Home Overview Data Services What's New

Team Operations Documentation Publications
  

SABER Temperature and CO2 Retrieval Accuracies

SABER temperature profiles are retrieved in two ways. The first is the so called “operational” retrieval algorithm that retrieves temperature using the measured 15µm CO2 limb vertical profile emission and a CO2 vertical mixing ratio profile calculated using the WACCM model (Model CO2 provided by Dan Marsh, NCAR, private communication). A second SABER algorithm referred to as the “2-channel” algorithm, simultaneously retrieves temperature and CO2 vertical profiles over the 70km to 110km range for temperature and 70km to 125km range for CO2 using measured 15µm and 4.3 µm CO2 limb vertical profile emissions. This latter algorithm has been applied to all daytime measurements to obtain self-consistent T/CO2 retrievals in the MLT region. The 2-channel T/CO2 data are available in the Level2C folder in a NetCDF format at the SABER GATS website where all SABER data reside. The methodology used and first results are presented in a paper by Rezac et al. (2015a). Results of a comprehensive CO2 validation study using the SD-WACCM and ACE-FTS data are presented in a validation paper by Rezac et al. (2015b) which is in review. Here we summarize important information on using the temperature and CO2 products.

1) The daytime (SZA < 80) measurements have been processed on a continuous basis since 2002 up to the current time.

2) The operational and 2-channel kinetic temperature (Tk) profiles differ but their differences are usually within the estimated error bars. It should be noted that the operationally retrieved Tk profile has a higher vertical resolution due to application of a retrieval interleaving process, which is not applied for the 2-channel inversion. The 2-channel temperature retrieval is always performed with the currently retrieved CO2 VMR profile obtained during the global iteration and as a result, the temperature in the polar summer mesopause region is typically slightly colder than the operational temperature.

3) The 2-channel CO2 retrieval is constrained below 80km to avoid large CO2 and hence large Tk variability (See Table 3 in Rezac et al., 2015a).

4) The retrieval for polar summer latitudes (>60o) below 90km provides only a mean for the 65-90km region. Above 90 km the inversion proceeds normally without any constraints. See Rezac et al., (2015a) for a detailed description of the feature and reasoning for the mean retrieval. This constraint does NOT apply to any other season or latitude region.

5) Tables 1, 2 and 3 give the SABER combined operational and 2-channel single profile Tk uncertainties, the operational single profile Tk uncertainties and the 2-channel single profile Tk and CO2 total uncertainties respectively. These errors are based on studies reported in papers by Remsberg et al. (2008), Garcia-Comas et al. (2008), Rezac et al. (2015a,b) and our additional studies. Uncertainties discussed in the earlier papers were based on the v1.07 data version but the operational Tk uncertainties used in the tables here include consideration of changes introduced by the v2.0 algorithm. There are no changes in v2.0 relative to v1.07 that would impact precision but overall accuracy can be impacted due to changes in inputs to the NLTE model including:

  • v2.0 uses retrieved [O] values for all events whereas v1.07 used retrieved [O] only for day and for SZA < 85 degree.
  • v2.0 uses CO2 from an updated WACCM model and is purely WACCM whereas v1.07 uses scaled WACCM results to match a CO2 trend model.
  • Some reaction rates were changed in the CO2 vibrational temperature model.

Changes to coefficients used for instrument electronic gain switches, the Local Thermodynamic Equilibrium (LTE) algorithm and the procedure used to merge LTE and Non-LTE results, lead to small v2.0 versus v1.07 (<2 K) differences below 80km. These changes result in a more reliable and more self-consistent product for v2.0. Also, errors in the tables are representative of all environments except for polar summer mesopause events where systematic errors above 80 km are expected to be higher due to higher uncertainty in the vibrational temperatures calculated for these very cold conditions.

Table 1: SABER combined operational and 2-channel** single profile Tk total uncertainty

T Retrieval*

15 km

20 km

30 km

40 km

50 km

60 km

70** km

80** km

90** km

100** km

110** km

Temp. prec. (K)

0.3

0.3

0.3

0.6

0.6

0.7

1.0

1.4

3.3

5.4

10.5

Temp. systematic (K)

1.4

1.2

0.7

1.5

1.9

1.9

2.8

2.7

4.5

9.0

27.5

Temp. accuracy (K)

1.4

1.3

0.8

1.6

2.0

2.0

3.0

3.0

5.6

10.5

29.4

**2-channel Tk retrieval starts at 70km; Starting at 70km, precision and systematic entries are the averages of the operational and 2-channel values In Tables 2 and 3. Accuracy is the RSS of precision and systematic values.

Table 2: SABER operational single profile Tk* total uncertainty

Operational T Retrieval**

15 km

20 km

30 km

40 km

50 km

60 km

70 km

80 km

90 km

100 km

110 km

Temp. prec. (K)

0.3

0.3

0.3

0.6

0.6

0.7

1.0

1.8

3.6

6.7

15.0

Temp. systematic (K)

1.4

1.2

0.7

1.5

1.9

1.9

1.5

1.4

4.0

5.0

25.0

Temp. accuracy (K)

1.4

1.3

0.8

1.6

2.0

2.0

1.8

2.3

5.4

8.4

29.2

*The WACCM CO2 results are used at all altitudes for the operational Tk retrieval.

Table 3: SABER 2-channel single profile Tk and CO2 total uncertainty

2-Channel Tk Retrieval

70 km

80 km

90 km

100 km

110 km

125 km*

Temp. prec. (K)

1

1

3

4

6

Temp. systematic (K)

4

4

5

13

30

RSS Syst. & Prec. Tk (K)

4.1

4.1

5.8

13.6

30.5

2-Channel CO2 Retrieval

70 km

80 km

90 km

100 km

110 km

125 km*

CO2 VMR prec. (%)

1

1

1

1

2

2

CO2 VMR systematic (%)

15

15

12

21

32

22

RSS Syst. & Prec.

CO2VMR (%)

15

15

12

21

32

22

* CO2 VMR above 110 km is based on WACCM T, O, O(1D). The error at 125 km will become smaller as the non-LTE influence of O(1D) on the CO2 populations becomes weaker.

Figure 1 shows typical altitude-latitude cross-sections of Tk differences (2-channel Tk minus operational Tk) for the month of May as an example. Figure 2 shows a comparison of the mean SABER and ACE temperature profiles and their differences (operational Tk minus 2-channel Tk) for the period 2004 to 2012. The top panel in Figure 2 is for March to September and the lower panel is for January to June. Figure 3 shows the zonal distribution of the 2-channel retrieved CO2 to WACCM VMR point-by-point ratio for a typical month of May.






References:

Garcıa-Comas, M., M. Lopez-Puertas, B.T. Marshall, P. P. Wintersteiner, B. Funke,1 D. Bermejo, Pantaleón, C. J. Mertens, E. E. Remsberg, L. L. Gordley, M.G. Mlynczak, and J.M. Russell III, Errors in SABER kinetic temperature caused by non-LTE model parameters, VOL. 113, D24, doi: 10.1029/2008JD010105, 2008.
Link to Article

Remsberg, E. E., B. T. Marshall, M. Garcia-Comas, D. Krueger, G. S. Lingenfelser, J. Martin-Torres, M. G. Mlynczak, J. M. Russell III, A. K. Smith, Y. Zhao, C. Brown, L. L. Gordley, M. J. Lopez-Gonzalez, M. Lopez-Puertas, C.-Y. She, M. J. Taylor, and R. E. Thompson , Assessment of the quality of the Version 1.07 temperatureversus-pressure profiles of the middle atmosphere from TIMED/SABER, JGR, VOL. 113, D17101, doi:10.1029/2008JD010013, 2008.
Link to Article

Rezac, L., A. Kutepov, J. M. Russell III, A. G. Feofilov, J. Yue, and R. A. Goldberg, Simultaneous retrieval of T(p) and CO2 VMR from two-channel non-LTE limb radiances and application to daytime SABER/TIMED measurements, J. Atm. And Solar-Terrestrial Physics, 130-131(2015), 23–42, May, 2015(a).
Link to Article

Rezac, L., Y. Jian, J. Yue, J. M. Russell III, A. Kutepov, R. Garcia, K. Walker, P. Bernath, The global distribution of CO2 volume mixing ratio in the mesosphere and lower thermosphere from SABER,
In Review, J. Geophys. Res., July, 2015(b).




Home Overview Data Services Publications What's New Operations Documentation Team