Table of contents

An Absolute Radiation Detector (ARD) has been designed and constructed. After evaluation it will replace the existing UK primary national standard cryogenic radiometer, with an improved uncertainty. The ARD has been designed to measure black-body and laser radiation with an uncertainty approaching 10^-5. From these measurements it will be possible to determine the fundamental constant, the Stefan-Boltzmann constant, confirming the radiometer as an absolute detector, and link this determination to the SI unit of luminous intensity, the candela, and other detector-based radiometric units. Thus source- and detector-based optical units will be tied to an invariant physical quantity ensuring their long-term stability. This paper describes some of the design considerations to enable the ARD to meet its objectives.

Photometric units in Australia are currently derived from power measurements made with a roomtemperature electrical-substitution radiometer. This basis has been transferred to a cryogenic radiometer. The cryogenic radiometer was used to calibrate the absolute responsivity of a four-element, planar, transmission trap detector (measuring the transmission loss each usage), at laser wavelengths. Modelling of the quantum defect then provided a reference for absolute spectral responsivity throughout the photometric wavelength range. This was used to measure the responsivity of a number of photometers at the mercury green line and to calculate correction factors for CIE Illuminant A. The candela so derived essentially agrees with our existing unit, maintained on a set of lamps. The trap detector was also used to determine the polarization properties of the monochromator.

A single-element, windowless InGaAs photodetector has been calibrated against a cryogenic radiometer using a diode-laser source of wavelength 1296.9 nm (standard uncertainty u = 0.4 nm). This source differs from the gas lasers usually used with cryogenic radiometers in that the beam has an elliptical profile, is invisible to the naked eye, is not polarized, and is less spatially coherent. We describe the simple adaptations we have made to use this source with the cryogenic radiometer. In particular, we use a fibre-optic delivery system to prepare the beam, and an anti-reflection-coated window to admit the radiation. The relative uncertainty in this method of calibration is estimated to be 0.08% (1σ). The measured response (0.9752 A/W) differs from that obtained from the existing laboratory scale by 0.08%.

This paper presents the first results of parametric amplification obtained at the Istituto Elettrotecnico Nazionale G. Ferraris (IEN) at a wavelength of 633.5 nm in a crystal of lithium iodate, pumped by an argon laser at a wavelength of 351 nm and injected by a diode laser at 788.2 nm in the noncollinear phase-matching case. Amplified and spontaneous parametric signals around 633 nm were measured in a photon-counting setup, at different powers of both the argon and the diode laser. This experiment is preliminary to the absolute radiance measurement of a thermal source.

A standard of illuminance responsivity has been realized on a set of photopically-corrected detectors. It is based on the use of a cryogenic radiometer for the measurement of radiant power in terms of electrical standards and on a spectral responsivity scale established on silicon photodiodes in a "trap" configuration to evaluate the photopic response of the detectors. The spectral responsivity of the photometers was obtained from the trap detector by comparison using a spectrophotometer beam which underfilled both. The spatial responsivity of the photometers was found to be non-uniform by up to 1 % and this is a significant source of uncertainty in the realization. The measurements of aperture area used visual identification of the edges and were corrected for a bias in the visual technique that was demonstrated by checks with a similar-sized ring gauge robust enough for contact measurement.

A new design for a filter radiometer based on a trap detector, a set of temperature-controlled filters, and a precision aperture is described. This filter radiometer can be used to realize high-accuracy scales for illuminance, luminous intensity, and spectral irradiance. The new filter radiometer is an improved version of our earlier designs. It has been improved in such a way that the filter can be easily and reliably changed. This makes it more suitable for spectral irradiance measurements, where lamps usually have to be measured at several wavelengths. The results of the characterization of the filter radiometer equipped with a lambda filter using two different methods are presented. The results are in agreement at the level of 0.3%.

In a new facility, the use of thermal radiators such as FEL lamps and very-high-temperature black bodies as standards of spectral irradiance has been extended for the first time down to 200 nm, thus covering the whole air-ultraviolet range. This improvement is based on a significant increase of the throughput and signal-to-noise ratio of a spectroradiometer and on a large reduction in the effects from stray light within this new calibration facility. The reliability of this facility is demonstrated in a comparison between calibrations of spectral irradiance using a black body with different radiometric temperatures, and in a comparison of the calibration of an FEL lamp and a deuterium lamp with the existing scale. The uncertainties achieved are as low as or better than those of the standard facility in this spectral range.

Using a new, high-temperature black-body radiator developed by the All-Russian Institute for Optophysical Measurements (VNIIOFI), Moscow, and applying newly developed methods and equipment, the black-body-based spectral irradiance scale of the Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, has been extended and its uncertainties have been reduced compared with previous results. The pyrolytic-graphite Planckian radiator BB3200pg, which can be operated at temperatures up to over 3300 K, is presented and characterized along with modifications to optimize uniformity and stability. The determination of the radiometric black-body temperatures is already well established by measuring the weighted irradiance using filter detectors without imaging systems. This method is compared and verified with results based on self-consistent spectral irradiance measurements using fast-scanning spectroradiometers. In addition, a laser-based method for the determination of the radiometric temperature is presented and compared with conventional procedures of radiationtemperature measurement. The implementation of this additional method into the radiometric chain of the PTB is illustrated. Improvements in the uncertainty budget of the spectral irradiance scale are discussed.

This paper describes an experimental setup, based on an optical-fibre system, to calibrate infrared photodiodes at 1300 nm and 1550 nm by comparison with a cryogenic radiometer. An integrating-sphere radiometer used as a transfer standard for optical-fibre power measurements has been characterized and calibrated, obtaining an uncertainty of less than 0.1% at these wavelengths.

The set of cryogenic radiometers in the Physikalisch-Technische Bundesanstalt (PTB) at Braunschweig was recently completed by the addition of a modified LaseRad II radiometer. The advantages of the improved facility for measurement of optical radiant power are described. The paper outlines the radiometric calibration chain of the PTB, including a cryogenic laser and broadband radiometer. The link of radiometric units to the International System of Units (SI) volt and ohm in the PTB is presented. Results of current direct international comparisons of cryogenic radiometers and an internal comparison are summarized. The results demonstrate the degree of international equivalence at the lowest level of uncertainty available. The international agreement is within 2 × 10^-4.

A series of black-body radiation-temperature measurements has been made over the temperature range 1380 K to 3100 K using two different designs of pyrolytic-graphite black bodies with calculated emissivities of 0.999. All measurements were performed at the All-Russian Research Institute for Optophysical Measurements (VNIIOFI) during June 1997. A filter photometer from the VNIIOFI, broadband glass-filter detectors from the Physikalisch-Technische Bundesanstalt (PTB) and narrowband interference-filter-based radiometers from the National Physical Laboratory (NPL) were used to perform the temperature measurements in either radiance or irradiance mode using both black bodies. Across the entire temperature range, the NPL and PTB instruments showed consistent results with both black bodies and differing geometrical arrangements. Results from the VNIIOFI photometer were also broadly consistent for a wide temperature range.

To implement the detector-based radiometric scale in the new Medium Background Infrared (MBIR) facility at the National Institute of Standards and Technology (NIST), we have developed an electrical-substitution cavity radiometer that can be operated just above liquid-nitrogen temperature. This MBIR active cavity radiometer (ACR) utilizes a temperature-controlled receiver cone and an independently temperature-controlled heat sink. Being a thermal-type detector, low noise and drift of the radiometer signal depends mainly on low-noise temperature control of the receiver and heat sink. Using high critical-temperature (T_c) superconducting thin-film temperature sensors in the active control loops, we have achieved closed-loop temperature controllability of better than 10 µK at 89 K for a receiver having an open-loop thermal time constant of about 75 s. For a flux level of 1 µW to 10 µW, the rms noise floor over a measurement cycle time is below 20 nW. This is the lowest noise level yet reported for a liquid-nitrogen-cooled electrical-substitution radiometer, and it is the first demonstration of the use of high-T_c superconductors in such a radiometer. Potential uses for this ACR in the MBIR facility include absolute measurement of the broadband radiance of large-area 300 K cryogenic black-body sources, and absolute measurement of the spectral radiance of laser-illuminated integrating spheres for improved spectral responsivity measurements of infrared transfer standard radiometers.

New experimental configurations of a system to measure absolute infrared (IR) spectral radiance using correlated photons are presented. The method has the remarkable feature that it allows spectral radiance to be measured in the IR in an intrinsically absolute manner, using only uncalibrated, visible detectors. Previous tests of this method have demonstrated its feasibility by making radiance measurements at wavelengths as long as 5 µm with relative standard uncertainties of less than 2 %. A new configuration of the previously tested LiIO₃ down-conversion crystal is proposed that allows measurements with better efficiencies and lower uncertainties. A second proposal to measure radiance at 8 µm using a new IR crystal is also discussed.

We have constructed a new ultraviolet (UV) radiometry facility at the Synchrotron Ultraviolet Radiation Facility (SURF II) at the National Institute of Standards and Technology (NIST). The facility combines a high-throughput normal-incidence monochromator with an absolute cryogenic radiometer (ACR) optimized for UV measurements to provide absolute detector-based radiometric calibrations in the spectral range 125 nm to approximately 320 nm. The system can be configured for spectroradiometric calibration of photodetectors with completely automated mapping of detector spatial and angular responsivity. The facility can easily be adapted for other spectroradiometric measurements including transmittance and reflectance and the spectral range can be extended to about 50 nm.

The National Institute of Standards and Technology (NIST) has developed an infrared (IR) radiant power response scale for the spectral range from 2 µm to 20 µm. The scale has been realized on a cryogenic bolometer. The response of this bolometer has been determined by multiple ties to a primary standard, the NIST High Accuracy Cryogenic Radiometer (HACR). These transfers were made using three intermediate detectors over three different spectral regions. In addition, an independent method was used to determine the relative response of the bolometer over the entire spectral range. An IR spectral response scale over the entire 2 µm to 20 µm range was determined from these measurements, with the interlocking nature of the multiple independent measurements providing improved overall confidence and reduced uncertainties. The overall scale combined standard uncertainty for the entire spectral range of 2 µm to 20 µm is of the order of 0.25 %, with the typical combined standard uncertainty of an individual measurement made with the bolometer being ~0.8 %.

The National Institute of Standards and Technology (NIST) candela is maintained via standard photometers which have shown long-term stability of better than 0.1 % per year. The detector-based method has allowed us to reduce the uncertainties of calibrations and to expand the range of calibration capabilities. There have been several new developments in photometry at the NIST. A high-illuminance calibration facility has been developed. Four temperature-controlled standard photometers, tested for linearity and thermal effects at high illuminances, are used with a high-pressure xenon arc source to provide the illuminance scale at levels up to 100 klx. A flashing-light photometric unit (lx s) has been realized using four standard photometers equipped with current integrators, and using two independent methods. Calibration services for illuminance meters at high illuminance levels and flashing-light photometers are now available at the NIST. Ongoing new projects include the development of a detector-based luminous-flux calibration facility using a 2.5 m integrating sphere, realization of the total spectral radiant-flux scale, and the development of a reference spectroradiometer for colorimetry of displays. Calibration services are planned in these areas of photometry and colorimetry in the near future.

The Facility for Advanced Radiometric Calibrations (FARCAL) at the National Institute of Standards and Technology (NIST) in Gaithersburg, MD, comprises the Medium Background Infrared (MBIR) facility and the Training for Calibration Expertise in Radiometry (TraCER), the latter being a classroom setting. The MBIR is a 120 cm by 180 cm vacuum chamber with an internal cold shroud and an integral roll-out table. The chamber will contain a black-body source and an absolute cryogenic radiometer (ACR). Both components are mounted on the roll-out table. Measurements will be performed in the MBIR facility that cannot be accomplished using existing facilities, including the NIST Low Background Infrared (LBIR) facility and the room-temperature non-vacuum radiometric facilities. Initially the MBIR facility will be used for 80 K medium background measurements and 300 K ambient background measurements.

Silicon n-on-p photodiodes with 100 % internal quantum efficiency have been studied in the 160 nm to 254 nm spectral range. Preliminary values have been determined for the quantum yield of silicon at these wavelengths. Using these values, a trap detector is presented for absolute flux measurement in this region. The stability under intense 193 nm irradiation, a property of importance in lithography and in photorefractive keratectomy, has been measured, and the diodes tested were found to be several orders of magnitude more stable than p-on-n diodes tested by other investigators at this wavelength. Spatial nonuniformities of the n-on-p diodes were found to be less than 1 % at wavelengths of 254 nm and 161 nm.

A bilateral comparison of photometric units maintained at the All-Russian Research Institute for Optophysical Measurements (VNIIOFI) and at the Physikalisch-Technische Bundesanstalt (PTB) was carried out within a period of less than one year. The main goal of the comparisons was to obtain updated knowledge of the relationship between the magnitudes of the units of luminous intensity, luminous flux and photometric responsivity as maintained at the two national metrological institutes. The results of these comparisons show good agreement well within the stated measurement uncertainties.

A mode-locked titanium sapphire laser, tuneable over the region from 680 nm to 1000 nm, has been used to generate monochromatic laser radiation over the region from 250 nm to 450 nm by second and third harmonic generation, with power outputs of between 50 mW and 200 mW above 350 nm and around 30 mW between 250 nm and 300 nm. The (pulsed) radiation was assessed as follows for its suitability as a source for the calibration of detectors in continuous-wave (cw) applications. (i) Radiation at 356.41 nm was used to calibrate a trap detector for spectral responsivity against the National Physical Laboratory (NPL) cryogenic radiometer, and this value was compared with the one obtained using a cw laser at the same wavelength: the values differed by 0.06%. (ii) The linearity of response of the trap detector was determined at 362.0 nm. It was found to be linear to within 0.01% for powers up to 1.5 mW. The laser was then used to calibrate a trap detector for spectral responsivity against the cryogenic radiometer over the region from 350 nm to 360 nm, with an estimated standard uncertainty of less than 0.1%.

Improving the accuracy of the measurement of ultraviolet (UV) radiation is one of the most challenging problems facing radiometrists today. The difficulties arise in part from the stability and performance of the optical components required to make the measurements, such as sources, detectors, filters and diffusers, as well as from the relative intensity of the UV radiation compared with the visible background. A particular application of note is the need to measure changes in the level of terrestrial solar UVB radiation. This is required to assess the effects of ozone depletion and the potentially damaging impact on the biological ecosystem. To have a meaningful global measurement programme, uncertainties of around 1 % to 2 % are required. However, this is less than or close to the quoted uncertainties of primary UV spectral irradiance scales maintained by national standards laboratories. This paper describes work being undertaken at the National Physical Laboratory (NPL) to improve the measurement of UV radiation, in terms of establishing both primary spectral emission scales and improved transfer standards. Most of this work is centred around improving the design and characterization of filter radiometers. The paper presents details on the performance of the components used to construct the filter radiometers, as well as that of the complete devices. Calibration of the complete filter radiometers has been carried out using radiation from both monochromators and tunable lasers, and a comparison of the results obtained by the two methods is presented.

Using spectrally dispersed synchrotron radiation of continuously tuneable wavelength as delivered by the ultraviolet (UV) and vacuum ultraviolet (VUV) calibration facility of the Physikalisch-Technische Bundesanstalt (PTB) at the electron storage ring BESSY I in Berlin, various types of silicon photodiode have been examined for their radiometric performance in the 120 nm to 600 nm spectral range. Their absolute spectral responsivity was determined with a typical relative uncertainty of 0.7 % using the synchrotron-radiation cryogenic electricalsubstitution radiometer, SYRES, as primary detector standard. Particular emphasis has been given to the study of radiation-damage effects at wavelengths below 250 nm. In addition, the reflectance of the photodiodes was measured to determine their internal quantum efficiency. Using a physical model for the internal losses, the mean energy to create an electron-hole pair in silicon was derived.

A facility for the monochromator-based calibration of transfer standard detectors using a cryogenic radiometer has been set up at the National Research Council of Canada (NRC). A brief status report is given on this facility, which has undergone some modifications since it was first set up in 1994. Measurements so far have concentrated on the spectral range 450 nm to 700 nm. The wavelength calibration of the monochromator used with this facility is discussed in detail. Two types of transfer radiometer have been developed for use with the cryogenic radiometer. The original radiometers use a single photodiode and incorporate a quartz window. The second type of radiometer is a windowless three-detector trap using special 18 mm properties of these two types of transfer radiometer are compared: spectral responsivity, uniformity, polarization sensitivity, and linearity. Results of calibrations of these radiometers over one to two years are presented. These measurements suggest that the long-term reproducibility of the calibrations is approximately 0.025% for a single radiometer, but close to 0.01% for a group of radiometers. The transfer radiometers can be used to calibrate accurately other detectors using auxiliary apparatus. Measurements are presented which suggest that the transfer uncertainty is close to 0.01% on average. The accuracy of the cryogenic-radiometer-based calibrations can be estimated by linking them to the international comparison of spectral-responsivity scales carried out at the Bureau International des Poids et Mesures in 1994 under the auspices of the Consultative Committee for Photometry and Radiometry (CCPR). From this comparison, it is estimated that the overall accuracy of the cryogenic radiometer calibrations is better than 0.05% in the range 450 nm to 700 nm.

We have developed a direct optical method for aperture area measurement, based on constant, known irradiance over the aperture. The method is described and results of test measurements are given, including wavelength dependence of the effective aperture area. Systematic uncertainty components are studied experimentally to reach a relative uncertainty of 10^-4 for aperture sizes down to 3 mm in diameter. The most important advantages of the direct optical method are that it does not require any reference aperture, it can be applied to apertures of any shape, and the calibration setup is similar to the actual use of the apertures in filter radiometers.

National Institute of Standards and Technology (NIST)-traceable deuterium arc (D₂) lamps are used in the Shuttle Solar Backscatter Ultraviolet (SSBUV) programme as radiometric standards in the 200 nm to 250 nm wavelength region. These lamps are inherently less stable than the quartz-halogen tungsten filament (FEL) lamps used in the programme for calibration at longer wavelengths. In order to monitor changes in the D₂ lamp output, the SSBUV D₂ lamp irradiance measurements have been compared with FEL lamp irradiance measurements. Techniques have been developed to reduce noise in the SSBUV measurements and to analyse the stability of the FEL lamps. The D₂ lamp irradiance drifts have been found to be as large as -13% at 200 nm relative to 250 nm over a period of about seven years. These independently-derived D₂ lamp irradiance drifts agree with the NIST measurements of the SSBUV lamps, confirming the long-term stability of the NIST working standards for lamp irradiance in the 200 nm to 250 nm wavelength region. The D₂ lamp irradiances used for the SSBUV calibrations are those reported by the NIST, corrected for the lamp drift at the time of the instrument calibration. These derived D₂ lamp irradiances not only fill long-term gaps for which no updated NIST calibrations are available, but also allow any short-term changes to be traced in situ from a NIST calibration to the user's laboratory, thus improving the accuracy of the SSBUV instrument's calibration.

The Shuttle Solar Backscatter Ultraviolet (SSBUV) programme uses 1000 W, quartz-halogen, tungsten coiled filament (FEL) lamps as the primary radiometric calibration source in the 250 nm to 405 nm wavelength region. The National Institute for Standards and Technology (NIST) recommends using the product of a fifth-order polynomial and a Planck function as a fitting function for interpolating the NIST-calibrated FEL lamp irradiances to users' wavelengths. The NIST fitting procedure can result in more than a 1% overshoot in some data sets. This problem is related to the unconstrained properties of the fifth-order polynomial. Overshooting during lamp irradiance interpolation gives rise to corresponding errors in the SSBUV calibration and solar spectral irradiance measurements. Errors of this magnitude are unacceptably large for the SSBUV, and we have therefore developed a new interpolation scheme. A new fitting procedure has been developed to replace the NIST procedure in the SSBUV calibrations. Although it uses fewer fitting parameters, the new fitting function provides a better fit than the NIST fitting function does. The overshoot problems experienced with the NIST procedure do not exist in the newly interpolated lamp irradiances. We have also investigated the properties of the NIST FEL lamp interpolation scheme in the visible and near-infrared ranges, finding that the overshoot problem experienced in the ultraviolet extends into these regions, with potential errors of the order of 0.5% to 1%.

This article gives the results of the measurements of the Institut National de Metrologie (INM) LaseRad cavity reflectance. These measurements were performed at several laser wavelengths: 325 nm, 488 nm, 514 nm, 543 nm, 633 nm and 1550 nm. The comparison of our measurements at 633 nm with those carried out by Cambridge Research & Instrumentation, Inc. (USA) five years ago shows satisfactory agreement. The discrepancy between these two measurements is within the measurement uncertainties in the range of 5 × 10^-5. The change in the reflectance with wavelength is not negligible. The reflectance varies from 2.8 × 10^-4 to 1.10 × 10^-4, according to the wavelength.

Josephson tunnel junctions have recently proved to be fairly sensitive photodetectors over a broad range of wavelengths. This paper shows the results obtained with Nb/Al-AlO_x/Nb Josephson junctions at 4.2 K irradiated via an optical fibre and laser light at λ = 633 nm and 690 nm. The responsivity of the devices is about 50 V/W and is linear over several decades. Moreover, stacks of two Josephson junctions are shown to have twice the responsivity of a single junction. A thermal model of the device is considered and the responsivity is calculated as a function of the temperature derivative of the energy gap, the thermal parameters of the junction, and the modulation frequency of the incident beam.

An experimental setup has been developed at the Istituto Elettrotecnico Nazionale G. Ferraris (IEN) for the absolute calibration of photodetectors in the photon-counting regime by means of correlated photons in parametric fluorescence. A nonlinear crystal of lithium iodate is pumped by an argon-ion laser at 351.2 nm, generating a broadband source of correlated photons from the incoming wavelength to the far-infrared. Photomultipliers and solid-state, single-photon detectors are employed, operating in the photon-counting regime with very low dark counts. Quantum efficiency is measured at 633 nm. The results show the correlated photons to be a promising technique for radiometry. The advantages of this as opposed to a conventional method at low power levels are shown.

A new, fast method for broadband secondary calibrations of photodiodes against a calibrated reference diode has been developed. It is based on a Fourier-transform spectrometer (FTS) and does not require any exchange of the light source, grating or higher-order filter. With the present set-up, it takes only 10 minutes to measure the spectral responsivity ratios of three unknown photodiodes compared with a reference photodiode from 200 nm to 1000 nm. The estimated uncertainty of 0.05% of the responsivity ratios was confirmed by independent measurements using intensity-stabilized laser radiation at twelve wavelengths between 238 nm and 1000 nm. The combination of the uncertainties of the new secondary FTS calibration method and those of the reference diode's primary calibration at the cryogenic radiometer shows that an efficient, practicable way has been found to provide customers with photodiodes as working standards of spectral responsivity for wavelengths from 1000 nm to 200 nm with uncertainties of 0.05% to 0.2% (near-infrared and visible to ultraviolet).

The ultraviolet (UV) instability of silicon photodiodes is a very complex phenomenon. The change in spectral responsivity after UV exposure can depend on the level of irradiance and total radiant exposure, duration of exposure, wavelength of the UV radiation, the type of photodiode, and the way the photodiode was stored and used before UV exposure. For a systematic investigation, the responsivity change must be determined not only at the wavelength of UV exposure but throughout the whole spectral range in which the photodiode is used. Obviously, a fast method is needed to perform corresponding measurements over a wide range of parameters. Therefore a new, fast and accurate technique based on Fourier-transform spectroscopy was used to study the UV stability of silicon photodiodes in detail. Results are presented for different types of silicon photodiodes (Hamamatsu S1337 and S5227 diodes, a novel PtSi-n-Si Schottky photodiode, and UVG and AXUV photodiodes from International Radiation Detectors).

The study reported here was carried out as part of an international comparison of cryogenic radiometers, using silicon trap detectors as transfer instruments. Nonlinearity measurements are part of a more general characterization process of these detectors, including, among other parameters, spatial uniformity, temperature coefficient, sensitivity to beam polarization, and stability. When calibrated against cryogenic radiometers, the uncertainty in their calibration is of the order of 1 part in 10⁴. The accuracy in the nonlinearity measurements should therefore be better than this value.

During an international comparison of radiation temperature measurements performed at the AllRussian Institute for Optophysical Measurements (VNIIOFI), Moscow, in June 1997 by participants from the VNIIOFI, the National Physical Laboratory (NPL), Teddington, and the Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, detailed measurements of the characteristics of two different types of high-temperature black-body source took place. Both of the black bodies consisted of a pyrolytic-graphite cavity and differed only in the design of the electrodes, which were either axial or coaxial. All investigations were carried out covering a temperature range from 1380 K to 3100 K. The electrical characteristics of the black bodies were investigated at all temperature points; and measurements of the temperature drift were performed to determine the stability of the systems in constant-current mode and in optical-feedback control mode. The uniformity of the black-body radiation field was measured in the radiance mode by using narrowband, interference-filter-based radiometers with imaging optics from the NPL and in the irradiance mode by using broadband-filter detectors from the PTB to scan the irradiated area. The results confirm the suitability of the BB3200 for both radiance and irradiance measurements at national metrological institutes.

Based on extensive investigations of almost 100 high-power quartz-halogen lamps covering thirteen different types, seven different manufacturers and different batches with a total operating time of about 27000 h, a new type of lamp (1000 W T6 modified FEL-type lamp, Osram Sylvania Inc.) has been selected and its operating conditions have been optimized. This lamp, which after seasoning is about ten times more stable than the best lamps used before, is recommended as a reference and transfer standard of spectral irradiance in the whole range between 250 nm and 3 µm. While excellent short-term and long-term stability over operating times of several hundreds of hours can be achieved, the disadvantages of the new standard lamp lie in a long seasoning time (80 h to 100 h) and a noticeable sensitivity of the seasoned lamps to shock and vibrations. Typical results of the characterization, calibration and reproducibility of the new standard lamp are presented and discussed in order to illustrate its performance and to demonstrate how to use it under optimal operation conditions.

Band filter radiometry has recently become popular. Its application for detector-based spectral-irradiance measurements requires improved characterization of standard lamps. This paper presents an approximation function for the spectral irradiance of NPL/GEC and FEL lamp data, reported in the 1990 international comparison measurements of the Consultative Committee for Photometry and Radiometry (CCPR). Characterization with a minimum number of parameters, applicable within the visible spectral range, is attempted. Since the spectral emissivity of tungsten is nearly linear in the spectral range from 400 nm to 900 nm, the irradiance of the standard lamps was approximated with three parameters, representing a linear polynomial multiplied by the Planck function. The same linear polynomial was used for each lamp of the same type, and was fitted only for temperature and intensity. The average difference between the measured and the approximated data was found to be less than 0.085 % for the NPL/GEC lamps and less than 0.095 % for the FEL lamps. This means that the relative spectral emissivity of each lamp of the same type is practically the same. The temperature fitting was characterized by the average difference between the measured and approximated data in the ultraviolet and infrared spectral ranges.

We present results of measurements obtained with a new monochromator-based cryogenic radiometer facility at NMi/VSL. General design considerations of the facility and associated transfer detectors are discussed. At present, Si detectors can be calibrated between 330 nm and 1200 nm at power levels ranging from 5 µW to 80 µW. A vacuum trap detector of the reflective type has been constructed, which accepts an optical beam in conjunction with spot sizes up to 5 mm in diameter. To check the performance of the system, the trap detector has been calibrated directly against the cryogenic radiometer using lasers at 488.0 nm, 543.4 nm and 632.8 nm and subsequently using the monochromator at the same wavelengths. The agreement is better than 0.02 %.

High-power quartz-halogen lamps have been in use as transfer standards for a long time. However, even the best lamps available are not stable enough to match the increasing requirements stemming from the progress of measuring techniques. In particular, long-term stability in the ultraviolet-B region has to be improved. In order to achieve higher stability, it is necessary to change the operating conditions from constant-current to a detector-controlled system. The new OMTec transfer-standard system, consisting of a control unit including power supply and a lamp detector unit, is described. Its main features are: long-term stability with a deviation of the spectral radiant flux from the calibration values of less than 3 × 10^-5 h^-1 during nearly 400 hours of operating time throughout the spectral range from the ultraviolet-B to the near-infrared; continuous check on stability by monitoring and viewing the measured detector photocurrents; reduction of the required seasoning time by a factor of three thus enhancing the lifetime of the lamp; storage of lamp and calibration data in the lamp detector unit for external use.

Infrared remote sensing has become an important tool for studies of Earth resources, the environment and related issues. It is now well recognized that measurements in this field have to be accurately traceable to SI radiometric units in order to be comparable world-wide and reliable in the long term - in particular if the measurements are to detect small changes of key quantities. To provide a highly stable standard source of infrared radiation, the Physikalisch-Technische Bundesanstalt (PTB), Germany, has investigated the uncertainties of its double-heatpipe large-area black body for radiometric calibrations in the wavelength range from 1 µm to 50 µm. The spectral emission of the black body can be traced either to the International Temperature Scale ITS-90 or to the radiation thermometry cryogenic radiometer, with small uncertainties of about 50 mK at a black-body temperature of about 1000 C. The total relative standard uncertainty of the black body used as an irradiance standard in the infrared is less than 10^-3 for wavelengths between 1 µm and 15 µm and at most a few times 10^-3 for wavelengths up to 50 µm. With a more sophisticated treatment of diffraction corrections and cryogenic shielding of background radiation, the uncertainties can be reduced to a few parts in 10⁴ for high-precision applications.

Originally developed for application with laser beams, trap detectors are widely used as transfer standards in modern monochromator radiometry. However, a non-optimized beam focus applied to trap detectors can result in significant spectral errors. This paper shows that a defocused and improperly aligned beam entering a Si reflection trap detector can cause an apparent modification of the measured relative spectral responsivity. This is true especially in the ultraviolet (UV) region of the spectrum, where the measured relative differences are of the order of several percent due to the significant influence of the direct band-band transitions of Si on the reflectance of the Si photodiode. The spectral range from 248 nm to 600 nm was investigated in detail. Calculated curves, simulated by a computerized geometrical model, were fitted to the measured data. The geometrical model uses experimental spectral reflectance data.

Trap detectors are useful transfer standards well-suited to modern absolute radiometry. A potential problem with them is that cryogenic radiometers allow only small-diameter laser beams to enter the absorber cavity. This can significantly increase nonlinearity effects of the trap detectors even at radiant power levels below 1 mW. The results of nonlinearity investigations at 633 nm of Si reflection trap detectors are presented. Four types of trap were studied: the commercially available QED100 and QED200, and constructed traps using Hamamatsu S1227-1010BN and S1337-1010BN photodiodes. The quantum efficiency was investigated as a function of radiant power as well as of the bias voltage and beam diameter.

This work analyses the reflectance of silicon photodiodes as a function of incident beam polarization, angles of incidence and acceptance, and thickness of the passivating layer. The contributions of these effects to the polarization sensitivity of real trap devices are considered. An analytical model is used to calculate changes in responsivity due to polarization in traps with angular misalignments and traps constructed from different types of silicon photodiode. The results show that sensitivity to polarization should be considered when high-accuracy measurements are required.

One of the major uncertainties associated with radiance and irradiance measurements stems from the aperture system and a knowledge of the aperture area. High-quality apertures, with a roundness of 0.1 µm, can now be manufactured by diamond-turning techniques. The quality of these high-precision apertures is such that their geometrical diameter can then be measured with an uncertainty of 0.05 µm using the internal-diameter measuring machine developed at the National Physical Laboratory (NPL), UK. Optical techniques have also been developed at the NPL and the World Radiation Center (WRC), Switzerland, so that the ratio of the radiometric areas of two apertures can be determined and this ratio compared with the geometrical ratio as calculated from their diameters. Results are presented of these comparisons using apertures from the NPL and the WRC.

State-of-the-art in the application of the Monte Carlo method (MCM) to the computational problems of optical radiometry is discussed. The MCM offers a universal technique for radiation transfer modelling based on the stochastic approach. Developments of the original MCM algorithms and software for calculation of effective emissivities of black bodies, absorption characteristics of cavity radiometers and photometric properties of integrating spheres are used for designing advanced optical instruments. The capabilities of the developed software are illustrated by several examples. The techniques of convergence improvement and special time-saving algorithms are outlined.

The Absolute Integrating-Sphere Method was developed and introduced by the NIST for the realization of the luminous-flux unit using an integrating sphere rather than a goniophotometer. The total luminous flux of a lamp inside the sphere is calibrated against the known amount of flux introduced into the sphere from the external source through a calibrated aperture. The key element of this method is the correction for the spatial nonuniformity of the integrating sphere, which has been made possible by a technique using a scanning-beam source. This method is to be applied directly to routine calibration measurements of luminous flux. A new luminous-flux calibration facility with a 2.5 m integrating sphere has been recently completed at the NIST and will allow the calibration of test lamps with no need for luminous-flux standard lamps. Calibrations will be performed based on the illuminance measurement of the external source by standard photometers. This method makes the luminous-flux calibration a detector-based procedure, thereby eliminating the uncertainties associated with the use of standard lamps while achieving lower uncertainties by shortening the calibration chain. The characteristics of the new integrating sphere obtained by computer simulation and measurements are reported.

A radiometric source facility is being constructed with narrowband, widely tuneable lasers covering the wavelength range 200 nm to 12 µm. The laser beam will pass into an integrating sphere and the output will be large-area, highly uniform, and monochromatic. This facility will be used to make a wide variety of radiometric measurements, such as radiant power, radiance, irradiance, transmittance and reflectance.

InSb radiometers have been developed and characterized for use as transfer standards at the National Institute of Standards and Technology (NIST). The InSb radiometers will hold a portion of the recently realized infrared spectral response scale of the NIST and will serve as working standards for the calibration of test detectors. These radiometers will measure radiant power or irradiance from 2 µm to 5.4 µm. They have a 17° field-of-view. The InSb detectors were selected for high shunt impedance to achieve high sensitivity. The transimpedance-gain, voltage-gain, and loop-gain of the current-to-voltage converters were optimized for optical radiation chopped at 40 Hz. At the peak response, the noise-equivalent-power is 0.6 pW in an electrical bandwidth of 0.3 Hz. The ratio of the dc ambient-background current to the ac noise-equivalent current is 10⁷. Spatial and angular responsivities, linearity, stability, and flashing were also characterized.

For almost a decade, the National Institute of Standards and Technology (NIST) has supplied to its customers calibrated photodiode standards and special tests of photodetectors for absolute spectral responsivity from 200 nm to 1800 nm. During this time spatial responsivity measurements have been made on several dozen Hamamatsu silicon S1337-1010BQ photodiodes. We have found that the spatial responsivity changes with wavelength, sometimes significantly as the wavelength approaches the bandgap. The most significant changes appear to be caused by defects in the photodiode material and are not apparent over most of the wavelength region where the photodiode operates. The change in spatial uniformity with wavelength can significantly contribute to measurement uncertainties. These measurements have been repeated and the spatial responsivities have remained constant over several years. Measurements have also been made on other types of Si, GaN, Ge, and InGaAs photodiodes. The measurement equipment, method, and results are presented.

The uncertainty in the measurement of aperture area can limit high-accuracy radiometric and photometric measurements. Relative total uncertainties in some measurements have now been determined at or below the 0.1% level, making substantially smaller aperture-area measurement uncertainties necessary. The National Institute of Standards and Technology (NIST) has recently implemented an absolute aperture-area measurement facility and a relative aperture-area measurement facility; the two facilities together are designed to determine aperture areas with low uncertainty. The absolute instrument measures the aperture-area using optical edge detection, along with high-precision positioning of the optical edge relative to the sensor, resulting in an expected relative combined standard uncertainty of less than 10^-4. The relative instrument uses optical flux transfer to compare aperture areas and also has an expected relative combined standard uncertainty of less than 10^-4. The absolute instrument will be used to measure the area of standard apertures for use with the relative instrument.

This paper deals with the radiative calibration aspects of high-heat-flux sensors using black-body radiation. In the last two years, several heat-flux sensors were calibrated up to 50 kW/m² using a 25 mm diameter aperture variable-temperature black body and a reference room-temperature electrical-substitution radiometer. Tests on a typical Schmidt-Boelter heat-flux sensor showed long-term repeatability of calibration is within 0.6%. Plans for extending the present calibration capability to 100 kW/m² are discussed.

A portable thermal-infrared transfer radiometer (TXR) has been developed for use in comparisons and scale verifications of sources used to calibrate thermal-infrared (TIR) channels of the National Aeronautics and Space Administration's (NASA's) Earth Observing System (EOS) flight instruments. The TXR is designed to measure the radiance temperature of large-area black-body sources in cryogenic vacuum environments, either at the National Institute of Standards and Technology (NIST) or at the EOS instrument-calibration facilities. It can be operated in ambient conditions of room temperature and pressure, or in EOS instrument thermal/vacuum chambers at temperatures as low as 77 K. The TXR is a liquid-nitrogen-cooled filter radiometer with two channels: one centred at 5 µm based on a photovoltaic InSb detector, and the other centred at 10 µm based on a photovoltaic HgCdTe (MCT) detector. The spectral, spatial and temporal characterization of the TXR using state-of-the-art NIST ambient-infrared instrumentation is reported.

We have measured the thermal conductivity of polyimide rod at temperatures from 2.4 K to 8 K, and the specific heats of polyimide rod and a black paint at 2.4 K and 4.4 K. It is shown that the responsivity and time constant of cryogenic radiometers can be significantly improved by the use of a polyimide film as the heat link. This study will help future development of highly sensitive absolute cryogenic radiometers.

This paper describes a general, stability-measurement method for the short-term stability, or repeatability, of radiometers using fixed-point black bodies as radiation sources. A germanium photodiode was used as the detector in the radiometer. The radiance during the melts and freezes was measured five times for a total of 14 h to 18 h with the radiometer for each black body. The undercooling of the freeze was used as a reference signal to establish a common temporal scale for the five data sets. The average of the standard deviations of the five measurements at each time determined the component of the standard uncertainty associated with the short-term stability. This uncertainty includes the instability of the radiometer and the black-body radiance as a function of time. Four black bodies were measured with the radiometer. The repeatability of the output signal was found to be 0.006% to 0.002% in the optical power range from about 2 nW to 2 µW. The repeatability determined by using this method may be considered as an upper limit of the short-term repeatability of the germanium radiometer.

A low-temperature, composite silicon bolometer was tested for use as a highly sensitive infrared detector in spectrometric setups. The sensitivity of the detector was calculated as a function of the cryogenic temperature, the continuous background, and chopped-signal radiation. Based on this result, a new technique was used to characterize the linearity of bolometer sensitivity. The characteristic feature of this technique is that the bolometer is exposed to superimposed beams of chopped and continuous radiation. The intensity of both beams can be varied over a wide range, closely simulating the actual operation of the bolometer. Since the effect of the changing background radiation is not distinguishable from the effect of a shift in the reference temperature of the Dewar, or of other environmental factors, they can be modelled together by continuous radiation alone. The results are given. This technique can be used for any thermal detector if both the ac and dc electrical signals corresponding to the chopped and background radiation, respectively, are available together or separately. The correction factors refer only to the actual resistance of the bolometer. These factors determined by the new method for every combination of the two electrical signals can be applied even with unstable reference temperature or environmental factors. The same setup was used to determine the frequency response of the bolometer. The results show good agreement with the calculated value of sensitivity as a function of frequency.

Filter radiometers are used as transfer standards in many radiometric measurements, and essentially comprise a detector, filter (interference or coloured glass) and high-precision aperture. An initial study into the use of Fourier-transform (FT) spectrometry has resulted in the attainment of medium accuracy (around 0.3% to 0.7%) calibration of assembled filter radiometers in a period of a few hours. This contrasts with a day for characterization using a grating instrument and the weeks required for high-resolution, high-accuracy, laser-based calibration. The technique measures the response of the radiometer to be calibrated against that of a standard detector, which in this case is a trap detector.

A new primary spectral responsivity scale has been established in the ultraviolet (UV) and near-infrared (NIR) spectral regions (200 nm to 400 nm and 0.9 µm to 2.5 µm) at an uncertainty level of less than 0.5% (2σ) over most of this range. The new scale is maintained on solid-state transfer standard detectors which were calibrated traceable to the NPL cryogenic radiometer via spectrally flat pyroelectric detectors using monochromatic radiation from both laser and monochromator sources. The UV scale was established on GaAsP detectors while the NIR scale is based on InGaAs (0.9 µm to 1.64 µm) and modified InSb/sphere detectors (1 µm to 2.5 µm). This paper discusses the choice of the new transfer standard detectors, describes how the new scale was established, and analyses the component uncertainties which contribute to the total uncertainty associated with the new scale.

This paper describes the establishment of a high-accuracy, infrared, spectral responsivity scale in the 8 µm to 12 µm atmospheric window through the calibration of specially designed HgCdTe/sphere detectors. The scale has been established using intensity-stabilized laser radiation from a grating-tuned CO₂ laser directly with the primary standard cryogenic radiometer of the UK National Physical Laboratory (NPL). The factor-of-10 improvement in the uncertainty over the NPL's previous spectral responsivity scale in this spectral region has been achieved mainly through the use of improved transfer standard detectors and the direct calibration of these detectors against the cryogenic radiometer. This paper describes these new transfer standards, the intensity-stabilized CO₂ laser facility, its use with the NPL cryogenic radiometer and the improved methodology.

Near-ambient-temperature black-body sources are routinely used for calibration in terms of radiance of a variety of infrared instruments such as those used in remote sensing and thermal imaging. The black-body radiance is usually determined by reference to a measured temperature and a calculated effective emissivity. The temperature is measured with one or more contact thermometers positioned close to the emitting black-body surface. In this case traceability to the International System of Units (SI) is to the kelvin through the ITS-90. This paper describes an alternative, more direct method based on the use of absolutely calibrated filter radiometers. These filter radiometers form part of a new facility called AMBER (Absolute Measurements of Black-body Emitted Radiance) which has been designed to determine the radiance of an ambient-temperature black body with an uncertainty of about 0.1% (which corresponds to a radiance temperature difference of 25 mK at 4 mum) and a resolution of 0.001% (0.3 mK). The facility obtains its traceability to the SI directly through radiometric standards in the form of a cryogenic radiometer rather than through the ITS-90.

This paper discusses the performance of a range of new detector-based transfer standards for the dissemination of source-based scales. The transfer standards presented include both detector-stabilized and -monitored sources of spectral irradiance and radiance as well as a more radical absolute transfer standard spectrometer. The primary aim in developing the new transfer standards is to improve traceability and accuracy to the International System of Units (SI) while at the same time reducing the cost. Some generic remote-sensing applications are considered as examples.

An integrating-sphere source developed by MTL Systems, Inc. was calibrated for them by a commercial standards laboratory traceable to the National Institute of Standards and Technology (NIST), resulting in values of spectral radiance in the 400 nm to 2500 nm spectral region. The spectral radiance of the MTL sphere source was then measured using three filter radiometers, also NIST-traceable, that were developed for the Earth Observing System (EOS). In the visible and near-infrared, the values from two Si-photodiode filter radiometers and the supplied calibration values agree within their mutual uncertainties, but in the short-wave infrared, the agreement between the filter-radiometer results and the calibration values is less satisfactory. The reason for this discrepancy is not understood, and additional measurements should be performed.

The Multi-angle Imaging SpectroRadiometer (MISR) is to be launched with the Earth Observing System EOS-AM1 spacecraft in 1999. Demanding specifications include a requirement that the instrument be calibrated, and placed on an accurate radiometric scale, to within 3% (1σ) uncertainty, for incident radiances near the upper end of a camera's dynamic range. Contributing factors to the uncertainty in radiance measurements include signal-to-noise ratio, goodness of fit of the calibration data to a quadratic equation, and quality of the experimental conditions (i.e. range and number of radiometric levels used to provide the calibration). These sources, as they are identified, are flagged as contributing to the absolute, relative band-to-band, relative camera-to-camera, or relative pixel-to-pixel radiance uncertainties. This paper summarizes the products and results from the uncertainty analysis of radiometric calibrations for the MISR. It also provides an example of a geophysical product retrieval that makes use of these data.

Uncertainty in the transfer of power responsivity from a cryogenic radiometer to a detector arises from several factors. These include the transmittance and scatter of the Brewster window present in the absolute power measurement and the relative positions in the laser beam of the detector and radiometer. To investigate these, the responsivity of a three-element silicon photodiode reflectance trap has been determined against a cryogenic radiometer at 488 nm. The responsivity measurement was performed in two different positions along the path of the irradiating laser beam: firstly in front of the radiometer's Brewster window, and secondly in the plane of the radiometer's cavity. The resulting responsivities are compared in terms of the significant differences between each measurement technique and the uncertainty budget of the two calibrations. In addition, the uncertainty in and feasibility of measurements made in vacuum behind the Brewster window and in the cavity plane are assessed.

The uncertainty of measurements is critical in furthering the development of radiometric science and acquiring an increased understanding of its principles. Proper analysis of measurement uncertainties provides an indication of how well the measurements have been made, and also helps assess the results of comparisons. Uncertainties are necessary to quantify statements concerning traceability to the International System of Units. Three examples in the field of radiometry are presented for evaluation. Employing statistical theorems, the combined standard uncertainty of the signal integral or the temperature is derived.

The ScaRaB instruments are a series of Earth Radiation Budget radiometers of which the first flight model (FM1) was launched on a METEOR satellite on 25 January 1994. A second model (FM2) was launched on the Russian Resurs satellite in July 1998. Results obtained from FM1 indicated the need for more accurate ground calibration for FM2 and the spare model. Besides spectral response characterization (from 0.2 µm to 50 µm), channel gains and absolute filtered radiance of the on-board lamps and on-board black bodies have been measured, using different reference sources: laboratory black bodies, an absolutely calibrated integrating sphere, and the Sun. The black bodies ensure the validity of the calibrations in the long-wave spectral domain, while the Sun and the integrating sphere check the short-wave spectral domain. After a brief description of the ScaRaB instrument and its main characteristics, this paper describes the procedures and results, including the associated uncertainties, for the measurements of the broadband channel gains: these need to be accurately calibrated and compared. Points of interest and also weaknesses of the three methods are compared in order to lead to the most realistic uncertainty budget.

The balance between the incoming solar radiation from the Sun and the outgoing reflected and scattered solar radiation and thermal-infrared emission from the Earth, the so-called Earth Radiation Budget (ERB), provides information on the fundamental energy source of the climate system [1]. To fulfil global coverage and sampling requirements, the ERB measurements have to be made from space. Broadband measurements are necessary because all spectral regions in both the solar and infrared contribute to the Earth's radiative fluxes. The Geostationary Earth Radiation Budget instrument (GERB) will be launched on the Meteosat Second Generation (MSG) geostationary satellite in the year 2000. Both short-wave (0.32 µm to 4 µm) and total (0.32 µm to 30 µm) radiance measurements will be made, with long-wave (4 µm to 30 µm) data obtained by subtraction. The accuracy requirements, 1% short-wave and 0.5% long-wave, are challenging but consistent with previous radiation-budget measurements. The instrument will be characterized and calibrated on the ground prior to launch. The radiometric calibration is performed using two black bodies and a lamp-illuminated, integrating-sphere source in the near field. In this paper we describe the calibration algorithm and procedure.

Laboratory measurements of the bidirectional reflectance distribution function (BRDF) of diffusely reflecting samples are required to support calibration in the Earth Observing System (EOS) programme of the National Aeronautics and Space Administration (NASA). To assess the ability of the instrument calibration laboratories to perform accurate BRDF measurements, a round-robin comparison with the National Institute of Standards and Technology (NIST) as the central laboratory was initiated by the EOS Project Science Office. The comparison parameters, which include measurement wavelength, spectral bandwidth, illumination and viewing geometry, sample type and alignment, and data format, were selected in consultation with the participants. The participants were selected based on their roles as metrology laboratories with direct connections to EOS or other international Earth remote-sensing satellite programmes. This paper briefly describes the format of the comparison, the status of the first round, and some preliminary results.

A series of three interagency comparisons of solar-ultraviolet-monitoring spectroradiometers was conducted at Table Mountain outside Boulder, Colorado. The primary purpose of these comparisons was to compare the solar-ultraviolet irradiances measured by all the instruments during synchronized spectral scans. In addition, selected instrument parameters important for these measurements were characterized. Experimental techniques were developed to perform all these characterizations in the field. The comparisons demonstrated that the spectral responsivity of an instrument must be determined at its field location with standard lamps calibrated from a common scale of spectral irradiance. Using such a common scale and appropriate convolution techniques, the measured irradiances from synchronized scans agreed to within 2 %.

The SOLar STellar Irradiance Comparison Experiment (SOLSTICE) is a three-channel spectrometer designed for measuring the solar ultraviolet (UV) irradiance from 119 nm to 420 nm with a spectral resolution of 0.1 nm to 0.3 nm. The SOLSTICE is aboard the NASA Upper Atmosphere Research Satellite (UARS), which was launched on 12 September 1991. The degradation of the SOLSTICE sensitivity is primarily tracked in-flight by measuring a set of bright, early-type stars with the same optics and detectors and by changing only slit sizes and integration times. While the Sun changes by 1% in the near-UV and by as much as a factor of 2 in the far-UV, early-type main-sequence stars are not expected to change by more than 1% in the UV for long time periods. The ensemble average of the SOLSTICE stellar observations indicates that these stars are indeed stable to 2% or better. Since the launch of the UARS, the SOLSTICE sensitivity has decreased by a few percent per year. We attribute the degradation primarily to ageing of the photomultiplier tubes for all three channels and to diffusion of layers in the broadband interference filters for the F and N channels. There appears only minor degradation associated with optical contamination, mainly because of the strict use of low-outgassing materials in the SOLSTICE instrument and maintenance of class 10000 clean rooms and oil-free vacuum systems for all pre-flight testing of the instrument.

We discuss a technique for validation of instrument radiometric calibration using the surface of Antarctica. The proposed method involves comparing remotely-sensed radiances with predictions from a radiative transfer model, and is applicable to polar-orbiting instruments measuring in the wavelength range 335 nm to 650 nm. Coincident ground measurements are not necessary. We have estimated uncertainties in the method using existing ground measurements and Total Ozone Mapping Spectrometers (TOMS) data. Significant sources of uncertainty include ground-measurement accuracy (1 %), seasonal and spatial surface variation (0.5 %), and radiative-transfer modelling (2 %).

A joint campaign was held at Lunar Lake Playa, Nevada (USA) in June 1996 to evaluate the accuracy of reflectance-based, vicarious calibrations of Earth Observing Systems (EOS). Four groups participated in the campaign and made independent measurements of surface reflectance and atmospheric transmittance on five different days. Each group predicted top-of-the-atmosphere radiance for several bands in the 400 nm to 2500 nm spectral range. Analysis of the data showed differences of the order of 5% to 10% throughout the spectral region under study. Further study revealed that the major sources of discrepancy are differences in procedures and assumptions in finding the reflectance of field references used to determine the surface reflectance of the test site. Differences caused by varying radiative transfer codes and aerosol assumptions were found to be a relatively small error source, owing to the high reflectance and low turbidity of the test site. Differences in the solar irradiance values used by separate groups were found to be significant, but can be overcome by agreeing on a standard solar irradiance data set. The results from this campaign were used to plan a follow-up campaign in June 1997 that included developing a set of laboratory measurements to characterize the field radiometers which measure surface reflectance, and obtaining a consistent set of reference-panel reflectance factors. The expectation is that disagreement in absolute radiances at the top of the atmosphere generated by these field methods will be reduced to less than 3% if further cooperative work between groups is carried out to develop approaches which will account better for reference panel calibration, the consistent use of atmospheric characterization and radiative transfer codes.

A new approach is being investigated to facilitate operational monitoring of post-launch satellite sensor performance. Airborne hyperspectral data are used to provide spatially extended ground-reference data applicable to the spectral bands of many sensors. Two rangeland sites in Alberta, Canada, were used for the first data acquisitions to test the method. This paper focuses on the data-acquisition program and the resulting data sets.

A National Institute of Standards and Technology (NIST)-designed radiometer is being used to monitor the spherical integrating sources (SISs) used to radiometrically calibrate the Enhanced Thematic Mapper Plus (ETM+), an Earth-imaging sensor on the Landsat-7 satellite. This six-channel Landsat Transfer Radiometer (LXR) uses witness filter samples from the first four bands of the ETM+ along with two 10 nm bandpass filters centred at 441 nm and 662 nm. The ETM+ band 1 is 450 nm to 515 nm; band 2 is 525 nm to 600 nm; band 3 is 630 nm to 690 nm and band 4 is 775 nm to 900 nm. This monitor aids in characterizing the ETM+ stability with time and also provides an alternate radiometric scale for calibration. LXR measurements, which are taken off-axis while the ETM+ views the SIS on-axis, have documented changes as large as 7% in the SIS outputs in some of the ETM+ bandpasses. The ETM+ responses to the SIS have changed comparably, demonstrating ETM+ stability to within about 1%. LXR measurements have also shown that the linearity of the ETM+ response to radiance is about an order of magnitude better than that possible without the simultaneous measurements.

An observation strategy is described that uses the Moon to monitor the radiometric stability of Earthorbiting remote-sensing instruments without requiring precise lunar-radiance measurements over the full range of lunar cycles. The Moon can be used as a constant irradiance reference for monitoring on-orbit shifts in instrument responsivity if (i) the Moon is observed at equivalent phase angles, (ii) the Moon is observed at equivalent sensorMoon libration angles, (iii) the Moon is observed at equivalent Sun-Moon libration angles, and (iv) the Moon is observed at equivalent scan rates and aspect angles. Irradiance changes due to changes in Sun-to-Moon and Moon-to-sensor distance can be precisely computed from ephemerides using inverse square scaling. A preliminary model of libration effects being developed by the United States Geological Survey from Clementine lunar-orbiter measurements was used to evaluate the effects of lunar libration in the 400 nm to 1000 nm spectral region. If the satellite platform can be rotated to view the Moon from a position in the orbit that exactly provides the required lunar phase angle, lunar-irradiance levels for 48 of the 75 lunar cycles in the 1994-1999 epoch examined can be matched to within 0.1 %. All lunar cycles in the epoch have at least one match to within 0.27 %. A pixel integration procedure that minimizes calibration uncertainties due to sensor pointing and optics contamination is presented and evaluated for the NASA Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS). The total combined relative uncertainties between suitable lunar reference pairs are expected to be less than 1 %.

We demonstrate the calibration and error estimation for three inexpensive pushbroom imaging spectrometers that operate in the visible/near-infrared range. The instruments were all built at the Naval Research Laboratory from commercial, off-the-shelf components. Although they are similar in basic design, their differences reflect the different objectives for which they were built. Each camera presents specific difficulties in calculating corrections to achieve uniformity and to convert to physical units. The calibration errors of one instrument are dominated by uncertainties in the integrating sphere, particularly in the short wavelength region. The errors of one of the other instruments, equipped with an intensifier, are dominated by camera noise.

Total column ozone (O₃) amount and an aerosol index are among the geophysical properties currently retrieved from Total Ozone Mapping Spectrometer (TOMS) measurements. The accuracy with which each is retrieved depends in part on the characterization of diffuser plates used in the pre-launch instrument calibration. We present here the retrieval sensitivities to errors in the measured reflectance per steradian of these diffuser plates. In particular, errors in the wavelength dependence of the reflectance are much more important than absolute reflectance errors. We derive limits for the uncertainty of initial diffuser measurements based on reasonable goals for retrieval uncertainties: a relative standard uncertainty of 1% in total ozone and a standard uncertainty of 0.5 in aerosol optical depth. The maximum allowed 1σ uncertainty in diffuser reflectance is 0.4% for ozone and 0.25% for optical depth, assuming that errors at different wavelengths are uncorrelated. We also examine TOMS diffuser measurements made at the Goddard Space Flight Center (GSFC) in an attempt to evaluate existing diffuser uncertainty components for O₃ and aerosol retrievals.

SUMER, the Solar Ultraviolet Measurements of Emitted Radiation instrument, is part of the payload of the Solar and Heliospheric Observatory (SOHO) which began its scientific mission in January 1996. SUMER takes spatially and spectrally highly-resolved images of the Sun in the spectral range from 46 nm to 161 nm. It investigates structures and associated dynamic processes of the solar atmosphere from the chromosphere through the transition region to the inner corona, over a temperature range from 1 × 10⁴ K to 2 × 10⁶ K. Absolute radiance measurements, which are used to derive plasma temperatures and densities in the solar atmosphere, require that the spectral responsivity of the instrument be known. We describe the radiometric pre-launch calibration of SUMER, which is traceable to the electron storage ring BESSY, and the measurements performed to confirm and extend the laboratory calibration during the first year of operation. These measurements put SUMER in a position to measure at certain wavelengths solar vacuum-ultraviolet spectral radiance, and also irradiance, with a relative uncertainty of 0.15 (1σ).

The Shuttle Solar Backscatter Ultraviolet (SSBUV) experiment flew eight Space Shuttle missions between 1989 and 1996, in order to maintain and validate the calibrations of a number of satellite-borne ozone-monitoring instruments. The SSBUV sensitivity drifted during each mission. Reversible drifts resulted from the outgassing of the optics, while non-reversible drifts were the consequence of changes induced by solar UV radiation. These changes varied from mission to mission, but were typically less than 1% near 400 nm, increasing to 5% to 10% near 200 nm. In addition to using traditional calibration data, the SSBUV solar irradiance data were used to assess radiometric sensitivity changes. In the near- and mid-UV, short-term solar irradiance variations can be accurately modelled using the Mg II index. Therefore, during the course of each relatively brief Shuttle flight, the Mg II index-corrected solar irradiance provides a ''constant candle'' and any changes in the observed solar irradiance result from sensitivity drift. Using the Sun as a source, SSBUV sensitivity drifts were modelled via an iterative approach as the product of linear functions of outgassing time and accumulated solar exposure time. The resulting time-dependent corrections were validated using the conventional calibration data and used in conjunction with the absolute radiometric sensitivity determined in the laboratory to process the solar irradiance data. It is shown that the application of this technique reduces the uncertainty in the SSBUV solar measurements associated with radiometric sensitivity drifts to approximately 1% at 200 nm.

The determination of the total solar irradiance (TSI) from the SOHO/VIRGO experiment is made by first correcting the time series for all a priori known influences, and then correcting the data for instrumental degradation, using the back-up instruments PMO6-VB and DIARAD-R. The long-term behaviour of PMO6-VA shows an exponential decrease in its sensitivity with a time constant of τ = 390 days and a 1/e amplitude of 528 × 10^-6, whereas DIARAD-L exhibits a general increase in its sensitivity of about 0.25 × 10^-6/day combined with an early exponential decrease of 405 days and an 1/e amplitude of 65 × 10^-6. After correcting PMO6- VA and DIARAD-L for their degradation, the VIRGO TSI shows a minimum around July 1996 and has been increasing since then by about 0.4 W m^-2.

The SOLSPEC instrument was built to measure the solar spectral irradiance from 200 nm to 3000 nm. The instrument participated in the three ATLAS missions. It consists of three spectrometers, which measure the solar spectral irradiance in the ultraviolet, visible and infrared domains. The ultraviolet and visible solar spectra obtained during the ATLAS 1 mission have been presented elsewhere. Measurements during the second flight are presented in this paper. Most of the ATLAS 1 results are confirmed by the ATLAS 2 measurements, but some improvements are obtained in certain wavelength domains owing to better calibration procedures and selection of measurements. The data gathered during the first two flights by the SSBUV, SUSIM and SOLSPEC spectrometers are compared and discussed.

The SOLSPEC instrument has been built to carry out absolute solar spectral irradiance measurements from space. It was first flown in December 1983 on mission STS 9 (Spacelab 1), from August 1992 to June 1993 on board the free-flying EURECA platform with mission STS 46 (deployment) and STS 57 (retrieval), then flown three times more with the ATLAS missions STS 45, 56 and 66. Further applications are foreseen in combination with the International Space Station Alpha (ISSA). SOLSPEC covers a wavelength range from 180 nm to 3000 nm split into three separate spectrometer channels (ultraviolet, visible and infrared) and has been calibrated pre- and post-flight using a black body at the Landessternwarte Heidelberg with temperatures up to 3300 K. In this paper the calibration concept is presented, together with a short instrument description and suggestions for improvements in order that higher black-body temperatures can be reached.

A portable transfer radiometer covering the spectral range 400 nm to 900 nm has been built and calibrated. This radiometer was designed to measure the output of spherical integrating sources with large apertures such as those used to calibrate spacecraft sensors for the Earth Observing System (EOS) of the National Aeronautics and Space Administration (NASA). The radiometer is a simple, robust, narrowband multifilter instrument using a silicon trap detector. The temperature of the filters, apertures, detectors and electronics is controlled slightly above ambient. There are no imaging optics and the radiometer throughput is controlled by Invar-spaced apertures. The radiometer has been calibrated by two methods. First, the radiometer was calibrated in an irradiance mode by use of a National Institute of Standards and Technology (NIST) FEL lamp. Second, a Solar-Radiation-Based Calibration (SRBC) was carried out, in which the radiometer viewed a panel, calibrated with a bidirectional reflectance distribution function (BRDF), illuminated by the Sun. Differences between the NIST calibration and the SRBC method for a recent solar spectrum are quite small: less than 2.1% for the seven bands between 412 nm and 868 nm, and well within the estimated uncertainties for the two calibration methods.

A spaceborne spectral irradiance monitor (SIM) is being developed at the Laboratory for Atmospheric and Space Physics, University of Colorado, to measure solar spectral irradiance and its variation for the NASA Earth Observing System (EOS) SOLar STellar Irradiance Comparison Experiment (SOLSTICE). The SIM consists of two independent and identical prism spectrometers. Each channel is equipped with an electrical-substitution radiometer and additional photodiodes to provide spectral coverage from 0.3 µm to 2.0 µm with a resolving power greater than 30. One of the spectrometers is used daily, while the other is used on a lower duty cycle to account for instrument degradation caused by solar exposure. The prism is rotated on a flex pivot and driven with a voice-coil motor. Closed-loop drive control and wavelength calibration are achieved by using a concave mirror on the prism table to focus a second solar beam onto a linear charge-coupled device (CCD) located at the focal plane of the prism. Radiation-induced changes in prism transmission are measured in flight by using one of the spectrometers to transfer monochromatic light to the other spectrometer. Visible and infrared photodiodes on a flip-arm measure the ratio of the beam into and out of the prism, thus measuring the prism transmission as a function of wavelength.