Limits imposed on millimeter- and submillimeter-wave limb sounders by continuum emissions
Stefan Buhler,Veronika Eyring, Tobias Wehr, and Klaus Kunzi
Institute of Remote Sensing, Faculty of Physics, University of Bremen, Bremen, Germany
Abstract. MASTER, SOPRANO, and PIRAMHYD are scenarios for future satellite borne limb sounders, investigated by the European Space Agency. They are planned to operate in the millimeter and submillime- ter spectral range, allowing the retrieval of atmospheric mixing ratio proles of several molecular species. One criterion for the scientic value of such an instrument is the altitude range where the retrieval of mixing ratios is possible. The present work is a characterization of the low altitude limits of the retrieval ranges of all three instruments. The most important limiting factor at low altitudes is the high continuum background, mainly due to water vapor, but also due to oxygen and nitrogen.
1. Introduction
The European Space Agency is investigating several scenarios for future satellite borne limb sounders in the millimeter and submillimeter spectral range. Studied in particular are the instruments MASTER, SOPRANO, and PIRAMHYD. MASTER is planned to be an in- strument in the millimeter range focusing on the upper troposphere and on exchange processes between the tro- posphere and the stratosphere, whereas SOPRANO is planned to be an instrument in the submillimeter range focusing on stratospheric chemistry. PIRAMHYD is planned to measure at even smaller wavelengths than SOPRANO, its main target species being the OH rad- ical. All three instruments are planned as passive ra- diometers, detecting the atmosphere's thermal radia- tion. Each instrument has several windows or bands, i.e., frequency ranges where spectra are recorded a list of bands can be found in Table 1.
Retrieval and detectability simulations were carried out in order to quantify the capabilities of the di er- ent instruments. The atmospheric scenario used for the simulations was the mid-latitude summer scenario of the FASCODE package Anderson et al., 1986]. The simu- lation of instrument characteristics such as scan modes, antenna patterns, receiver noise gures, band widths, and band positions was based on the requirements doc- uments for MASTER Lamarre, 1995a] and SOPRANO Lamarre, 1995b]. For the PIRAMHYD instrument such a document was not yet available, therefore the re- quirements for SOPRANO were used, adapted to higher frequencies where necessary.
2. Detection versus Retrieval, Limiting Factors
Roughly speaking, a species is detectable at a given altitude if the signal it generates is above the noise level of the spectrometer. This notion of detectability is not the same as the notion of retrievability, the latter im- plying that it is possible to distinguish this particular signal from other signals. Detectability is a necessary condition for retrievability. Moreover, detectability is the more general notion of the two, since it does not depend on the retrieval method.
What now are the factors that limit detection and retrieval at low altitudes? Due to continuum emissions the atmospheric opacity increases with decreasing alti- tude, even in the so called window regions well away from strong oxygen and water vapor spectral lines. At some altitude in the upper troposphere or lower strato- sphere the limb path gets completely opaque. The in- strument can not `see' below this altitude. This puts a physical limit on the detectability of all trace gas species. Furthermore, it can not be taken for granted that a retrieval is possible down to the lowest detec- tion altitude. Pressure broadened lines blend together and there is a large continuum background. This makes the interpretation of the spectrum more dicult than at high altitudes. Another limiting factor can be the receiver bandwidth it has to be large enough to allow the distinction of the target line from the spectral back- ground. This may require one, two, or more full widths at half maximum of the target line at the lowest re- trieval level. In the upper troposphere, bandwidths of up to 10GHz are necessary.
An important additional factor is the mixing ratio of the target species. If this is too small, the species will not be detectable, in spite of otherwise good conditions.
3. Continuum Models
Since continuum absorption causes the high opacity in the troposphere, good continuum models are nec- essary if one wants to simulate the radiative transfer at low tangent altitudes. The predominant absorber is clearly water vapor (see also Figure 1). Two di erent water vapor absorption models were compared in or- der to get an estimate of the error associated with such models. One was the model of Liebe et al. 1993], the other the model of Clough et al. 1989]. At MASTER 1
2
Figure 1.Left plot: The noise equivalent volume mix- ing ratio (NVMR) and actual VMR of N2O as observed with band B of MASTER. Right plot: The NVMR of ClO as observed with band A1 of SOPRANO the actual VMR of ClO for undisturbed and disturbed chemistry.
In each plot two di erent NVMR proles are displayed, one for the realistic case, and one for an atmosphere without water vapor. The di erence between the cases shows that indeed water vapor is the main cause for the high tropospheric opacity. The reason why the NVMR rises again at high altitudes is the limited scan range.
MASTER scans up to 20, SOPRANO up to 50km.
frequencies, between 200 and 400GHz, the two mod- els show di erences only of about 5% at SOPRANO frequencies, between 500 and 1000GHz, the discrepan- cies can reach up to 20%. At PIRAMHYD frequencies, 2.5THz and 3.5THz, the Liebe model can not be used.
For the nitrogen continuum an empirical term was used, identical to the one in the model of Liebe et al.
1993]. The oxygen continuum was omitted, because no parameterization for frequencies above 150GHz was found in the literature extrapolation of the model of Rosenkranz 1993] indicates that it is only a minor con- tribution.
4. Noise Equivalent Volume Mixing Ratio, Retrieval Method
The notion of detectability, as dened above, natu- rally leads to the denition of a noise equivalent volume mixing ratio (NVMR). It is the volume mixing ratio (VMR) that generates a signal equal to the noise equiv- alent temperature (NET) of the receiver. However, the NET depends on the width of the spectrometer chan- nels. To take this into account, an algorithm was em- ployed that selects the best possible channel width for each detection altitude. This has the e ect that broader channels are used at low altitudes, where the spectral lines are broader. The algorithm is described in detail in Buehler et al.1996]. Two examples of NVMR proles are displayed in Figure 1.
The retrieval method was optimal estimation with
Figure 2. Simulation of an N2O retrieval from band B of MASTER. Simultaneously retrieved were ozone and water vapor, both not shown. Left plot: True pro-
le, a priori prole, and retrieved prole. Horizontal bars indicate a priori error and retrieval error (smooth- ing+noise). Right plot: The averaging kernel functions for this retrieval. Each function gives the contributions of the true prole to the retrieved prole at the alti- tude indicated to the right of the peak. The sum of the functions is also displayed.
Newtonian iteration as described by Rodgers 1976, 1990].
This method was selected because it greatly facilitates the error analysis and it does not require that the for- ward model be linear. When attempting retrievals at low altitudes, the continuum background has to be taken into account. Several strategies to this end were compared Buehler et al., 1996]. The strategy that works best is simultaneous water vapor retrieval, even if no water vapor lines are within the receiver band. This strategy has the advantage that it prevents the large uncertainty in the continuum parameterization from af- fecting the trace gas retrievals. It was used for all sim- ulations.
5. Two Examples
5.1. Retrieval of N2O from MASTER Band B
An example for the capabilities of MASTER is the retrieval of N2O from band B of this instrument. The left plot in Figure 1 shows the NVMR and the actual VMR. It proves that in this case the detectability at low altitudes is determined solely by the high tropospheric opacity which leads to a sharp rise of the NVMR at an altitude of approximately 10km. The corresponding retrieval simulation is displayed in Figure 2. A com- parison of Figures 1 and 2 shows that the retrieval is possible almost down to the detection limit.
3
Figure 3. A summary of detection altitude ranges for the SOPRANO instrument for di erent atmospheric scenarios. The two horizontal lines indicate the scan tangent altitude range. Above each group of bars the band name and the molecular species are indicated also indicated is the number of spectra that have to be integrated in order to achieve this range. The integration time for a single spectrum at one tangent altitude is 0.3s.
Table 1. A Comparison of Low Altitude Detection Limits Versus Low Altitude Retrieval Limits for a Mid-Latitude Summer Atmosphere
Instrument Band Freq. Species Detection limit Caused by Retrieval limit
GHz] km] km]
MASTER A 203 H2O a opacity 4
N2O 8 opacity 8;10
B 301 O3 opacity 10
N2O 10 opacity 10
C 322 H2O opacity 6;8
D 345 CO 10 opacity 10;12
HNO3 12 VMR 12;18
SOPRANO A1 501 O3 VMR 12;14
N2O opacity 10
ClO 24/12b/c VMR 24;30/b/20;26c
BrO 18c VMR 20cd
A2 504 CH3Cl 12 opacity 12;14
H2O opacity 8
B1 626 HCl 14 VMR 16;18
B2 628 HOCl 20c VMR 24cd
C 637 CH3Cl 12 opacity 14
E 852 NO 14c VMR 26;28ce
F1 952 NO 16c VMR 26;28ce
PIRAMHYD A 2514 OH 26/c VMR 26;30/20;24c
B 3551 OH 22/c VMR 26;30/20;22c
The column `Caused by' indicates the limiting factor for the detectability. Cases where this is `opacity' are similar to the N2O example in Section 5.1, cases where this is `VMR' are similar to the ClO example in Section 5.2.
The bands not listed here are reserved for temperature retrieval. Some bands in the scenarios are meant to be alternatives for the nal instruments, so the SOPRANO bands A2/C, E/F1, and the PIRAMHYD bands A/B.
The standard altitude resolution for the retrieval was 2km.
aThe `' means: not calculated.
bFor a polar atmosphere with disturbed chemistry.
cFrom 100 integrated spectra.
dWith 4km altitude resolution.
eThis discrepancy can be explained by the special shape of the NO prole.
4
Figure 4. Simulation of a ClO retrieval from band A1 of SOPRANO. Simultaneously retrieved were H2O, O3, BrO, and N2O, all not shown. Left plot: True pro-
le, a priori prole, and retrieved prole. Horizontal bars indicate a priori error and retrieval error (smooth- ing+noise). Right plot: The averaging kernel functions for this retrieval. Each function gives the contributions of the true prole to the retrieved prole at the alti- tude indicated to the right of the peak. The sum of the functions is also displayed.
5.2. Retrieval of ClO from SOPRANO Band A1
A second example shall illustrate the capabilities of SOPRANO. The right plot in Figure 1 shows two dif- ferent VMRs of ClO, one for standard conditions, and one for disturbed polar chemistry. Also displayed is the NVMR for detection with band A1 of SOPRANO. This is a good example for the importance of the species' VMR for the detectability. Under disturbed chemistry conditions the detection is possible down to 12km, un- der normal condition only down to 24km. Tropospheric opacity, on the other hand, is not an issue for this species the VMR gets too small for detection before the region of high opacity is reached. A retrieval simu- lation for the undisturbed case is displayed in Figure 4.
6. Results and Conclusion
A summary of detection altitude ranges for the SO- PRANO instrument is given in the bar chart Figure 3.
Table 1 shows a comparative summary of low altitude detection limits versus retrieval limits for all bands of MASTER, SOPRANO, and PIRAMHYD. The de- tectability of all but one MASTER species is limited by the tropospheric opacity, as in the case of the N2O example in Section 5.1. For SOPRANO the limiting factor is more often the low mixing ratio of the target species at low altitudes, as in the case of the ClO exam- ple in Section 5.2. Retrieval of OH from PIRAMHYD
also su ers from this problem, since the mixing ratio of OH in the lower stratosphere is extremely low.
Acknowledgment. The work described here was car- ried out as part the ESTEC study `The impact of contin- uum emissions in the mm and sub-mm spectral range', ES- TEC/Contract No 10998/94/NL/CN. The nal study report is available at ESTEC Buehler et al., 1996].
References
Anderson, G., S. Clough, F. Kneizys, J. Chetwynd, and E. Shettle, AFGL atmospheric constituent pro les (0{
120km), Tech. Rep. TR-86-0110, AFGL, 1986.
Buehler, S., V. Eyring, and H. Kuellmann, The impact of continuum emissions in the mm and sub-mm spec- tral range, nal report, Tech. rep., ESTEC/Contract No 10998/94/NL/CN, 1996.
Clough, S., F. Kneizys, and R. Davies, Line shape and the water vapor continuum,Atmos. Res.,23, 229, 1989.
Lamarre, D.,MASTER Requirements, Version: 1, Revision:
1, 1995a.
Lamarre, D., SOPRANO Requirements, Version: 1, Revi- sion: 2, 1995b.
Liebe, H., G. Huord, and M. Cotton, Propagation modeling of moist air and suspended water/ice particles at frequen- cies below 1000 GHz, AGARD 52nd Specialists' Meeting of the Electromagnetic Wave Propagation Panel, Palma De Mallorca, Spain, 17-21 May.
Rodgers, C., Retrieval of atmospheric temperature and com- position from remote measurements of thermal radiation, Reviews of Geophysics and Space Physics, 14(4), 609{
624, 1976.
Rodgers, C., Characterization and error analysis of pro les retrieved from remote sounding measurements, J. Geo- phys. Res.,95(D5), 5587{5595, 1990.
Rosenkranz, P. W., Absorption of microwaves by atmo- spheric gases, in Atmospheric Remote Sensing by Mi- crowave Radiometry, edited by M. A. Janssen, chap. 2, John Wiley and Sons, Inc., 1993.
S. Buhler, V. Eyring, T. Wehr, and K. Kunzi, Insti- tute of Remote Sensing, University of Bremen / FB1, p.o. box 330440, 28334 Bremen, Germany sbuehler, vroni, tobias, kunzi@atm.physik.uni-bremen.de