GSICS VIS/NIR subgroup report David Doelling (sub-lead-facilitator), and all other VIS/NIR members GSICS Annual Meeting, Darmstadt, Germany, March 24-28, 2014

Outline An example of how calibration impacts retrievals Calibration accuracy through time Reference sensor and traceability to future instruments and methods Comparison of visible calibration techniques Coordination of GPRC for common short and long term goals Proposed forward progress This years goals

CERES fluxes MODIS C5 DCC stability CERES measures the Earth’s broadband shortwave and longwave fluxes MODIS cloud properties aid in converting measured CERES radiances to fluxes, since the SW BRDF is scene based, absolute calibration is tied to CERES instrument The current CERES Edition 3 products uses a combination of MODIS C4 and C5 radiances CERES users try to correlate cloud/aerosol/land retrievals with CERES fluxes to evaluate climate change MODIS C5 DCC stability Wu et al. 2013

The effect of MODIS band 1 calibration on retrieved optical depth Compute global monthly mean optical depth from MODIS pixel level retrievals Deseasonalize to determine trend over time A 2.5% MODIS calibration drift has resulted in a ~8% decrease in cloud optical depth Which may be falsely interpreted as climate change Collection 6 stable to 1%/decade GSICS calibration goal is to improve and harmonize cloud/aerosol/land retrievals +0.4 +0.4 Terra-MODIS C5 Aqua-MODIS C5 +0.0 +0.0 http://ceres.larc.nasa.gov/order_data.php -0.4 2000 2005 2012 -0.4 2002 2007 2012

VIS/NIR Calibration Milestones 1980 ISCCP 5% calibration accuracy “The Earth is more stable than your sensor” Bill Rossow 1995 to 2000 First solar diffusers for VIS bands ERS2-ATSR2, TRMM-VIRS, Terra-MODIS 2006 MODIS band 1 C5 calibration 2%/decade trend 2010 MODIS band 1 C6 calibration 1%/decade trend 2014 Predicted VIIRS stability Desert Calibration stability 1%/decade DCC calibration stability 0.5%/decade (Bhatt et al. 2014) Predicted anthropogenic SW forcing, 0.3%/decade 2025 CLARREO/TRUTHS absolute calibration accuracy 0.3% 2K Solar Cycle amplitude 0.1%/11years

Reference Calibration Reference Instrument or reference calibration tied to radiative transfer (RT) predicted radiances? How traceable are the radiative transfer (RT) radiances Dependent on RT method and inputs, how traceable in time are they? The inputs were based on MODIS cloud properties, which are tied then to the MODIS calibration Either way we agree on reference instrument for traceability Okuyama. GSICS 2009 web meeting RT based Reference Instrument based

VIS/NIR reference instrument Ideal traceable on orbit calibration hyper-spectral instrument such as the future CLARREO/TRUTHS If the calibration is traceable on orbit you need no overlap between successive instruments You can monitor change with an equivalent instrument years in the future Now given todays instruments The data must be freely available, user friendly, and widely used, very familiar with the community (retrievals well documented) Must be a well characterized sensor and stability referenced through a solar diffuser and/or moon Navigation, instrument radiometric noise, polarization knowledge Must be traceable (inter-calibrated) with future instruments

Traceable inter-calibration if the VIS reference instruments can be faithfully inter-calibrated, then the absolute calibration of a future instrument can be reference back in time Eventually future instruments will have excellent absolute and SI traceable calibration Factors that increase the confidence of inter-calibration Overlapping successive or follow on operational instruments, such as NPP/JPSS-VIIRS, Metop-SG-MetImage Length of overlap period Similar spectral channels, known SRFs Similar (equator crossing time) well maintained (non drifting LEO) orbits, to observe entire dynamic range over all earth targets

Reference Instrument The reference instrument absolute calibration is secondary as long as the calibration is very stable and well characterized However it is important that the absolute calibration can be traced to the reference instrument for all contemporary operational visible calibration methods This allows the comparison of calibration methods This allows uniform cloud/aerosol/land retrievals over multiple platforms both spatially and temporally This allows future calibration enhancements to improve the entire record GSICS goal is to verify the operational instrument stability so that the reference instrument calibration can be transferred effectively

Visible Calibration Methods No traceable reference VIS hyper-spectral instrument is available, therefor a multiple calibration approach is taken Agreement among methods validates the resulting calibration Each method has its advantages and disadvantages Hewison, GSICS annual meeting 2009

VIS calibration technique comparison (not comprehensive) GEO Method Stability at TOA SBAF Traceable to reference Sampling Historical application Lunar •Geological variability • Earth view optics? ROLO model at MODIS/VIIRS phase angles •Monthly •Long phase and libration period Dependent on sampling DCC •0.5%/decade (VIIRS) • tropopause Nearly flat spectrally •DCC raymatch •Domain mean Large sample collective method Reference calibration difficult Rayliegh RT depends Wind speed Chlorophyll RT at surface, good for interband Reference/RT compare Use coincident MODIS .8µm Need 0.8µm reflectivity Desert 1.0%/decade (VIIRS) • surface Need observed hyper-spectral SADE with atmosphere Not available over all GEO domains Long term invariance not known Liquid Water Clouds Direct compare RT model MODIS cloud properties Dependent on reference orbit Need coincident reference obs Ray-match Needed for clear-sky and low clouds MODIS radiances

GSICS GEO VIS ATBDs ATBD is the first step to implement VIS method within GSICS VIS ATBDs in GSICS library, effort started in 2011 Liquid Water Clouds (JMA) – Arata Okuyama Liquid Water Clouds (SNU) B.J. Sohn Lunar (NOAA) – Fred Wu Stellar (NOAA) – Fred Wu DCC (NASA) – David Doelling DCC (SNU) – B.J. Sohn (MODIS/GEO) Ray-matching (NASA) – David Doelling Combining Methods (EUMETSAT) – Tim Hewison VIS ATBDs TBD Rayleigh (CNES) draft status? - Bertrand Fougnie Desert (CNES) – Patrice Henry Desert (EUMETSAT) – Sebastien Wagner Sunglint (NOAA) Andy Heidinger

Implementation of VIS methods Fred Wu, GSICS web meeting 2009 Common VIS method goals Common effort Common product (requires more GPRC commitment) Common algorithm (creates more uniformity) Common VIS calibration product goals Quantify the bias Correct the bias (benefits the retrieval community) Diagnose the bias Common VIS record goals Common algorithm during reference sensor record Common algorithm for reference sensor and historical record Common GPRC goals Best to coordinate various GPRC goals and commitments

VIS calibration method product timeline Calibration method timeline Year 1: Agree on calibration approach from existing methods or a combined approach Year 2: GPRC’s implement common approach – have verification dataset to ensure uniform implementation Year 3: GPRC’s deliver submission/demonstration phase products to GSICS – formatted, documented, peer reviewed, user tested Year 4: GPRC’s deliver pre-operational/operation phase products, version control, made public, accepted Can method implementations be staggered to accelerate products? Fangfang Yu, GPPA workflow 2013 2012 2013 2014 2015 2016 2017 2018 DCC Lunar Rayleigh?

VIS calibration product timeline Do we spend time to modify method for historical instruments? Do we spend time to modify method for LEO instruments? How will these efforts be redirected when the next generation GEOs are launched in a few years? These will have their calibration stability tied to solar views The methodology will be straight forward once either CLARREO or TRUTHS hyper-spectral radiances are available for calibration Probably at least 10 years from now

Goals/Discussion for this year DCC Agree on methodology (Version 1) Product, bias monitoring and path to demonstration product Determine DCC reference calibration tied to MODIS and SBAF Lunar Agree on methodology that applies uniformly to all Reference ROLO with MODIS at MODIS phase angles Improve ROLO with Plaiedes? Rayleigh Requires 0.86µm channel, use coincident MODIS 0.86µm? Prepare for next generation GEO Explore next technique Desert (IVOS), liquid water, combining methods