Millimetre wave measurements have been performed at IRF between 1996 and 2000 on a campaign basis by the Institute of Meteorology and Climate Research, Forschungszentrum und Universität Karlsruhe, using the 268-280 GHz radiometer MIRA 2 [Berg, 1998]. Since January 2001, the Swedish Institute of Space Physics (Institutet för rymdfysik, IRF) operates its own ground based millimetre wave radiometer performing measurements continuously throughout the year.
The IRF radiometer has been built in collaboration with the millimetre wave group in Karlsruhe. Designed for the observation of thermal emission lines of stratospheric trace gases between 195 and 233 GHz, the instrument is designed to measure O3, ClO, CO, N2O and HNO3 in a continuous mode 24 hours a day throughout the year. Additionally, measurements of the tropospheric transmission and tropospheric water vapour columns are routinely carried out. The inversion of the measured spectra to retrieve profiles and columns of the atmospheric constituents is performed by the Karlsruhe millimetre wave group.
The radiometer at IRF has an obvious advantage in the location above the polar circle with only a short polar night. This is favourable for studying the evolution of the polar stratospheric winter chemistry both inside and outside the polar vortex and to investigate early ozone loss. With a detailed study of the potential vorticity of the polar vortex using the Equivalent Latitude Method [Nash, 1996] it is possible to identify whether measurements have been taken inside or outside the Arctic vortex. Since ground-based measurements over a longer time span like a week observe a substantial part of the inside vortex air mass, they provide some kind of vortex averaged data.
A continuous time series can be obtained due to the fact that millimetre wave observations are hardly impaired by changing weather conditions. Moreover, the Kiruna winter troposphere is usually very cold and dry. For instance, the tropospheric transmission at around 195 GHz from January to March 2002/03 was around 60% ±14%. The continuous measurements provide detailed ozone data that can be used for both, monitoring purposes, but also process and case studies.
In the microwave region, the received power from faint thermal emission lines in the atmosphere is very small. In order to detect spectral lines of this low intensity with microwave radiometers there are two observation methods available, the total power method and the reference beam method. In total power mode the radiometer receives the atmospheric signal (typically between 100 and 270 K) and compares it to a cold calibration load (usually liquid nitrogen at 77 K) and a hot calibration load at ambient temperature at around 293 K. This method has an uncertainty because of nonlinearities due to the detector performance at different temperatures. This effect is accounted for with the reference-beam method as described by Parrish . The advantage of this method is that nonlinearities of the detector as well as variations in the gain of the amplifiers are minimized and can be neglected. Instead of measuring the total power of the atmospheric signal only, the atmospheric signal is compared to a reference source at a brightness temperature similar to that of the received atmospheric signal. From the difference between the spectrum in signal and reference beam the information for the retrieval is obtained.
However, rather than deploying a second signal from the atmosphere at a higher elevation angle as reference signal (as suggested by Parrish), the IRF radiometer uses a rotatable wire grid which blends the brightness temperature of two calibration loads at high and low temperature [Krupa, 1998]. Since the receiver is only sensitive in one polarisation, the varying contribution of the hot and cold load to this particular polarisation synthesizes any brightness temperature between the hot and cold load brightness temperature. Thus, a reference signal can be chosen that matches the power level of the atmospheric signal. This provides the opportunity to choose optimal elevation angles for the atmospheric measurements even under changing weather conditions. Moreover, the reference beam is not contaminated by signatures of emitting trace gases, so the difference Signal – Reference does not subtract an unknown part of the molecule’s signature in the atmospheric signal. With the atmosphere as reference beam this effect has to be considered in the retrieval process.
The millimetre wave radiometer at IRF has been developed in collaboration with the Forschungszentrum Karlsruhe. It has been designed and built for measurements of O3, ClO, CO, HNO3 and N2O in the frequency range from 195 to 233GHz. Additionally, the tropospheric transmission and an estimate for the tropospheric water vapour column can be obtained from the retrieval.
A conventional Schottky diode mixer is deployed, cryogenically cooled to about 35 K. The receiver noise temperature in Single Side Band (SSB) mode is about 1800 K (SSB).
The mixer is pumped by a frequency multiplier (210 – 225 GHz) and, converts the incoming atmospheric signal into a first intermediate frequency of 8 GHz with a total bandwidth of about 1.2 GHz. After further downconversion to the second IF of 2.1 GHz, the signal is coupled to the spectrometer. Spectral analysis is performed by an acousto-optical spectrometer with 1024 channels providing an effective spectral resolution of about 1.2 MHz. For balanced calibration an internal adjustable reference load is used with a cold load at about 125 K and a hot load at ambient temperature (293 K). A publication with a detailed description of the system is in preparation.Since 2007 KIMRA is equipped with a Fast-Fourier-Transform spectrometer working at the first IF (8GHz). With the very high resolution of 100kHz the mesospheric CO emiission line at 230GHz can be observed simultaneously with the ozone observations.
The radiometer is equipped with a periscope-like mirror system enabling measurements in any direction above the horizon. However, it pointed northward for all measurements presented in this paper.
The measured spectra are integrated in order to reduce the noise and thus to improve the S/N ratio. The integration time depends on the tropospheric conditions and is typically about 0.5 – 2 hours for a single profile of stratospheric ozone. In situations of high tropospheric transmission even a shorter integration time of 5 to 10 min is sufficient for a successful retrieval. The varying content of water vapour in the troposphere leads to absorption of the atmospheric signal while it produces an offset in the background signal. These two effects have to be taken care of in the data retrieval. The contrast between line signal and background in the measured spectra varies between 5 K and 25 K, depending on the tropospheric conditions.
The radiative transfer model, described in Kopp , uses merged HITRAN 96 [Rothman et al.,1998] and JPL [Pickett et al., 1998] spectral data as supplied by the BErnese Atmospheric Multiple Catalog Acess Tool (Beamcat) [Feist, 2003]. Daily pressure and temperature profiles of the National Centers for Environmental Prediction, NCEP, [Kanamitsu, 1989] are merged with ground temperature data as measured at IRF (available on http://tempo.irf.se/cgi-weather/weather_page) and are then used for the forward calculations. For the retrieval a modified “Optimal Estimation Method” [Rodgers, 1976] is used.
The retrieval yields vertical profiles of the volume mixing ratio (vmr) of ozone and CO in the altitude range from about 18 to 80 km. Using the full width at half maximum (FWHM) of the averaging kernels as a criterion, a vertical resolution ofof the ozone profiles of at best 7 km (at 25 km altitude) for ozone and 15km for CO can be achieved. As an example, the left hand side of Figure 1 shows the averaged vertical resolution of the ozone profiles for 20 January 2003 as calculated from the averaging kernels. The right hand side of Figure 1 shows the sensitivity of the instrument for ozone estimated from the sums of the averaging kernels at given altitudes levels. The instrument has a sensitivity for ozone of at least 75 % in the vertical range from about 15 to 55 km. The uncertainty in the retrieved profiles due standing waves and systematic errors amounts to at least 1 ppmv [Kopp, 2000], errors due to thermal noise are mostly negligible due to the integration of the measured spectra.