1. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 1 - Future altimetry design From impact studies to operational metrics or the reverse ? G.Dibarboure J.Dorandeu P.Escudier

2. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 2 - Introduction Should early impact studies define operational metrics or the reverse ? Framework : support to future mission design (ESA, CNES, Eumetsat, Ifremer) –Definition of Sentinel-3, Phasing options for the Jason tandem –Figure of merits for future altimetry concepts : wide-swath, large constellations… –Impact of payload changes : noise level reduction, cost reduction… Exploratory studies : –Iterative process (mission concept  performance assessement) –CLS’ Toolbox for Mission Analysis Testing and Optimisation (TOMATO) –Two types of analysis : mission analysis & OI based OSSEs Need : simple yet compelling results –Subtle payload differences ?  Metrics used must give a clear answer –Conflicting mission objectives (e.g.: OC vs Altimetry on S3) ?  Altimetry metrics must be convincing for decision-makers  E.g: 15% of additional variability observed is not convincing for neophytes Illustration of the DUACS approach through two ongoing studies : Post-EPS reference orbit & High-resolution altimetry

3. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 3 - Example 1 : Post-EPS reference orbit Question asked by Eumetsat : which orbit should be used for future reference missions ? –Reference orbit (T/P, Jason) is exceedingly aggressive –Onboard anomalies and failure (Jason-1 has burnt most redundant safeties) –The motivations behind the TP orbit choice are no longer major constraints Many factors to take into account : –History and existing time series –Sampling capability, Aliasing issues –Error budget (e.g. : POD performance vs orbit parameters) –Many applications : climate, mesoscale, ice monitoring, hydrology… –Mission cost (launch, operations, mission lifespan) –And other altimetry missions ! Unlikely to find a single perfect orbit, so the study rationale is to : –Sort out (many) bad options  first filter : no time wasted –Analyze orbit candidates with more details  second filter process : when in doubt, trash it –Keep only a handful of interesting orbits for the community to check out

4. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 4 - Post-EPS : Aliasing analyses First filter : orbit geometry and base properties –Acceptable altitude and inclination range –Repetitive –Acceptable repeat (sub)cycle duration –Optimal to host a tandem of 2+ altimeters Second filter : tidal components –Must allow tidal wave observation within 3 to 5 years (aliasing under control) –Tidal components must be separable within a reasonable time span Basic selection leads to 1400 options Drastic separability requirements : 0 option Trade-offs  many options with different pros/cons

5. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 5 - Post EPS : multi-satellite sampling analysis Geometrical sampling analysis (no model, no OI) –Observation quality (correlation between structure and observation) –Ability to detect mesoscale changes in NRT –Observation isotropy (e.g.: currents mapping, crossovers) –Structure monitoring/tracking capability –… Protocol validation on historical missions After this screening process : 12 candidates interesting

6. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 6 - Post-EPS - Output of step 1 : first orbit selection Altitude (km) Inc (deg) Cycle (days) Exact repeat cycle (days) Tide Aliasing Tide separabili- ty (years) Optimi- sation for 2 satel- lites Sub- cycle (days) S1 aliasing (days) rev/day Scores from ‎ 2. 5 A878_i66_c10878.73166109.901936 K1 included 3.5 yno1100.9713+9/107.001 A1150_i72_c111150.317721110.917843No K14 yyes5132.8913+2/11105.02 A964_i74_c19964.879741918.860482No K15 yyes3135.1813+13/19 116.03 1 A835_i75_c19835.619751918.860551No K15 yyes1135.2514+1/19 116.03 1 A1076_i68_c111076.855681110.904653 no K1, but alias K1>2cpy afterall 6yyes3114.3613+4/1119.01 A801_i71_c22801.857712221.810438 K1 inluded 6 yno7115.0514+3/2219.011 A1361_i65_c111361.612651110.905338 K1 included 6 yearsyes3115.2012+7/1118.01 A912_i70_c11912.147701110.905479 K1 included 8 yearsyes5115.3713+9/1118.01 A822_i68_c15822.474681514.858551 K1 included 8 yearsyes1105.0414+1/1517.01 A1104_i76_c161104.802761615.895629K1<2cpy-no3152.313+5/16104.02 A923_i67_c9923.3656798.915522K1>2cpy-no4105.5313+7/9103 A926_i67_c13926.436671312.878103K1>2cpy-no4105.6413+10/13214.01 Initial selection (tidal filter nominal) Additional selection (relaxed tidal aliasing requirements, except on 4-9 cpy climate band)

7. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 7 - Post-EPS : Mesoscale sampling capability (1/2) Analysis performed from OI OSSE based on Mercator simulations Mercator « reality »  Observation simulated  OI used to reconstruct Reconstruction error gives access the sampling capability

8. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 8 - Post-EPS : Mesoscale sampling capability (2/2) Once suboptimal options are removed, the mapping process offsets uneven sampling  minor differences Sampling error on U/V vary by ~10% in non coordinated tandems Impact of orbit inclination on sampling isotropy still still visible after mapping (especially combined with high-inclination S3) Three good candidates (T/P-like with ~10% more data thanks to lower altitude): results coherent with geometrical analysis SWOT orbit 22d is not the best option to host (only) a traditional altimeter Any contribution from GODAE would be useful to complete this study (model-based OSSE, metrics suggested…)

9. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 9 - Example 2 : High resolution altimetry Explore the benefits of a 24 satellite constellation (Nadir only) –Next generation of Iridium = altimeter payload passengers ? –Cost minimized (minimal payload, error budget tradeoffs) Comparison to a global wide-swath altimeter observation Impact of noise reduction (AltiKa, doppler altimetry, SWOT) First step : geometrical analysis –To provide a first quantification of the benefits –To tune the altimeter payload distribution on the 66 potential Iridium slots –To explore multiple time+spatial scales (e.g.: meteo, mesoscale, submeso) Second step : OI impact study on (sub)mesoscale –Needed to quantify the impact on currents and vorticity and the HF or short scale specific error Work performed with support from CNES and Ifremer Up : Jason cycle / sub-cycle scanning pattern Down : Space/Time scale observation limit

10. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 10 - Constellation detection and monitoring skill (left : 150km, right : 20km) High-resolution altimetry : geometrical analysis Instantaneous correlation between one snapshot and past altimetry data (realistic correlation model 150km/15d, arbitrary snapshot from day 12)

11. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 11 - Model EKE : ES Reality used : POP or Earth Simulator (ES) outputs –Configurations analysed 1 to 4 sats, 24 Iridium, SWOT –Realistic error levels on simulated observations Ongoing work : –Actual mapping reconstruction error (H,U/V, Vorticity) –First step : crude mapping parameters (100km, 5 to 10d) –Separation of error on HF/LF content (time & space) –Separation of error from mapping limitations & sampling limitations –For POP content SWOT sampling is good and Iridium excellent –For ES, SWOT temporal sampling is more problematic, but correlation scales must be revisited High-resolution altimetry : OI impact study Model EKE : Los Alamos 1/10° 2000 cm²/s²

12. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 12 - High-resolution altimetry : impact of noise level Starting question : how does the altimeter data high frequency error (instrument noise, processing error…) affect the power spectrum ? Earth Simulator output  Sampled along altimetry ground tracks (50 days of ideal obs) Variable white noise  Realistic observations Consistent with spectrum slope of actual data in Gulf Stream (-3.4 for [90-200 km] for 2.5 to 3cm HF content) Impact of SWOT roll+baseline error : far range spectrum is K -3 and increasing to K -11/3 as the data get closer to the Nadir position in swath Reducing the high-frequency error is important : Ka-Band, Doppler, SLOOP project processing… Noise std Spectrum slope (50 to 200 km) Spectrum slope (80 to 200 km) 3 mm-3.72-3.69 1 cm-3.59-3.70 2 cm-3.21-3.60 3 cm-2.90-3.24 3 cm 1 cm 3 mm SSH power spectrum (Jason-2 simulated data from ES reality + variable HF error level) K -5 or K -11/3 ?

13. 2nd GODAE Observing System Evaluation Workshop - June 2009 - 13 - Summary and Conclusion Overview of ongoing studies –Long term questions : new reference orbit, impact of HF error, high-resolution sampling… –Short answers : 3+ good orbits, reduce the noise, attractive 24 satellite concept (complement to SWOT ) Two-types of studies carried out by CLS : mission analysis & OI impact studies –Excellent way to explore unusual configurations or numerous variants (sort out poor options) –Some metrics are more convincing for decision-makers than classical results E.g. : mesocale can observed in real time with 12 satellites is a stronger message than 4% mesoscale observation error removed with +8 sats –Full science content must be consolidated afterwards (finer quantification once the general concept is nailed down) Past impact studies helped define current operational metrics (DUACS : K.P.I) Conversely any operational metric can be deployed for such a demonstration This logic is applicable to GODAE models : design/impact studies  operational metrics For future concepts, we need to be consistent with future operational metrics What routine-to-be metrics should be used to help design future observing systems ? So what will be requirements of GODAE models in 2018+ ?