CIPS PMC Level 2 Data: Orbit-by-Orbit Cloud ParametersLast Updated August 20201. IntroductionVersion 5.20 CIPS polar mesospheric cloud (PMC) Level 2 data files consist of measurements ofcloud parameters on an orbit-by-orbit basis. These files are provided for quantitative analyses ofthe CIPS retrievals at high spatial resolution. For those users interested mainly in averaged quantities, the CIPS team also provides level 3C “summary” files – these are files that contain one PMCseason each of orbit-by-orbit and daily-averaged quantities binned in 1-degree latitude bins (separate binning for ascending and descending node data). More information can be found in the level3C documentation.For an overview of the CIPS measurements and a detailed description of the version 4.20 PMCretrieval algorithm, readers are referred to Lumpe et al. [2013; http://dx.doi.org/10.1016/j.jastp.2013.06.007]. The scientific validity of the CIPS v4.20 data was established through its use ina variety of scientific and validation analyses. The v4.20 cloud frequencies and albedo were foundto agree very well with coincident measurements from the Solar Back Scatter Ultraviolet (SBUV2) instruments [Benze et al., 2009; 2011]. Baumgarten et al.  analyzed CIPS data obtainedin close coincidence with ground-based lidar measurements and found good agreement in the cloudbrightness observed by these two very different methods. The detailed spatial structures observedby CIPS have been used to study mesospheric gravity waves [Chandran et al., 2009; 2010, 2012]and planetary waves [Merkel et al. 2009], while the CIPS ice water content has been used byStevens et al.  to analyze the effect of tidal signatures on PMCs. CIPS cloud frequencieswere used in Karlsson et al.  to connect southern hemisphere (SH) PMC variability with thebreakdown of the wintertime SH stratospheric polar vortex. The SH intra-seasonal PMC variabilityobserved by CIPS was also used to investigate inter-hemispheric coupling in Karlsson et al. .Stevens et al.  used the CIPS observations of PMC frequency and albedo in July 2011 tohelp demonstrate a causal link between the occurrence of very bright clouds and the main engineexhaust from the space shuttle’s final flight.This document describes the CIPS PMC v5.20 level 2 data products and provides guidance fordata users. The v5.20 PMC dataset is currently being validated by comparison with the CIPS v4.20data. Results will be summarized and published in a forthcoming paper, which is currently in preparation. Preliminary analyses show the following trends in v5.20 data compared to v4.20: slightlyhigher cloud frequency, due to increased cloud detection sensitivity; albedos are shifted slightlyhigher, 1-2 10-6 sr-1 on average; particle radius is lower by up to 5 nm on average; ice watercontent (IWC) is also slightly higher, by 10-15 g/km2, consistent with the albedo changes. Basedon these initial comparisons we consider the CIPS v5.20 PMC data products to be valid for scientific analysis with the caveats described below.The publicly available Level 2 data set consists of three NetCDF data files and three png imagefiles for each orbit 1. Documentation and IDL software tools to read the level 2 NetCDF files are1A fourth file type, with extension etc.nc, is archived at the NASA Space Physics Data Facility (SPDF). Thesefiles contain extra variables that the retrieval development team desired, but that are not part of the publicly distributed data products. For v05.20r05, the variables contained in the file are:significance rayleigh: Chi-square significance of the rayleigh parallel component of the residual scattering profile.1
available for download from the AIM web site. NetCDF readers for other software packages areavailable elsewhere (see, for instance, re.html). These data files are:(1) Geolocation, including variables such as latitude, longitude, time, etc. The file nameextension is cat.nc.(2) Cloud properties, including albedo, particle radius, and ice water content. The file nameextension is cld.nc.(3) Cloud phase function, containing cloud albedo vs. scattering angle. The file name extension is psf.nc.(4) Orbit-strip image of cloud albedo. The file name extension is alb.png.(5) Orbit-strip image of particle radius. The file name extension is rad.png.(6) Orbit-strip image of ice water content. The file name extension is iwc.png.The compressed geolocation, cloud property and phase function NetCDF files are 2, 5, and 13MB in size, respectively. Uncompressed file sizes are approximately three times larger due to thesignificant fraction of fill (NaN) values in these files (see below). Most users of level 2 data willnot need the cloud phase function file; it is provided mainly for users who wish to re-derive suchparameters as cloud particle radius using independent algorithms.Variables in files (1) through (3) are described in tables at the end of this document. The v05.20Level 2 data are reported on an equal area Lambert projection grid. Resolution elements are 7.5km 7.5 km in the nadir and become elongated away from nadir but remain 56.25 km2 in totalarea (this will be rounded to 56 km2 in the remainder of this document).There are 15 orbits per day. Data arrays in the level 2 files provide cloud properties in each 56km2 resolution element (hereafter referred to as a level 2 “pixel”), with array dimensions corresponding to the number of elements in the along-track and cross-track directions. Each array element thus corresponds to a unique location (latitude and longitude) that is observed up to ten timeswith different observation geometries, and thus scattering angles (see Lumpe et al.  for adescription of the viewing geometry and measurement approach). For convenience in data handling, the arrays span the entire bounding box defined by a CIPS orbit, typically consisting of 215,000 elements. However roughly half of these elements correspond to locations where nomeasurements are made and therefore have fill values.significance nonrayleigh: Chi-square significance of the rayleigh orthogonal subspace of the scattering profile.sd n: Noise component of the directional albedo uncertainty.sd s: Small scale additional directional albedo uncertainty. This error characterizes the additional average retrievalerror in the rayleigh background subtraction after accounting for the noise and the large scale zonal variability in theozone.rad i: Smoothed in fill value for the radius for instances when a good radius cannot be retrieved, but good radiusretrievals exist nearby. This is particularly useful for orbit strip edges that have bad scattering angle sampling.alb i: Cloud albedo consistent with rad i above. This is used to produce cloud albedo images for which large discontinuities are less common in regions of bad sampling (like orbit strip edges).2
The CIPS level 2 PMC retrievals use all Level 1A image data poleward of 30 degrees latitude. AllLevel 1A pixels at solar zenith angles (SZA) greater than 95 degrees, as well as occasional individual camera images at various SZA, are eliminated to screen out isolated scattered light artifactsthat appeared starting in 2012 as the orbit geometry changed. The range of scattering and viewangles observed for each location changes along the orbit track. For uniformity and comparison toother data sets, the cloud albedos reported in files (2) and (4) are normalized to 90 scattering angleand nadir (0 ) view angle. The view angle correction is accomplished by removing the sec(θ) geometry factor to account for the view angle dependence in path length (where θ, the view angle, isthe angle between the satellite and zenith directions, as measured from the scattering volume). Thescattering angle correction is accomplished by obtaining the best fit of the observed phase function(albedo vs. scattering angle) to a set of assumed scattering phase functions that are constrained bylidar data (for a comprehensive discussion see Lumpe et al. ; also see Hervig et al. and Baumgarten et al. ). We make the assumptions that the ice particles have an axial ratioof 2 and a distribution width that varies approximately as 0.39 radius for radii up to 40 nm andthen stays fixed at 15.8 nm for larger particles. The albedo at 90 scattering angle from that bestfit is the value to which the view angle correction is applied.From 2007 through 2015 CIPS operated as originally intended, executing a measurement sequencethat provided 27 images per orbit in each camera (30 for the PX camera) at a 43-second cadence.Science data images were obtained only in the summer polar regions, covering approximately 8000km along the orbit and 900 km in the cross-track direction. This operational sequence is known asSummer Pole Imaging. Since February 11, 2016 CIPS has been operating in variations of what iscalled continuous imaging (CI) mode. In this scenario the image cadence is decreased (typically 2– 3 minutes) and images are spread out over the entire sunlit portion of the orbit. This change wasmade in response to the evolution of the AIM orbit, which has precessed from its original midnight(ascending) equator crossing orientation through a terminator (full sun) phase and is now approaching a noon equator crossing (i.e., the satellite is flying backwards relative to its originalorientation). The CI modes allow CIPS to make year-round, global measurements of stratosphericgravity waves while still providing PMC measurements in the summer polar region. The trade-offis that the decreased measurement cadence limits the spatial overlap between consecutive images,so that the scattering profiles derived for each Level 2 pixel contain fewer independent measurements compared to the pre-CI mode data (the number of data points in the scattering phase functionis denoted by NLAYERS in the Level 2 data files; see Table 1 and the discussion below). Decreasing NLAYERS has implications for the retrieval of particle radius and IWC, as explained below.2. Orbit Strip ImagesUsers interested in a quick, qualitative view of the data for a particular orbit should download thealbedo image png files, which are type (4) in the list above. In defining the color scale for eachimage, the plotting routine imposes a limit on the albedo of 2 10-6 sr-1 as a lower threshold forplotting; thus any clouds dimmer than this will not appear. The upper plotting threshold is set sothat 1% of the pixels are saturated, unless this threshold is less than 10-5 sr-1, in which case thethreshold is set to 10-5 sr-1. Since color scales for these png files are determined uniquely for eachorbit, these images should not be used to compare cloud brightness from one orbit to the next. Forthat purpose, users should download the NetCDF files. The particle radius and ice water contentimages – types (5) and (6) above – are made using the same data screening as is used for the albedoimages.3
Figure 1. CIPS PMC albedo for orbits 11893 on 1 July 2009 (top) and 14632 on 1 January 2010 (bottom).This sampling is representative of the summer pole imaging sampling from May of 2007 to February of2015. The discontinuity indicated by the yellow rectangle in the top panel is due to a spacecraft roll, whichwas required for AIM SOFIE pointing from the NH 2008 PMC season through the SH 2009-2010 season. By1 January 2010 the effects of the roll on sampling had diminished.Figure 1 shows examples of Level 2 orbit strip albedo images for both the northern hemisphere(NH, top) and southern hemisphere (SH, bottom) during the Summer Pole imaging mode. Thesemeasurements are representative of CIPS observations in the middle of the PMC season from Mayof 2007 through February of 2016. Latitude lines are drawn in ten-degree increments from 80 tothe lowest latitude observed. Each orbit strip nominally consists of overlapping measurementsfrom 27 different 4-camera scenes. As can be seen from these two examples, the CIPS imagescontain a wealth of information on the cloud structures, which can be analyzed quantitatively usingthe level 2 NetCDF files.In March of 2016 CIPS began its Orbit-wide Continuous Imaging mode, in which images wereacquired throughout both the day and night sides of the orbit. The night side images cannot be usedfor PMC retrievals, however. The satellite beta angle, which is the angle between the satelliteorbital plane and the vector pointing from the satellite to the sun, changed rapidly from early 2016through 2018 (approaching -90 in 2017). Because of this, the orientation of the CIPS imagesvaried throughout this time period. Examples of Level 2B orbit strips for 1 July 2016, 1 January2017, and 1 July 2018 are shown in Figure 2. Early in 2017 AIM entered a “full sun” period, inwhich the beta angle was between negative 68 and negative 90 , which meant that the spacecraftwas in continual sunlight. This necessitated changes in the operations, which previously had reliedon the timing of satellite sunrise. Unfortunately, this resulted in an inability to acquire pre-seasoncalibration images for the NH 2017 and SH 2017-2018 seasons, so no PMC data are available forthese seasons.In November of 2018 CIPS began its Sunlit Continuous Imaging mode, in which images wereacquired throughout the sunlit side of the orbit, with none taken on the night side. This resulted in4
more images being available for PMC retrievals. Examples of the level 2 PMC orbit strips for 1January 2019, 1 July 2019, 1 January 2020 and 1 July 2020 are shown in Figure 3.Figure 2. CIPS PMC albedo for orbits 50189 on 1 July 2016 (top), 52969 on 1 January 2017 (middle), and61228 on 1 July 2018 (bottom). Continuous Imaging operations, where images were acquired throughoutthe sunlit hemisphere, began in March 2016 and continues to the present. From March 2016 throughOctober 2018, images were acquired on the day and night sides of the orbit, but only dayside images arescientifically useful. This figure illustrates the variable orientation of PMC images as the satellite beta anglechanged during this time period. No PMC data are available for the NH 2017 or SH 2017-2018 PMC seasonsbecause of a lack of calibration data.Figure 3. CIPS PMC albedo for orbits 64004 on 1 January 2019, 66752 on 1 July 2019, 69534 on 1 January2020, and 72280 on 1 July 2020 (top to bottom). From November 2018 to the present, CIPS images areacquired only on the dayside of the orbit, so more scientifically valid images are available on each orbitthan during the time period from March 2016 through October 2018.5
Sections 2.1 and 2.2 describe two anomalies in the albedo png file images of which users shouldbe aware. If images are found to exhibit suspicious behavior that is not described here, we wouldvery much appreciate being informed (please send an email to [email protected]).2.1. Edge ArtifactsOn occasion, artifacts near the cross-track edges of the orbits appe
The CIPS level 2 PMC retrievals use all Level 1A image data poleward of 30 degrees latitude . All Level 1A pixels at solar zenith angles (SZA) greater than 95 degrees, as well as occasional indi-vidual camera images at various SZA, are eliminated to screen out isolated scattered light artifacts that appeared starting in 2012 as the orbit geometry changed. The range of scattering and view ...