Assessment of Sampling Sufficiency for Low-Cost Satellite Missions: Application to PREFIRE
1. Introduction
The Arctic is transitioning to a warmer and wetter state (Boisvert and Stroeve 2015) at about twice the global rate (Screen and Simmonds 2010). Changes in low-frequency circulation modes and the intensities and distributions of baroclinic waves, as well as a moister and warmer lower troposphere, contribute to reductions in Arctic sea ice and Arctic amplification (e.g., Luo et al. 2017). Boisvert et al. (2015a) showed with a combination of Atmospheric Infrared Sounder (AIRS) data and trajectory modeling that approximately 10% of moisture is contributed locally with the vast majority advected from lower latitudes pointing to the key role of Arctic moisture intrusions associated with baroclinic waves. Cullather et al. (2016), Devasthale et al. (2016), Boisvert and Stroeve (2015), Kay et al. (2008), Johansson et al. (2017), and Taylor et al. (2015) and other studies have demonstrated linkages between increased cloud coverage, water vapor mixing ratio, surface temperature, atmospheric regime shifts, and increased frequencies and intensities of moisture intrusions. Vavrus et al. (2012) discuss that a wetter, cloudier and stormier Arctic in the twenty-first century is anticipated.
The meteorological mechanisms driving the rapid change in the Arctic are still debated, in part due to the absence of satellite observations within the far-infrared (FIR) bands of Earth’s emission spectrum (Harries et al. 2008). A moistening Arctic suggests that the “dirty” window regions within the FIR (e.g., Delamere et al. 2010; Turner and Mlawer 2010) will increase in opacity over larger spatial extents and longer time periods. This increased opacity is expected to peak in the summer season and into the early fall sea ice minimum and corresponds to climatological increases in water vapor (e.g., Serreze et al. 2012). Observations, models, and reanalysis datasets provide widely varying estimates of surface-based upward and downward longwave radiation flux estimates (L’Ecuyer et al. 2015) that motivate satellite benchmarks of the FIR. Temperature, water vapor, and cloud properties remain problematic in the latest generation of reanalyses (Graham et al. 2019).
Peterson et al. (2019) show that surface temperature (Tsfc) contributes more to the spectral clear-sky outgoing longwave radiation (OLR) and greenhouse effect (GHE) trends than atmospheric temperature (T) and water vapor mixing ratio (q) profiles. However, Peterson et al. (2019) also show that, for certain geographical regions and time periods, both T and q contribute more substantially to changes in FIR bands. An absence of observations of surface spectral emissivity (εν) in the FIR additionally contributes to uncertainties in the infrared surface and top of atmosphere (TOA) radiation budget. Huang et al. (2016) use a synthetic database of εν aggregated by band in the NCAR Community Atmosphere Model, version 5 (CAM5), to show that changes in global clear-sky and all-sky OLR range from −0.7 to −1.5 W m2 when including “realistic” band-by-band εν; regionally, these changes can be much larger. The sensitivity of TOA OLR to uncertainties in FIR εν is much more significant in dry atmospheres when column water vapor (CWV) ≤ 1.0 mm (Feldman et al. 2014).
The rapid change in the Arctic, combined with the lack of a FIR capability in the current and projected (National Academies of Sciences, Engineering, and Medicine 2018) NASA Earth science fleet, has created an urgency for a cost-effective approach to observing the FIR. Nearly 60% of Arctic emission occurs at wavelengths longer than 15 µm that have never been systematically observed. The NASA Earth Ventures program provides a mechanism for focused short-term efforts that add value to the understanding of the Earth system. With the selection of the Polar Radiant Energy in the Far Infrared Experiment (PREFIRE) mission in early 2018 (https://www.jpl.nasa.gov/news/news.php?feature=7054), NASA will systematically observe the FIR (15–54 µm) spectral region for at least two seasons at 0.86 µm spectral sampling using low-cost CubeSats carrying miniaturized thermal infrared spectrometers (TIRS). PREFIRE will quantify spatial and temporal variability in FIR spectra throughout the expected range of Tsfc and CWV encountered at both poles.
The success of the PREFIRE mission is predicated on its ability to sample a majority of the instantaneous environmental regimes that compose the Arctic and Antarctic climate. We seek to establish the sampling coverage of PREFIRE mission scenarios across a range of surface types (land, ocean, sea ice, land ice, glacier ice, snow) and for polar orbit launch scenarios around 98° inclination. Given the important role clouds play in polar climate, PREFIRE sampling characteristics are estimated separately for clear-sky and all-sky conditions spanning the expected ranges of Tsfc and CWV at both poles. Clear-sky sampling determines the observational characteristics for clear-sky OLR and GHE, εν and CWV retrievals, while all-sky sampling determines the observational characteristics for all-sky OLR and GHE, cloud properties, and the cloud mask.
Most of the literature regarding space–time sampling characteristics of satellite datasets has primarily addressed mean biases and errors for (i) specific choices of time binning and grid spacing (e.g., Astin 1997), (ii) satellites flown through discrete times within the diurnal cycle (e.g., Salby and Callaghan 1997), (iii) satellite simulators “flown” inside of GCMs (e.g., Lin et al. 2002), and (iv) multiple satellite constellations compared to single satellites (e.g., North et al. 1993). However, the mean biases vary according to the geophysical quantity of interest. For example, Guan et al. (2013) showed that cloud water and upper tropospheric humidity are inherently more sensitive to orbital sampling biases than 2-m temperature and humidity.
The space–time sampling for specific quantities of interest will play a secondary role in this initial PREFIRE investigation. Rather, we seek a deeper understanding of measurement sampling sufficiency: how much of the “observational space” in the polar regions is covered for a given observational scenario? The sampling will be assessed within realistic ranges of Tsfc and CWV for polar regions, as the FIR spectrum is highly sensitive to these two variables. This initial study is entirely focused on the polar regions rather than the tropics (e.g., Stephens et al. 2016; Kahn et al. 2016) or mountainous regions (e.g., Feldman et al. 2014) where the FIR plays a key role in modulating Earth’s radiation budget.
NASA Aqua’s AIRS and Advanced Microwave Sounding Unit (AMSU) are used to define the “climatology” with respect to an “observational space” composed of Tsfc and CWV dimensions. Subsets of this observational space that approximate potential PREFIRE observing swath and orbit scenarios are then compared to the full observational space defined by AIRS/AMSU. Previously, Sèze and Rossow (1991) describe the exploratory use of time-cumulated histograms versus those for a more restricted range of time, space, and geophysical extent. In this work, the time-cumulated histograms are analogous to the climatology, while the restricted histograms are analogous to the potential PREFIRE orbit configurations under consideration. There are multiple similarities between PREFIRE and AIRS that include approximate footprint size, overlapping spectral regions, and likely orbital inclinations. With the launch of Aqua on 4 May 2002 and the nearly continuous observational record through present day, the AIRS/AMSU datasets are ideal for estimating the observational coverage of PREFIRE.
This paper is organized as follows. In section 2, observations from the Aqua satellite and validation results are described. In section 3, the method in which the Aqua data are subsampled to several potential PREFIRE configurations is outlined. In section 4, the primary results of the sampling investigation are detailed. Finally, in section 5, the main findings and implications for PREFIRE observations and algorithm development are discussed.
2. Data
The Aqua satellite is an observing linchpin of the global water and energy cycles with nearly continuous pole-to-pole coverage since launch (Parkinson 2003), providing a realistic representation of the climatology of environmental states in the polar regions. Aqua provides a good benchmark for PREFIRE prelaunch sampling studies because of the long period of observation, AIRS’ stable radiometry (Aumann et al. 2019), the wide observing swaths that enable nearly daily global coverage of Earth’s surface and atmosphere, and extensive validation efforts.
There are 90 AIRS infrared fields of view (FOVs) and 30 AMSU fields of regard (FORs) in the cross-track direction of the AIRS/AMSU swath. An array of 3 × 3 AIRS FOVs are coregistered within one AMSU FOR (Lambrigtsen and Lee 2003) and serves as the basis of the cloud-clearing process used to retrieve Tsfc, CWV, and effective cloud fraction (ECF), among other properties such as T and q profiles and several trace gas species (Chahine et al. 2006). As the loss of AMSU-A2 channels in September 2016 led to the degradation of several geophysical retrievals from this time period onward (Yue and Lambrigtsen 2017), this investigation is restricted to a 14-yr period starting 1 September 2002 and ending 31 August 2016. As the infrared-only version of the retrieval is available after the loss of AMSU-A2, only the combined level 2 (L2) standard IR/microwave (MW) cloud-cleared products (AIRX2RET) is used henceforth. The four parameters used include effective cloud fraction (ECF), Tsfc, total CWV, and MW surface classification (Fig. 1).
Example 6-min AIRS/AMSU data granule 146 on 16 Nov 2002 centered on Greenland for several fields described in section 2: (a) brightness temperature (Tb) at 1231 cm−1 (K), (b) surface classification (low-emissivity snow was not observed in this granule), (c) total column water vapor (CWV, mm), (d) surface temperature (Tsfc, K), (e) cloud mask (cloud is purple, clear is white), and (f) total effective cloud fraction (ECF) where ECF = 0.01 is used to differentiate clear and cloud in (e). The three boxes in (a) approximate the subsamples for the 103° (red), 98° (green), and 93° (black) PREFIRE orbits for 15-footprint swaths. For the 8-footprint swath, those are drawn from the middle portion of each box.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Example 6-min AIRS/AMSU data granule 146 on 16 Nov 2002 centered on Greenland for several fields described in section 2: (a) brightness temperature (Tb) at 1231 cm−1 (K), (b) surface classification (low-emissivity snow was not observed in this granule), (c) total column water vapor (CWV, mm), (d) surface temperature (Tsfc, K), (e) cloud mask (cloud is purple, clear is white), and (f) total effective cloud fraction (ECF) where ECF = 0.01 is used to differentiate clear and cloud in (e). The three boxes in (a) approximate the subsamples for the 103° (red), 98° (green), and 93° (black) PREFIRE orbits for 15-footprint swaths. For the 8-footprint swath, those are drawn from the middle portion of each box.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Example 6-min AIRS/AMSU data granule 146 on 16 Nov 2002 centered on Greenland for several fields described in section 2: (a) brightness temperature (Tb) at 1231 cm−1 (K), (b) surface classification (low-emissivity snow was not observed in this granule), (c) total column water vapor (CWV, mm), (d) surface temperature (Tsfc, K), (e) cloud mask (cloud is purple, clear is white), and (f) total effective cloud fraction (ECF) where ECF = 0.01 is used to differentiate clear and cloud in (e). The three boxes in (a) approximate the subsamples for the 103° (red), 98° (green), and 93° (black) PREFIRE orbits for 15-footprint swaths. For the 8-footprint swath, those are drawn from the middle portion of each box.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
The ECF variable (CldFrcStd in AIRX2RET file) is reported for one or two layers within each AIRS footprint (Kahn et al. 2014). In this investigation, the two layers are summed into a single value of total ECF. The ECF was shown to be sensitive to thin clouds with values of ECF as low as 0.02–0.05 when compared against the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) coincidences (Kahn et al. 2008), although larger uncertainties are typical for infrared-based cloud amount estimates over Antarctica (e.g., Wang and Key 2005; Bromwich et al. 2012; Lubin et al. 2015). ECF is generally consistent with other satellite sensors (Nasiri et al. 2011), and short-term temporal variations are correlated to Arctic moisture intrusions or subseasonal variability in CWV (Devasthale et al. 2016; Boisvert and Stroeve 2015; Cullather et al. 2016).
Surface temperature (TSurfAir in AIRX2RET files) is fundamental to understanding Arctic climate change and rapid storm processes and is a key dimension in this investigation. Among the most challenging areas for AIRS Tsfc retrievals is the cold Antarctic plateau. Boylan et al. (2015) describe comparisons of coincident AIRS Tsfc retrievals to dropsondes over Antarctica, with 55% of the coincidences falling within ±5 K and 84% within ±10 K. For the remainder of Earth’s surface, a mostly universal ±1 K bias for Tsfc is found across most latitude bands for both ocean and land when compared against coincident radiosondes, with standard deviations between 1 and 4 K (Dang et al. 2012). The spatial patterns and secular trends in AIRS Tsfc exhibit realism within different seasons (Devasthale et al. 2016) and are largely consistent with other Tsfc datasets (Susskind et al. 2019). Over sea ice Tsfc has been validated with coincident observations during several field campaigns with differences generally between 2 and 4 K (Boisvert et al. 2015b, 2020; Taylor et al. 2018). Over the Greenland ice sheet Tsfc has been validated with coincident surface station data with differences near 2 K (Boisvert et al. 2017), and potentially larger biases in the presence of temperature inversions (Hearty et al. 2018). Therefore, the AIRS Tsfc variable realistically captures not only seasonal variations, but weather-scale variations and spatial gradients on the order of a few kelvins and larger for land, ocean, sea ice, and land ice surfaces. As Tsfc is retrieved on the AMSU FOR, the retrievals of Tsfc are duplicated across all nine AIRS FOVs and are matched to values of total ECF.
CWV (TotH2OStd in the AIRX2RET files) is fundamental to understanding how frequently PREFIRE will observe “open,” “dirty,” or “closed” atmospheric windows. Biases in CWV are occasionally high or low depending on the cloud amount and cloud regime within the sounding FOR (Fetzer et al. 2006). Roman et al. (2016), Alraddawi et al. (2018), and Perez-Ramirez et al. (2019) show that AIRS is biased slightly wet in the coldest and driest Arctic observations, but by and large performs well against Global Navigation Satellite System (GNSS) surface stations and the Maritime Aerosol Network (MAN). Similar to the Tsfc validation studies described earlier, any nonzero biases and standard deviations would suggest some potential “smearing” of CWV across a few bins in the histograms (see section 3). As with Tsfc, the retrievals of CWV are duplicated across all 9 AIRS FOVs.
The MW surface class variable (MWSurfClass in the AIRX2RET files) contains eight surface classifications: coastline, land, ocean, new sea ice, multiyear sea ice, new snow, old snow, and glacier ice. Surface classification is inferred from the land fraction at the AMSU FOR and channel-dependent emissivities from several AMSU channels (Grody 1993; Olsen 2017). Within each AMSU FOR, coastline implies 50%–99% liquid water coverage, land implies less than 50% liquid water coverage, while ocean implies >99% liquid water coverage. New snow and old snow differences are caused by a complex set of factors including age, snow grain size, and melt–refreeze of snow. New snow is typical for small grains and higher liquid water content, while old snow is typical for refrozen snow with larger grain sizes and lower liquid water content. New sea ice and multiyear sea ice differences are caused by a complex set of factors including age and ice thickness. Glacier ice is differentiated from sea ice and snow from very low MW emissivity values (Grody 1993). The AMSU and AIRS spatial coregistration (Lambrigtsen and Lee 2003) and instantaneous time matching make them ideal for footprint-scale consistency and filtering/binning of data.
In section 4, the snow and ice categories defined by MW surface classification are combined together into single snow and ice categories. This simplification is motivated by a need to quantify the enhancement in the sampling rates for combinations of surfaces with similar spectral infrared emissivities. Huang et al. (2016, 2018) show calculations that stronger similarities are found between various ice surfaces than with snow surfaces with regard to spectral infrared emissivity.
3. Methodology
An example AIRS 6-min data granule in Fig. 1 illustrates the subsampling strategy. The outlines over the 1231 cm−1 brightness temperature (Tb) in Fig. 1a depict how the AIRS swath is subsampled to approximate possible PREFIRE orbit and swath characteristics. For the Arctic, the right third of the granule corresponds to a PREFIRE inclination of 93° (black), the middle third is 98° (green), and the left third is 103° (red). In the Antarctic, the order is reversed such that the left third of the granule is designated 93°, the middle third is 98°, and the right third is 103° (not shown). Surface classification, CWV, Tsfc, and ECF are shown in Figs. 1b–d and 1f, respectively. For the cloud mask (Fig. 1e), changes in the assumed thresholds of clear-sky (ECF ≤ 0.01) and cloudy-sky (ECF > 0.01) lead to relatively modest changes in counts but have little material impact on the results that follow. To approximately reproduce the expected spacing between successive PREFIRE footprints in the cross-track direction, every other AIRS FOV across the swath is skipped, leaving a total of 45 AIRS FOVs, with 15 AIRS FOVs within each third (93°, 98°, and 103°) of the swath.
Using Sèze and Rossow (1991) as a guideline, the “climatology” is defined for the total CWV and Tsfc dimensions during a 2-yr period starting 1 September 2002 and ending 31 August 2004. The climatological space of Tsfc and CWV is generally insensitive for longer time periods, while a 1-yr climatology will be sensitive to interannual variability. The sensitivity of the climatology to the time period selected is discussed in section 4g. The clear-sky and all-sky histograms for the Arctic (60°–90°N) and Antarctic (70°–90°S) regions are shown in Fig. 2 and summarized in Table 1. The higher counts in the Arctic are due to the wider latitude range considered. A minimum of 20 counts are required for a given CWV and Tsfc bin to be included in the statistics, and to appear on the color scale in Fig. 2 within the histograms that follow. The two-dimensional histograms are divided into 50 bins each ranging from 0 to 10 mm (0.2 mm bin width) and 210 to 310 K (2 K bin width) for CWV and Tsfc, respectively. Both coarser and finer binning (25 and 100, respectively) was tested; however, the general characteristics of the histograms are mostly insensitive in this range of bin widths. Finer binning may not be justified in light of the AIRS/AMSU CWV and Tsfc uncertainty characteristics discussed in section 2, while coarser binning begins to smear out features that exhibit strong gradients in the counts, such as very dry values of CWV, and Tsfc values near 273 K. The climatological histograms display higher counts in cloudy-sky than in clear-sky within the Arctic region. As expected, the cloudy-sky histogram is somewhat warmer and moister than the clear-sky histogram. While the higher counts for the cloudy-sky histogram translate to more bins in the Arctic histograms, that is not the case in the Antarctic since clear-sky has a higher number of counts yet fewer bins as these observations are concentrated within the coldest and driest bins of the histograms.
The 2-yr climatological sampling of CWV and Tsfc for the Arctic (60°–90°N) (a) clear-sky and (b) cloudy-sky regions, and Antarctic (70°–90°S) (c) clear-sky and (d) cloudy-sky regions using the 45-footprint-wide AIRS/AMSU swath (skipping every other footprint). The 2-yr climatology extends from 1 Sep 2002 until 31 Aug 2004. There are 50 bins each in the CWV and Tsfc dimensions with bin winds of 0.2 mm and 2 K, respectively. A natural log scale is used as the counts span several orders of magnitude across bins.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
The 2-yr climatological sampling of CWV and Tsfc for the Arctic (60°–90°N) (a) clear-sky and (b) cloudy-sky regions, and Antarctic (70°–90°S) (c) clear-sky and (d) cloudy-sky regions using the 45-footprint-wide AIRS/AMSU swath (skipping every other footprint). The 2-yr climatology extends from 1 Sep 2002 until 31 Aug 2004. There are 50 bins each in the CWV and Tsfc dimensions with bin winds of 0.2 mm and 2 K, respectively. A natural log scale is used as the counts span several orders of magnitude across bins.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
The 2-yr climatological sampling of CWV and Tsfc for the Arctic (60°–90°N) (a) clear-sky and (b) cloudy-sky regions, and Antarctic (70°–90°S) (c) clear-sky and (d) cloudy-sky regions using the 45-footprint-wide AIRS/AMSU swath (skipping every other footprint). The 2-yr climatology extends from 1 Sep 2002 until 31 Aug 2004. There are 50 bins each in the CWV and Tsfc dimensions with bin winds of 0.2 mm and 2 K, respectively. A natural log scale is used as the counts span several orders of magnitude across bins.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
A summary of total counts, and the number of Tsfc and CWV bins with >20 counts depicted in Fig. 2.
The counts for clear-sky and cloudy-sky for the Arctic and Antarctic in Fig. 2 are summed into a single histogram for “all-sky” and serve as the climatology from which various statistical subsets are derived. While the histograms are restricted to clear-sky or encompass all-sky conditions, additional subsampling is applied to (i) individual seasons (three month time periods for SON, DJF, MAM, and JJA) within individual years or averaged over multiple years, (ii) multiple seasons (6-month time periods that start on 1 September, 1 December, 1 March, or 1 June), (iii) single or multiple orbit CubeSat configurations that approximate the PREFIRE “threshold” and “baseline” mission scenarios (see Table 2), and (iv) variable swath widths with a primary focus on the nominal PREFIRE eight-footprint swath.
Orbital configurations and time period of observation and the PREFIRE notional “baseline” and “threshold” missions. The PREFIRE baseline mission requires two CubeSats and a mission length of 1 year. The two CubeSats are not necessarily required to be in formation and could have different orbital inclinations and altitudes. The PREFIRE threshold mission requires one CubeSat and a mission length of 6 months. The additional orbital scenarios are included to gain a more complete understanding of potential sampling variability. The options for the baseline and threshold missions are necessarily limited by the Aqua swath geometry.
As interannual variability of polar weather and climate is substantial (Serreze et al. 2012), a 1-yr PREFIRE baseline mission suggests a “random draw” aspect to PREFIRE observational sampling. Therefore, the 14 years of data in the AIRS/AMSU record are used to calculate 14 separate baseline missions from which averages and standard deviations are calculated. These data are then presented in a series of dot-and-whisker plots in section 4. Last, new and old snow surfaces are combined into an “all snow” category, and new, multiyear, and glacier ice surfaces are combined into an “all ice” category.
4. Results
a. Two-dimensional histograms of CWV and Tsfc
The primary metric used to define PREFIRE sampling characteristics is the percentage of the AIRS/AMSU CWV–Tsfc climatology covered by different PREFIRE scenarios. The climatology (gray shading) for CWV and Tsfc within all-sky conditions, and also for each surface individually and all surfaces combined, is shown in Fig. 3 and Table 3. The subsampling representing PREFIRE observations (color scale) is restricted to 1 September 2002 to 31 August 2003 for an eight-footprint swath and a combined 98° + 93° two-orbit inclination. This shows that, even for a limited 1-yr time period and spatial sampling restricted to the Arctic, a total of 78% of all bins over all surfaces are sampled.
The 12-month PREFIRE baseline mission histograms for two-launch 98° + 93° orbits, with eight pixel-wide swaths, for the all-sky Arctic between 1 Sep 2002 and 31 Aug 2003. Each of the eight different surface types are shown separately: (a) coastline, (b) land, (c) ocean, (d) new sea ice, (e) multiyear sea ice, (f) new snow, (g) glacier ice, (h) old snow, and (i) all surfaces combined [the sum of (a)–(h)]. The gray area depicts the climatological sampling from the full AIRS/AMSU swath shown in Fig. 2 for each surface individually in (a)–(h) and for all surfaces together in (i). The climatological sampling is the sum of the clear-sky and cloudy-sky Arctic and Antarctic histograms in Fig. 2. The color scale is for bins sampled with counts >20.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
The 12-month PREFIRE baseline mission histograms for two-launch 98° + 93° orbits, with eight pixel-wide swaths, for the all-sky Arctic between 1 Sep 2002 and 31 Aug 2003. Each of the eight different surface types are shown separately: (a) coastline, (b) land, (c) ocean, (d) new sea ice, (e) multiyear sea ice, (f) new snow, (g) glacier ice, (h) old snow, and (i) all surfaces combined [the sum of (a)–(h)]. The gray area depicts the climatological sampling from the full AIRS/AMSU swath shown in Fig. 2 for each surface individually in (a)–(h) and for all surfaces together in (i). The climatological sampling is the sum of the clear-sky and cloudy-sky Arctic and Antarctic histograms in Fig. 2. The color scale is for bins sampled with counts >20.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
The 12-month PREFIRE baseline mission histograms for two-launch 98° + 93° orbits, with eight pixel-wide swaths, for the all-sky Arctic between 1 Sep 2002 and 31 Aug 2003. Each of the eight different surface types are shown separately: (a) coastline, (b) land, (c) ocean, (d) new sea ice, (e) multiyear sea ice, (f) new snow, (g) glacier ice, (h) old snow, and (i) all surfaces combined [the sum of (a)–(h)]. The gray area depicts the climatological sampling from the full AIRS/AMSU swath shown in Fig. 2 for each surface individually in (a)–(h) and for all surfaces together in (i). The climatological sampling is the sum of the clear-sky and cloudy-sky Arctic and Antarctic histograms in Fig. 2. The color scale is for bins sampled with counts >20.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
A summary of Tsfc and CWV bins with >20 counts depicted in Figs. 3–5. The left column is the number of bins for all-sky, which is the sum of clear-sky and cloudy-sky in Fig. 2.
Figure 3 illustrates that the coastline has many of its observations near freezing temperatures which suggests complex subfootprint features over a mix of bare land, ocean, and frozen surfaces. The ocean has less spread and fewer total bins than land despite the fact that there are twice as many counts compared to land; the ocean Tsfc is constrained to a narrower range than land Tsfc. New sea ice is found at warmer Tsfc than multiyear sea ice. New snow is spread over more of the total climatological space than old snow although the sampling rate is very similar (74% and 76% for new and old snow, respectively). As expected, glacier ice and old snow categories are the coldest and driest of all the surfaces. The ranges of CWV, Tsfc, the relative numbers of counts, and their positioning within the climatological observational space are consistent with expectations for an arbitrary Arctic year.
The Antarctic region is shown in Fig. 4 and Table 3. Generally speaking, the Antarctic sampling is sparser than the Arctic in most categories except for glacier ice. Observations for coastline and ocean are 32% and 46%, respectively, which is consistent with the 70°–90°S sampling range set by PREFIRE mission requirements that prioritize high-latitude data. By design these Antarctic histograms fill in PREFIRE sampling for colder and drier sets of bins than the Arctic histograms. The observations are complementary as the two polar regions would be obtained simultaneously during the mission. Histograms in Figs. 3 and 4 are summed and then shown in Fig. 5 and Table 3. The total polar sampling (85%) is a significant increase over the Antarctic (62%) and Arctic (78%) sampling alone highlighting the value of PREFIRE’s combined sampling of both poles. In particular, individual surface classes are found in different ranges of CWV and Tsfc within the Arctic and Antarctic histograms such that a larger sampling for the total polar region is obtained (Fig. 5).
As in Fig. 3, but for the Antarctic.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
As in Fig. 3, but for the Antarctic.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
As in Fig. 3, but for the Antarctic.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
As in Figs. 3 and 4, but for the total polar sampling (sum of Figs. 3 and 4).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
As in Figs. 3 and 4, but for the total polar sampling (sum of Figs. 3 and 4).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
As in Figs. 3 and 4, but for the total polar sampling (sum of Figs. 3 and 4).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
b. Clear- versus cloudy-sky sampling
The clear-sky and all-sky conditions for the 98° + 93° two-orbit inclination and eight-footprint-wide swaths are summarized in Figs. 6a and 6c, respectively. There is approximately a 10%–15% increase in the sampling rates for all-sky conditions compared to clear-sky conditions. In the case of ocean, the increase between clear-sky and all-sky conditions is much larger (>30%) because of the dominance of cloudy skies over the oceans. As there is a larger surface type diversity in the Arctic region considered, there is a tendency for more bins to be filled in with sufficient observations within the Arctic compared to the Antarctic, except for glacier ice, where Antarctica has a greater number of observations. The year-to-year standard deviation is generally a few percent or less for most surfaces, and is slightly higher in the Antarctic than the Arctic.
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for each surface type, combined snow and ice categories, and all combined surface types for 1-yr, two-orbit 98° + 93° and clear-sky for (a) 8-footprint and (b) 15-footprint swaths, and all-sky for (c) 8-footprint and (d) 15-footprint swaths. The mean values and standard deviations are calculated from the 14-yr AIRS/AMSU observing record for 1 Sep 2002 to 31 Aug 2014.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for each surface type, combined snow and ice categories, and all combined surface types for 1-yr, two-orbit 98° + 93° and clear-sky for (a) 8-footprint and (b) 15-footprint swaths, and all-sky for (c) 8-footprint and (d) 15-footprint swaths. The mean values and standard deviations are calculated from the 14-yr AIRS/AMSU observing record for 1 Sep 2002 to 31 Aug 2014.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for each surface type, combined snow and ice categories, and all combined surface types for 1-yr, two-orbit 98° + 93° and clear-sky for (a) 8-footprint and (b) 15-footprint swaths, and all-sky for (c) 8-footprint and (d) 15-footprint swaths. The mean values and standard deviations are calculated from the 14-yr AIRS/AMSU observing record for 1 Sep 2002 to 31 Aug 2014.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
With respect to individual surface classes, as well as all surfaces, all snow and all ice combined, the “random draw” effect of the time of year—and the particular year in which PREFIRE will launch—should have no material impact on the bin percentage covered within the climatology for the baseline mission. This does not suggest, however, that the areas within the histograms will not shift significantly from year to year. In particular, seasonal variations in themselves will be substantial because of changes in surface types, and Tsfc and CWV within each type. Some of this effect will be mitigated by the combination of Antarctic and Arctic sampling together into a total polar category as opposing seasons will “fill out” more of the histogram space.
c. Impact of swath width
To assess trades within instrument and CubeSat constellation designs, the sampling footprints in the AIRS/AMSU swath are expanded. This expansion provides statistical evaluation of having a single instrument with nearly twice the areal coverage, or two instruments flown in formation. The clear-sky and all-sky conditions for the 98° + 93° two-orbit inclination and 15-footprint swath are summarized in Figs. 6b and 6d, respectively. A 6%–9% increase in sampling rate is gained from 8-footprint increasing to 15-footprint swaths in clear-sky for all surfaces (7%), all snow (6%), and all ice (9%). For all-sky, the increase is very similar for the aforementioned categories. For individual surface types, the increases are as large as ~10% with similar standard deviations between the two swaths, implying that the “random draw” aspects of 8-footprint and 15-footprint swaths are very similar.
d. Summary statistics for one year of sampling
The clear-sky and all-sky sampling for all snow and all ice surfaces are summarized for 1-yr time periods of sampling starting on 1 September in Figs. 7a and 7b. For both the all ice and all snow surface classes, there are slight preferences for the 98° + 93° two CubeSat launch over the 103° + 98° two CubeSat launch, as 98° + 93° spans more of the high-latitude snow- and ice-covered surfaces. The slight preference for 98° + 93° is a little larger within clear-sky by 1%–2% for all ice surfaces compared to all snow, but this is not the case in all-sky. There is also a slight preference for 98° in clear-sky when compared to 93° and 103° for the single CubeSat launch, but there is either no preference or perhaps a slight preference for 93° in all-sky. The reasons for the slight differences in orbital preference of clear-sky and all-sky is not entirely obvious but could be related to cloud amount differences with respect to different surface types. Overall, there is a ~15% increase in the sampling rates for all-sky when compared to clear-sky.
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for all ice and all snow surfaces starting 1 Sep for (a),(b) 12, (c),(d) 6 (SONDJF), and (e),(f) 3 (SON) months for (a),(c),(e) clear-sky and (b),(d),(f) all-sky. Three single-launch CubeSat orbital inclinations (93°, 98°, 103°) and two-launch CubeSat scenarios (93° + 98°, 98° + 103°) are considered for 8-footprint swaths. Additional 3- and 6-month time periods that start on 1 Dec, 1 Mar, or 1 Jun are within a few percent of 1 Sep and are not shown. The 1-yr, two CubeSat “baseline” mission of PREFIRE (Table 1) is represented by the 93° + 98° and 98° + 103° scenarios in (a) and (b). The 6-month, one CubeSat “threshold” mission of PREFIRE (Table 1) is represented by 103°, 98°, and 93° in (c) and (d).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for all ice and all snow surfaces starting 1 Sep for (a),(b) 12, (c),(d) 6 (SONDJF), and (e),(f) 3 (SON) months for (a),(c),(e) clear-sky and (b),(d),(f) all-sky. Three single-launch CubeSat orbital inclinations (93°, 98°, 103°) and two-launch CubeSat scenarios (93° + 98°, 98° + 103°) are considered for 8-footprint swaths. Additional 3- and 6-month time periods that start on 1 Dec, 1 Mar, or 1 Jun are within a few percent of 1 Sep and are not shown. The 1-yr, two CubeSat “baseline” mission of PREFIRE (Table 1) is represented by the 93° + 98° and 98° + 103° scenarios in (a) and (b). The 6-month, one CubeSat “threshold” mission of PREFIRE (Table 1) is represented by 103°, 98°, and 93° in (c) and (d).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Dot-and-whisker plots for the Arctic, Antarctic, and combined polar for all ice and all snow surfaces starting 1 Sep for (a),(b) 12, (c),(d) 6 (SONDJF), and (e),(f) 3 (SON) months for (a),(c),(e) clear-sky and (b),(d),(f) all-sky. Three single-launch CubeSat orbital inclinations (93°, 98°, 103°) and two-launch CubeSat scenarios (93° + 98°, 98° + 103°) are considered for 8-footprint swaths. Additional 3- and 6-month time periods that start on 1 Dec, 1 Mar, or 1 Jun are within a few percent of 1 Sep and are not shown. The 1-yr, two CubeSat “baseline” mission of PREFIRE (Table 1) is represented by the 93° + 98° and 98° + 103° scenarios in (a) and (b). The 6-month, one CubeSat “threshold” mission of PREFIRE (Table 1) is represented by 103°, 98°, and 93° in (c) and (d).
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
e. Summary statistics for 3 or 6 months of sampling
The clear-sky and all-sky sampling for all snow and all ice surfaces are summarized for six months (SONDJF) and three months (SON) in Figs. 7c–f. A decrease from 12 to 6 months will lead to a 5%–10% decrease in sampling rates, and likewise an additional 5%–10% decrease in sampling rates from 6 to 3 months. There is little sensitivity to whether the time period starts in September, December, March, or June, as these differences are at most a few percent (not shown). This is an encouraging development as it implies that the time of year for the launch of a short-term (Venture class) mission like PREFIRE will have little material impact of the sampling of the climatological histograms using the total number of bins sampled as the guiding metric. As mentioned previously, while it is expected that the area within the CWV and Tsfc dimensions may exhibit significant differences from season to season; even in a 3- or 6-month mission, the total polar statistics (Arctic + Antarctic) should mitigate the impacts from seasonal differences.
f. Impacts from periodic PREFIRE data dropouts
With the increasing usage of low-cost access to space options such as CubeSats, there is general concern about communications challenges. We assess the potential impacts of periodic data dropouts by withdrawing either one out of every four, one out of every three, and one out of every two orbits to simulate the impacts of unexpected bandwidth limitations, other types of downlink problems, or the effects of heavy housekeeping data downloads on sampling (Fig. 8). For a given color, the solid dot furthest to the right within each surface category is for the full sampling coverage. The other three symbols in each row are the mean values for one out of four, three, and two orbit dropouts. The results of Fig. 8 are limited to SON 2002 as all other seasons and years yield very similar statistics (not shown). There is a 7%–8% reduction in climatological sampling coverage for all polar regions if one out of two orbits are lost at a periodic rate. Assuming the data dropouts occur regularly and frequently, they will have little impact on the total sampling rates. Interestingly, some surfaces over Antarctica show even lower losses (less than 5%), while the coastline and land surface classes in the Arctic show losses as high as 10%. Thus, there could be as much of a factor of 2 difference in the dropout rates for individual surface types if dropouts are considered separately for the Arctic or Antarctic. This is likely a consequence of the earlier point made that the Antarctic bins have higher counts (redundancy) within a tighter range of bins.
Impacts of orbit dropouts by withdrawing every fourth, third, and second orbit for SON 2002 in all-sky conditions for a single CubeSat 98° orbit inclination and eight-footprint swaths for the Arctic, Antarctic, and total polar regions. Each group of four dots depicts the spread from the data dropouts on the sampling rates, with the solid circle farthest to the right represented by Fig. 5. The solid square is for one out of four, the open circle is for one out of three, and the open square is for one out of two orbital dropouts. Other seasons, sky types, and orbital inclinations are similar and are not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Impacts of orbit dropouts by withdrawing every fourth, third, and second orbit for SON 2002 in all-sky conditions for a single CubeSat 98° orbit inclination and eight-footprint swaths for the Arctic, Antarctic, and total polar regions. Each group of four dots depicts the spread from the data dropouts on the sampling rates, with the solid circle farthest to the right represented by Fig. 5. The solid square is for one out of four, the open circle is for one out of three, and the open square is for one out of two orbital dropouts. Other seasons, sky types, and orbital inclinations are similar and are not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Impacts of orbit dropouts by withdrawing every fourth, third, and second orbit for SON 2002 in all-sky conditions for a single CubeSat 98° orbit inclination and eight-footprint swaths for the Arctic, Antarctic, and total polar regions. Each group of four dots depicts the spread from the data dropouts on the sampling rates, with the solid circle farthest to the right represented by Fig. 5. The solid square is for one out of four, the open circle is for one out of three, and the open square is for one out of two orbital dropouts. Other seasons, sky types, and orbital inclinations are similar and are not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
If the data dropouts are clustered in time, the sampling rates may be further reduced if the dropouts are concentrated within the early or latter portions of the season. To quantify this potential effect, the year-to-year all-sky sampling rates restricted to the months of September and October only are shown in Figs. 9a and 9b, respectively (November is not shown as it is very similar to October). For September with respect to the total polar sampling, the values range between 55% and 60% from year to year which is very similar to one out of two orbit data losses. For October, this drops to 50%–55%. The greater sampling in September is consistent with a wider range of Arctic Tsfc and CWV values observed compared to October, which is cooler and drier than September, while Antarctica still is generally very cold and dry. Similar results are found for the Arctic and Antarctic regions separately (Fig. 9). Thus, for the most extreme data loss scenario (30 or 60 continuous days of data loss), the sampling is at most a few percent less than periodic data losses for every other orbit.
Sampling rates for all-sky conditions during (a) September and (b) October over the 14-yr observational record for the Arctic, Antarctic, and total polar sampling at a 98° orbital inclination and eight-footprint swath. November is very similar to October and is not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Sampling rates for all-sky conditions during (a) September and (b) October over the 14-yr observational record for the Arctic, Antarctic, and total polar sampling at a 98° orbital inclination and eight-footprint swath. November is very similar to October and is not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Sampling rates for all-sky conditions during (a) September and (b) October over the 14-yr observational record for the Arctic, Antarctic, and total polar sampling at a 98° orbital inclination and eight-footprint swath. November is very similar to October and is not shown.
Citation: Journal of Atmospheric and Oceanic Technology 37, 12; 10.1175/JTECH-D-20-0023.1
Last, when considering the impact of abnormally high Arctic sea ice loss (i.e., 2007 and 2012) on sampling characteristics, it is important to distinguish sea ice loss (pixel counts) versus sampling rates. Despite the reduced ice cover, the sampling rates over all ice surfaces averaged together are generally indistinguishable from the variability over the 14-yr observational record. For example, the sampling difference of all ice surfaces between September 2003 (41%) and 2012 (35%) was relatively modest compared to the 47% reduction in pixel counts from 2003 (278 873) to 2012 (149 147). However, the change in the sampling rates are somewhat varied depending on the type of ice surface. Glacier ice decreases from 32% to 29%, new sea ice from 28% to 22%, and multiyear sea ice from 38% to 21% between September 2003 and 2012. Therefore, in years with low sea ice, there is much less impact on sampling rates than the counts themselves if all ice surfaces are taken together, and the sampling rates overall are somewhat higher for the all-ice category. If new sea ice, multiyear sea ice, and glacier ice are considered separately, the sampling rates may vary significantly from year to year, primarily for multiyear sea ice, and are significantly lower than the all-ice category.
g. Sensitivity to the baseline climatology
Changes in the sampling rates from a climatology based on the 2-yr period starting 1 September 2002 and ending 31 August 2004, versus a 2-yr period starting 1 September 2014 and ending 31 August 2016 are negligible. The total number of bins with >20 counts is 1828 and 1858 bins for clear-sky (1839 and 1866 for cloudy-sky) in the earlier and latter periods, respectively, for a total difference of 30 (27) bins. Nearly all bins in the latter period are located within the same two-dimensional CWV and Tsfc space depicted in Fig. 2 and is therefore not shown for reasons of brevity. As a result, we conclude that the sampling rates derived in this work are insensitive to the time period of the baseline climatology.
However, this does not capture the full story of sampling shifts between the two time periods as the counts could move significantly within the histograms. This potential shift in the population of the counts is estimated by calculating mean values of CWV and Tsfc for each surface type within clear-sky, cloudy-sky, and all-sky. Tables 4 and 5 list the mean values of CWV and Tsfc, respectively, for the histograms depicted for the earlier climatological period in Figs. 3–5 (gray shading) and also for the latter period (not shown in figures). Although the time period of the climatology has little bearing on the sampling rates, upon inspection of Tables 4 and 5, important shifts in the mean values of CWV and Tsfc appear to be related to shifting counts within surface types and confirm a tendency toward a cloudier combined total polar region in later years. Interestingly, the CWV is approximately constant—or shows slight reductions—between the time periods within clear and cloudy skies when treated as separate entities. One exception is a slight increase in CWV by 0.1–0.2 mm for snow surfaces (Table 4). The shift toward cloudier skies in the percentages of counts (Table 6), which are generally moister for all surfaces by a factor of 2 except for ocean, explains the slight increase in CWV for all-sky conditions. The lack of a strong increase in CWV is consistent for both AIRS/AMSU and the Moderate Resolution Imaging Spectroradiometer (MODIS) (Chang et al. 2019).
The mean values of CWV (mm) for clear-sky, cloudy-sky, and all-sky histograms for total polar sampling within the climatological time periods of 1 Sep 2002 to 31 Aug 2004, and 1 Sep 2014 to 31 Aug 2016.
As in Tables 4 and 5, but for the percentage of the total number of counts. The total number of all-sky, all-surface counts for 2002–04 is 197 097 180 and for 2014–16 is 202 274 060. Each reported percentage is with respect to these total counts.
The changes in Tsfc are another matter. Between the two climatological time periods considered, the Tsfc increases by +0.44 K in clear-sky while it is +0.59 K in cloudy-sky; however, the all-sky increase is larger than either at +0.87 K. The larger all-sky increase in Tsfc is explained by the increased cloudiness (Table 6), where cloudy-sky is generally warmer than clear-sky for the same surface type. The difference in Tsfc is largest between clear-sky and cloudy-sky snow and ice surfaces, while the time period differences are most pronounced for snow surfaces. To summarize, the increases in Tsfc over the Aqua observing record—unlike CWV—are attributable to (i) a combination of increased Tsfc within each surface type and their clear-sky and cloudy-sky categories treated as separate entities, and (ii) a shift in the observations toward a cloudy-sky (Table 6).
5. Summary and discussion
PREFIRE will sample FIR (15–54 µm) spectra in both polar regions for the first time. This study introduces a new approach for assessing the sampling coverage of PREFIRE for a range of likely launch scenarios in the context of CWV, surface temperature, and surface type (land, ocean, sea ice, glacier ice, snow). We describe experiments where the potential sampling coverage for PREFIRE is tested separately for clear-sky and all-sky conditions spanning the expected climatological ranges of surface temperature (Tsfc) and column water vapor (CWV) for typical polar conditions. As miniaturized technological capabilities rapidly advance, opening the door to increased use of SmallSats and CubeSats (Stephens et al. 2020), it remains an open question as to whether these low-cost platforms could offer similar coverage in time and space as present-day flagship missions such as NASA’s Aqua satellite. The sampling approach that is described in this work serves as an example of how one may determine whether these sensors are up to the task of meeting specific science observing goals.
To this end, Aqua’s Atmospheric Infrared Sounder (AIRS) and Advanced Microwave Sounding Unit (AMSU) are used to define the climatological “observational space” for Tsfc and CWV histograms. Subsets of this observational space approximate potential PREFIRE launch scenarios and are then compared to the full observational space defined by AIRS/AMSU. There are several similarities between PREFIRE and AIRS that include approximate footprint size, overlapping spectral regions, and likely orbital inclinations that lend themselves to a straightforward use of AIRS to estimate the range of conditions that PREFIRE will observe. With the launch of Aqua on 4 May 2002, and the nearly continuous observational record through present day, the AIRS/AMSU datasets are ideal for defining potential PREFIRE sampling characteristics.
The climatological observational space is defined with respect to the total CWV and Tsfc dimensions during a 2-yr period starting 1 September 2002 and ending 31 August 2004. The clear-sky and all-sky histograms for the Arctic (60°–90°N) and Antarctic (70°–90°S) regions are shown separately and then combined into a total polar region histogram. Further subsampling is applied to (i) individual seasons (3-month time periods for SON, DJF, MAM, and JJA) within individual years or averaged over multiple years, (ii) multiple seasons (6-month time periods that start on 1 September, 1 December, 1 March, or 1 June), (iii) single or multiple orbit CubeSat launch configurations that reflect the PREFIRE “threshold” and “baseline” mission scenarios, and (iv) variable swath widths with a primary focus on the likely PREFIRE eight-footprint-wide swath. The year-to-year variability in the polar weather and climate implies some randomness in the sampling characteristics. Therefore, the 14-yr Aqua record defines the extent to which the “random draw” of the launch date and mission length meaningfully impact the observational statistics. The primary findings are as follows:
The sampling rates of the histograms increased by 5%–15% by summing together the Arctic and Antarctic regions into a total polar region
With respect to the climatological histograms of CWV and Tsfc, the single-year (12-month) sampling rates over all surface classes for the total polar (Arctic + Antarctic) clear-sky is ~75%, and for all-sky it is ~85%
A decrease from 12 to 6 months for a randomly chosen time period decreases the sampling rates by an additional 5%–10%
A decrease from 6 to 3 months for a randomly chosen time period further decreases the sampling rates by an additional 5%–10%
There is a slight preference for a 98° orbit over a 93° or 103° orbit for a single CubeSat launch
There is a slight preference for 93° + 98° over 103° + 98° for a two-orbit CubeSat launch
There is a modest 7%–8% reduction in sampling from full temporal data coverage to data dropouts of one out of every two orbits
An increase in swath width from 8 to 15 footprints increases the sampling rates by a modest 6%–9%
The differences for the total polar sampling between the four individual seasons (SON, DJF, MAM, JJA) are a few percent or less
Differences in sampling rates are negligible using either the first or last two years of the Aqua time period considered as the baseline climatology
CWV for all surfaces in all-sky conditions increases by only 0.05 mm between the two climatological time periods
Tsfc for all surfaces in all-sky conditions increases by +0.87 K between the two climatological time periods
This statistical study demonstrates that sufficient (i.e., >50%) sampling can be accomplished with a miniaturized thermal infrared spectrometer deployed in a CubeSat platform. Wide-band, spectrally resolved sampling of 50%–85% of the climatological space spanning the Tsfc and CWV variables is not only feasible, but also affordable and implementable within an Earth Ventures class mission.
This investigation was limited to the polar regions of Earth from 60° to 90°N and 70° to 90°S that are the primary observational targets of PREFIRE. In the event that additional data are obtained at lower latitudes, further analysis on sampling rates in the clear-sky and all-sky tropics (e.g., Kahn et al. 2016) and over high-altitude mountainous regions (e.g., Feldman et al. 2014) is warranted. The high values of CWV in the tropics will render the surface essentially unobservable in the FIR from space. However, free-tropospheric tropical humidity profiles are complex and vary across orders of magnitude, suggesting that the FIR portion of the spectrum is highly variable within the tropics. Over high-latitude mountainous regions, the values of CWV are much lower and the surface will be frequently observed in the FIR spectral region (Feldman et al. 2014).
This initial PREFIRE sampling investigation paves the way for detailed investigation of the sampling required to address specific mission science objectives and requirements for level-2 products. These follow-on studies are ongoing and will be reported elsewhere.
Acknowledgments
All authors were supported by the PREFIRE project at JPL and the University of Wisconsin under Award 80NSSC18K1485. We thank three reviewers for their insightful comments and suggestions. Part of this research was carried out at the Jet Propulsion Laboratory (JPL), California Institute of Technology, under a contract with the National Aeronautics and Space Administration.
Data availability statement
AIRS and AMSU data were obtained through the Goddard Earth Services Data and Information Services Center (http://daac.gsfc.nasa.gov/).
REFERENCES
Alraddawi, D., and Coauthors, 2018: Comparison of total water vapor content in the Arctic derived from GNSS, AIRS, MODIS, and SCIAMACHY. Atmos. Meas. Tech., 11, 2949–2965, https://doi.org/10.5194/amt-11-2949-2018.
Astin, I., 1997: A survey of studies into errors in large scale space-time averages of rainfall, cloud cover, sea surface processes and the Earth’s radiation budget as derived from low Earth orbit satellite instruments because of their incomplete temporal and spatial coverage. Surv. Geophys., 18, 385–403, https://doi.org/10.1023/A:1006512715662.
Aumann, H. H., S. Broberg, E. Manning, and T. Pagano, 2019: Radiometric stability validation of 17 years of AIRS data using sea surface temperatures. Geophys. Res. Lett., 46, 12 504–12 510, https://doi.org/10.1029/2019GL085098.
Boisvert, L. N., and J. C. Stroeve, 2015: The Arctic is becoming warmer and wetter as revealed by the Atmospheric Infrared Sounder. Geophys. Res. Lett., 42, 4439–4446, https://doi.org/10.1002/2015GL063775.
Boisvert, L. N., D. L. Wu, and C.-L. Shie, 2015a: Increasing evaporation amounts seen in the Arctic between 2003 and 2013 from AIRS data. J. Geophys. Res. Atmos., 120, 6865–6881, https://doi.org/10.1002/2015JD023258.
Boisvert, L. N., D. L. Wu, T. Vihma, and J. Susskind, 2015b: Verification of air/surface humidity differences from AIRS and ERA-Interim in support of turbulent flux estimation in the Arctic. J. Geophys. Res. Atmos., 120, 945–963, https://doi.org/10.1002/2014JD021666.
Boisvert, L. N., J. N. Lee, J. T. M. Lenaerts, B. Noël, M. R. van den Broeke, and A. W. Nolin, 2017: Using remotely sensed data from AIRS to estimate the vapor flux on the Greenland ice sheet: Comparisons with observations and a regional climate model. J. Geophys. Res. Atmos., 122, 202–229, https://doi.org/10.1002/2016JD025674.
Boisvert, L. N., T. Vihma, and C.-L. Shie, 2020: Evaporation from the Southern Ocean estimated on the basis of AIRS satellite data. J. Geophys. Res. Atmos., 125, e2019JD030845, https://doi.org/10.1029/2019JD030845.
Boylan, P., J. Wang, S. A. Cohn, E. Fetzer, E. S. Maddy, and S. Wong, 2015: Validation of AIRS version 6 temperature profiles and surface-based inversions over Antarctica using Concordiasi dropsonde data. J. Geophys. Res. Atmos., 120, 992–1007, https://doi.org/10.1002/2014JD022551.
Bromwich, D. H., and Coauthors, 2012: Tropospheric clouds in Antarctica. Rev. Geophys., 50, RG1004, https://doi.org/10.1029/2011RG000363.
Chahine, M. T., and Coauthors, 2006: The Atmospheric Infrared Sounder (AIRS): Improving weather forecasting and providing new insights into climate. Bull. Amer. Meteor. Soc., 87, 911–926, https://doi.org/10.1175/BAMS-87-7-911.
Chang, L., G. Gao, Y. Li, Y. Zhang, C. Zhang, Y. Zhang, and G. Feng, 2019: Variations in water vapor from AIRS and MODIS in response to Arctic sea ice change in December 2002–November 2016. IEEE Trans. Geosci. Remote Sens., 57, 7395–7405, https://doi.org/10.1109/TGRS.2019.2913149.
Cullather, R. I., Y.-K. Lim, L. N. Boisvert, L. Brucker, J. N. Lee, and S. M. J. Nowicki, 2016: Analysis of the warmest Arctic winter, 2015–2016. Geophys. Res. Lett., 43, 10 808–10 816, https://doi.org/10.1002/2016GL071228.
Dang, V. T., and Coauthors, 2012: AIRS/AMSU/HSB version 6: Level 2 performance and testing report. California Institute of Technology Jet Propulsion Laboratory Rep., 197 pp., http://disc.sci.gsfc.nasa.gov/AIRS/documentation/V6_L2_Performance_and_Test_Report.pdf.
Delamere, J. S., S. A. Clough, V. H. Payne, E. J. Mlawer, D. D. Turner, and R. R. Gamache, 2010: A far-infrared radiative closure study in the Arctic: Application to water vapor. J. Geophys. Res., 115, D17106, https://doi.org/10.1029/2009JD012968.
Devasthale, A., J. Sedlar, B. H. Kahn, M. Tjernström, E. J. Fetzer, B. Tian, J. Teixeira, and T. S. Pagano, 2016: A decade of spaceborne observations of the Arctic atmosphere. Bull. Amer. Meteor. Soc., 97, 2163–2176, https://doi.org/10.1175/BAMS-D-14-00202.1.
Feldman, D. R., W. D. Collins, R. Pincus, X. Huang, and X. Chen, 2014: Far-infrared surface emissivity and climate. Proc. Natl. Acad. Sci. USA, 111, 16 297–16 302, https://doi.org/10.1073/pnas.1413640111.
Fetzer, E. J., B. H. Lambrigtsen, A. Eldering, H. H. Aumann, and M. T. Chahine, 2006: Biases in total precipitable water vapor climatologies from Atmospheric Infrared Sounder and Advanced Microwave Scanning Radiometer. J. Geophys. Res., 111, D09S16, https://doi.org/10.1029/2005JD006598.
Graham, R. M., and Coauthors, 2019: Evaluation of six atmospheric reanalyses over Arctic sea ice from winter to early summer. J. Climate, 32, 4121–4143, https://doi.org/10.1175/JCLI-D-18-0643.1.
Grody, N. C., 1993: Remote sensing of the atmosphere from satellites using microwave radiometry. Atmospheric Remote Sensing by Microwave Radiometry, M. A. Janssen, Ed., Wiley, 259–334.
Guan, B., D. E. Waliser, J. F. Li, and A. da Silva, 2013: Evaluating the impact of orbital sampling on satellite–climate model comparisons. J. Geophys. Res. Atmos., 118, 355–369, https://doi.org/10.1029/2012JD018590.
Harries, J., and Coauthors, 2008: The far-infrared Earth. Rev. Geophys., 46, RG4004, https://doi.org/10.1029/2007RG000233.
Hearty, T. J., J. N. Lee, D. L. Wu, R. Cullather, J. M. Blaisdell, J. Susskind, and S. M. J. Nowicki, 2018: Intercomparison of surface temperatures from AIRS, MERRA, and MERRA-2 with NOAA and GC-Net weather stations at Summit, Greenland. J. Appl. Meteor. Climatol., 57, 1231–1245, https://doi.org/10.1175/JAMC-D-17-0216.1.
Huang, X., X. Chen, D. K. Zhou, and X. Liu, 2016: An observationally based global band-by-band surface emissivity dataset for climate and weather simulations. J. Atmos. Sci., 73, 3541–3555, https://doi.org/10.1175/JAS-D-15-0355.1.
Huang, X., X. Chen, M. Flanner, P. Yang, D. Feldman, and C. Kuo, 2018: Improved representation of surface spectral emissivity in a global climate model and its impact on simulated climate. J. Climate, 31, 3711–3727, https://doi.org/10.1175/JCLI-D-17-0125.1.
Johansson, E., A. Devasthale, M. Tjernström, A. M. L. Ekman, and T. L’Ecuyer, 2017: Response of the lower troposphere to moisture intrusions into the Arctic. Geophys. Res. Lett., 44, 2527–2536, https://doi.org/10.1002/2017GL072687.
Kahn, B. H., and Coauthors, 2008: Cloud type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount. Atmos. Chem. Phys., 8, 1231–1248, https://doi.org/10.5194/acp-8-1231-2008.
Kahn, B. H., and Coauthors, 2014: The Atmospheric Infrared Sounder version 6 cloud products. Atmos. Chem. Phys., 14, 399–426, https://doi.org/10.5194/acp-14-399-2014.
Kahn, B. H., X. Huang, G. L. Stephens, W. D. Collins, D. R. Feldman, H. Su, S. Wong, and Q. Yue, 2016: ENSO regulation of far- and mid-infrared contributions to clear-sky OLR. Geophys. Res. Lett., 43, 8751–8759, https://doi.org/10.1002/2016GL070263.
Kay, J. E., T. L’Ecuyer, A. Gettelman, G. Stephens, and C. O’Dell, 2008: The contribution of cloud and radiation anomalies to the 2007 Arctic sea ice extent minimum. Geophys. Res. Lett., 35, L08503, https://doi.org/10.1029/2008GL033451.
Lambrigtsen, B. H., and S.-Y. Lee, 2003: Coalignment and synchronization of the AIRS instrument suite. IEEE Trans. Geosci. Remote Sens., 41, 343–351, https://doi.org/10.1109/TGRS.2002.808246.
L’Ecuyer, T. S., and Coauthors, 2015: The observed state of the energy budget in the early twenty-first century. J. Climate, 28, 8319–8346, https://doi.org/10.1175/JCLI-D-14-00556.1.
Lin, X., L. D. Fowler, and D. A. Randall, 2002: Flying the TRMM satellite in a general circulation model. J. Geophys. Res., 107, 4281, https://doi.org/10.1029/2001JD000619.
Lubin, D., B. H. Kahn, M. A. Lazzara, P. Rowe, and V. P. Walden, 2015: Variability in AIRS-retrieved cloud amount and thermodynamic phase over West versus East Antarctica influenced by the SAM. Geophys. Res. Lett., 42, 1259–1267, https://doi.org/10.1002/2014GL062285.
Luo, B., D. Luo, L. Wu, L. Zhong, and I. Simmonds, 2017: Atmospheric circulation patterns which promote winter Arctic sea ice decline. Environ. Res. Lett., 12, 054017, https://doi.org/10.1088/1748-9326/aa69d0.
Nasiri, S. L., V. T. Dang, B. H. Kahn, E. J. Fetzer, E. M. Manning, M. M. Schreier, and R. A. Frey, 2011: Comparing MODIS and AIRS infrared-based cloud retrievals. J. Appl. Meteor. Climatol., 50, 1057–1072, https://doi.org/10.1175/2010JAMC2603.1.
National Academies of Sciences, Engineering, and Medicine, 2018: Thriving on Our Changing Planet: A Decadal Strategy for Earth Observation from Space. National Academies Press, 716 pp., https://doi.org/10.17226/24938.
North, G. R., S. S. P. Shen, and R. Upson, 1993: Sampling errors in rainfall estimates by multiple satellites. J. Appl. Meteor., 32, 399–410, https://doi.org/10.1175/1520-0450(1993)032<0399:SEIREB>2.0.CO;2.
Olsen, E. T., Ed., 2017: AIRS/AMSU/HSB version 6 level 2 product user guide. California Institute of Technology Jet Propulsion Laboratory Rep., 148 pp., https://docserver.gesdisc.eosdis.nasa.gov/repository/Mission/AIRS/3.3_ScienceDataProductDocumentation/3.3.4_ProductGenerationAlgorithms/V6_L2_Product_User_Guide.pdf.
Parkinson, C., 2003: Aqua: An Earth-observing satellite mission to examine water and other climate variables. IEEE Trans. Geosci. Remote Sens., 41, 173–183, https://doi.org/10.1109/TGRS.2002.808319.
Pérez-Ramírez, D., and Coauthors, 2019: Precipitable water vapor over oceans from the Maritime Aerosol Network: Evaluation of global models and satellite products under clear sky conditions. Atmos. Res., 215, 294–304, https://doi.org/10.1016/j.atmosres.2018.09.007.
Peterson, C. A., X. Chen, Q. Yue, and X. Huang, 2019: The spectral dimension of Arctic outgoing longwave radiation and greenhouse efficiency trends from 2003 to 2016. J. Geophys. Res. Atmos., 124, 8467–8480, https://doi.org/10.1029/2019JD030428.
Roman, J., R. Knuteson, T. August, T. Hultberg, S. Ackerman, and H. Revercomb, 2016: A global assessment of NASA AIRS v6 and EUMETSAT IASI v6 precipitable water vapor using ground-based GPS SuomiNet stations. J. Geophys. Res. Atmos., 121, 8925–8948, https://doi.org/10.1002/2016JD024806.
Salby, M. L., and P. Callaghan, 1997: Sampling error in climate properties derived from satellite measurements: Consequences of undersampled diurnal variability. J. Climate, 10, 18–36, https://doi.org/10.1175/1520-0442(1997)010<0018:SEICPD>2.0.CO;2.
Screen, J. A., and I. Simmonds, 2010: The central role of diminishing sea ice in recent Arctic temperature amplification. Nature, 464, 1334–1337, https://doi.org/10.1038/nature09051.
Serreze, M. C., A. P. Barrett, and J. Stroeve, 2012: Recent changes in tropospheric water vapor over the Arctic as assessed from radiosondes and atmospheric reanalyses. J. Geophys. Res., 117, D10104, https://doi.org/10.1029/2011JD017421.
Sèze, G., and W. B. Rossow, 1991: Time-cumulated visible and infrared radiance histograms used as descriptors of surface and cloud variations. Int. J. Remote Sens., 12, 877–920, https://doi.org/10.1080/01431169108929702.
Stephens, G. L., B. H. Kahn, and M. Richardson, 2016: The super greenhouse effect in a changing climate. J. Climate, 29, 5469–5482, https://doi.org/10.1175/JCLI-D-15-0234.1.
Stephens, G. L., and Coauthors, 2020: The emerging technological revolution in Earth observations. Bull. Amer. Meteor. Soc., 101, E274–E285, https://doi.org/10.1175/BAMS-D-19-0146.1.
Susskind, J., G. A. Schmidt, J. N. Lee, and L. Iredell, 2019: Recent global warming as confirmed by AIRS. Environ. Res. Lett., 14, 044030–10, https://doi.org/10.1088/1748-9326/aafd4e.
Taylor, P. C., S. Kato, K.-M. Xu, and M. Cai, 2015: Covariance between Arctic sea ice and clouds within atmospheric state regimes at the satellite footprint level. J. Geophys. Res. Atmos., 120, 12 656–12 678, https://doi.org/10.1002/2015JD023520.
Taylor, P. C., B. M. Hegyi, R. C. Boeke, and L. N. Boisvert, 2018: On the increasing importance of air-sea exchanges in a thawing Arctic: A review. Atmosphere, 9, 41, https://doi.org/10.3390/atmos9020041.
Turner, D. D., and E. J. Mlawer, 2010: The Radiative Heating in Underexplored Bands Campaign. Bull. Amer. Meteor. Soc., 91, 911–924, https://doi.org/10.1175/2010BAMS2904.1.
Vavrus, S. J., M. M. Holland, A. Jahn, D. A. Bailey, and B. A. Blazey, 2012: Twenty-first century Arctic climate change in CCSM4. J. Climate, 25, 2696–2710, https://doi.org/10.1175/JCLI-D-11-00220.1.
Wang, X., and J. R. Key, 2005: Arctic surface, cloud and radiation properties based on the AVHRR Polar Pathfinder dataset. Part I: Spatial and temporal characteristics. J. Climate, 18, 2558–2574, https://doi.org/10.1175/JCLI3438.1.
Yue, Q., and B. Lambrigtsen, 2017: AIRS V6 test report supplement: Performance of AIRS+AMSU vs. AIRS-only retrievals. California Institute of Technology Jet Propulsion Laboratory Rep., 109 pp., https://docserver.gesdisc.eosdis.nasa.gov/repository/Mission/AIRS/3.3_ScienceDataProductDocumentation/3.3.5_ProductQuality/V6_Test_Report_Supplement_Performance_of_AIRS+AMSU_vs_AIRS-Only_Retrievals.pdf.
