The Calibration Home Base for Imaging Spectrometers

The Calibration Home Base (CHB) is an optical laboratory designed for the calibration of imaging spectrometers for the VNIR/SWIR wavelength range. Radiometric, spectral and geometric calibration as well as the characterization of sensor signal dependency on polarization are realized in a precise and highly automated fashion. This allows to carry out a wide range of time consuming measurements in an e cient way. The implementation of ISO 9001 standards in all procedures ensures a traceable quality of results. Spectral measurements in the wavelength range 380–1000 nm are performed to a wavelength uncertainty of ±0.1nm, while an uncertainty of ±0.2nm is reached in the wavelength range 1000–2500 nm. Geometric measurements are performed at increments of 1.7μrad across track and 7.6μrad along track. Radiometric measurements reach an absolute uncertainty of ±3% (k=1). Sensor artifacts, such as caused by stray light will be characterizable and correctable in the near future. For now, the CHB is suitable for the characterization of pushbroom sensors, spectrometers and cameras. However, it is planned to extend the CHBs capabilities in the near future such that snapshot hyperspectral imagers can be characterized as well. The calibration services of the CHB are open to third party customers from research institutes as well as industry. *Cite article as: DLR Remote Sensing Technology Institute. (2016). The Calibration Home Base for Imaging Spectrometers. Journal of large-scale research facilities, 2, A82. http://dx.doi.org/10.17815/jlsrf-2-137


Instrument Scientists:
-Johannes Brachmann, IMF, DLR Oberpfa enhofen phone: +49 8153 281427, email: Johannes.Brachmann@dlr.de-Andreas Baumgartner, IMF, DLR Oberpfa enhofen phone: +49 8153 281402, email: Andreas.Baumgartner@dlr.de-Peter Gege, IMF, DLR Oberpfa enhofen phone: +49 8153 281242, email: Peter.Gege@dlr.deAbstract: The Calibration Home Base (CHB) is an optical laboratory designed for the calibration of imaging spectrometers for the VNIR/SWIR wavelength range.Radiometric, spectral and geometric calibration as well as the characterization of sensor signal dependency on polarization are realized in a precise and highly automated fashion.This allows to carry out a wide range of time consuming measurements in an e cient way.The implementation of ISO 9001 standards in all procedures ensures a traceable quality of results.Spectral measurements in the wavelength range 380-1000 nm are performed to a wavelength uncertainty of ±0.1 nm, while an uncertainty of ±0.2 nm is reached in the wavelength range 1000-2500 nm.Geometric measurements are performed at increments of 1.7 µrad across track and 7.6 µrad along track.Radiometric measurements reach an absolute uncertainty of ±3 % (k=1).Sensor artifacts, such as caused by stray light will be characterizable and correctable in the near future.For now, the CHB is suitable for the characterization of pushbroom sensors, spectrometers and cameras.However, it is planned to extend the CHBs capabilities in the near future such that snapshot hyperspectral imagers can be characterized as well.The calibration services of the CHB are open to third party customers from research institutes as well as industry.

Introduction
The Calibration Home Base (CHB) is an optical laboratory developed and operated by the German Aerospace Center (DLR) Oberpfa enhofen for the calibration of (airborne) hyperspectral sensors and eld spectrometers.Radiometric and spectral characterization of cameras is carried out as well.The CHB is made available to public through the user service Optical Airborne Remote Sensing and Calibration Home Base (OpAiRS) (for contact details see Sec. 6).
The CHB was partly funded by the European Space Agency (ESA) to establish a calibration facility for the airborne imaging spectrometer APEX (Itten et al., 2008;Schaepman et al., 2015), but is used for other optical sensors as well.NEO HySpex VNIR-1600 and SWIR-320me sensors, which are owned by DLRs Remote Sensing Technology Institute (DLR-IMF) are regularly calibrated in the facility (Lenhard, Baumgartner, & Schwarzmaier, 2015).It is the only facility in Europe which allows a precise characterization of the radiometric, geometric and spectral properties of bulky and heavy instruments up to 500 kg (including mechanical interface) in the wide spectral range 380-2500 nm.All standard measurements in the CHB are routinely carried out and are completely automized.However, the CHB is also open to requests for special measurement campaigns, if communicated.Pushbroom spectrometers have long been a standard design for airborne and satellite imaging spectrometers.These are in the main focus of the calibration activities in the CHB.They are, in e ect, line scanners that need to be moved such that successively recorded data can be assembled to ight lines.Thus, two orthogonal directions are distinguished: Across-track data are recorded simultaneously in a direction orthogonal to the movement of the instrument.The total viewing angle in this direction is called eld of view (FOV).Along-track data is successively recorded line-by-line in direction of the sensor movement.The viewing angle of each detector element is called instantaneous eld of view (IFOV).The IFOV is similar to the along-track viewing angle.
In this paper, a short description of the CHB and its capabilities is given.A more complete description with technical details of the status of the facility as it was in 2009 can be found in Gege et al. (2009).Upgrades to the facility that have been made since are pointed out throughout the text.The paper is structured as follows: In Sec. 2 standard measurements are described.A description of the laboratory and measurement setups is given in Sec. 3. To be characterizable by standard procedures in the CHB sensors must comply with the speci cations summarized in Sec. 1.It is to be noted that the required speci cations cover a very wide parameter range.Thus, almost all current imaging spectrometers will meet these requirements.In Sec. 6 a list of publications with context to recent activities in the CHB is mentioned.
2 Standard CHB Measurements

Spectral Measurements
Spectral measurements are performed with the monochromator setup described in Sec.3.3.
The following sensor parameters are derived: • Spectral Response Function SRF k,x (λ ) of selected pixels (x) and selected channels (k).Gives a normalized signal vs. wavelength λ .• Center wavelength λ k,x of selected channels and selected pixels.Gives the median of SRF k,x (λ ).
• Spectral smile of a selected channel.Gives the center wavelength λ k,x vs. the pixel number x.
• Spectral sampling distance of selected channels and selected pixels.Gives the wavelength di erence |λ k+1,x − λ k,x | of adjacent channels.• Spectral range of selected pixels.Gives the wavelength di erence |λ N,x − λ 1,x | of rst and last channels.
• Full Width at Half Maximum (FWHM) of selected channels and selected pixels.Gives the wavelength interval corresponding to 1/2 SRF k,x (λ ).The FWHM is identical to the spectral resolution.

Geometric Measurements
Geometric measurements on imaging spectrometers as well as cameras are performed with the collimator setup described in Sec.3.4.See Fig. 2 for the choice of a coordinate system.Images of targets (slits of various sizes) are moved in small steps across the detector elements.
If the sensor under investigation is a camera, the image of a target is geometrically similar to the target itself.However, for a pushbroom spectrometer across track (spatial direction) are di erent from along track (spectral direction of each frame) measurements.
The following sensor parameters can be derived: • Line Spread Function across track (LSF Y (α)) and along track (LSF X (β )): normalized signal of a detector element vs. across track angle α and along track angle β , respectively.• Center coordinates (across track angle α x , along track angle β x ) of selected pixels (pixel number x): angles corresponding to the signal at the median of LSF Y (α) and LSF X (β ).• Across track sampling distance: angle di erence |α x+1 − α x | of adjacent pixels.
• IFOV of selected pixels across track.This corresponds to an angle interval of 1/2 LSF Y (α) and 1/2 LSF X (β ), respectively.The IFOV is de ned as the FWHM.• FOV across track and along track.FOV Y is the across track angle di erence |α 1 − α N | of rst and last pixel, FOV X is identical to IFOV X .• Modulation transfer functions are calculated from measured line spread functions.

Radiometric Measurements
The task of radiometric calibration is the conversion of sensor signals from sensor units C (digital numbers, DN) to physical units L (spectral radiance, W/(m 2 nm sr)).For these measurements, our radiometric standard (Schwarzmaier et al., 2012) and two integrating spheres are available.The setup for radiometric measurements is described in Sec.3.5.The following measurements that are connected to a sensor's radiometric response are carried out in the CHB: • Absolute radiometric response at one xed illumination r k,x (λ ) of selected pixels (number x) and all channels (number k) in the spectral range 380 -2500 nm.• Relative radiometric response at one xed illumination r k,x /r k,x of all pixels and all channels in the spectral range 380 -2500 nm.

Polarization Dependency
Polarization sensitivity of sensors in the wavelength range 380 -2500 nm is investigated with the setup shortly described in Sec.3.6.

Laboratory Infrastructure
The optical laboratory has a size of 12.8 ×5.5 m with a height of 8 m and can be darkened.The entrance door (2.3 m width, 4.10 m height) and a crane, which is mounted on the ceiling, allow for an easy handling of larger equipment.Air-conditioning keeps the temperature at 20 ± 1 • C and meta-data such as temperature, pressure, and humidity is recorded.Fig. 1 shows a picture of the facility and the main measurement equipment.Detailed descriptions of the components are given in the following.

Calibration Bench
The calibration bench (see Fig. 1) is made of granite.It provides a working area of 3×1.6 m.The surface was polished to a maximum deviation from atness of 1.24 µm at the area where the folding mirror (see Sec. 3.2) is attached., downward looking to a mirror 4 , which re ects either the beam for geometric 5 or for spectral 6 measurements into the instrument.This 'folding mirror' can be tilted in across-track direction of the sensor in order to set the angle of incidence, and it can be moved in the horizontal direction to meet the entrance aperture.The sensor is attached and aligned on the calibration bench using an adapter.Adapters can be supplied by customers or the general purpose adapter shown in Fig. 3 will be made available.The latter is designed to carry loads of up to 150 kg and is described in detail in Gege et al. (2009).

Folding Mirror
A central component on the calibration bench is a mirror which performs coupled linear and rotary movements, called the folding mirror (see Figs. 1, 2).A summary of the technical speci cations is given in Gege et al. (2009).
The mirror has a usable area of rectangular shape with axis lengths of 174 × 114 mm 2 .It is made of Zerodur, the surface is silver to minimize changes of the polarization properties of the re ected radiation.
Mirror movements are realized very precisely by means of two air-bearing stages: a rotary stage is mounted on top of a linear stage.The linear stage de nes the Y-axis of the calibration bench's coordinate system (see Fig. 2).The folding mirror assembly can be moved along the Y-axis by ±375 mm

Monochromator Setup
For spectral measurements a monochromator generates a spectrally narrow-band beam of light which is re ected by the folding mirror at well-de ned angles into the entrance aperture of the instrument.The setup is illustrated in Fig. 4. For the sake of simplicity the setup is illustrated with a linear arrangement of all components.
A QTH lamp (LOT-Oriel: 15 V, 150 W) powered by a stabilized DC power supply is used as a radiation source 1 .After ltering out short wavelengths by use of a long pass lter (order lter) 2 the radiation enters the monochromator, whose main components are the entrance slit 3 , grating 5 and exit slit 7 .The nearly monochromatic beam exiting the monochromator can be attenuated using a neutral density lter that is additionally mounted at 2 .A recent update to the setup is the insertion of a short multimode optical ber into the beam path at 9 .This ensures that the beam entering the sensor after passing the folding mirror does not contain spectral -spatial correlations.The monochromator (Oriel MS257TM) is an asymmetrical Czerny-Turner design.It is equipped with a turret that can hold up to 4 gratings to cover di erent spectral ranges.The gratings are changed by turning the turret under software control.Spectral measurements require that the beam coming from the monochromator over lls both the sensor's entrance aperture and its IFOV.In standard con guration the beam size at the detector entrance http://dx.doi.org/10.17815/jlsrf-2-137 aperture is approximately 7 cm, while an IFOV of up to 8.5 mrad is illuminated.However, if required, the mirror -optical ber ( 9 -10 ) combination can be adapted to a sensor and measurement task.

Collimator Setup
The collimator setup is used for geometric calibration supplying a nearly parallel beam formed by a lamp-slit-collimator combination.A schematic sketch of the measurement setup is given in Fig. 5.For the sake of simplicity the setup is illustrated with a linear arrangement of all components, while in reality the components 1 and 2 are mounted on the calibration bench at 90 • relative to the others (see Fig. 2).
A lamp 1 illuminates a turnable wheel 2 .The wheel can be rotated in steps of 0.17 mrad using a rotary stage.Three slits of 50, 100 and 1000 µm width and 10 mm length are mounted radially on the wheel, three similar slits with identical dimensions are aligned tangentially with respect to the axis of rotation.These slits are used as calibration targets.They can be moved in Y-direction by turning the folding mirror in small angle intervals and in the X-direction by moving the slit.Slits are located at the focal plane of a collimator 4 , whose parabolic mirror 3 forms a beam of almost parallel light.This beam is re ected from the folding mirror 5 into the entrance aperture of the sensor 6 .The parabolic mirror has a focal length of 750 mm and produces a beam with divergence ( s/ f ), where s is the width of the slit.
For across track measurements a slit mounted tangentially on the wheel (corresponding to the sensor's X-direction) is used as the calibration target.Due to its height of 10 mm the beam has a divergence of 13 mrad in the along track direction, which over lls the IFOV of typical imaging spectrometers completely in the along track direction, but only partially in the across track direction due to the narrow slit width (0.07 mrad for the 50 mm slit).Imaging spectrometers disperse the radiation and project it in the wavelength direction on the detector array.Since the lamp emits a broad-band spectrum, an illuminated line is formed on the detector array by the di erent wavelengths.Measurements are performed by moving this line in sub-pixel steps in the across track direction over individual detector elements (pixels) using the folding mirror.In this way pixel response is determined as a function of the across track angle.
For along track measurements a slit mounted radially on the wheel is used as calibration target.Its height of 10 mm produces a beam of 13 mrad divergence in the across track direction which illuminates a number of adjacent pixels.Consequently, a broad band of spectrally dispersed light is projected on the focal plane.Due to the narrow slit width each illuminated pixel receives light in the along track

Radiance Standards
Two integrating spheres are available for radiometric measurements (see Fig. 1).The sensor is mounted during the measurements on top of the frame either above the small sphere or above the large sphere, depending on the measurement task.The radiance spectrum of both spheres is traceable against German national standard.In order to achieve traceability, the radiance spectrum is constantly monitored against our radiometric standard (RASTA, calibrated at Physikalisch Technische Bundesanstalt) (Schwarzmaier et al., 2012) by the use of a transfer spectrometer (SVC HR1024i).The resulting radiometric spectral (k=1) uncertainty is 2% in the spectral range 400-2500 nm.
The large sphere is used for sensors with large apertures.It has a diameter of 1.65 m and an aperture of 55 × 40 cm 2 which can be reduced to 20 × 30 cm 2 for higher radiance.It is equipped with 18 QTH lamps of di erent power, which are operated independently using 18 stabilized DC power supplies.
The radiant exitance can be changed in various steps from 57 to 1524 W/m 2 by using di erent lamp combinations.The sphere o ers a stability of better than 0.5% during a typical measurement period of 20 minutes.
The small sphere's angular integrated exitance is 2309 W/m 2 .The expanded uncertainty U(k = 2) is 1% from 390 to 1700 nm.It increases towards shorter and longer wavelengths (Gege et al., 2009).The sphere has a diameter of 50 cm and an aperture of 4 × 20 cm.At the standard distance (sensor aperture at plane of CHB adapter ange) the sphere illuminates a FOV of ±12.5 • .Four 100 W QTH lamps are operated using a stabilized DC power supply.

Polarizor Setup
To analyze polarization sensitivity of the sensor under test, a wire grid polarizer (Moxtek UBB01A, open aperture = 9 cm) is mounted after the exit port of the large integrating sphere.Using the latter, all pixels and channels can be characterized simultaneously.A motorized rotation stage (Newport URS150BPP) is used to set angles of the linearly polarized radiation.
Hyperspectral snapshot or single frame sensors o er the possibility to simultaneously acquire hyperspectral data in two dimensions.Recently, these rather new spectrometers have arisen much interest in the remote sensing community.Di erent designs are currently used for "smaller scale" observation such as by use of small unmanned aerial vehicles.In this context the CHBs measurement capabilities will be extended such that a standard measurement procedure for these new sensors will be implemented.

Figure 3 :
Figure 3: General purpose adapter available for sensor mounting. 1Opening for sensor ba e (covered); 2 Flanges of di erent size to mount sensor; 3 Sliding rule to read the X setting; 4 Meters to read the settings of β and γ; X , β , γ : Hand wheels to align the indicated axes.