ESA Climate Change Initiative PhaseII Sea Surface Temperature

ESA Climate Change Initiative Phase-II Sea Surface Temperature (SST) www. esa-sst-cci. org Phase I results and Phase II plans for climate research Nick Rayner CMUG integration meeting, 2 -4 th June 2014

Optimised impact of ATSRs in SST CDR § New Multisensor Matchup System (MMS) supported techniques to cross-reference AVHRRs to ATSRs ● This gives AVHRR density of sampling … ● … with improved accuracy, stability and independence from ATSRs Daily independent “climate-quality” coverage ATSR Reprocessing for Climate SST CCI Phase I results and Phase II plans for climate research SST CCI CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 2

SST CCI products The three ESA SST CCI LT version 1. 0 products assessed are: o o o ATSR. SSTs from ATSR instruments in L 3 U format at 0. 05° latitude by 0. 05° longitude resolution covering 1991 – 2010. (Hereafter, SST CCI ATSR. ) AVHRR. SSTs from AVHRR instruments in L 2 P format at Global Area Coverage (GAC) resolution covering 1991 – 2010. (Hereafter, SST CCI AVHRR. ) Analysis. Satellite-only SST-depth L 4 daily analysis created by OSTIA system from SST CCI ATSR and SST CCI AVHRR products at 0. 05° latitude by 0. 05° longitude resolution covering 1991 – 2010. (Hereafter, SST CCI analysis. ) These are utilised over the period 1991 -2010. SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 3

Uncertainty Estimation § URD established no commonality of spatio-temporal resolution required by climate users § Therefore, SST CCI provides data at a fundamental resolution (4 km / 0. 05 deg; daily) and a tool to create lower-resolution data § Uncertainty estimates should be provided with every SST § At L 2 P / L 3 U this is estimated as part of the retrieval process § For uncertainty estimates to be available for lower-res outputs of the tool, need to separate components of uncertainty that respond to averaging differently § First time these have been estimated and provided in an EO product (? ) § View uncertainty estimates as part of product as much as the SSTs themselves § Uncertainty estimates are therefore validated in their own right SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 4

Phase II plans: Sea surface temperature user workshop on uncertainties Met Office Hadley Centre, Exeter, UK, 18 -20 th November 2014 The aims of the meeting are to: Exchange information about uncertainties in sea surface temperature (SST) observations; Create new expert SST users who, through publication of their work, can inspire others to take uncertainty information into account; Source requirements from SST users on uncertainty information and other aspects of the Climate Data Records; Spread best practice through a follow-on meeting report or journal article. SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 5

Global, monthly average SST anomaly (relative to average for 1985 -2007) Overall variability is similar except for erroneous variability in the 1990 s in the SST CCI AVHRR and ATSR products. The SST CCI products are more consistent after 1996. Integration meeting, Met Office, Exeter SST CCI Phase I results and Phase II plans for climate research Trend of SST CCI products is at CMUG upper end of comparison range 2 June 2014 Page 6

Stability SST CCI 95% confidence interval (m. K year ) for 1991 – 1995 SST difference between SST CCI products and the GTMBA -1 Day SST CCI ATSR SST CCI AVHRR -137. 9<trend<-2. 4 105. 9<trend<462. 3 SST CCI ATSR -13. 6<trend<60. 1 -7. 4<trend<36. 8 SST CCI 95% confidence interval (m. K year-1) for 1995 – 2010 Day Night SST CCI AVHRR -12. 3 < trend < -7. 4 -2. 0 < trend < 2. 0 SST CCI ATSR 0. 7 < trend < 3. 2 -1. 4 < trend < 6. 4 Both SST CCI analysis SST CCI Phase I results and Phase II plans for climate research Both -1. 8<trend<22. 1 SST CCI AVHRR SST CCI analysis GCOS target is 3 m. K/year Night CMUG Integration meeting, Met Office, Exeter 0. 1 < trend < 3. 2 2 June 2014 Page 7

Phase II option 8: Use of Argo network Develop and test new ways to assess the stability of the SST CCI products Assess whether or not Argo network offers satisfactory stability reference data set for the 21 st century Assess SST CCI products for artificial trends and step changes through comparison to moored buoys and the Argo network May allow assessment of stability globally – currently only possible in the tropical Pacific. SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 8

Coupled model evaluation Simulated SST CCI minus Daily OI SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter Simulated minus SST CCI Simulated minus Daily OI 2 June 2014 Page 9

North Indian Ocean Spurious variability in North Indian Ocean in SST CCI AVHRR product Thought to be due to CLAVR-X cloud detection passing desert dust SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 10

Phase II option 11: passive microwave SST production and impact assessment Development of microwave SST products from AMSR-E and AMSR-2 Blending with SST retrievals from infrared sensors to create L 4 analysis, 2002 -2016 May improve retrievals in challenging regions with high aerosol/dust loading or in persistently cloudy regions. SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 11

Regional phenomena – Dipole Mode Index Overall variability is similar in the SST CCI and comparison with some differences. Peak seen in comparison data sets in 2009/10 missing in SST CCI products SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Some peaks to trough variability higher in SST CCI products Page 12

Improved feature resolution Jun–Dec 1997 Cold area off Indonesia is much colder in SST CCI products In situ only data sets have much weaker cold anomaly: e. g. Had. SST 3, ERSST, Kaplan, COBE Similar improvements are seen in the Tropical Atlantic SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 13

Exploration of heat transport in ocean models via Tropical instability Waves Provision of daily mean information in SST CCI analysis is a better natural comparator to the simulated daily data than SSTfnd or the daily “means” provided in the Daily OI. Courtesy Tim Graham (submitted to Ocean Modelling) SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 14 Improved agreement during the El Nino period

Simulation of tropical storm track density The tropical cyclone tracks may be influenced by the higher SSTs in the SST CCI analysis compared to those in the Reynolds et al Daily OI. Courtesy Malcolm Roberts (submitted to SST CCI Phase I results and Phase II plans for climate research CLIVAR Exchanges) increase in theslightly tracks shifted frequency ofelongated tropical and southwards, cyclonemore tracks slightly frequent CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 15

New background covariance errors: Results • Improvements to analysis method (the OSTIA system): background errors and improvement to numerical convergence • Improved accuracy of SST CCI analysis • Improved resolution of small-scale ocean features without introducing unrealistic observational noise. Old OSTIA Updated OSTIA Drifter o-b 0. 52(-0. 01) 0. 37(0. 00) AATSR o-b 0. 45(0. 04) 0. 37(0. 03) ARGO o-a 0. 47(0. 03) 0. 43(0. 04) SST CCI Phase I results and Phase II plans for climate research SST gradients in Gulf Stream in SST CCI analysis CMUG Integration meeting, Met Office, Exeter Observation-minus-background/analysis RMSE (bias) calculated for March 2012. 2 June 2014 Page 16

Phase II plans: dedicated climate modelling Malcolm Roberts, Met Office Hadley Centre In the third year of the project, we will run simulations using the latest version of the Met Office Hadley Centre model at a spatial resolution of about 25 km. This will be achieved by running two dedicated experiments where the atmosphere is driven by (i) the SST analysis used as standard for such experiments, possibly the Reynolds et al Daily OI and (ii) reprocessing 3 of the SST CCI L 4 analysis. These simulations will be made for about 30 years of the 1981 -2016 period, using exactly the same model version, allowing a robust assessment of the impact. It will be necessary prior to the analysis to choose specific processes or locations to analyse according to our assessment of where the greatest impact is likely to be. SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 17

Phase II: Principal challenges § To extend the SST CCI climate data record (CDR) before and after the period of the Along-Track Scanning Radiometers (ATSRs); § To evolve the SST CCI CDR throughout the data period… § To evolve the existing prototype system into an implementation that is : ● sustainable in the long term ● able to harness the best scientific SST R&D ● able to provide necessary performance: – a ‘nimble’ improvement cycle for reprocessing – high-performance capacity to store and process relevant data flows now and in the coming era of Sentinel missions SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 18

R&D Passive Microwave Sensors SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 19

Summary (see also CAR at http: //www. esa-sst-cci. org/? q=webfm_send/104) SST CCI Phase I products demonstrated to be of use in various climate research applications Challenges remain, e. g. noise in AVHRR retrievals in Arabian Sea arising from intermittent desert dust episodes; inability to assess stability outside of the tropical Pacific; the need to create a stable record back to 1981 In Phase II, we will: host a user workshop to discuss needs for and presentation of uncertainty information; perform 30 -year 25 km resolution climate model simulations, driven by the SST CCI analysis product to explore the impact of improved resolution on aspects of atmospheric variability; automate aspects of the climate assessment framework to facilitate the development of a sustainable system; continue to encourage users of the SST CCI products to provide feedback on their use of the data SST CCI Phase I results and Phase II plans for climate research CMUG Integration meeting, Met Office, Exeter 2 June 2014 Page 20