A version 2 processing of data from two ozone monitoring
instruments on Suomi NPP, the OMPS nadir ozone mapper and the OMPS nadir
ozone profiler, has now been completed. The previously released data were
useful for many purposes but were not suitable for use in ozone trend
analysis. In this processing, instrument artifacts have been identified and
corrected, an improved scattered light correction and wavelength registration
have been applied, and soft calibration techniques were implemented to
produce a calibration consistent with data from the series of SBUV/2
instruments. The result is a high-quality ozone time series suitable for
trend analysis. Total column ozone data from the OMPS nadir mapper now agree
with data from the SBUV/2 instrument on NOAA 19 with a zonal average bias of
NASA has been measuring ozone from space since the launch of the Backscatter Ultraviolet (BUV) instrument on Nimbus 4 in 1970. The series of follow-on instruments, SBUV (Solar Backscatter Ultraviolet) and TOMS (Total Ozone Mapping Spectrometer) on Nimbus 7 and SBUV/2 instruments on NOAA 9, 11, 14, 16, 17, 18, and 19 produced a long-term time series of global ozone observations. Under NASA's MEaSUREs (Making Earth System data records for Use in Research Environments) program, data from this series of instruments were re-processed to create a coherent ozone time series. Inter-instrument comparisons during periods of overlap as well as comparisons with data from other satellite- and ground-based instruments were used to evaluate the consistency of the record and make careful calibration adjustments as needed (McPeters et al., 2013). The result is an ozone data record suitable for trend studies that we designated the Merged Ozone Data (MOD) time series (Frith et al., 2014). Ozone instruments on the Suomi NPP spacecraft and the planned series of JPSS (Joint Polar Satellite System) spacecraft will now be used to continue this series of measurements in order to document long-term ozone change.
The Suomi National Polar-orbiting Partnership (Suomi NPP) is a joint NOAA–NASA mission that collects and distributes remotely sensed land, ocean, and atmospheric data to the meteorological and global climate change communities. Suomi NPP was launched 28 October 2011. The Ozone Mapper Profiler Suite (OMPS) on NPP consists of three instruments – the ozone total column nadir mapper (NM), an instrument similar to the TOMS and OMI ozone mapping instruments, the nadir profiler (NP), an instrument similar to the SBUV and SBUV/2 profilers, and the limb profiler (LP), an instrument that measures the ozone vertical distribution using light scattered from the Earth's limb. Details of the OMPS instruments and mission are given by Flynn et al. (2006).
The purpose of the version 2 processing of data from the two OMPS nadir sensors, which is the subject of this paper, is to correct various instrument artifacts and to apply an updated calibration that will be consistent with data from earlier instruments. Only the reprocessed version 2 data from the two nadir instruments will be discussed here. While some comparisons with data from the limb profiler will be shown in this paper, detailed LP validation results will be discussed in other papers.
The OMPS nadir mapper (NM) is a nadir-viewing, wide-swath,
ultraviolet–visible imaging spectrometer that provides daily global
measurements of the solar radiation backscattered by the Earth's atmosphere
and surface, along with measurements of the solar irradiance. It shares a
telescope with the OMPS nadir profiler (NP) spectrometer. A dichroic filter
splits light from the telescope into two streams. Most of the 310–380 nm
light is transmitted to the NM instrument, while most of the 250–300 nm
light is reflected to the NP instrument. The transition between reflection
and transmission occurs between 300 and 310 nm, the wavelength overlap
region. The detector for each instrument is a 340 pixel
Unlike the heritage TOMS instruments which measured ozone using a
photomultiplier detector at six discrete wavelengths (from 306 to 380 nm,
depending on the instrument), the NM instrument measures the complete
spectrum from 300 to 380 nm at an average spectral resolution of 1.1 nm. The
OMPS NM sensor has a 110
Each orbit of NM data measures a swath of total column ozone: 35 individual ozone measurements (see example near the Equator) are made for each scan line.
The OMPS nadir profiler (NP) has a 16.6
The goal of the version 2 processing is to produce ozone data sufficiently accurate to be used to continue the Merged Ozone Data (MOD) time series. This time series is a unified multi-instrument ozone data set created by merging data from a series of SBUV and SBUV/2 instruments beginning with the original BUV instrument launched on Nimbus 4 in 1970 and extending to the SBUV/2 instrument on NOAA 19, which continues to operate. Data from these instruments were recently reprocessed as version 8.6 with a consistent calibration to create a coherent ozone time series (McPeters et al., 2013). The MOD data set created from this series is described in detail by Frith et al. (2014). Figure 2 shows the MOD fit to data from three recent SBUV/2 instruments, on NOAA 16, 18, and 19, for which good data are available during the OMPS observation period. Comparison with ozone from ground networks shows that total ozone in the MOD series is consistent to within about a percent for the recent data. Data from the OMPS NP and NM instruments will be used to extend this MOD data record.
OMPS ozone will be compared with MOD (merged ozone data) ozone
created by merging data from recent SBUV/2 instruments. Monthly average
ozone for 60
In the version 2 processing we use the latest version of the Level 1 data, the data set of calibrated radiance measurements from NM and NP that implements a refined calibration for both instruments (Seftor et al., 2014) and corrects for several instrument effects. Both the NM and NP L1b data now use an improved set of calibration coefficients that exhibit smoother wavelength-to-wavelength behavior and provide a wavelength registration that accounts for intra-orbital (for the NM) and intra-seasonal (for the NP) shifts that were identified in analysis of the data. A small bandpass error in the NP instrument near 295 nm was corrected, and errors in the pre-launch calibration measurements in the dichroic transition region (300–310 nm) for both instruments were identified and corrected. The daily dark current correction has been refined for each instrument.
Soft (in orbit) calibration techniques were used to refine the instrument calibration. The NM pre-launch calibration of the 331 nm channel, which is used to determine reflectivity, was not adjusted at nadir since the measured radiance over ice matched the expected radiance (determined from other instruments such as Earth Probe TOMS and OMI) to within 1 %. Cross-track adjustments to this channel to “flatten” the 331 nm reflectivity calculation over ice were then determined and applied. Similarly, the nadir radiance at 317 nm, which is the channel used to determine ozone, was not changed; the off-nadir radiances were then adjusted to take out any cross-track ozone dependence. The 317 and 331 nm NM nadir radiances are also used in the NP algorithm retrieval, with no adjustments applied. For the NM radiances at 312 nm, which are used in the NP algorithm but not in the NM algorithm, an adjustment was determined and applied to minimize the final retrieval residuals. Similarly, the NP 306 nm radiances were adjusted to minimize the final residuals. The calibrations were not explicitly adjusted to agree with the NOAA 19 SBUV/2 calibration, so NOAA 19 comparisons can be used for validation.
The algorithm used to retrieve total column ozone from the NM is very similar to the v8.5 algorithm used in the processing of data from Aura OMI instrument as described by Bhartia (2007) and Bhartia et al. (2004). The basic algorithm uses two wavelengths to derive total column ozone, one wavelength with weak ozone absorption (331 nm) to characterize the underlying surface and clouds, and the other at a wavelength with strong ozone absorption (317 nm). The ozone retrieval algorithms for both the NP and NM instruments now use the Brion–Daumont–Malicet ozone cross sections (Brion et al., 1993) to be consistent with other data sets in the MOD time series.
The NP retrieval algorithm uses 12 discrete wavelengths to retrieve ozone profiles employing Rodgers' optimal estimation technique (Bhartia et al., 2013). It is very similar to the v8.6 algorithm used to reprocess the SBUV and SBUV/2 data sets (McPeters et al., 2013) used in the MOD time series. While the vertical resolution of an OMPS NP ozone retrieval is somewhat coarse in comparison with the LP sensor, about 8 km resolution in the stratosphere, NP provides valuable data for the continuation of the historical SBUV/2 ozone data record, and for validation of the OMPS LP retrievals.
The accuracy and stability of the OMPS ozone data record has been evaluated through comparisons with ground-based observations and comparisons with other satellite data sets. The worldwide network of Dobson and Brewer stations has been used for years for ground-based validation of total column ozone. For satellite validation of total ozone, comparisons with the MOD data set are used as a primary standard for this evaluation. Validation of profile ozone (in Sect. 5) will use data from balloon sondes, data from the currently operating SBUV/2 instrument on NOAA 19, and data from the microwave limb sounder (MLS) on the Aura spacecraft.
Figure 3 compares average ozone from 52 ground-based Brewer and Dobson stations in the Northern Hemisphere with coincident observations of ozone measured by the NM instrument over the individual stations (Labow et al., 2013). Comparison with ozone from the NOAA 19 SBUV/2 is also shown (in blue) since these data are the basis of much of the NM and NP validation. Northern Hemisphere comparisons are shown because the network density is much better in the Northern Hemisphere than in the Southern Hemisphere, and comparisons in a single hemisphere will illuminate any seasonally dependent errors. Such comparisons have been shown to be capable of detecting instrument changes over the long term of a few tenths of a percent (McPeters et al., 2008). The comparison covers the period from April 2012 through the end of 2016. Figure 3 shows that the agreement of NM total ozone is mostly within half a percent. The linear fit in Fig. 3 shows that OMPS NM has very little drift in ozone relative to the ground observations (0.8 % per decade) and an average bias of less than 0.2 %.
A comparison of OMPS NM ozone (in black) and NOAA 19 SBUV (in blue) with average ozone from an ensemble of 52 Northern Hemisphere Dobson and Brewer stations. A linear fit to the NM data is also shown. Weekly mean percent difference of satellite ozone minus ground-based ozone is plotted.
For average ozone in the 60
The comparison of ozone from the NM instrument with ozone from the MOD
(merged ozone data set) time series shown in Fig. 4 illustrates the
improved accuracy of the version 2 processing. The monthly zonal average
ozone, area weighted for the latitude zone from 60
A similar plot for the OMPS nadir profiler shows that the large bias in the released version 1 data is reduced in the version 2 processing.
Figure 5 is the same plot but for total column ozone measured by the NP
instrument. NP total column ozone is derived by integrating the retrieved
ozone profiles. In principle, this should be more accurate over a broad
range of solar zenith angles than ozone derived from the limited wavelength
range of the NM instrument. Here the average relative bias of about
Figure 6 shows the latitude dependence relative to MOD of the version 2
ozone from the mapper and from the profiler. Figure 6b plots ozone
averaged for five Marches from 2013 through 2016, while Fig. 6a
shows the percent difference from MOD for the same months. The latitude
dependence of ozone varies by season so it is useful to examine individual
months, and latitude coverage is maximum near an equinox. The NM instrument
has very little latitude dependence except at the highest southern latitudes
where ozone is low. The NP instrument has the bias as noted in Fig. 5 and
likewise has little latitude dependence at low to midlatitudes. The higher
ozone (by 2 % to 3 %) for retrievals at latitudes greater than
50
In version 2 the 4-year average of March ozone latitude
dependence (2013–2016) is shown in
The long-term behavior of ozone as a function of altitude is in some ways
more interesting than the behavior of total column ozone because it can be
used to confirm the accuracy of various model predictions. However, the
accuracy of these measurements is more difficult to validate (Hassler et
al., 2014). Data from the ozone sonde network can be used to validate the
profile in the troposphere and lower stratosphere, while satellite data can
be used to validate the middle to upper stratospheric results. There are
ground-based measurements of the ozone vertical distribution by lidar and by
microwave sounders, but such measurements are very sparse. There are
An average of ozone sonde data from Hilo, Hawaii, is compared with
OMPS NP version 2 ozone profiles for coincident days, with percent difference
plotted in
The NP ozone anomaly, the difference from NOAA 19 ozone, for midlatitudes and low latitudes is shown as a function of time for total column ozone, the lower stratosphere, and the upper stratosphere. Ozone from the version 1 processing (in red) and the version 2 processing (in green) is shown.
Looking at ground-based comparisons of ozone in the lower stratosphere
first, Fig. 7 compares NP ozone profiles with ozone measured by ECC ozone
sondes from one station, Hilo, Hawaii (20
OMPS NP version 2 June zonal average ozone profiles (2012–2016) compared with NOAA 19 SBUV/2 profiles, MLS profiles, and profiles from the OMPS LP. OMPS NP version 2 percent differences from N19, MLS, and LP are plotted on the right.
The time dependence of the version 2 ozone anomaly relative to NOAA 19 shown for low to midlatitudes.
The time dependence of the version 2.0 ozone anomaly relative to NOAA 19 shown for high latitudes.
For the middle to upper stratosphere, monthly zonal mean comparison with
other satellite observations of the ozone vertical distribution is the best
approach for evaluating the accuracy of the version 2 NP results. Figure 8
shows the time-dependent difference of NP from the NOAA 19 SBUV/2 retrievals
averaged over low to middle latitudes (40
The time series of tropospheric ozone shown for four locations. Tropospheric ozone derived by subtracting OMPS LP stratospheric ozone from NM total column ozone is shown in the blue solid curve, while tropospheric ozone derived by subtracting MLS stratospheric ozone from OMI total column ozone is shown in the dashed red curve.
Ozone agreement as a function of altitude is shown in Fig. 9 where ozone
in low to middle latitudes is averaged for five Junes from 2012 through
2016. Selecting a single month for this comparison allows us to see any
seasonal effect that might be suppressed in the annual average. As will be
shown later, there are seasonal variations in NP ozone at high latitudes.
The stratospheric ozone mixing ratio is plotted for OMPS NP version 2, for NOAA
19 SBUV/2, for the Aura microwave limb sounder (MLS) (Froidevaux et al., 2008),
and for the OMPS limb profiler (LP). The right panel shows the agreement of
the OMPS NP version 2 ozone profile with each of the three other profile
measurements by plotting the percent difference from each. Agreement is
almost always within
The NP version 2 ozone has a somewhat different behavior at low to midlatitudes
than at high latitudes. The ozone anomaly, the percent difference of NP
ozone from the NOAA 19 SBUV ozone, is shown for low to midlatitudes
(
Ziemke et al. (2011, 2014, and references therein) have shown that tropospheric ozone can be derived by subtracting stratospheric ozone from total column ozone. This technique has most recently been applied by subtracting stratospheric ozone measured by the Aura MLS instrument from total column ozone measured by the Aura OMI instrument. The OMI/MLS tropospheric ozone time series currently spans over 12 years and has been a central data product for each of the BAMS State of the Climate Reports since 2013 and will be used in the upcoming international Tropospheric Ozone Assessment Report.
The OMPS ozone measurements can also be used to calculate tropospheric ozone and continue the current OMI/MLS time series of measurements should either of the Aura instruments fail. Because the OMPS instrument suite includes both a total ozone mapper (NM) and a limb profiler (LP), a similar technique can be applied as with OMI/MLS. Figure 12 shows the tropospheric ozone time series for two locations in the tropics, Java and Brazil, and two locations at northern midlatitudes, Beijing and Washington DC. In each case the red dashed curve shows tropospheric ozone derived by subtracting MLS stratospheric ozone from OMI total column ozone. For comparison, the blue solid curve shows the same tropospheric ozone derived by subtracting stratospheric ozone from the OMPS LP from total column ozone from the NM. While there are some small differences the overall agreement is quite good. Data on tropospheric ozone from the NP plus LP combination can be used to continue the tropospheric ozone time series.
The OMPS nadir mapper (NM) has proven to be a very stable instrument.
Comparison with a network of 52 Northern Hemisphere ground-based Dobson and
Brewer instruments shows very good agreement over the four years of
operation, agreeing within
The nadir profiler (NP) has likewise been very stable. NP total column ozone
has a time dependence of only 0.5 % per decade relative to MOD or NOAA 19.
The bias of
Ozone data from these instruments can now be considered “trend quality” –
usable to extend the data record from previous instruments to create an
accurate time series. Data from NP at latitudes above 50
NPP OMPS version
2 data are now available online from the Goddard
DISC:
The authors declare that they have no conflict of interest.
The OMPS nadir profiler and nadir mapper were built by Ball Brothers for flight on the joint NASA–NOAA NPP satellite. We thank the many people who have worked over the years to understand the behavior of the OMPS instrument. The Ozone Processing Team has carefully maintained the calibration of the nadir instruments through both hard and soft calibration techniques. Edited by: Diego Loyola Reviewed by: two anonymous referees