in association with GOV ST F. Hernandez M. Balmaseda, Y. Fujii, K. - PowerPoint PPT Presentation

CLIVAR-GSOP report in association with GOV ST F. Hernandez M. Balmaseda, Y. Fujii, K. Haines, T. Lee, Y. Xue • Outcomes from the ongoing ORA-IP project • New real time ORA intercomparison GOVST 5 Meeting, Beijing, 13-17 October 2014

CLIVAR-GSOP/GODAE OceanView Ocean Reanalysis Intercomparison (ORA-IP, 2012-2014)  Reanalysis production is an on-going activity, following the feedbacks and outcomes of GSOP 2006-2009  New vintages are produced approximately every 5 years  Improved quality controlled observations (XBT corrections, Argo corrections and black lists)  Improved and extended forcing fluxes  Improved models and methods  We need to assess uncertainties among ocean reanalyses (through intercomparison and validation with independent data) due to model errors and bias, and observing system reliability over time  benefits of the ensemble approach both to improve the estimation of the signals and to provide uncertainty ranges  We need to facilitate the use of ocean reanalyses by other communities  We need to prepare for quasi-real time monitoring of the ocean Courtesy of M. Balmaseda GOVST 5 Meeting, Beijing, 13-17 October 2014

More than 20 participating ORA’s and observed products: • some coupled • from 1° to ¼° resolution • different models, forcing, DA See a summary at http://www.clivar.org/sit es/default/files/Exchange s/Exchanges_64.pdf Balmaseda et al, The Ocean Reanalyses Intercomparison Project (ORA-IP) JOO, accepted 2014 GOVST 5 Meeting, Beijing, 13-17 October 2014

Ocean Heat Content DA altimetry since 1993 by Matt Palmer 0-300m 0-700m Less dispersion near the surface, in particular after 2002 (Argo) 0-1500m 0-4000m

Correlation REAENS vs alti-GRACE Steric Sea Level (SSL) by Andrea Storto ORA water mass representation impact ! • The Intercomparison has helped to identify errors in some reanalyses products. • The Intercomparison has also helped to identify errors in some GRACE products. • Ensemble of reanalyses (REAENS) outperforms obs-only products (OAENS) (although comparison is not fair, since OAENS has less ensemble members). • Partition btw haline/thermal component less clear among ORAs, as well as Cx REAENS – Cx OAENS contribution at depth of the SSL trend Contours indicate 95% confidence level

Sea Level by Fabrice Hernandez • GSLR not assessed • Good performance of ORA assimilating satellite altimetry • What can we learn from ORA when ORA-EM smooth out noisy signals altimetry derived products look more • Valuable for regional sea reliable ? Correlation of ORA-EM (detrended and no seas. Cycle) level indices Against Tide Gauges SL index validation against SL-CCI • 97-98 Niño, Kelvin Waves • 3-4 mm/y negative trend (strengthening Trade Winds) • Consistent pattern, small spread SL index (0-12°N, 84-108°W)

Mixed Layer Depth (MDL) by Takahiro Toyoda Validation of MLDs from syntheses without model (EN3v2a, ARMOR3D) and ensemble mean of 17 Strongly dependent on forcing errors reanalyses (ENSMEAN) Differences from MILA-GPV and vertical mixing, technical issues deduced from individual TS profiles of Argo data (x)Month-(y)latitude diagram for temporal (2005-2011) and zonal mean values • Issues on the way MLD is computed, monthly time scales and vertical averaging: Smaller negative biases due to higher vertical resolution in the reanalyses • Negative biases in syntheses without model due to vertical, horizontal (within a grid) and temporal averaging of profiles (c.f., de Boyer Montegut, 2004) • Model biases in ensemble members largely canceled out in ENSMEAN

Surface Fluxes and Transports by Maria Valdivieso Time mean Global net surface Heat Flux and increment corrections Interannual Std ORA integrated assessment Most ocean model products have positive bias into the ocean (mean net surface heat flux into the ocean). The bias is often smaller than observational products, e.g., ISCCP/OAFlux and NOC2.0 Negative contribution of assimilation increments The bias is comparable than atmospheric (removing heat from the ocean on global reanalyses in some cases. average) Interannual variations are usually few Wm -2 Total neat heat flux still positive 2 W/m 2, ,smaller than the bias. consistent with net ocean warming

Sea Ice Mean March Sea Ice Thickness: 2007 by Greg Smith (Predictor for seasonal sea ice extent) GFDL UR025. GLOSEA5 GMAO 4 Large discrepancies among ORAs Too thin Mean March Sea Ice Thickness: 2007 MOVEG2 CMCC MOVECOR ICESa E t Too ERAN G2V3 thick LIM too thick CICE too thin Large variability central Arctic/Siberia

Summary and Outlook 1. Open Assessment of products • Ensemble mean appears to be a robust estimation 1. Intercomparison brought to assess the reliability of multi • Ensemble spread useful to estimate structural uncertainty model ensemble approach • Some relevant indices have been defined NEXT step: Dissemination of results in scientific literature. GODAE Special JOO paper accepted, Clim Dyn. Contribution ongoing Issue in JOO. Summary paper Special Issue in Clim Dyn. Individual contributions 2. We need to facilitate data access and usage: • Data repository of data entering the intercomparison (unified grid and format) 2. Under discussion: CMIP like repository (only ENS or all • Data repository with ensemble mean and spread. With a version ORA?) number to assess progress in the future. ORAIP v1 3. Highly expected from some GOV participant… ongoing 3. Monitoring of relevant indices still pending. acions 4. Balance between “Ensemble of All System” versus “Best Systems” needs to be addressed. 4. Multi-model ensemble approach versus ensemble from individual systems: common issues with Native Class 1 Courtesy of M. Balmaseda consensus forecasting approach

Real-Time Ocean Reanalyses Intercomparison Y. Xue, Y. Fujii, M. Balmaseda proposal Extend CLIVAR-GSOP/GODAE OceanView Ocean  Reanalyses Intercomparison Project (ORA-IP) into real time Assess uncertainties in temperature analysis of  tropical Pacific in support of ENSO monitoring and prediction Explore any connections between gaps in ocean  observations and spreads among ensemble ORAs Articulate needs for sustained ocean observing  systems in support of TPOS2020 Monitor signal-to-noise ratio in the global ocean  temperature, 300m heat content, depth of 20C isotherm Yan Xue Climate Prediction Center 11

http://www.cpc.ncep.noaa.gov/products/GODAS/multiora_body.html Duplicating with 1992- 6 OOS, joining FOAM (UK-Met) 2013 climatology and PSY3 (Mercator)

GODAS - The ensemble mean (ensemble spread) can be used to measure signal (noise). - The signal-to-noise (SN) ratio is relatively low in the western (central- eastern) Pacific where negative (positive) anomalies presented. - The low signal-to-noise ratio may be partially attributed to the sparse observations in those regions.

Signal, Noise and Signal-to-Noise Ratio (1985-2013)

Influences of ocean observations on spread among ocean reanalyses

Warm Water Volume Index Derived From Ensemble Mean of Ocean Reanalyses MJ 97 MJ 14 MJ 82 MJ 06 MJ 91 MJ 09 MJ 02 Jun 2002 Jun 1991 (DJF NINO3.4=+1.1) (DJF NINO3.4=+1.6) Jun 2006 (DJF NINO3.4=+0.7) Jun 1982 Jun 2014 Jun 1997 (DJF NINO3.4=+2.2) (DJF NINO3.4=+2.2) Jun 2009 (DJF NINO3.4=+1.6) - Warm Water Volume averaged in May-June 2014 is similar to that in May-June of 2009, 2006 and 1991. However, the pattern of subsurface temperature anomaly averaged in 5S-5N in Jun 2014 is mostly similar to Jun 1991.

Upper 300m Heat Content Anomaly Averaged in 5S-5N 82/83 97/98 91/92 14/15 06/07 12/13 09/10 02/03

CLIVAR-GSOP together with GOV • The GOV OSEval-TT workshop is hosting CLIVAR-GSOP discussions next december • The ORA-IP is a success, we are learning a lot from it (obs, model, forcing, DA limits and errors), and it will continue • The starting real time ORA intercomparison is: – Proposed to be endorsed and supported by GOV IV-TT – Participants are operational centres involved in GOV – The climate monitoring and ocean state reporting activity corresponds to what GOV OOC wanted to implement by participating to the CLIVAR-GSOP ORA- IP project • Near Real Time Ocean climate monitoring: – Could it be a GOV showcase, in association with OOPC? – There is an obvious link with seasonal prediction (at least for ENSO), which was not addressed specifically inside GOV – Status of the ocean observing system to be linked with OSEval-TT GOVST 5 Meeting, Beijing, 13-17 October 2014