Arctic SST Algorithms Validation of 6 operational SST

Arctic SST Algorithms • Validation of 6 operational SST products in Arctic • High latitude algorithm developments (CCI round robin) • Validation of CCI products • Conclusions/Challenges • Future work

Arctic conditions Arctic Ocean is a challenging region for SST: • Persistent cloudiness • Sea Ice • Complex atmosphere • Few in situ data • Extended periods with twilight, day and night only -> These issues make SST retrievals challenging

Validation of operational products

In situ coverage • Majority of validation results from Nordic Seas and Barents sea • I. e. Validation results NOT representative of inner Arctic In situ coverage from Høyer et al, 2012

Validation results • 6 operational products • Solar elevation angles with AATSR, Metop-A and AMSR-E

High latitude algorithms within CCI

Regional vs global coefficients, Arctic Ocean • Daytime : NLSST • Nighttime: SST_3. 7 • Sensors: NOAA 17, 18, 19 and Metop-A (2006 -2010) • Results from independent test data set. • NOAA-19 has much fewer match-ups than others. • Largest improvements for daytime algorithms • Reduced regional bias in most cases compared to global algorithms

Atmospheric temperature effects Arctic Ocean • Much more temperature inversions in Arctic compared to Southern Ocean. • Tinv = (Tair_900 – Tair_surf) correlated with SST error Southern Ocean

Atmospheric profiles • High variability in Arctic associated with anomalous atmospheric profiles, not the case in the Southern Ocean • Both for negative and positive outliers Good (blue) : < -+0. 5*RMS Bad (red) : > -+2*RMS Tair_900 - Tair_surf

CCI product validation for high latitudes

High latitude validation results • L 2 and L 3 U products validated • Arctic > 60 deg N • Southern Ocean < 50 deg S • Limited data available for several sensors Number of Match-ups Spatial coverage, AVHRR-12 + 17

Overall results • • • Satellite – in situ Median and stddev Generally small biases Significant negative bias for AVHRR 18 Larger stddev in Arctic than Southern Ocean Timeline of match-ups

Solar zenith angle dependency • AATSR dependence upon water vapour and solar zenith angle • Cold summer bias for AVHRRs

TCWV dependency • AATSR dependence upon water vapour and solar zenith angle • Cold summer bias for AVHRRs

Summary/Challenges • SST products (AATSR, AVHRRs and AMSR-E) have in general larger errors in the Arctic, compared to Global and Southern Ocean performance • Biases generally depend upon Solar Zenith angle and TCWV • AVHRR biases found in operational products as well as ESA CCI products • Regional AVHRR coefficients can improve biases, largest improvements in daytime algorithms. • Limited in ”reference” situ observations in inner Arctic • Arctic NWP profiles with ”large SST errors” are more humid and warmer than profiles from ”low SST errors”

Future work • Continue the validation and error characterisation of Arctic SST products. • Write up the paper on CCI high latitude algorithms • Collect in situ data set for high Arctic validation • Look for alternative SST products: • Develop MW OE Sea Ice and SST processor for AMSR-E and AMSR 2 • Develop and validate Metop-A SST + IST product

High latitude DMI-ISAR deployments • • 7 week deployed at ODEN icebreaker Autonomous deployment July-November 2013 • • • Planned Activ circumpolar expedition (2014, 2015) RAL line, Denmark-Greenland