Temperature

CLIMCAPS temperature retrieval is described. Includes key questions, preparing for applications, and relevant product field names.

Key Questions

CLIMCAPS retrieves profiles of temperature in units [K] on a fixed vertical grid with 100 pressure levels. These are the standard pressure levels used in the stand-alone AIRS radiative transfer model, known as SARTA (Strow et al., 2003). We select a subset of channels for the CLIMCAPS temperature retrieval from the long-, mid- and short-wave infrared (IR) bands. Refer to the channel selection chapter for more details.

How can I access temperature retrievals?

CLIMCAPS temperature retrievals are part of the main Level 2 product file that is generated and archived by the NASA Goddard Earth Sciences Data and Information Services Center (GES DISC).

There are temperature retrievals at every cloud-cleared retrieval scene (~50 km at nadir and ~150 km at edge-of-scan), twice a day from each instrument ascending and descending orbit.

Can I use CLIMCAPS T(p) retrievals for studying climate trends?

We caution against using satellite sounding retrievals in calculating long-term trends without careful consideration for the influence from their a-priori and systematic sources of uncertainty in the measurements. In this chapter, we focus on the fact that T(p) retrievals depend on spectral channels primarily sensitive to CO2 emissions. This means that the retrieval of long-term trends in temperature – on the order of ~0.1 K per decade – depends on accurate knowledge of CO2 decadal patterns, which is difficult to know globally. In turn, the retrieval of CO2 mixing ratio strongly depends on knowledge of temperature at every scene (See Section 1).

In CLIMCAPS V2 we take a different approach to AIRS V7 by having a reanalysis model, MERRA-2, as the a-priori fortemperature. AIRS V7 uses a non-linear statistical regression (Milstein and Blackwell, 2016; Susskind et al., 2014) as the temperature a-priori that is calculated at run-time using all IR channels, with the effect that the a-priori at each footprint is independent of its neighbors. There is, thus, no a-priori spatial structure. MERRA-2 temperature, on the other hand, has strong spatial structure and meso-scale gradients for temperature in the troposphere and stratospheres. With a MERRA-2 as thetemperature a-priori, the CLIMCAPS temperature retrieval inherits this spatial, temporal and vertical structure from its MERRA-2 a-priori.

1. Overview

1.1 Combined IR and MW temperature retrieval

There are two chapters relevant to our discussion here; CLIMCAPS flow diagram as well as the CLIMCAPS water vapor (H2O) retrieval. Similar to H2O, CLIMCAPS retrieves temperature in two stages; first with a set of microwave-only (MW) channels (mw_air_temp) using the method discussed in (Rosenkranz, 2001, 2003) and then with a combination of microwave and IR channels (MW+IR) using the method discussed in (Smith and Barnet, 2019, 2020). The MW+IR temperature retrieval itself has two steps, as illustrated in the flow diagram. Note that only the final MW+IR retrieval step is written to the product file as air_temp.

We explain this two-step MW+IR retrieval approach in Smith and Barnet (2019, 2020) but can briefly summarize the two primary reasons. (1) The retrieval of H2O and trace gas species from an IR measurement depends on knowledge of temperature at the target scene. In CLIMCAPS, we retrieve temperature first, followed by H2O and trace gases species in the order as depicted in the flow diagram. Having an estimate of temperature that is consistent with the MW+IR measurement ensures robust trace gas retrievals. Once we can account for scene-specific trace gases, we retrieve temperature a final time. (2) The CLIMCAPS T(p) retrieval is useful for performing internal quality checks on the final product. For example, one quality check tests if the final retrieved temperature near the surface deviates significantly from an MW-only estimate of temperature. If the root mean square is > 1.5 K, then the retrieval is rejected even if all other checks were successful. This test does not change the state of the retrieved values, but instead identifies and flags failed retrievals.

The MW+IR temperature retrieval uses a subset of channels. We documented the subset of IR channels we selected for temperature in the channel selection chapter for the first and second retrieval steps. As far as the MW channels go, they vary between retrieval stages and instruments as detailed in Table 1.

Table 1: Channel selection from AMSU and ATMS for each of the two CLIMCAPS retrieval stages, microwave only (MW-only), and a combined MW and IR (MW+IR) retrieval. See CLIMCAPS flow diagram for methodology and IR channel selection chapter for the IR channel subsets.

Instrument (platform)

MW-only retrieval

MW+IR retrieval

Channel number (total number of channels)

Channel number (total number of channels)

AMSU (Aqua)

3, 6, 8, 9, 10, 11, 12, 13, 14, 15 (10)

3, 6, 8, 9, 10, 11, 12, 13, 14 (9)

ATMS (SNPP, JPSS-1)

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 (22)

3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (13)

 

The CLIMCAPS file contains a suite of diagnostic metrics with which to evaluate retrieval quality. In Figure 1, we plot the degrees of freedom (DOF) for temperature on 1 April 2016. The DOFs has a strong latitudinal dependence with highest values in the Tropics, with a maximum around ~3.6.

Figure 1: Information content as ‘degrees of freedom’ for CLIMCAPS-SNPP T(p) retrievals.

Figure 1: Information content as ‘degrees of freedom’ for CLIMCAPS-SNPP temperature retrievals (air_temp_dof) from full spectral resolution cloud cleared CrIS radiances for an ascending orbit (13h30 local overpass time) as a global equal-angle grid on 1.5° resolution, close to single footprint size in the lower latitudes at edge of scan. We did not apply any quality control filtering (air_temp_qc) since the averaging kernels (ave_kern/air_temp_ave_kern) from which DOF is derived are unaffected by the quality of the retrieval. DOF, instead, characterizes the potential a sounding system has in retrieving a target variable (Smith and Barnet, 2020).

Figure 2 and Figure 3 show the CLIMCAPS averaging kernels, temperature profile retrievals and their associated errors. We plot these for CLIMCAPS-SNPP on 1 April 2016. Figure 2 shows the mean profiles (with standard deviation error bars) for the Tropics (30°South to 30°North), and Figure 3 for the North Polar region (> 60°North). We used the diagonal vector of the averaging kernel matrix as the representation of the maximum sensitivity at each pressure level. The blue line (Figure 2 and Figure 3, left) is the mean of the diagonal vectors in each latitudinal zone, respectively.

Note how there are fewer vertical error bars on the blue line compared to the retrieval (orange line, center) and error (yellow line, right) profiles. This is because the 2-D averaging kernel matrices are written to the product file on the trapezoid pressure layers to save space. The retrieval and its error covariance matrix are, however, written out on the standard 100 pressure layers as 1-D arrays.

Figure 2: A diagnosis of CLIMCAPS-SNPP T(p) retrievals for the North Polar latitudinal zone [>60°N] on 1 April 2016.

Figure 2: A diagnosis of CLIMCAPS-SNPP temperature retrievals for the North Polar latitudinal zone [>60°N] on 1 April 2016. Each solid line represents the mean zonal profile and the error bars are the standard deviation at each pressure level. [left] CLIMCAPS temperature averaging kernel matrix diagonal vector from netCDF field ave_kern/air_temp_ave_kern that indicates the pressure levels at which CLIMCAPS has sensitivity to the true state of temperature in the atmosphere. [middle] CLIMCAPS CO profile retrieval from netCDF field air_temp [K]. [right] CLIMCAPS retrieval error from netCDF field air_temp_err [K] represented here as percentage [air_temp_err]/[air_temp]*100. CLIMCAPS uses an empirical a-priori error estimate and is represented by the thick grey line. In addition, CLIMCAPS damps T(p) by 20-25% with respect to MERRA-2 to improve the retrieval estimation of trace gases. A Bayesian Optimal Estimation retrieval system (like CLIMCAPS) typically reduces the a-priori error in all successful retrievals. In calculating these mean profiles, we filtered out all retrievals where air_temp_qc(*,i,j) ≥ 1. We plot these profiles using the pressure level array from air_pres*100 in hPa units.

Figure 2: A diagnosis of CLIMCAPS-SNPP T(p) retrievals for the Tropical zone [30°S to 30°N] on 1 April 2016.

Figure 3: Same as Figure 2 but for the Tropical zone [30°S to 30°N].

The averaging kernel diagonal vectors for the North Polar zone (Figure 3) have less information content and vertical variability than those in the Tropics (Figure 2). This is consistent with the DOF shown in Figure 1, where information content of temperature is a strong function of latitude.

1.2 MW-only T(p) retrieval

The CLIMCAPS system has a MW-only step that retrieves temperature (mw_air_temp), H2O column density (mw_h2o_vap_mol_lay), liquid water path (mw_h2o_liq_mol_lay), and surface emissivity (surf_mw_emis) using the method developed by (Rosenkranz, 2001, 2006), a sequential optimal estimation that uses a MW-only radiative transfer model as described by (Rosenkranz, 2003; Rosenkranz and Barnet, 2006). The liquid water path and surface emissivity retrieved variables are propagated into subsequent CLIMCAPS retrieval steps while temperature and H2O are written to the file as MW-only retrievals for use in research. Note that there are no corresponding averaging kernels for these MW-only retrievals.

MW-only estimates of temperature may be useful for certain applications where cloud clearing has failed due to uniform clouds or difficult surface conditions. However, the MW-only retrieval has a lower vertical resolution than the combined IR+MW CLIMCAPS retrieval, we caution against combining these in analyses without careful consideration.

2. Preparing temperature retrievals for applications

2.1 Boundary layer adjustment

CLIMCAPS temperature uses a standard 100-level pressure grid to retrieve atmospheric variables from the Earth’s surface to the top of atmosphere­­. This pressure grid is required by radiative transfer models (SARTA for CLIMCAPS) to accurately calculate top of atmosphere hyperspectral IR radiances.

CLIMCAPS uses the exact same pressure grid at every scene on Earth and accounts for surface pressure as a separate variable during radiative transfer calculations. CLIMCAPS V2 uses MERRA-2 surface pressure as input. The retrieved profiles are, however, reported on the 100-level grid as a means to standardize the output. It is important that you adjust the bottom level, i.e. that pressure level intersecting the Earth surface as identified by air_pres_nsurf in the CLIMCAPS netCDF file, to accurately reflect the temperature at the surface. We refer the reader to the procedure described in a different chapter.

2.2 Temperature a-priori

CLIMCAPS employs MERRA-2 (Gelaro et al., 2017; GMAO, 2015) as a-priori for its temperature, water vapor and ozone retrievals. CLIMCAPS converts MERRA-2 temperature profiles from their 72 pressure levels to the standard 100 retrieval levels (air_pres). Additionally, CLIMCAPS interpolates MERRA-2 profiles spatially and temporally to match the measurements. These interpolated MERRA-2 profiles are written to the CLIMCAPS product file as fg_air_temp, fg_h2o_mol_lay and fg_o3_mol_lay.

We describe the benefits of employing MERRA-2 as a-priori in Section 2.2.3 of Smith and Barnet (2019) and also explain how we derived the a-priori error covariance matrix depicted here. The reader should compare this section with the same one in the CLIMCAPS water vapor chapter that goes into more detail. Note that MERRA-2, being a reanalysis model, has spatial correlation in its temperature field, which will propagate into the CLIMCAPS temperature retrievals giving them smooth spatial gradients.

Figure 4: Empirical a-priori error covariance matrix used in CLIMCAPS V2 H2O retrievals as described in Smith and Barnet (2019).

Figure 4: Empirical a-priori error covariance matrix used in CLIMCAPS V2 H2O retrievals as described in Smith and Barnet (2019).

We depict the square root of the diagonal vector of the CLIMCAPS temperature a-priori matrix (Figure 4) as the grey profile in Figure 2 and Figure 3. Why, then, does the retrieval error profile (yellow in the same figures) have a zig-zag pattern? The simple answer is that this pattern emerges as a numerical artifact caused by our data compression methods.

Figure 5: CLIMCAPS V2 smoothing error, measurement error and retrieval error covariance matrices as described in Smith and Barnet (2019).

Figure 5: CLIMCAPS V2 smoothing error, measurement error and retrieval error covariance matrices as described in Smith and Barnet (2019).

The a-posteriori, or retrieval, error covariance matrix (Figure 5c) is the sum of the smoothing error (Figure 5a) and measurement error (Figure 5b) covariance matrices. The square root of the diagonal vector from the a-posteriori matrix (Figure 5c) is the yellow profile in Figure 2 and Figure 3.

2.3 Consider using CLIMCAPS temperature products for these applications

CLIMCAPS temperature is a useful comparison to model results and useful for examining past weather events. For example, Figure 6 shows an example of a cold air outbreak over the Eastern Unites States that is associated with expansion of the polar vortex. When the jet stream weakens, cold arctic air can migrate southwards and cause below-average surface temperatures. These “polar vortex” events can occur frequently during the boreal winter (AAAS, 2001).

Figure 6: CLIMCAPS-SNPP T(p) at 500 hPa retrievals [K] from full-spectral resolution CrIS on 18 March 2019 from the ascending SNPP orbit.

Figure 6: CLIMCAPS-SNPP temperature at 500 hPa retrievals [K] from full-spectral resolution CrIS on 18 March 2019 from the ascending SNPP orbit. We filtered out all retrievals where air_temp_qc(*,i,j) > 1, which are shown as missing values.

3. CLIMCAPS product field names relevant to temperature applications

Within the netCDF files, the following fields are relevant for temperature studies. Each CLIMCAPS file contains 45 scanlines along track (atrack) and 30 FOR along each scanline, or across track (xtrack). With temperature profiles retrieved at each FOR on 100 pressure levels (air_pres), the arrays have dimensions [atrack, xtrack, airs_pres].

3.1 Retrieved variables

  • air_temp MW+IR retrieved temperature profile.
  • mw/mw_air_temp Air temperature profile from the MW-only step.

3.2 Retrieved surface variables

  • surf_temp: MW+IR retrieved surface skin temperature.
  • mw/ mw_surf_temp: MW-only retrieved surface skin temperature.
  • surf_ir_emis: MW+IR retrieved IR surface emissivity.
  • mw/mw_surf_mw_emis: MW-only retrieved surface emissivity.
  • surf_ir_refl: Retrieved IR surface reflectivity.

3.3 Derived variables

  • surf_air_temp: Near-surface temperature as retrieved MW+IR temperature (air_temp) at surface pressure.
  • tpause_temp: Tropopause temperature as retrieved MW+IR temperature (air_temp) at tropopause height.
  • mw/mw_surf_air_temp: Near-surface air temperature as MW-only retrieved temperature (mw_air_temp) at surface pressure.
  • cld_top_temp: Cloud top temperature derived as the temperature (air_temp) at cloud top pressure (cldfrac_tot).

3.4 Quality metrics

  • Ave_kern/air_temp_ave_kern temperature: averaging kernel matrix.
  • *_dof: the trace of the averaging kernel matrix as a measure of the number of pieces of information about the methane profile provided by the physical retrieval step. Degrees of freedom indicate the number of distinct vertical levels that the algorithm has sensitivity. For T(p), this is typically 3.6 or lower.
  • *_err: Optimal-Estimation retrieval error, as the diagonal vector of the a-posterior error matrix.
  • *_qc: profile quality control metrics ranging from 0 = good, 1 = suspect, 2 = bad.

3.5 A-priori

  • fg_air_temp: Air temperature profile from MERRA-2 spatially, temporally and vertically interpolated to CLIMCAPS footprint and pressure grid (air_pres).
  • fg_surf_air_temp: MERRA-2 first guess for near-surface temperature.
  • fg_surf_temp: MERRA-2 first guess for surface skin temperature.
  • clim_surf_ir_refl: a-priori IR surface reflectivity.

4. References

  • AAAS. Stratospheric polar vortex influences winter cold, researchers say. Retrieved April 7, 2020, from http://www.eurekalert.org/pub_releases/2001-12/uoia-spv120301.php, 2001.
  • Gelaro, R., McCarty, W., Suárez, M. J., Todling, R., Molod, A., Takacs, L., Randles, C. A., Darmenov, A., Bosilovich, M. G., Reichle, R., Wargan, K., Coy, L., Cullather, R., Draper, C., Akella, S., Buchard, V., Conaty, A., da Silva, A. M., Gu, W., Kim, G.-K., Koster, R., Lucchesi, R., Merkova, D., Nielsen, J. E., Partyka, G., Pawson, S., Putman, W., Rienecker, M., Schubert, S. D., Sienkiewicz, M. and Zhao, B.: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2), Journal of Climate, 30(14), 5419–5454, doi:10.1175/JCLI-D-16-0758.1, 2017.
  • GMAO: MERRA-2 inst3_3d_asm_Nv: 3d,3-Hourly,Instantaneous,Model-Level,Assimilation,Assimilated Meteorological Fields V5.12.4, , doi:10.5067/WWQSXQ8IVFW8, 2015.
  • Rosenkranz, P. W.: Retrieval of temperature and moisture profiles from AMSU-A and AMSU-B measurements, IEEE Transactions on Geoscience and Remote Sensing, 39(11), 2429–2435, doi:10.1109/36.964979, 2001.
  • Rosenkranz, P. W.: Rapid radiative transfer model for AMSU/HSB channels, IEEE Transactions on Geoscience and Remote Sensing, 41(2), 362–368, doi:10.1109/TGRS.2002.808323, 2003.
  • Rosenkranz, P. W.: Cloud liquid-water profile retrieval algorithm and validation, Journal of Geophysical Research, 111(D9), doi:10.1029/2005JD005832, 2006.
  • Rosenkranz, P. W. and Barnet, C. D.: Microwave radiative transfer model validation, Journal of Geophysical Research, 111(D9), doi:10.1029/2005JD006008, 2006.
  • Smith, N. and Barnet, C. D.: CLIMCAPS observing capability for temperature, moisture, and trace gases from AIRS/AMSU and CrIS/ATMS, Atmos. Meas. Tech., 13, 4437–4459, https://doi.org/10.5194/amt-13-4437-2020, 2020.
  • Smith, N. and Barnet, C. D.: Uncertainty Characterization and Propagation in the Community Long-Term Infrared Microwave Combined Atmospheric Product System (CLIMCAPS), Remote Sensing, 11(10), 1227, doi:10.3390/rs11101227, 2019.
  • Wargan, K., Kramarova, N., Weir, B., Pawson, S. and Davis, S. M.: Toward a Reanalysis of Stratospheric Ozone for Trend Studies: Assimilation of the Aura Microwave Limb Sounder and Ozone Mapping and Profiler Suite Limb Profiler Data, Journal of Geophysical Research: Atmospheres, 125(4), doi:10.1029/2019JD031892, 2020.