Detection and Characterization of Polar Mesospheric Clouds with OMI Matthew DeLand Science Systems and Applications, Inc. (SSAI) International OMI Science Team Meeting June 7, 2007

Introduction How are mesospheric clouds different? Only observed in summer (June-August in the Northern Hemisphere, December-February in the Southern Hemisphere). Mostly seen at high latitudes (> 50º). Very high altitude (80-85 km). Composed of water ice crystals (not vapor). I will use polar mesospheric cloud (PMC) in general and for satellite observations, noctilucent cloud (NLC) for ground-based observations. Noctilucent is historic term, but PMC is more generally descriptive.

Structure of the Atmosphere Layers are defined by temperature minima and maxima. Troposphere = normal weather. Stratosphere = ozone layer. PMCs occur at the coldest place on Earth (120 K has been measured!). Polar summer mesopause Cold temperatures at summer mesopause caused by dynamics (gravity wave breaking  upwelling, adiabatic cooling). More?

Basic PMC Information Typical particles (water ice) are 20-50 nm in size. Originate from meteor “smoke”? Cannot be observed visually during daytime (optical depth < 10-4). Occurrence frequency and brightness have seasonal cycle, increase in magnitude towards higher latitude. Physical structure observed on scales from few km to hundreds of km. Short-term variations (minutes) show effects of atmospheric waves.

Why Study PMCs? PMC formation and brightness are very sensitive to mesospheric temperature and water vapor content. Increased carbon dioxide (CO2) in lower atmosphere  warmer troposphere  colder mesosphere  more and brighter PMCs. Increased methane (CH4) in troposphere  more H2O in stratosphere  transport to mesosphere  more and brighter PMCs. Thus, long-term changes in PMC occurrence frequency and brightness may indicate climate change. PMC occurrence frequency and brightness are strongly influenced by solar activity, but mechanism is not fully understood. Plus, they’re really cool!! - Observational data in this region (polar latitude, mesospheric altitude, summer season) are not sufficient to identify underlying trends in either temperature or water vapor.

NLC Visual Observation Shortly after sunset (or before sunrise), the observer is in darkness, but the NLC is still in sunlight (“noctilucent” = night shining). NLC over Finland, photo by Pekka Parviainen Observation method does not work at latitudes > 65 degrees because Sun doesn’t get far enough below horizon. NLC sightings are very rare below 50 degrees latitude. Very few NLC observations in Southern Hemisphere (no land at appropriate latitudes).

Satellite Measurements All instruments prior to AIM not designed with PMC measurements in mind  some compromises in data sampling or quality are inevitable. Various observation techniques are represented here: limb scattering (UV and visible), occultation (visible and IR), nadir backscattering (UV). First regular measurements and climatology derived from SME data in early 1980s. SNOE used same technique in 1998-2004. SBUV technique developed by Gary Thomas and Rich McPeters in 1991 with Nimbus-7, later extended to other SBUV/2 instruments. Occultation measurements (SAGE II, HALOE) have long records, but limited sampling and variable latitude coverage. New limb scattering data from OSIRIS, SCIAMACHY provide expanded spectral coverage.

SBUV/2 Detection Method Useful in UV (λ < 300 nm) where Earth albedo is (relatively) low. Identify PMC using enhancement above background, spectral dependence of 5 shortest wavelengths. Uncertainty in derived background due to ozone variability limits PMC detections to brightest 15% of overall population. NOAA-16 SBUV/2 data Original technique developed for Nimbus-7 SBUV in 1991. SBUV, SBUV/2, SCIAMACHY; OMI, CIPS/AIM. “low” albedo = few * 1.0e-04. Figure shows albedo data at 252 nm, progression of measurement angles and local time for nominal orbit. Asterisk = screened data point. DeLand et al. [2003]

Typical SBUV/2 PMC Results Typical day of measurements. Each box represents 32-second scan. Orbits are separated by ~25 deg. longitude. PMC detections shown by filled boxes, with color scale indicating brightness represented by 252 nm albedo. Orbital geometry allows some locations (~60, ~70 degrees N in this case) to have two observations ~10-14 hours apart. Repeat cycle of orbit for specific location is ~7 days.

Seasonal Variations Plot shows daily + smoothed occurrence frequency for single season in different latitude bands. Detections begin 10-15 days before solstice, end ~60 days after solstice. Hemispheric average reaches ~15% at peak, but significant latitude dependence can be seen. Limb viewing instruments see much higher frequency values with dark background. Daily variability, intermediate term variations are visible, particularly at high latitudes. Long time series with many seasons shows differences in shape, amplitude, timing of seasonal variations. How to identify possible secular and/or periodic variations? Try to use quantitative measure (brightness).

DeLand et al. [2007, J. Geophys. Res.] Long-Term Variations Combine albedo data from multiple SBUV instruments to examine secular and periodic behavior. Adjust data for local time effects. Results show increasing brightness with time at all latitudes, both hemispheres. Multiple regression fits also find anti-correlation with solar activity (0-1 year phase lag), stronger response at higher latitudes. SBUV data from multiple instruments can be combined and examined in latitude bands. This figure includes local time adjustment. Anti-correlation with solar activity (stronger at higher latitudes). PMC response lags solar activity by 0-1 years (consistent with HALOE). Increasing brightness with time (smallest at lower latitudes; approximately same magnitude in both hemispheres; better results with local time adjustment). DeLand et al. [2007, J. Geophys. Res.]

OMI Advantages for PMC Detection Smaller pixels (13x48 km2 in UV1 data) increase chance of being filled with PMC  more contrast to background. Cross-track observations can show horizontal structure over large geographic regions. First regular measurements up to 90º latitude. Overlapping measurements at high latitude allow study of short-term (orbit to orbit) variability, planetary waves. Spectral information with each pixel holds potential to address particle size questions. Many more opportunities for coincidence analysis with ground-based data.

Basic Detection Algorithm Adapt SBUV/2 approach. Use 265 nm as shortest wavelength. Initial tests use all pixels, 5 wavelengths (same number as SBUV/2), sample day of Level 1B data. Process each swath independently due to cross-track albedo variations. Raw albedo data show clear structure (nadir swath shown). PMC brightness is higher than SBUV/2. Shortest available wavelength varies between swaths, but 265 nm is always present. Spline each spectrum to 0.5 nm grid for consistent processing. Data represents one orbit, but background derived from full day of measurements. Latitude-dependent noise threshold (one element of detection tests) not yet “tuned” to OMI data  low latitude detections not necessarily accurate. Absolute brightness needs appropriate spatial binning for quantitative comparison with SBUV/2 results, but basic magnitude is good. Along-track sample density is 15*SBUV/2  structures with scale < 100 km can be seen clearly.

OMI Results: Single Orbit Combine data from single swath and all orbits for entire day to calculate background (like SBUV/2). Latitude coverage is 40-90 degrees. Results shown are for all swaths on single orbit. Continuity between adjacent swaths is reasonably good. Attempted to show variation in pixel size, but algorithm not fully working. Brightness increase for outer swaths is more than expected from Mie phase function variation at backscattering angles. Number of PMC detections from one orbit comparable to full season from SBUV/2 instrument!

OMI Results: Multiple Orbits Look for continuity of geographic regions between consecutive orbits. Note gap north of Alaska-Canada border. High latitude locations (>70 degrees) will have multiple consecutive overpass orbits every day  many chances to examine short-term variability.

Single Day – All SBUV/2 Results from four separate instruments shown together. No attempt made to resolve overlaps  total of 230 detections could be considered oversampling. Typical brightness is 4-8 units below 70 degrees latitude, 8-14 units at higher latitudes (few detections up to 20+ units). Note extended feature over Alaska, bright patch over central Russia (65 N, 115 E).

Single Day - OMI Limit this plot to swaths 9-22 to minimize effects of pixel expansion, cross-track brightness increase. Substantial overlap of consecutive orbits still observed above 75 N. Low latitude detections (< 60 degrees) are suspect because algorithm has not been “tuned” using OMI data. Even with these limitations, notice huge increase in number of detections. More samples have brightness in 15-30 unit range  demonstrates better pixel filling.

Coincidence Analysis Low latitude NLC (< 50º) detections increasing in frequency, but still rare. Would like quantitative evaluation of these clouds with satellite data. NLC and SBUV-type PMC observations not simultaneous by definition. SBUV/2 orbit tracks are far apart at mid-latitudes (22º longitude = 1000 km)  hard to get useful samples. Work in progress for 2003-2005 seasons using Canadian-American NLC network and SBUV/2 data. OMI cross-track coverage and smaller pixels will allow tighter windows on geographic, temporal coincidence tests.

Courtesy of J. Tvedtnes and M. Taylor [Utah State Univ.] Can Am Network + SBUV/2 Explain symbols for satellites, ground stations. Note angular separation between orbit tracks at lower lattude. This version of SBUV/2 processing only extends down to 50 N (~US-Canada border). Point out Glen Ullin. Courtesy of J. Tvedtnes and M. Taylor [Utah State Univ.]

OMI at Single Location Glen Ullin makes NLC observations as part of Can Am network. NOAA-18 “pixels” for same date shown as dark boxes. Next OMI orbit (5236) has two detections in almost identical locations from right-hand edge of swath. SBUV/2 data in upper sample are slightly too faint, but pass all other PMC detection tests  possible faint PMCs?

Further OMI Algorithm Work Calculate revised PMC detection thresholds using OMI data. Implement adjustment for cross-track albedo variations (wavelength-dependent?). Modify background calculation to analyze multiple swaths together. Test algorithm performance with alternate wavelength sets, spectral averaging. Create larger spatial bins for quantitative comparisons with SBUV/2 instruments. Analyze PMC detections for particle size information. - Research proposal will be submitted to NASA this summer.

Conclusions OMI data can be used to continue the unique PMC database developed from SBUV and SBUV/2 instruments. The improved measurement capabilities of OMI will give tremendous advances in our understanding of PMC morphology and evolution. Aura MLS temperature and water vapor profiles can provide valuable background information. OMI measurements can specifically address the recent increase in low-latitude PMC detections. OMI data will help to validate measurements from the AIM (Aeronomy of Ice in the Mesosphere) satellite, launched 25 April 2007 to study PMCs.