6g. Activity on GSICS UV Projects

6g. Activity on GSICS UV Projects
L. Flynn, NOAA With contributions from members of the NOAA and NASA Ozone Teams and Eumetsat GSICS UV Subgroup members The IEEE Geoscience and Remote Sensing Society/Standards Committee (GRSS/SC) is sponsoring a new standard: P Standards for Characterization and Calibration of UV-SWIR Hyperspectral Imaging Devices.

Disclaimer "The contents of this presentation are mine personally and do not necessarily reflect any position of the US Government or the National Oceanic and Atmospheric Administration."

Outline Satellite Measurement Comparison Approaches UV Projects
Best Practices and Key Data Sets for Calibration Reference Solar Spectra and Comparisons Comparisons of Effective Reflectivity and Aerosol Indices Comparisons of Initial Measurement Residuals for Profile Wavelengths

Match-Up Comparisons We would like to expand the use of matchup comparisons for UV instruments. Approaches include: Chasing Orbits (Opportunistic Formation Flying) S-NPP and EOS-Aura have 16-day repeat cycles but one makes 227 orbits and the other 233 so once every 16 days they are flying with orbital tracks within (360/14)*110/(14*16*2) ~ 6 km of each other, 15 minutes apart. For NOAA-19 and S-NPP, the matchups are every 12 days – (360/14)*110/(14*12*2) ~ 9 km. LEO vs LEO Simultaneous Nadir Overpass (and its non-simultaneous No-Local-Time-Difference zonal means) LEO underflights of GEO and L-1 instruments – Coincident Line-of-Sight Observations. (GOME-2 vs. SEVIRI, OMPS vs. TEMPO) Zonal or regional Means (including ascending/descending repeat coverage – S1*α1 = S2* α2)

Chasing Orbits for S-NPP and NOAA-19 POES:
adjusting STAR re-processed OMPS V8PRO to agree with SBUV/2 results

Simultaneous Nadir Overpass and No Local Time Difference Comparisons
Latitude range for zonal mean Latitude, Degrees North ____ GOME-2 Descending - - - OMPS Ascending Local Time, Hours

No Local Time Difference Comparisons, NOAA-17 SBUV/2 & NOAA-18 SBUV/2
May-August 2010, 69 N to 73 N, Daily Zonal Mean Total Ozone, DU NOAA-18 SBUV/2 --<>-- NOAA-17 SBUV/2 Daily Time Series

White Paper on Ground-based Characterisation
White Paper still in drafting stage – contributions and/or offers to author sub-sections welcome! Proposed table of contents Accuracy, sensitivity and repeatability Sources / commissioning Thermal and pressure environment / stability and characterization Instrument components Detector level Noise PRNU/PPG SMEAR Etaloning Stray-light Grating and alignment (ISRF) Spectral assignment Spectral stability Pointing and Spatial stability (ISRF/PSF) Spatial and spectral aliasing Radiometric and spectral scene in-homogeneity errors. Detector co-registration (overlap) Polarisation sensitivity Radiometric response Sources Geometry Diffuser characterisation Degradation and contamination ? Two Related Items: Agenda Item 10e on Friday: Planning a workshop The IEEE Geoscience and Remote Sensing Society/Standards Committee (GRSS/SC) is sponsoring a new standard: P Standards for Characterization and Calibration of UV-SWIR Hyperspectral Imaging Devices. David Allen of NIST is an organizer.

UV Solar Reference project
We now have solar measurements from GOME, MetOP(-A, -B & -C) GOME-2, OMI, SCIAMACHY, S-NPP & NOAA-20 OMPS, NOAA(-9, -16, 17, -18 & -19) SBUV/2, SSBUV & SSI with corresponding wavelength shift and solar activity patterns and bandpasses for most of them. We have requested SBUS solar spectra from CMA and will receive TropoMI spectra from ESA soon. We have reference solar spectra from SAO (including the newer spectrum but not the newest), KNMI and Woods Hole. We also have spectra and activity patterns from a solar model.

Solar Measurement Comparisons to KNMI Proxy
Relative differences between SSBUV8, N19 SBUV/2, SNPP OMPS NP and KNMI Proxy. Run code to find Mg II index and vertex wavelength 2%

SSBUV Differences with and without Solar activity adjustment

Mg II Index GOME-2 Metop-B and -A
Model in place of measurement

UV Reflectivity Channel Comparisons 335 nm to 400 nm
We are working to create time series of reflectivity and aerosol index statistics for a latitude/longitude box in the Pacific for the S-NPP & NOAA-20 OMPS and the Metop (-A, -B & -C) GOME-2. We are using the reprocessed S-NPP OMS SDRs and will coordinate this work with EUMETSAT to use the best GOME-2 Level 1. We will also work with NASA as they are looking at a “Dual Residual” method comparing radiance/irradiance ratios to forward model results to compare OMI and OMPS. This method is effectively a vicarious calibration method using TOMRad forward model results to create double differences between instruments for derived effective reflectivity and absorbing UV aerosol retrievals.

NASA DRCM Dual Residual Comparison Method:
To directly compare co-located sun-normalized radiance measurements between UVB Instruments using DRCM to account for differences in sensor characteristics and viewing conditions. (TOMRad Forward model and instrument bandpass and viewing conditons.)

From Proposal by Colin Seftor (SSAI) & Glen Jaross (NASA)

From study by C. Seftor and G. Jaross

Project Background Version 8 Total Ozone Algorithm Description Target area, cross-track segregation, weekly Reflectivity Results (1-percentile) Aerosol Index Results (average)

Background: Effective Reflectivity Project
The aim is to produce over-pass comparisons of UV/Vis sensors for specific target sites or regions in use by the community. As a first step, summaries of methods and results for target sites currently in use will be collected. We will compare measurements at reflectivity channels from 330 nm to 500 nm. Ice, desert and open ocean targets. Absolute Radiance/Irradiance check; Track variations over time. Reflectivity range/distribution, 1-percentile, Deep Convective Clouds (DCC) Wavelength Dependence – Aerosol Indices, Clean atmospheres Complications Viewing and Solar angle considerations Sun Glint Surface pressure Partially cloudy scenes Polarisation Inelastic Scattering Turbidity, chlorophyll Compare Global monthly surface reflectivity data bases Goals Agreement at 1% on cloud free scene reflectivity for 340 nm. Desert, Equatorial Pacific, Polar Ice. Agreement at 1% on aerosol index – wavelength dependence of reflectivity. Long-term reflectivity channels at 0.5% stability

Version 8 Total Ozone Algorithm
The algorithm makes two key assumptions about the nature of the BUV radiation. Firstly, we assume that the BUV radiances at wavelengths greater than 310 nm are primarily a function of total O3 amount, with only a weak dependence on O3 profile that can be accounted for using a set of standard profiles. Secondly, we assume that a relatively simple radiative transfer model that treats clouds, aerosols, and surfaces as Lambertian reflectors can account for most of the spectral dependence of BUV radiation, though corrections are required to handle special situations. The algorithm uses measurements at 12 channels to estimate the effective reflectivity and create absorbing aerosol and SO2 indices. A radiative transfer lookup table created by using TOMRad and standard ozone profiles is used to match the viewing conditions and an ozone absorbing channel measurement.

Pacific Box Statistics
The lines on the next slide show weekly 1-percentile effective reflectivity, total column ozone and aerosol index values (measurement residuals for wavelengths in the 360-nm range using effective reflectivity calculated for the 331-nm range) for the V8 algorithm for all the data in a latitude/ longitude box in the Equatorial Pacific versus cross-track view position. We expect the reflectivity minimum to be between 4% and 6% for open ozone, and we expect the aerosol index values to be approximately zero N-values for this region of the globe. The cross-track variations for positions around position #10 are related to sun glint effects. Consistent variations versus cross-track are due to calibration biases across the instrument CCD array.

Cross-Track Internal Consistency for OMPS
 Weekly Effective Reflectivity values for the V8 algorithm for March 2016 for all the data in a latitude/ longitude box in the Equatorial Pacific versus cross-track view position, 17 is nadir. We expect the 1-percentile values to be approximately 5% for this region of the globe. The cross-track variations for positions 8 to 15 are related to sun glint effects. Weekly Aerosol Index values for the V8  algorithm for March 2016 for all the data in a latitude/ longitude box in the Equatorial Pacific versus cross-track view position, 17 is nadir. We expect the aerosol index values to be approximately zero N-values for this region of the globe. The cross-track variations for positions 8 to 15 are related to sun glint effects.

Weekly Averages over the Target Area
Weekly values for February 2017 for OMPS V8 Total Ozone algorithm products fading to weekly values for February We need to use the reprocessed SDR data and redo these figures.

We are examining the V8 TOMS algorithm reflectivity and Aerosol Index Values for an Equatorial Pacific Region for OMPS, OMI and GOME-2 Time Series of GOME-2 Aerosol Index (360 nm vs 331 nm) Equatorial Pacific Time Series of  GOME-2 1-percentile Reflectivity Equatorial Pacific Jumps are from NOAA-applied soft calibration adjustments to the operational products. GOME_2 MetOP-B

Summary, Questions & Future
How stable are the values year-to-year? What factors produce the most instability? What should the 1-percentile values be? Can the method be used for absolute calibration? NASA uses the minimum reflectivities over land be used for sunglint FOVs? They screen for aerosols? We plan to generate and compare Equatorial Box time series for OMI, OMPS and GOME-2. Can we develop a V8TOz tool to allow similar computations for other instruments’ measurements? For the Comparisons of Effective Reflectivity and Aerosol Indices project, we have continued to use the Equatorial Pacific box as a region to generate our soft calibration adjustments. We are using minimum land reflectivities and comparisons to other products to check these adjustments especially for sun glint contaminated FOVs. We will be reprocessing all of the OMPS Nadir Mapper Version 8 Total Column Ozone products for the first five years and we will use the Pacific region to check the stability of the products. Talk 4.b in this meeting will cover these topics.

UV for wavelengths below 310 nm
We continue to monitor SBUV/2 and OMPS initial measurement residuals at and are working with NASA on research to reconcile the initial measurement residuals for SBUV(/2) and OMPS measurements to create a more homogeneous measurement data set for ozone profile retrievals.

By L. Flynn, Z. Zhang & C.T. Beck
Using Version 8 Ozone Profile Algorithm initial and final residuals to track calibration drift and estimate biases between instruments. By L. Flynn, Z. Zhang & C.T. Beck

Outline Project Overview SBUV/2 CDR Example
Drifting Orbit (SZA) Examples Operational Examples Radiance Comparison Ideas Contribution Function Equivalences

Initial Measurement Residual Project
The purpose of this project is to use initial measurement residuals from the Version 8 ozone profile retrieval algorithm to compare channels from 240 nm to 290 nm. (Note, this will require modification of the first guess creation to use consistent total ozone starting values as inputs.) Ascending/descending equivalent channel ideas will be used with hyperspectral measurements. Zonal mean and other matchup criteria will be used both to establish offsets and track relative drifts. Expand SBUV(/2) results to other sensors (OMPS, SBUS, OMI, GOME-2) Monitor time dependence for multiple instruments. Goals Agreement at 2% for Profile channels by using the Version 8 A Priori Profiles with TOMRad Tables and single scattering.

Outline of an Approach for Comparisons of radiance/irradiance ratios from 240 nm to 300 nm
Double Difference using Climatology: Compute the measurement residuals using a forward model with the effective scene reflectivity of the clouds and surface determined from longer channel measurements, and the ozone profile prescribed by the Version 8 a priori climatology. Use viewing geometries and bandpasses are as reported for each instrument. Compare residuals for channels λ1 and λ2 where S1*α1 = S2* α2, where S values give the path lengths and α values give the ozone absorption cross sections. That is, works with pairs of wavelengths where the measurement contribution functions are similar. Perform comparisons (statistical trade off in quantity of matchups vs. quality) Simultaneous nadir overpass matchups Zonal means (and No-local-time-difference zonal means) Opportunistic formation flying / Chasing orbits Benign geographic regions (e.g., Equatorial Pacific Box) Ascending/descending zonal means (In the Summer hemisphere, the same latitude is observed twice so one can obtain a set of internal comparisons.) Forward model and measurements V8 SBUV/2 forward model and A Priori as transfer for Viewing conditions Complications from real diurnal variations in the ozone profiles Complications if best ozone product values differ and initial residuals are used Measurement residuals’ correlation with scene reflectivity for longer wavelengths can disclose stray light contamination. I believe a similar method was used to check the GOME long-term calibration by assuming stable equatorial ozone profiles.

Long-term Inter-calibrated Initial Measurement Residuals for SBUV/2

Inter-calibrated Initial Residuals for SBUV/2 with SZA Drift
The figures show the initial measurement residuals for three profile wavelengths (Top 288 nm, Middle 292 nm, and Bottom 298 nm) for the V8PRO product for the equatorial daily zonal means (20N to 20S). The two sets of data are for the NOAA-16 SBUV/2 and the NOAA-17 SBUV/2. The units are N-values (~2.3%). The Version 8 algorithm a priori ozone profiles and forward model have been used to allow direct comparison of the radiance/irradiance ratios for the two instruments. NOAA-16 was an afternoon satellite and NOAA-17 was a morning satellite during this period. By the end of the record, the NOAA-16 satellite was in a late afternoon orbit.

Operational Initial Residuals for SBUV/2 with SZA Drift and Calibration Adjustments

Operational Initial Residuals for SBUV/2 & OMPS with Operational Adjustments

Matching orbit on 3/20/2013 for S-NPP OMPS and NOAA-19 SBUV/2

V8Pro Initial Residuals along Chasing Orbit
Red and Black OMPS (Before and After), Green SBUV/2. Jumps at 30N/S and 60 N/S where climatologies switch latitude bins. Adjustments: ; 274 0.3; ; ; ; 298 1.1; ; ; 3% 254 nm 3% 3% 274 nm 283 nm 288 nm 3% 3% 292 nm 3% 298 nm Changes at 30N/S and 60N/S are changes in profile climatologies. The adjustments for OMPS V8PRO 1.0 0 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1 1.0 1.0 1 1.0 1 0.451 Statistics NOAA19 Vs OMPS V8PRO over Pacific box after adjustment the average NOAA19 reflectivity is: the average OMPS reflectivity is: the average NOAA19 stp1oz is: the average OMPS stp1oz is: the average NOAA19 stp2oz is: the average OMPS stp2oz is: the average NOAA19 aerosol is: the average OMPS aerosol is: the average NOAA19 stp3oz(bsttoz) is: the average OMPS stp3oz(bsttoz) is: 3% 3% 312 nm 3% 306 nm 302 nm Changes at 30N/S and 60N/S are changes in profile climatologies.

V8Pro Layer Ozone, Bottom to Top
Bottom to top, Black OMPS (After Adjustment) and Green SBUV/2. Jumps at 60S and 60N are changes in profile climatologies.

Pseudo-Channels in the UV from 250 nm to 300 nm
As the SZA or SVA increases, the contribution functions shift up. One can find combinations (linear?) of radiances for longer channels that can represent (capture the response to ozone changes) a measurement at a shorter channel at SZA=0 and SVA=0. (MW Pseudo-Channel Ideas) We can compare instruments measuring at different viewing geometries or times of day. This can help to determine both internal and external biases. Diurnal ozone variations will present an involved complication. Changing channel emphasis can introduce wavelength-dependent biases.

Range of absorption /atm. -cm (DU = milli-atm
Range of absorption /atm.-cm (DU = milli-atm.-cm, 300 DU is an average total column.) Rayleigh scattering like 1/lambda^4, Optical depth is 1.4 for the full atmosphere RS at 290 nm. Other absorption about 600 nm, see limb talk by J. Larsen. Power law for ozone and drop in absorption. Stops at peak in stratosphere. Good for us, but not for measurements. For long wavelengths, ozone layer and atmosphere. Start off simple, will get to problems later. Looking ahead at retrievals: Pairs of absorbing and non-absorbing channels. DOAS uses small scale relative variations. Atmospheric ozone provides a shield from solar ultraviolet (UV) irradiance by absorbing almost all of the UV irradiance below 290 nm and by absorbing most of the irradiance from 290 to 310 nm. The UV absorption properties are expressed in terms of cross sections which relate the decrease in transmitted flux to the amount of ozone in an atmospheric layer. Figure 1 gives a plot of the ozone absorption as a function of wavelength. (The absorption also depends on the air temperature and a representative value was used to create this graph.) Sunlight is also scattered, both toward the surface and back toward space, by the atmosphere. The most important type of scattering for UV wavelengths is Rayleigh scattering. Differential scattering of visible wavelengths of sunlight is responsible for the blue sky.

273 nm 253 nm 318 nm 340 nm Figure 6.a. Normalized Single Scattering Contribution Functions for 12 wavelengths at [253,273,283,288,292,297,302,306,313,318,331,340] nm for a 325 DU total column ozone profile for Solar Zenith Angle θ0 = 30°.

273 nm 318 nm 340 nm Figure 6.b. Normalized Single Scattering Contribution Functions for 12 wavelengths at [253,273,283,288,292,297,302,306,313,318,331,340] nm for a 325 DU total column ozone profile for Solar Zenith Angle θ0 = 70°.

Comparison Considerations
Different spectral and spatial resolution Forward models can remove these dependencies Chasing orbits If orbital periods are slightly off, then beat frequency matchups are better. SNO for AM with PM (+product comparisons?) No-local-time difference zonal means Asc/Desc Langley –> S1*alpha1 = S2*alpha2

S1*α1 = S2* α2, Si = 1 + sec(SZAi) for nadir viewing
252 nm 273 nm 306 nm 302 nm Figure 2. Normalized weighting functions for SBUV/2 measurements for a standard 325 DU (0.325 atm.-cm.) ozone profile for two SZAs (Eight wavelengths from 252, 273, 283, 288, 292, 298, 302 to 306 nm) The normalized single-scattering weighting functions, which show how sensitive the measurements are to ozone changes as functions of pressure, are given for eight wavelengths (252, 273, 283, 288, 292, 298, 302, and 306 nm) in Figure 2. The two sets of curves from top to bottom are for the wavelengths from short to long. This shows, for example, that the measurement by a space-based instrument looking down on a sunlit atmosphere of the radiance at 252 nm is primarily affected by changes in the ozone above 2.0 mbar while the measurement at 306 nm is affected by changes in ozone throughout the stratosphere. Ozone profile retrieval algorithms make use of this physical difference in the spectral measurements to obtain estimates of vertical ozone profiles. The information at wavelengths measured by a BUV instrument shifts upward as the Solar Zenith Angle (SZA) or the Satellite Viewing Angle (SVA) increases and the path length of photons through the atmosphere increases. For the example in Figure 2, at 70 SZA, the 273 nm measurements responds to changes in the ozone vertical profile in almost the same manner as the 252 nm measurements do at 30 SZA. Notice that the second highest dotted curve and the highest dashed curve almost coincide. The eight wavelengths in Figure 2 are the standard profiling wavelengths for normal SBUV/2 observations. They provide good coverage of the stratosphere. The gradual decrease in the weighting function (the lack of sharp cutoffs) limits the resolution of the vertical profile information. Even these eight wavelengths have some redundancy in their information and it remains to be seen how much new information will be obtained by denser spectral coverage. Hyperspectral algorithms for profile retrievals are under development. The duplication of measurement information both spectral and spatial, should help to provide internal consistency and calibration checks. It will also help with methods for radiance comparisons between measurements for Cal/Val work. Limb viewing produces a geometric magnification of layer response compared to nadir viewing, but with at the cost of a large increase in the optical path. S1*α1 = S2* α2, Si = 1 + sec(SZAi) for nadir viewing

Summary, Questions and Future
Initial measurement residuals can identify calibration biases between instruments. Provide tools to create initial residuals for other instruments given channels and bandpasses. Expand and formalize matchup techniques. Reprocess and Homogenize NOAA-16 through NOAA-19 SBUV/2 and OMPS NP for a post-2000 Ozone Profile CDR. Create invariant channel combinations under SZA & SVA changes. For the Comparisons of Initial Measurement Residuals for Ozone Profile Channels project, we have used this method to generate soft calibration adjustments to remove measurement bias between the NOAA-19 SBUV/2 and S-NPP OMPS NP. We will reprocess the first five years of OMPS NP measurements and compare the two records. Talk 4.a in this meeting will cover some of these results.

Background

Solar Spectra Project The purpose of the is project is to compare solar measurements from BUV (Backscatter Ultraviolet) instruments. The first step is to catalog high spectral resolution solar reference spectra and agree on a common one to use for the project. For each instrument, participants should provide the following datasets: Solar measurement for some date (wavelength scale, irradiance) Wavelength scale and bandpass (Δλ, # of points, bandpass centers, normalized bandpass weights) Synthetic spectrum from common reference (wavelength scale, irradiance) Synthetic for wavelength scale perturbations (±0.01 nm) from common reference (wavelength scale, irradiance) Synthetic from alternative reference spectra (wavelength scale, irradiance) Solar activity pattern (wavelength, relative change) Mg II index (if 280 nm is covered) Mg II Mg I (date, index) Ca H/K index (if 391 nm to 399 nm is covered) CA II and 396.8. Goals: Agreement at 1% on solar spectra relative to bandpass-convolved high resolution spectra as a transfer after identifying wavelength shifts and accounting for solar activity Long-term solar spectra drift and instrument degradation by using OMI solar activity pattern (with internal confirmation from Mg II Indices and scale factors)

Solar UV Measurement Project
High resolution solar reference spectra Reference high resolution solar Spectra (SOLSTICE, SIM, Kitt Peak, etc. – Everybody has a favorite. How do they compare?) Mg II Index time series, Scale factors at high resolution Instrument data bases Bandpasses, wavelength scales (Shift & Squeeze codes) Day 1 solar, time series with error bars (new OMI product) (Formats, Doppler shifts, 1 AU adjustments) Mg II Indices and scale factors at instrument resolution Reference calibration and validation papers Using the information from above we can compare spectra from different instruments and times. Composite Mg II solar activity index for solar cycles 21 and 22 Matthew T. DeLand, Richard P. Cebula DOI: /93JD00421 Creation of a composite solar ultraviolet irradiance data set Matthew T. DeLand, Richard P. Cebula DOI: /2008JA013401

Solar Reference Spectra & Their Use
Matthew DeLand Science Systems and Applications, Inc. (SSAI) Lanham, Maryland USA Slides of this presentation GRWG Meeting – UV Subgroup Darmstadt, GERMANY 27 March 2014

Introduction Solar irradiance is primary energy source for terrestrial system. Important to know absolute value and relative variations, for both integrated measurement and spectral dependence. Sun represents a valuable calibration source for satellite instruments: Easy access High intensity Relatively stable compared to on-board sources Many spectral features for wavelength calibration Some “features” also represent challenges for operational use: Exposure-dependent degradation (particularly in UV) Large dynamic range needed to co-exist with Earth view data Natural variability in UV region Complex spectral structure This presentation will briefly discuss some issues associated with using reference solar spectra.

Reference Spectra - Comments
No single instrument measures all spectral regions of interest (X-ray, UV, visible, IR) simultaneously. Thus, any “reference” spectrum is likely to be a composite of measurements from multiple instruments, using different bandpasses, probably taken at different times. Some reference spectra are constructed using measurements from one data set, but adjusted radiometrically through comparisons with another data set. Differences in spectral resolution are particularly important in the UV, where many Fraunhofer lines occur. The definition of “high” resolution can depend on the requirements of the user. Frequent sampling in the data set may not equal high spectral resolution. Check the background story before using the data.

Reference Spectra KNMI [Dobber et al., 2008] Covers 250-550 nm.
Hall and Anderson [1991] for nm, Kurucz et al. [1984] for nm. Original data have very high resolution (Δλ = 0.025, nm). Convolve with nm bandpass, sample at 0.01 nm. Adjust radiometric calibration to match lower resolution reference spectrum (average of UARS SUSIM data). SAO [Chance and Kurucz, 2010] Covers nm. Hall and Anderson [1991] for nm, Kurucz et al. [1984] for nm. Convolve with 0.04 nm bandpass, sample at 0.01 nm. Adjust radiometric calibration to match lower resolution reference spectrum (Thuillier et al.). ATLAS-1, ATLAS-3 [Thuillier et al., 2004] Covers nm. Rocket data (1994) for nm; UARS SUSIM and SOLSTICE for nm; UARS SUSIM and SOLSTICE, ATLAS SUSIM, SSBUV, SOLSPEC for nm; SOLSPEC for nm; SOSP for nm. UV: Smooth to 0.25 nm resolution, sample at 0.05 nm; Visible: resolution = 0.5 nm, sampling = nm. Average individual data sets together where available. Each reference spectrum represents single date. WHI (Whole Heliospheric Interval) 2008 [Woods et al., 2009] Covers nm. Rocket EVE + TIMED SEE for nm; SORCE SOLSTICE for nm; SORCE SIM for nm. Resolution = 0.1 nm in UV, ~ nm from near-UV to near-IR. Sampling = 0.1 nm throughout. Each reference spectrum represents narrow range of dates (~1 week). Reference spectra scaled to TIM TSI values.

Resolution and Sampling
Same spectral region ( nm) and absolute scale used for all panels in this figure. Effect of bandpass change between KNMI and SAO is apparent, even with same input data set. Satellite measurements (bottom two panels) have lower resolution. WHI spectrum shows change in original instrument resolution when SIM data begin at 310 nm.

Solar Activity Solar irradiance shortward of 300 nm varies on short-term (~27-day) and long-term (~11-year) time scales. Longward of 300 nm, only Fraunhofer lines show variations exceeding ~0.5% (magnitude of variability is resolution- dependent). Spectral dependence of solar variations appears to be very consistent for both short and long time scales. Therefore, we can use solar UV activity proxy (e.g. Mg II index) and wavelength-dependent “scaling factors” to estimate irradiance change with time.

Solar Cycle Variations
Spectral dependence shows distinct features due to atomic species in Sun. Variations generally less than 1% beyond ~300 nm. Good agreement between data sets from two solar cycles, semi-empirical model. DeLand and Cebula [2012]

Solar Variations – UV, Visible
Better OMI instrument resolution (0.4 nm) shows small features at Fraunhofer lines. Generally very good agreement between long- term changes (red) and short-term changes (black). Adapted from Marchenko and DeLand [submitted to Ap. J., 2014]

Scaling Factors Calculate max/min ratio for irradiance over solar rotation to minimize sensitivity change effects. Calculate ratio for same dates with Mg II index. Slope of linear regression fit gives “scaling factor”. Good dynamic range is preferred. Value of scaling factor at any wavelength depends on instrument resolution for both irradiance and proxy. Slope = 1.07(±0.04)

Conclusions Many high resolution reference solar spectra have been constructed for different purposes. Understanding how a reference spectrum was created helps to make it more useful. At wavelengths shorter than 300 nm, natural solar variations should also be considered for best results. Mg II proxy index can be used with (instrument-dependent) scaling factors to determine adjustments as a function of wavelength and time.

Analysis of Time Series of Solar Spectra
The OMI, GOME-2 and OMPS teams have generated models of their time series of solar measurements by using Solar activity With proxies (e.g., Mg II Indices) Directly estimating pattern over solar rotations Wavelength shifts With proxies (e.g., optical bench temperatures) Directly from fits of solar features Diffuser and instrument degradation With proxies (e.g., diffuser exposure times) From working and reference diffuser measurements From residual changes after identifying activity and wavelength changes Considering albedo changes over targets or compared to other sensors Many UV solar spectra can be modeled well with three terms Solar activity (Mg II Index with Scale factors) Wavelength shifts (wavelength shift patterns) Diffuser and instrument degradation Linear Regression or EOF analysis can help to identify these patterns.

Trending of OMPS measured solar data
Reference Diffuser Every six months Working Diffuser Every two weeks Nadir Mapper Nadir Profiler EUMETSAT September 2014

OMPS Nadir Profiler Solar Measurements
11 November 2018 Newest Degradation Component The working diffuser’s exposure is 13 times the reference exposure. Two years of measurements Compared to their average. 250 nm nm 250 nm nm Wavelength Shift Component Solar Activity Component The working diffuser for the OMPS has 13 times the exposure of the reference diffuser. Patterns are Mg II scale factors and track Solar activity. Wavelength shifts track optical bench annual thermal variations. Oldest 250 nm nm 250 nm nm

By L. Flynn, Z. Zhang, E. Beach, Y. Pachepsky
Vicarious calibration by using statistical properties for ozone, reflectivity and aerosol index products in a latitude/longitude box over the equatorial Pacific By L. Flynn, Z. Zhang, E. Beach, Y. Pachepsky

Adjustments using A, K, and Dy
The Averaging Kernel, A, is the product of the Jacobian of partial derivatives of the measurements with respect to the ozone profile layers, K, and the measurement retrieval contribution function, Dy: A = Dy # K For a linear problem, the retrieved profile, Xr, is the sum of the A Priori Profile, Xa, plus the product of the Averaging Kernel, A, times the difference between the Truth Profile, Xt, and Xa: Xr = Xa + A # [Xt – Xa] The measurement change, ΔM, is the Jacobian times a profile change, ΔX: ΔM = K # ΔX The retrieval change, ΔXr, is the contribution function times a measurement change, ΔM: ΔXr = Dy # ΔM Dy = SaKaT [KaSaKaT + SM]-1

+ ΔXr * Dy # ΔM Dy, measurement contribution function; deltaM, difference in annual tropical zonal mean initial measurement residuals. Comparison of actual differences in annual tropical zonal mean profiles retrieved by NOAA-16 and NOAA-17 SBUV/2 for 2003 with those predicted by the differences in their initial residuals. The “+” symbols are ΔXr computed directly and the * symbols are Dy ΔM.

On-ground calibration GOME-2 / Metop
Accuracy, sensitivity and repeatbility GOME-2 FM2 Metop-B Example: Radiometric response of the solar (irradiance) measurement port FM2: Campaigns 2004 / 2011 2004 2012 (Feb) 2012 (June) s1% 2004 2012 w.r.t (Feb)

GSICS UV Subgroup Report
Rose Munro, Larry Flynn

Selected GRWG-UV Subgroup Baseline Projects
Reference Solar Spectrum Aim: to evaluate the available reference solar spectra and make a recommendation for a reference solar spectrum for community use. Lead – Larry Flynn (NOAA) White Paper on Ground-based Characterisation of UV/Vis/NIR/SWIR spectrometers Aim: to prepare a white paper documenting best-practise for the on-ground calibration of UV/Vis/NIR/SWIR spectrometers based on in-orbit experience from relevant missions. Lead – Rosemary Munro (EUMETSAT) (transferred from R. Lang) Match-ups and Target Sites Aim: to produce over-pass comparisons of UV sensors for specific target sites in use by the community. As a first step summaries of methods and results for target sites currently in use will be collected. Lead – TBC. Cross-calibration below 300nm Aim: To devise new methods for comparison of wavelength pairs for different viewing geometries taking into account contribution function equivalence to allow radiometric performance comparisons for ozone profile wavelengths from 240 – 200 nm. Lead Larry Flynn (NOAA).

Reference Solar Spectrum – Status
Compare solar measurements from BUV (Backscatter Ultraviolet) instruments. Goals Agreement at 1% on solar spectra relative to bandpass-convolved high resolution spectra as a transfer after identifying wavelength shifts and accounting for solar activity. Long-term solar spectra drift and instrument degradation can also be analysed. Collaborative work has started well with participation growing to include more instruments and solar modellers. See talks in UV Subgroup session at this meeting: 6c - FY-3/TOU inter-calibration with GOME-2 and OMPS for solar diffuser correction (H. Wang, NSSC/CAS) 6f - Comparison among Reference Solar Spectra using TROPOMI Solar Measurements (M. Kang, Ewha Women's University) 6g - NOAA Update on Three GSICS UV Projects: Solar, Reflectivity and Residuals (L. Flynn, NOAA) Larry Flynn (Project Lead) will focus on a model to explain the OMPS Nadir Profiler solar measurements and provide an initial comparison to a synthetic spectrum – these will both be used to support inter-comparisons with other solar measurements.

Match-ups and Target Sites
Produce over-pass comparisons of UV/Vis sensors for specific target sites in use by the community. Goals Agreement at 1% on cloud free scene reflectance for 340 nm. Desert, Equatorial Pacific, Polar Ice. Agreement at 1% on aerosol index – wavelength dependence of reflectance. Long-term stability of 0.5% in reflectance channels Work being carried out by L. Flynn and colleagues, NOAA, ... Focussing on comparisons of Effective Reflectivity and Aerosol Indices in an Equatorial Pacific box as a region for generation of soft calibration adjustments. Using minimum land reflectivities and comparisons to other products to check these adjustments especially for sun glint contaminated FOVs. Reprocessing all of the OMPS Nadir Mapper Version 8 Total Column Ozone products for the first five years and will use the Pacific region to check the stability of the products. See talk: 6g - NOAA Update on Three GSICS UV Projects: Solar, Reflectivity and Residuals (L. Flynn, NOAA)

Cross-Calibration below 300nm
Methods Double Difference using Climatology: Perform comparisons (statistical trade off in quantity of matchups vs. quality): Forward model and measurements: Ongoing Activities Comparisons of Initial Measurement Residuals for Ozone Profile Channels This method will be used to generate soft calibration adjustments to remove measurement bias between the NOAA-19 SBUV/2 and S-NPP OMPS NP. The first five years of OMPS NP measurements will be reprocessed and compare the two records. See talk: 6g - NOAA Update on Three GSICS UV Projects: Solar, Reflectivity and Residuals (L. Flynn, NOAA)

6g. Activity on GSICS UV Projects

Similar presentations

Presentation on theme: "6g. Activity on GSICS UV Projects"— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

6g. Activity on GSICS UV Projects

Similar presentations

Presentation on theme: "6g. Activity on GSICS UV Projects"— Presentation transcript:

Similar presentations

About project

Feedback