collaboration with Clark Weaver (and A. da Silva) on aerosol products • include emissivity/reflectance model in forward model in NCEP radiance assimilation • include ocean mixed layer model (Li, Harris, Rienecker) • Collaborations on Assimilation methodology: • bias estimation (Carton, Keppenne) • methods to assimilate altimetry - noise models (Kaplan, GMAO) • ensemble generation (Kaplan, GMAO) • Collaborations on sources of model bias: • free-running coupled model (Carton, NCEP, GMAO) JCSDA 3rd Workshop on Satellite Data Assimilation

JCSDA: Ocean Data Assimilation