Recovering climate-quality coastal sea level measurements from satellite altimetry!

Transcription

1 UK Sea Level Meeting 215 In honour of Philip Woodworth! Liverpool March 215! Recovering climate-quality coastal sea level measurements from satellite altimetry! Francisco M. Calafat 1, Paolo Cipollini 1, Luciana Fenoglio-Marc 2, M. Joana Fernandes 3, Andrey Kurekin 4, Clara Làzaro 3, Graham Quartly 4, J. Benveniste 5! 1 National Oceanography Centre, U.K., 2 Technical University of Darmstadt, Germany, 3 University of Porto, Portugal, 4 Plymouth Marine Laboratory, U.K., 5 ESA/ESRIN, Italy!

2 The ESA Sea Level CCI project! Aims at providing long-term monitoring of the sea level Essential Climate Variable (ECV) with regular updates, as required for climate studies, and uncertainty estimates! SL CCI Phase 1 completed in Phase 213! GCOS Large-scale/global Sea Level Measurement Requirements!! CCI Phase1:! ~.5 mm/yr! Data/documentation available at:!

3 GMSL from satellite altimetry! Ablain et al, Ocean Science, 215!

4 Geographical variation of SL trends!

5 Phase 2 (214-17): more focus on the coast! Coastal Altimetry has improved dramatically in the last fey years! And its were impacts are felt!! Work being done:! Better screening of data! Improved corrections (especially Wet tropospheric correction) + coastal tides! Assessment of coastal retrackers! Batch (2-D) retracking of waveforms! Regional validation! coastalt

6 User requirements in the coastal zone! a necessary preamble to this work is a quantification of the requirements for accuracy and long-term stability for climatequality observations of sea level in the coastal zone! This was done by a survey carried out by NOC! we asked altimeter specialists, i.e. experts of the processing and/or analysis of altimetric data, drawn from the International Coastal Altimetry Community and from 14 different countries!

7 Survey the questionnaire! Coastal Sea Level Questionnaire v1. Apr 214 Coastal Sea Level Questionnaire v1. Apr 214 A short questionnaire on Requirements+for+climate1quality+monitoring+ of+coastal+sea+level+from+satellite+altimetry+ Prepared by Paolo Cipollini, National Oceanography Centre, UK, cipo@noc.ac.uk for the ESA Sea Level CCI Project, Phase 2 WP1 Why this questionnaire? Within Phase 2 of the ESA Sea Level CCI Project there is a specific task to update the User Requirements for climate-quality monitoring of sea level from satellite altimetry. Phase 1 of the project had summarized the requirements from different sources (including GCOS, WMO/WCRP, GOOS, OSTST, the Coastal Altimetry Community and the CCI s Climate Modelling User Group) in the following table 1 : Synthesis of target sea level requirements from Sea Level CCI phase 1. Observable Horizontal Temporal Accuracy Long-term Stability resolution resolution Global mean sea level Global mean one orbital cycle mm Decadal scale: <.3 mm/y Annual scale: <.5 mm/y Regional sea 5-1km weekly 1 cm < 1 mm/y level Mesoscale 15 km daily.5 cm (No strong requirements) One issue that requires a dedicated focus in Phase 2 is the coastal zone. The purpose of this questionnaire targeted to altimetry specialist and expert users of altimetry data is to help us to define specific requirements for altimetry in the coastal zone, in terms of: - Accuracy: congruence of the single value ( single = averaged over one space and time grid cell ) to the true value - Long-term stability: consistency over time of the instrument calibration and corrections Note that the requirements in question are those for climate applications i.e. where one uses repeated observations to derive some statistical properties of the phenomena. A simple example to illustrate this concept: a one-off observation of some extreme event does not belong to the climate category; 1 The full User Requirement Document that this table is taken from is available at 2 Individual global mean sea level values are obtained by geographically averaging sea surface heights measured over the ocean during an orbital cycle (1 days for Topex and Jason satellites; 35 days for ERS and Envisat). To reach a 2-4 mm accuracy, individual (1Hz) sea surface height measurements must be accurate to 1-2 cm. +and+then+just+a+few+questions:+please+answer+them+based+on+ your+own+experience.+ Note that for each question you are asked to specify TWO values: a THRESHOLD value ( = the MINIMUM value that makes that parameter usable for at least one climate application) a TARGET value ( = a nice-to-have value that will enable a fuller range of applications TARGET values should be STRINGENT but REALISTIC at the same time!) Let us first focus on a LOCAL product, i.e. sea level on a single grid cell in the coastal zone (say a 15 km x 15 km stretch along the coast) and with a time resolution (i.e. time average) of ONE MONTH. Q1) What level of ACCURACY of LOCAL altimetric measurements of sea level would be required? THRESHOLD cm TARGET cm Q2) What level of LONG-TERM STABILITY of LOCAL altimetric measurements of sea level would be required? On an ANNUAL SCALE: THRESHOLD mm/y TARGET mm/y On a DECADAL SCALE: THRESHOLD mm/y TARGET mm/y Then let us think of a GLOBAL COASTAL product, i.e. one generated by quality-controlling and averaging all the measurements in the global coastal strip (-15 km from coast) and with a time resolution of ONE MONTH. Q3) What level of ACCURACY of GLOBAL COASTAL altimetric measurements of sea level would be required? THRESHOLD cm TARGET cm Q4) What level of LONG-TERM STABILITY of GLOBAL COASTAL altimetric measurements of sea level would be required? On an ANNUAL SCALE: THRESHOLD mm/y TARGET mm/y On a DECADAL SCALE: THRESHOLD mm/y TARGET mm/y Space available for specific comments: Done, thanks! The results will be made available in the updated User Requirement Document (via ) and discussed at ESA symposia, OSTST Meetings and Coastal Altimetry Workshops.

8 Requirements expressed as! ACCURACY (cm)! STABILITY over 1-y period (mm/y)! STABILITY over 1-y period (mm/y)! for a LOCAL product! (single cell 15km x 15km x 1mth in the coastal zone)! for a GLOBAL COASTAL product! global QC-screened 1-mth average in 15-km coastal strip! We asked for! a THRESHOLD value (minimum to enable at least one application)! a TARGET value! 4 surveys handed out, 15 (38%) returned we can start making some basic statistics.

9 Requirements: accuracy! ACCURACY+(cm) Median+and+[range] LOCAL THRESHOLD 3. [1.%,%15.] TARGET 1. [.1%,%5.] GLOBAL+COASTAL THRESHOLD 1.8 [.5%,%5.] TARGET 1. [.1%,%3.] ACCURACY+(cm) First Third+quartile LOCAL THRESHOLD TARGET GLOBAL+COASTAL THRESHOLD TARGET.4 1.

10 Requirements: 1-year stability! STABILITY+over+1+year+(mm/y) Median+and+[range] LOCAL THRESHOLD 3. [.5%,%1.] TARGET 1. [.2%,%6.] GLOBAL+COASTAL THRESHOLD 1. [.3%,%5.] TARGET.5 [.1%,%2.] STABILITY+over+1+year+(mm/y) First Third+quartile LOCAL THRESHOLD TARGET GLOBAL+COASTAL THRESHOLD.6 2. TARGET.3 1.

11 Requirements: long-term stability! STABILITY+over+1+years+(mm/y) Median+and+[range] LOCAL THRESHOLD 1.5 [.3%,%5.] TARGET 1. [.2%,%3.] GLOBAL+COASTAL THRESHOLD.9 [.1%,%2.] TARGET.4 [.1%,%1.] STABILITY+over+1+years+(mm/y) First Third+quartile LOCAL THRESHOLD TARGET GLOBAL+COASTAL THRESHOLD.5 1. TARGET.2.5

12 Additions to the requirements table!

13 Improvement of Wet Tropospheric ( = water vapour) correction (Univ. Porto)! Two types of correction have been developed:! GPD combines valid on-board radiometer, GNSS and model values.! Revisited GPD corrections for ERS-1/2, Envisat, T/P, Jason-1, Jason-2. New correction for AltiKa.! significant coastal improvement for older missions (ERS-1/2 and T/P), moderate for J-1, not significant for J-2.! Dcomb combines all available data from imaging microwave radiometers (such as AMSR- E), GNSS and model values.! Mostly useful for missions with no on-board MWR (e.g. CryoSat). Also useful as independent correction for validation purposes.! Computed for CryoSat-2 and SARAL/AltiKa!

14 Example 1: variance reduction, GPD wrt previous ( composite ) correction! Variance differences (cm 2 ) function of distance from coast. Both all latitudes and excluding high-lats (polewards of 55 ) ERS-1! Topex/! Poseidon! ERS-2! Jason-1! Envisat! Jason-2!

15 Example 2: variance reduction, Dcomb correction for SARAL/AltiKa and CryoSat-2! Variance differences (cm 2 ) function of distance from coast SARAL DComb MWR DComb ECMWF-Op CryoSat-2 DComb ECMWF-Op _!

16 Data Screening & Specialized Retrackers! Dedicated screening and filtering of existing data/corrections at high-rate (2 Hz).! find the optimum trade-off keeping as many data as possible but preserving climate-quality! Performance assessment of specialized waveform retrackers!! including ALES! Epoch : gives range (therefore height) Maximum amplitude: related to wind speed Slope of leading edge: related to significant wave height Figure from J Gomez-Enri et al. (29)

17 NOC s ALES retracker! radar! waveform! Adaptive Leading-Edge Subwaveform (ALES) retracker! Passaro et al., Rem Sens Env., 214! Pass 1: identification of leading edge and initial estimate of SWH h s(1)! ALES algorithm for width of sub-waveform window (from Monte Carlo Simulations) sub-waveform! Pass-2: subwaveform retracking (Brown Model) The sub-waveform approach keeps the measurement focused on the pulse-limited footprint of the instrument à prevents any range, σ, h! non-homogeneities in the far-view field from degrading its s accuracy.! à very good for coastal zone and fine scales!! See poster by Passaro et al on application to the Danish Straits

18 New ALES retracker: examples with Jason-2 SWH =.75 m open ocean SWH = 1.65 m coastal ocean SWH = 9.5 m open ocean r r Envisat 416 Validation vs Trieste TG SGDR ALES CTOH Jason interleaved Latitude ( N) 2 SGDR 15 ALES CTOH Jason Latitude ( N) 2 1 SGDR.8.6 ALES.4 CTOH.2 Passaro et al, Rem. Sens. Env., 214 r Km Km Km

19 RMS (m) r RMS (m) r RMS (m) r Absolute RMS difference w.r.t. TG black dashed line: RMS difference between Mean Sea Surface along-track and at the Tide Gauge (using DTU1 MSS) Envisat 416 SGDR 1 ALES 5 CTOH Jason interleaved Latitude ( N) 2 SGDR ALES CTOH Jason Latitude ( N) 2 SGDR ALES CTOH Latitude ( N) Km Km Km Distance from coast Passaro et al, Rem. Sens. Env., 214

20 Validation now extended to SWH using SWH from wave buoys in the German Bight (Passaro, Fenoglio & Cipollini, IEEE Trans. Geoscience and Rem. Sens., 215) cycle % for correlation > Land Jason vs Helgoland WB Latitude ALES 2Hz ALES 1Hz SGDR 1Hz SGDR 2Hz Land Distance From Coast (Km) cycle % for correlation >.9 Jason vs Helgoland Latitude Distance From Coast (Km) STD w.r.t. buoy (m) Latitude STD w.r.t. buoy (m) Latitude bias Jas buoy (m).5 For Pee Latitude bias Jas buoy (m) Latitude à ALES represents a significant step forward in retracking altimetry in the coastal region (and works well also offshore!)

21 Batch retracking (PML)! a retracker that fits a model to multiple waveforms at once: 2-D retracker"!

22 Batch retracking: producing a smooth output! SSH weights SSH weights SSH weights The results of each block of processing is a polynomial in SSH, ζ; a continuous output may be produced by linear weighting between solutions

23 Batch retracking: early results! Window_size= 21 samples, step=11 samples Ongoing work: To evaluate performance of 2-D retracker,! including response to spurious data spikes!

24 Validation of SLCCI products over the Mediterranean! L. Fenoglio! Trend in sea level over altimetry period is smaller than global mean (2.2 +/-.4 mm/yr), Step-wise increase of sea level in altimeter period Sea level budget over GRACE period almost closed with steric and mass contribution. Trends are consistent. Mass component dominates.

25 UK Sea Level SpaceWatch! Fast-track study (Jan-Mar 15) funded by UK Space Agency! Aims to demonstrate a prototype of a sustainable longterm service providing systematically updated sea level advice around the UK from altimetry and tide gauges! Envisat 35-day / AltiKa Envisat 3-day Rolling annual sea level trend (mm/y) time + Holyhead TG! Workshop on 3 latitude NOC Southampton See poster 5

26 New ocean data from CryoSat-2! Since Apr 214 ESA have been distributing new CryoSat-2 Ocean data from dedicated processor! IOP Interim Ocean Products (3-day latency)! GOP Geophys Ocean Products (3-day latency)! These are being validated by NOC (under ESA contract) very good quality! good agreement with TG a flying tide gauge!! As time series grow we will assess stability for climate research and compare MSL with other missions!

27 Example GOP data at Chatham Island! Total sea level (cm) Tide gauge SSHA CryoSat 2 12/4/14 12/5/14 11/6/14 11/7/14 1/8/14 9/9/14 29/9/14

28 Example GOP data at Chatham Island! Detided sea level (cm) /4/14 12/5/14 Tide gauge SSHA CryoSat 2 11/6/14 RMSE CryoSat/TG = 3.1 cm! Jason-2/TG (not shown) = 3.9 cm! 11/7/14 1/8/14 9/9/14 29/9/14

29 GMSL on a very short time span.! Colorado University CryoSat 2 RMSE CryoSat/TG = 3.1 cm! Jason-2/TG (not shown) = 3.9 cm! GMSL (cm) May Jul Aug Oct Dec Jan 215

30 Conclusions! Accuracy of GMSL rate estimates from satellite is approaching GCOS requirements! requirements in coastal zone only marginally lower! ESA SL CCI instrumental to this result: lots of technical work on algorithms/corrections, plus validation! and now (Phase2) special focus on coastal zone! UK Sea Level SpaceWatch aims to contribute to uptake of sea level advice from altimeters on a national basis:! concept can easily be extended! CryoSat-2 (and AltiKa) giving excellent data.! With J-3, S-3, S-6 forthcoming, space-based GMSL measurements are guaranteed for many years. Can we detect acceleration in these data?!