Observing the Far-IR radiative properties of the Earth's atmosphere with the Tropospheric Airborne Fourier Transform Spectrometer (TAFTS): Instrument design and performance
J. E. Murray, A. A. Cańas, P.D Green, J. E. Harries, A.E. Last, P. Ade, G. Straine, A. Rochester, P. Brown, J. Pickering, M. Bolton, C. Cox,
Abstract
The development of a Far Infrared Spectrometer for use on aircraft has been undertaken. The aim is to improve the paucity of in-situ measurements in the crucial spectral region of the water vapour rotational band. The design and first results from the Tropospheric Airborne Fourier Transform Spectrometer, TAFTS, is presented here.
Keywords: HumBugs and liquorice all sorts
1. Introduction and Motivation
In order to understand how the Earth’s atmosphere regulates the climate system, radiative measurements are needed in the infra-red, IR, which dominates radiative cooling to space. The radiative emission spectral peak of the atmosphere at surface temperatures, in the tropics, is about 600 cm-1 (16 μm), this peak shifts to longer wavelengths at higher altitudes and lower temperatures. At the top of the troposphere the radiative peak is about 200 cm-1 (50 μm). Water vapour, which is the dominant greenhouse gas, is present throughout the troposphere and has an impact on the radiative balance of the atmosphere throughout the Far-IR and visible. The high oscillator strengths in the rotational band of water vapour (100-500 cm-1) is such that the atmosphere at lower levels in the troposphere is shielded by the column of water vapour above, and cooling to space from these lower levels is not significant. However, the water amount decreases significantly as we move to the mid-upper troposphere and the opacity of the atmosphere drops accordingly. The lower temperatures associated with these higher altitudes reduces the outgoing radiation that would otherwise escape the lower troposphere. Nether-the-less ~25% of radiative cooling to space occurs in the FIR, [1-4]. It is, therefore, evident that the distribution and variability of water vapour in the upper troposphere, UT, will have a significant impact on radiative cooling, but this distribution and variability is not well known and not well represented in models [5].
As well as the clear sky radiative properties of water vapour, the role of ice clouds on the FIR radiative properties of the atmosphere need to be addressed. Cirrus, predominantly UT clouds ranging from sub-visible to opaque sheets, have been shown to have a strong effect on the radiative cooling of the planet to space, [1-2,6]. Until recently, high spectral resolution infrared measurements of cirrus have covered the 10 – 130 cm-1[7] and 600 – 3000 cm-1[8-9] ranges. These have helped to improve the radiative models of cirrus and our understanding of the inconsistencies between broadband measurements and the broadband cirrus effect inferred from other spectral regions. However, although these measurements cover the majority of the outgoing longwave radiation, a significant proportion, as stated above, is contained in the poorly studied far-infrared (100-600 cm-1). The radiative properties of ice in the FIR, are poorly observed in the real atmosphere. It is known that the real refractive index of ice reaches a local maximum at ~400 cm-1 thus scattering from cirrus, which can be ignored or very simply treated in the window, is more significant in this region [10]. This region also coincides with a minimum in the imaginary component of the refractive index and hence low absorption. As there have been few measurements covering these spectral regions the community is currently reliant on un-validated modeling studies for the definition of the far infrared properties of cirrus. In-situ measurements of the cloud micro-physics and radiative properties in the FIR are crucial for validating and improving current ice cloud models. The influence of UT atmospheric humidity and clouds on the Earth’s climate system, are two of the most important uncertainties in our present understanding of how the climate system works [11]. To improve this situation, and increase the accuracy of the parameterisation with which cloud-radiation and humidity-radiation processes are described in General Circulation Models (GCM) we require accurate observations of how water vapour and clouds interact with the radiation field. Few high-resolution, clear-sky radiative cooling measurements within the troposphere, covering the 20 mm - 120 mm region, have been made. A novel radiometer, the Tropospheric Airborne Fourier Transform Spectrometer (TAFTS), has been developed at Imperial College, with funding from the UK Natural Environment Research Council, NERC. TAFTS aims to study the radiative properties of the upper troposphere. This paper gives details of the TAFTS instrument design and demonstrates its performance using data taken during a campaign in tropical Australia.
An overview of the instrument:
2. Instrument design
TAFTS is a differential, dual-input, polarising, Fourier Transform Spectrometer (FTS), of the Martin-Puplett kind [12]. This type of interferometer makes simultaneous differential measurements of spectrally resolved radiance from two input ports. We employ a sampling system that is particularly simple, and which yields diagnostic information on the vibration environment of the instrument as it affects the optical path. TAFTS covers a wide spectral range at high spectral resolution: currently it operates across two spectral bands covering 80 to 320 cm-1 and 320 to 600 cm-1 but can operate to 1000 cm-1 with the right detectors and optical filters. The instrument has a un-apodised resolution of 0.1 cm-1. The optical and detector system employed in TAFTS provides two differential signals within each spectral band. Each channel is DC-coupled allowing scene variations, during a scan, to be detected. The cryogenic optical design makes use of high-efficiency non-imaging components. The control computer employs a real-time operating system (real-time Linux). The instrument is designed to be fully autonomous or capable of remote control via the internet. Table.1 details the instrument specifications.
TAFTS is comprised of four sections. These four sections are: pointing optics (scene selection and calibration targets), interferometer (input combiner and beamsplitter, sampling laser moving mirror and associated drive), cryostat (cooled detector housing) and electronics control box are outlined in more detail below. Figure.1 shows a simple schematic of the optical layout of the instrument. Each of the two inputs can be steered to view either a hot or ambient temperature blackbody, or a sky view. The radiation from these views are steered into the interferometer housing and combined at the beam-combiner, PP1. PP2 is a second polarisor at 45o to the first and acts as our beam-splitter. L1, L2 and L3 are a combination of off-axis parabola and spherical mirrors which define the field of view and a virtual aperture cold stop (25 mm centred at the moving mirror, RT, zero path difference position). LP1 and LP2 are long-pass filters at 79 K and 4.2 K respectively. PP3 is the third polarisor, which acts as an analyser to separate the two, complementary, outputs of the interferometer [12]. Dichroic filters, DC, are used to transmit/reflect the radiation to our long-wave and short-wave detectors. These detector blocks consist of an off-axis mirror and hyperbolic concentrator, designed to reject off-axis radiation, and an integrating cavity within which is housed the photoconductor.
Interferometer / Martin-PuplettSpectral range / 80 - 600 cm-1
Spectral resolution / 0.1 cm-1 unapodised
Radiometric sensitivity / 1 mW m-2 sr-1 (cm-1)-1
Integrated sensitivity / 0.10 W m-2 (80 - 600) cm-1
Field of view / ± 0.80
Aperture diameter / 25 mm nominal
Optical input ports / 2: polarisation coded
Output signals / 4: two true differential pairs
Detectors / Ge:Ga and Si:Sb photo-conductors
Detector cooling / Liquid helium, 4.2 K
Cryostat hold time / 12 hours measured
Scan sampling / HeNe laser (2.5312 mm spatial interval)
Signal sampling rate / 4 channels, 20 kHz per channel
Digital Converters / 2 stereo S-D ADCs, 20 bit resolution
Data volume / 560 kBytes per scan, 806 Mbytes per hour
Data storage / Twin 18 Gbyte SCSI-2 hard disks
Control PC / AMPRO P5V, 266 MHz Pentium
Software/operating system / GNU C/ real time Linux
Power Consumption / < 1 kW from 220V ac, 400 Hz
Dimensions / 900 x 415 x 550 mm
Mass (main instrument/control box) / 90 kg/30 Kg
Table 1. TAFTS specifications and performance
Figure.1: Schematic of the TAFTS optical system.
2.1 Pointing Optics and blackbodies
The pointing optics box is unsealed, allowing unimpeded views of the sky and blackbody targets. No windows are used as the calibration targets are by necessity internal to the pointing optics box. Any variation in the window transmission properties cannot, therefore, be accounted for and factored into the calibration. Radiation is incident on the two input mirrors, one zenith and one nadir viewing. A balanced system of folding mirrors directs the light into the main interferometer whilst ensuring that each beam encounters the same number of reflections en-route. The pointing optics section also contains four on-board black bodies, see figure 1. Each input mirror can be steered to view the outside world, a heated black body or a near ambient temperature black body.
Each detector sees a signal from the two inputs. During a scan the interference detected, at each output from each input is in anti-phase, the recorded interferogram yielding a differential signal. The instrument can operate in two distinct modes:
(a) Fully differential, in which up-welling and down-welling radiation is observed simultaneously.
(b) Black-body referenced, in which one signal from the outside world (up-welling or down-welling) is measured against a black body (hot or ambient).
The instrument response function (the overall spectral responsivity) is determined by cross referencing internal calibration targets against each other. Each black body has two temperature sensors: one is a solid-state unit used to control built-in heaters, maintaining a reasonably constant temperature, whilst the second is a precision Platinum resistance thermometer element placed close to the emitting surface. This is used to make precise measurements of the actual temperature, ±0.5 K. The black surface is composed of a highly rugged, thermally conductive carbon-loaded epoxy compound into which have been molded pyramidal depressions to enhance its radiative properties [13]. The black body ‘effective emissivity’ is further improved using a cone shaped cavity, whose walls are machined with a saw-tooth pattern and coated with Nextel paint.
2.2 The Main Interferometer
After being steered from the pointing optics section radiation from the two inputs are combined using a polarising optic, PP1 fig.2, comprising a sheet of Mylar some 1.5 mm thick and 65 mm in diameter, onto which has been deposited a grid of copper wires of 1 mm width and 2 mm pitch, oriented in the vertical. The result of combining the two inputs using this grid is to orthogonally polarisation-code the two inputs, with respect to each other, before entering the interferometer. The polarising optics were supplied by the Astrophysics instrumentation group, Cardiff University. The beamsplitter, PP2, fig.2, is also a polarisor, whose grid is oriented at 45o to that of the beam-combining polarisor; this splits the FIR radiation into the two arms of the interferometer. Reflection in each arm is from a highly-precise back-to-back roof mirror manufactured as a monolithic block. This mirror block is moved through a distance of 25 mm using a geared-down micro-stepping motor. As the two arms are scanned differentially, and each arm is double-passed, a 4:1 optical gearing is achieved which provides the 100mm optical path scan required for 0.1 cm-1 resolution. The polarizing beam-splitting grid is deposited as a 40 mm strip along the centre of the 65 mm mylar substrate, the mylar above and below this strip is clear in the case of the combiner and semi-silvered for the beamsplitter. The semi-silvered section acts as a beam-splitter for the laser. The laser uses the same mylar substrate as the FIR beam so that vibrations in this substrate, which will cause path variation modulation, can be accounted for. However, the planarity of the beamsplitter is an issue for the laser, 633 nm. The poor planarity limits the observed laser fringes to one or two tilt fringes within the diameter of the beam. The fringe detector element is 1 mm2 and in a static environment does not cause a problem for the fringe counter. This is not a satisfactory condition and does make the sampling system more sensitive to vibrations when flying. A fringe detector and discriminator are placed within the interferometer to detect the laser interference signal. The scan takes 2 seconds to complete, giving a nominal laser fringe frequency of 80 kHz; this is divided down by a factor of 4 to generate a spatial sampling grid of about 2.5 mm pitch. For smooth and accurate travel, the roof-top mirror is housed within a mount whose travel is maintained true by three precision steel balls defining a plane, two mounted within opposing v-groves and one between a v-groove and lapped flat surface. A micro-stepping motor is geared to the roof-top block via a pulley system resulting in approximately 500,000 steps for a single scan, 50 nm/step. All scans are symmetric double-sided interferograms. The interferometer enclosure is a vacuum tight aluminium tank milled from a single billet. The input/output windows are 40 μm thick polypropylene and 40/25 mm in diameter, respectively. Polypropylene has excellent transmission in the spectral region under investigation. A vacuum tight tank is used for two reasons; to provide isolation of the system from acoustic noise (inducing microphonics in the thin film polarisors) and path dependent absorption during scans. Steering optics direct the re-combined beam too an off-axis diamond-machined paraboloid which in turn focuses the light into the cryostat optical system.
Figure.2 The interferometer optical plate: The pointing optics sit to the left, radiation is steered to the beam-combiner, P1, and then on to the beam-splitter, P2. The back-to-back rooftop mirror assembly, RT, can be seen in its carriage. The plain of travel is defined by two precision steel balls between opposing v-groves at bottom and a single ball between a v-grove and lapped flat at top. The moving carriage is held in position by the force exerted between rare earth magnets and iron bars. After recombination the beam is steer to optics bolted to the bottom of the plate and on to the cryostat. The laser is also situated on the lower level. The beam from this is steered through a beam expander, to the left, and into the interferometer.