2.1

WLS use cases in more detail

On-ground. During on-ground calibration the instrument will view a selection of on-ground calibration sources. In order to monitor potential changes of the instrument during the on-ground calibration period, regular monitoring measurements are done with the instrument internal sources, among which the WLS.

The calibration of the on-ground sources can in principle be transferred through the instrument to the on-board calibration sources. As the instrument is transferred from on-ground calibration conditions to in-flight operational conditions, the calibration assigned to the on-board sources can thus be transferred from ground to orbit. This assumes stability of the onboard calibration sources during this transfer (which is not the case for the radiance levels of a QTH, due to the difference of halogen convection in the lamp between 1 g and microgravity).

In-orbit. A WLS is used in-orbit to characterise instrument spectral features rapidly varying with wavelength such as detector etalon (caused by e.g., a protective silicon dioxide layer on the detector material)) and PRNU, and for relative detector electronic gain characterisation. The characterisation data are used for correction of instrument measurement data during on-ground processing.

Characterisation of (change of) spectral features is done by assuming spectral smoothness of the WLS. An appropriate polynomial is fitted to the detector (spatial and spectral) response to WLS illumination, following the low-frequency features of the instrument and the lamp. The measurements are then divided by the polynomial fit, removing the low-frequency features and leaving only high frequency features. The choice of polynomial will influence the turn-over frequency between “low” and “high”.

Relative detector electronic gain is applicable for OMI where gain switching was used during CCD read-out in order to better cover the expected dynamic range in the spectrum. WLS measurements at different gain settings are made. At a gain switch a jump in the spectrum will occur. Assuming spectral smoothness, the magnitude of the jump (i.e., the relative gain actor) can be determined from the smooth spectra adjacent to the jump. Alternatively, lamp reproducibility can be assumed and the ratio of measurements with and without gain switching will provide the relative electronic gain during switching.

Instrument monitoring. A WLS is used in-orbit to monitor charge transfer efficiency, detector pixel health, and for detector linearity checks. The monitoring data is only used for health monitoring and flagging data, it is not used for corrections of the data.

During Charge transfer Efficiency (CTE) monitoring, the efficiency of charge transfer through the CCD pixels is assessed. A known bright/dark transition will be smoothed out in the read-out direction due to a fraction of the charge lagging behind during a pixel charge transfer, resulting in a somewhat more blurred bright/dark transition than with perfect charge transfer. Long-term stability of the intrinsic sharpness of the bright/dark transition is assumed.

Detector pixel health is assessed by means of WLS measurements (effectively PRNU determination), where pixels exceeding pre-defined PRNU values are flagged as bad or dead.

Radiometric calibration. A special use case for a QTH WLS is applied in SCIAMACHY [9], [10]. The absolute radiometric calibration of the instrument as determined on-ground is transferred to the WLS, changes from ground to orbit are tracked using the WLS in combination with a correction for 1 g to microgravity, and ageing of the WLS during the mission is also modelled to maintain the absolute radiometric calibration.

2.2

Requirements on WLS

The WLS requirements of previous and current Earth observation hyperspectral imagers mentioned above can be summarised as follows:

In historic missions, the WLS has sometimes been used beyond its intended purposes. Predictability of the WLS played an important role there, e.g., the link between radiances at wavelengths over the entire spectral range of the instrument by means of the black (or grey) body curve of a QTH. Likewise, heat-up times and stabilisation times were characterised and optimised.

Future use of QTH’s in space instruments may become difficult. Since 1 September 2018 the EU has decided to stop halogen lamp production, the main reason being the energy savings when e.g., switching to LED. This has resulted in QTH producers not investing in or further supporting the production of these light sources. Next to that, QTH sources limited irradiance stability (hysteresis and ageing) and limited lifetime have been an issue in previous missions.

3.1

List of sources

Sun irradiance. The sun is often used as calibration source for earth observation instruments via the sun port. The sun is nuclear fusion powered and the final spectrum is a Fraunhofer black body spectrum. The TRL of this source is 9 / 9 (ground / flight proven). The irradiance stability of the sun is low, as there are surface dynamic effects (like flares and spots) that give random and periodic variation to the irradiance level. The radiance level via the instrument sun diffuser is close to earth radiance by design. The diffuser is however susceptible to degradation. The spectral coverage is good, but for the use cases the Fraunhofer lines are hampering the data. The need for a sun port and diffusers on the instrument introduces additional mass and volume.

The sun can be used very well for relative and absolute radiance characterisation.

Due to the Fraunhofer lines the sun is less useful for PRNU, pixel mapping and non-linearity.

Potential “new” application: Spectral calibration using Fraunhofer lines.

The spectral variability (stability) of the Sun is described extensively in literature, e.g. [11], [12]. The rotational variability of the solar spectrum ranges from about 0.5% at 280 nm to below 0.01% at long wavelengths (2600 nm). Variations are mainly caused by faculae and sunspots.

Earth radiance. Earth observation instruments measure the radiance level of the Earth nearly continuously. The data from the instrument and the knowledge of the Earth’s radiance can be used for monitoring of the instrument properties. The power source is nuclear, converted by the sun (fusion) to light (Fraunhofer black body). The light from the sun illuminates planet Earth with its atmosphere, generating a top of atmosphere final spectrum for the instrument to measure using its earth port. The spectral coverage is automatically good, but for monitoring hampered by Fraunhofer, Earth albedo, scattering, and atmospheric gas absorption lines. No additional mass or volume is needed for this source, just additional data processing.

Potential “new” application: Just as for the sun, a limited spectral calibration can be performed using the Fraunhofer spectrum.

Moon radiance. The Moon has been used as in-flight calibration source on various instruments, among them GOME, GOME-2 and SCIAMACHY, as well as geostationary instruments.

Viewing the Moon with the instrument is possible if the Moon passes through the instrument field of view, as is the case with geostationary instruments imaging the whole Earth. With scan mirrors or agile platforms, the line of sight of the instrument can be directed towards the Moon.

The spectral irradiance of the Moon has been characterised on-ground with traceability to SI by means of extensive lunar observations and correction for absorption of the atmosphere [13]. International collaboration on improving lunar models is ongoing and coordinated by GSICS (Global Space-based Inter-Calibration System, by the World Meteorological Organisation). One of the more recent publications combines timeseries of SCIAMACHY and GOME lunar measurements with lab measurements of lunar soil returned by the Apollo missions to improve lunar models and extend the lunar irradiance model from to 250 nm to 2500 nm, with a wavelength-dependent uncertainty on the order of 1% in the wavelength range between 500 and 1600 nm, and increasing outside of that range to about 5% [14].

As the Moon reflects the solar spectrum, the Fraunhofer lines in the latter will be present. This adds all the disadvantages of the solar spectrum to the lunar spectral irradiance. However, monitoring of the lunar albedo (which requires measurements of the solar irradiance) should allow comparison of the on-board diffuser and overall instrument performance with the lunar albedo, which is spectrally smooth and reproducible at the 10^-8 level per year [15].

This source can be used for characterizing and monitoring the instrument.

Potential “new” application: Spectral calibration using Fraunhofer lines and polarisation calibration. Spatial straylight measurements are also possible using the moon.

QTH lamp. The QTH lamp is being used frequently as the standard source for in orbit monitoring and characterisation of the instrument properties. Production of QTH lamps has diminished over the last few years and is expected to stop completely in the near future. Electricity as the power source is used to heat up a resistor to a high temperature (3000 K). The output spectrum is very predictable because of its black body spectral output. The TRL is 9, flight proven. Radiance stability is very good on the short- and long term. Spectral coverage is high and radiance levels are low in the UV, but excellent in all other wavelengths (VIS a bit too high). Space compatibility is high, although QTL lamps exhibit a ground to orbit effect due to gravity change. Power usage is low (up to 5W needed) and heat dissipation is high, almost all power is converted to conductive and radiative heat.

This source can be used for characterizing and monitoring the instrument.

Assuming a perfect black body of approximately 3000 K, the spectral stability of the QTH is coupled to the temperature of the black body or filament, and thus to the power source (excluding influence of the thermal environment of the lamp). Figure 1 shows the power source noise magnification factor as function of wavelength. E.g. a 1% increase in power supplied to the lamp will result in almost 5% increase in the UV and only 0.6% increase in the SWIR lamp output. The advantage of a black body is that fluctuations in one part of the spectrum are coupled to those at other wavelengths.

Figure 1.

Spectral stability of a 3000 K perfect black body. The power source noise magnification is the factor with which fluctuations in the power applied to the lamp are multiplied

Single narrowband LED. Earth observation instruments have been using narrowband LEDs for detector monitoring for decades, which is still the standard source for this purpose. The power source is electricity, which is converted by a light emitting diode (through electroluminescence) to narrow band light (typically 10 to 20 nm FWHM bandwidth). The TRL is for some centre wavelengths 9 and for other wavelengths between TRL 1 and 8. Market availability is very high for ground applications, and average and growing for flight. Radiance stability is very good, although LEDs are sensitive to temperature and current fluctuations. The radiance levels of LEDs are excellent for a wide variety of LED central wavelengths, and average to low for others. Spectral coverage when implementing only a single LED is low, due to the narrow bandwidth per single LED. The power efficiency of a LED is high in most cases and the heat dissipation needs to be taken into account. The physical size and mass of a LED and its power supply is small.

Using a single LED only a narrowband part of the spectral range of the instrument can be monitored and characterized.

New potential use: Linearity for the narrowband part of the detector.

Single white LED. Similar to the single narrowband LED, but with added phosphor (or quantum dots) for extended bandwidth. A phosphor is a substance that exhibits the phenomenon of luminescence; it emits light when exposed to some type of radiant energy. Typically, a narrowband UV LED is used to excite the phosphor. The emission spectrum of a single white LED has a bandwidth of a few 100nm. Space compatibility of the phosphor is to be investigated.

Due to the extended wavelength range of this single white LED with respect to the single narrowband LED, monitoring of instrument properties is possible with coverage of a higher number of pixels.

Potential new use: Linearity using LED power control.

Multiple narrowband LEDs. Similar to the single narrowband LED, but then implementing an elastic scatterer to combine multiple narrowband LEDs to create a broadband spectrum.

Due to the extended wavelength range of these combined narrowband LEDs, characterisation and monitoring of instrument properties is possible for the full instrument spatial and spectral range.

Potential new use: Spectral straylight by switching on and off individual single LED’s; LEDs as (reference) detectors, redundant set could measure main set; Linearity with full LED power control.

Multiple white LEDs. Similar to the single white LED, but then implementing multiple phosphors with multiple excitation narrowband LEDs to create a broadband spectrum. Space compatibility of phosphor is to be investigated.

Due to the extended wavelength range of these combined white LEDs, characterisation and monitoring of instrument properties is possible for potentially the full detector.

Potential new use: Linearity using LED power control; Limited spectral straylight monitoring when multiple phosphors are used.

White Laser (phosphor). Similar to the white LED, but then the phosphor excitation LED is replaced by a diode laser. This is done to decrease the illumination spot on the phosphor to create a bright spot. Typically being developed for automotive head lights.

Due to the extended wavelength range of the phosphor, characterisation and monitoring of instrument properties is possible for only a part of the detector pixels.

Potential new use: Linearity using LED power control

YAG rod LEDs. Similar to the white LED, but then the phosphor layer is replaced by a (doped) YAG rod. This is done to combine multiple excitation LEDs to create a bright broadband emitting surface at the ends of the rod. Typically being developed for home use and automotive head lights.

Due to the extended wavelength range of the phosphor, characterisation and monitoring of instrument properties is possible for only a part of the detector pixels.

Potential new use: Linearity using LED power control

Laser driven plasma. The laser driven light source is commonly used source for on-ground calibration due to its high colour temperature (up to 6500K), high UV output and small spot size of typically a few 100 micrometres. Electricity is converted to monochromatic light by the diode laser. The laser light is focused into a high-pressure gas bulb (typically xenon bulb) to maintain a plasma. The plasma start-up is guided by electrodes. The final spectrum is broadband black body like with superimposed gas emission lines. The TRL is 9 for on ground and 1 for flight. Market availability is very high for on ground, but 1 for flight (development needed). Radiance stability on ground is excellent with a bulb life of more than 10000 hours. Radiance levels are very high and spectral coverage is excellent. Space compatibility needs work. Power usage and dissipation of this source is high (roughly 100W for the smallest source).

Due to the excellent wavelength range and its high radiance stability, characterisation and monitoring of instrument properties is possible for all detector pixels, minus the xenon emission lines.

Potential new use: UV output is high, so better UV wavelength monitoring

Synchrotron (Microtron). High energy electrons, on the order of 100 MeV, emit synchrotron radiation (a form of Bremsstrahlung) when they are decelerated, accelerated, or deflected in a magnetic field. Synchrotrons are used as radiation sources, in particular for short-wave UV, X-ray and gamma radiation. In case only the UV part of the spectrum is needed, it suffices with lower electron energies than used for shorter wavelength radiation. E.g., PTB and NIST have access to synchrotron radiation sources for calibration purposes.

A smaller version of the synchrotron exists, usable only for electrons instead of any ion, known as a microtron. The dimensions of the microtron are much smaller than the synchrotron, on the order of one meter instead of tens to hundreds of meters, and even exist in portable versions. The main drawback of these types of sources as on-board calibration sources is the size, mass and power consumption.

Apart from its size and power consumption, synchrotron radiation sources seem quite ideal calibration sources, with well-defined spectral response and predictable radiance decay over time.

3.2

Source trade-off

The source trade-off was done by unweighted summation of the criteria scores for the source aspects and the source use cases. The scores were determined in a consistent way between the various sources, but it was in some cases difficult to properly quantify the score, mostly due to lack of firm numbers or performance details. For this reason, the QTH, Sun, Earth and Moon were also considered in the trade-off, to serve as reference for the scores attributed to new methods.

Table 1 shows the summary of the trade-off of the source aspects. The maximum number of points is 110, values above 80 points are marked in green. Table 2 shows the summary of source use cases. The maximum possible score is 70 points, values above 45 points are marked in green. Finally, Table 3 combines the totals of the previous two tables. The maximum for the combined scores is 180 points, those above 130 points are marked in green.

Table 1

Trade-off summary source aspects

Table 2

Trade-off summary source use cases

Table 3

Total scores of source aspects and source use cases combined

The QTH, multiple narrowband LEDs and multiple white (phosphor) LEDS have all columns green.

As can be seen in the total combined score table, the currently used sources are scoring very well for monitoring of the instrument characteristics. For the iWLS replacing a QTH, a combination of multiple narrowband LEDs offers the best performance in this trade-off.

4.1

Quantitative analysis for the Sentinel-5 use case

As an exercise to demonstrate the potential of the proposed iWLS, we design an implementation for Sentinel 5, used as a reference mission to calculate the performance of the iWLS. The Sentinel 5 calibration sub-assembly contains a QTH WLS, with specific performance requirements. The WLS is mainly used to derive the Pixel Response Non-Uniformity (PRNU) of the spectrometer detectors.

The wavelength ranges of the Sentinel 5 spectrometers are 270 nm to 310 nm for the UV1 spectrometer, 300 nm to 500 nm U2VIS spectrometer, 685 nm to 773 nm for the NIR spectrometer, 1590 nm to 1675 nm for the SWIR1 spectrometer, and 2305 nm to 2385 nm for the SWIR3 spectrometer.

The radiance levels the CAS is required to provide to the instrument are given in Table 4.

Table 4

Sentinel 5 CAS WLS radiance levels

Other requirements for the WLS are: The WLS should not introduce spatial features with a similar frequency as the pixel size; The peak to peak output variations as a function of angle shall be smaller than 0.05% over every combination of XX° (with XX = 8.3 degrees for UVN and XX = 16.6 degrees for SWIR) in the FoV after a subtraction of a 2nd order polynomial; The number of on-off cycles are at least once daily, plus on-ground use, plus margin (factor 2); The total burn time is 120 seconds operation + appropriate warm-up/stabilisation time, per on-off cycle; The stability is better than 1% over 120 s (drift + higher frequency effects).

All of these requirements are expected to be fulfilled with the proposed design, but attention should be given to the power supply, current control and thermal control of the LEDs.

To fulfil the wavelength range requirement and radiance levels, multiple LEDs need to be selected:

These LEDs are currently commercially available, though space qualification is likely missing. We approximate the LEDs in a box in combination with the hemispherical reflector (dome) by an integrating sphere of the size of the box. We assume the spectral reflectivity of the box, diffuser, and dome to be equal at ρ(λ) as function of wavelength λ.

The port opening of the dome is transferred to the effective port opening of the box, as any photons exiting the transmission diffuser are reflected back to the diffuser either by the box walls or the dome. We calculate the effective port fraction f of the integrating sphere as the port fraction of the dome: , where A_e is the area of the exit port of the dome, and is the area of a 4π steradian dome with radius r_d.

Since we directly view the diffuser of the box, we need to calculate the radiance of the LED box using the area of the box, A_b = 2(L_bW_b + L_bH_b + H_bW_b), where L_b,H_b, and W_b are the length, height, and width of the LED box, respectively. The dome sphere surface radiance then becomes

Φ_i(λ) is the sum of the input spectral irradiances of all the LEDs in the box.

4.2

Radiance levels using the draft implementation

We assume for simplicity’s sake a spectral distribution of the LEDs that is described by a Gaussian with centroid and width as given by the LED specification. The optical output power of the LED is equal to the integral of the spectral distribution.

Simulating the radiance spectrum of the iWLS, using a dome opening of 60 by 5 mm and aluminium coatings to calculate ρ(λ), we get the result shown in Figure 3. The asterisks connected by the black line show the requirement for Sentinel 5 WLS, the black curves show the combined contribution of the individual narrowband LEDs, and the blue curve shows the predicted output if a 5 W QTH were to be used instead of the LEDs.

The SWIR-3 LED is slightly too faint in this calculation but inserting three 2350 nm LEDs in the box would mitigate this. The maximum power of the LEDs can be tuned down to better match the requirement.

For reference, the same calculation with a 5W QTH at 3000 K filament temperature is shown in Figure 3. The QTH in this configuration has too low output in the UV, and more than 2 orders of magnitude more output than required in the NIR and SWIR.

Figure 3.

Simulated LED box + dome radiance (black curve) and requirement values (asterisks), with the addition of a 5 W QTH at 3000K (blue curve).