International Conference on Space Optics ICSO 2014 La Caleta, Tenerife, Canary Islands 7 10 October /cso _2014 ono ' r

Transcription

1 Internatonal Conference on Space Optcs ICSO 2014 La Caleta, Tenerfe, Canary Islands 7 10 October 2014 Edted by Zoran Sodnk, Bruno Cugny, and Nkos Karafolas /cso _2014 ono ' r 6 October 2014: La Palma Excurson 7-10 October 2014: I(SO 2014 Tenerfe, Canary Islands Portable traceablty soluton for ground-based calbraton of optcal nstruments Omar El Gawhary Marjn van Veghel Pepjn Kenter Natasja van der Leden et al. Internatonal Conference on Space Optcs ICSO 2014, edted by Zoran Sodnk, Nkos Karafolas, Bruno Cugny, Proc. of SPIE Vol , Q 2014 ESA and CNES CCC code: X/17/$18 do: / Proc. of SPIE Vol Q-1 Downloaded From: on 8/21/2018 Terms of Use:

2 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 PORTABLE TRACEABILITY SOLUTION FOR GROUND-BASED CALIBRATION OF OPTICAL INSTRUMENTS Omar el Gawhary 1, Marjn van Veghel 1*, Pepjn Kenter 2, Natasja van der Leden 1, Paul Dekker 1, Elena Revtova 1, Maurce Heemskerk 1, André Trarbach 1, Ramon Vnk 3, Domnc Doyle 3 1 VSL, Delft, The Netherlands; 2 Scence and Technology B.V., Delft, The Netherlands; 3 ESA ESTEC, Noordwjk, The Netherlands * Correspondng author: VSL, P.O. Box 654, 2600 AR Delft, The Netherlands; mvveghel@vsl.nl Abstract We present a portable traceablty soluton for the ground-based optcal calbraton of earth observaton (EO) nstruments. Currently, traceablty for ths type of calbraton s typcally based on spectral rradance sources (e.g. FEL lamps) calbrated at a natonal metrology nsttute (NMI). Dsadvantages of ths source-based traceablty are the nflexblty n operatng condtons of the source, whch are lmted to the settngs used durng calbraton at the NMI, and the susceptblty to agng, whch requres frequent recalbratons, and whch cannot be easly checked on-ste. The detector-based traceablty soluton presented n ths work uses a portable flter radometer to calbrate lght sources onste, mmedately before and after, or even durng nstrument calbraton. The flter radometer tself s traceable to the prmary standard of radometry n the Netherlands. We wll dscuss the desgn and realzaton, calbraton and performance verfcaton. I. INTRODUCTION The accuracy of optcal nstruments that are used on board of earth observaton (EO) satelltes has drect mpact on the qualty of the data obtaned wth them, and on the valdty of the conclusons drawn from that data [1]. For ths reason, consderable effort s spent on the radometrc calbraton of these nstruments before flght. The goal of these calbratons s to establsh a lnk between the nstrument readngs and the relevant unts wthn the Internatonal System of Unts (SI) [2]. The term whch expresses the exstence of such a lnk s traceablty. In order to acheve traceablty, reference standards have to be used whch are themselves traceable to the SI. Ths means that these reference standards are drectly or ndrectly calbrated aganst the prmary realzaton of the SI unts at a natonal metrology nsttute (NMI). For optcal calbratons, the reference standards frequently take the form of calbrated lamps. FEL lamps wth an NMI calbrated spectral rradance can be used as spectral rradance standards, to calbrate the spectral rradance response of an EO nstrument. They are however not very flexble: the dstance and current settng have to be dentcal to the dstance and current settng durng calbraton at the NMI, leavng no room for adaptng the rradance level. Moreover, the calbraton s only vald for a lmted number of burnng hours. Recalbraton, or even an ntermedate check, can only be performed by sendng the lamp back to the NMI. Ths results n loss of tme, wth the consequence that recalbraton wll not be performed as frequently as s perhaps desred. In between recalbratons, only assumptons can be made regardng the drft behavour of the lab, wth no way of drectly checkng these assumptons. No reference standard used for transferrng traceablty s completely mmune to drft, but there are mportant varatons n the level of long-term stablty that can be acheved. Wth a detector-based traceablty soluton, a much greater stablty s possble. Also, detectors wth good lnearty over a large range can be realzed, provdng a hgher degree of flexblty. In ths scheme, a source wll stll be necessary for calbraton of EO nstruments, but nstead of makng the source the carrer of the traceable scale, ths scale s ncorporated nto a portable detector. The detector allows for the on-ste calbraton of the source, mmedately before and after the source s used for the calbraton of the EO nstrument. The source calbraton can be done wth exactly the settngs needed for the nstrument calbraton and where necessary, ntermedate checks are possble. The most accurate radometrc detectors are electrcal cryogenc substtuton radometers (ESCRs) [3-5], used n combnaton wth a tuneable lght source to realze a spectral rradance scale. These types of detectors are ndeed used at NMIs as prmary standards. They are however nherently slow, bulky and requre cryogenc coolng, makng them unsuted as a portable standard. Sold-state detectors made of e.g. S, Ge or InGaAs (dependng on the desred wavelength range) are more compact and are successfully beng used as transfer detectors, n partcular n a so-called trap confguraton [6]. In a trap detector, several detectors are Proc. of SPIE Vol Q-2 Downloaded From: on 8/21/2018 Terms of Use:

3 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 arranged n seres, so that lght reflected from the frst detector hts the second detector and so on. The total reflecton from such a trap detector s well below 1%. For a gven wavelength, a trap detector shows excellent lnearty over a large spectral rradance range, as well as a hgh degree of polarzaton nsenstvty. To measure spectral rradance, a wavelength selecton mechansm has to be provded. A scannng monochromator, as typcally used wth an ESCR, s to slow and bulky for a portable system. An alternatve s to use a seres of narrow band-pass flters to buld a flter radometer [7-11]. A flter radometer measures the ntegrated spectral rradance n a seres of narrow bands. If the general shape of the spectrum s known, the spectral rradance at other wavelengths can be obtaned by -based nterpolaton. Obvously, ths procedure works best for relatvely smooth spectra, whch have no sharp features n between flter bands. Planckan, or quas-planckan spectra, such as those of a FEL lamp, are deal n ths regard. A flter radometer usng sold state (trap) detectors s thus a good choce for a portable spectral rradance scale to be used n combnaton wth a FEL lamp as a calbraton stmulus. In secton II below, we wll present our desgn for such a portable flter radometer. In secton III we wll descrbe how the realzed flter radometer has been calbrated usng our scannng spot method. The nterpolaton method used for constructng a contnuous spectrum from the detector response for each of the ndvdual flters wll be dscussed n secton IV. Secton V wll show the results of a performance verfcaton based on measurements of actual FEL lamps. We end wth a concluson and outlook n secton VI. II. HARDWARE DESIGN AND REALIZATION The flter radometer contans three essental elements that together ncorporate the portable spectral rradance scale: (1) a seres of flters to select dfferent wavelengths; (2) a detector to measure the optcal power transmtted by the flters; and (3) a precson aperture to defne the effectve area. An mportant desgn choce s the wavelength range over whch the flter radometer wll operate. The wavelength range targeted n future EO mssons (Tropom, Sentnel 4 and 5) s nm. It s not possble to cover ths entre wavelength range wth a sngle sold-state detector. For ths reason, the entre wavelength range s splt up nto three subranges, that wll each be covered by a separate detector. An overvew of ths dvson s gven n table 1. Per detector there s a dedcated set of flters spannng the specfc wavelength range. These flters are placed n a flterwheel so that they can be rotated n front of the detector one-by-one. Each detector has ts own precson aperture. The combnaton of precson aperture, detector, flterwheel and housng forms a modular unt, whch we wll call a detector-flter unt (DFU). A schematc drawng of such a DFU s shown n Fg. 1. The flter wheel houses up to sxteen ½ nch flters. One flter poston however has to be reserved for an nternal shutter to measure the dark current. The dameter of the precson aperture s 3.5 mm. The current from the detector s converted by a trans-mpedance amplfer on the DFU tself to a voltage n the range of 0 to 10 V. The DFU housng s made of black anodzed alumnum. The temperature nsde the housng s controlled by a combnaton of resstve heatng elements and PT100 temperature sensors to ± 0.1 C. The entre flter radometer features a carousel where three DFUs can be mounted, as shown n Fg. 2. In the current stuaton, only the UV-VIS DFU has been mplemented. The NIR and SWIR DFUs can be easly added once they are realzed. The carousel rotates the selected DFU n front of the openng n the flter radometer housng. Note that the openng aperture n the flter radometer housng s not the defnng aperture for the spectral rradance response; that s the aperture on the DFU. In fact, all elements that determne the spectral responsvty of the flter radometer n a specfc subrange are located wthn the respectve DFU, makng t suffcent to calbrate just the DFUs as ndependent elements. Subrange Wavelength nterval Detector type UV-VIS nm S trap detector NIR nm Ge trap detector SWIR nm InGaAs photodode Table 1. Wavelength subranges and detector types Proc. of SPIE Vol Q-3 Downloaded From: on 8/21/2018 Terms of Use:

4 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 Fg. 1. DFU desgn, seen from the front sde where the lght s ncdent (left) and from the back sde (rght). Depcted s the DFU for the UV-VIS subrange, whch features a S trap detector. The flter radometer housng fully covers the DFU carousel. It provdes mechancal stablty and thermal solaton. At the front of the housng there s a knematc mount to place an algnment target over the entrance aperture. The flter radometer tself rests on a trpod va another knematc mount. Adjustment screws allow for precse algnment of the flter radometer optcal axs wth respect to the source and the EO nstrument under test. Because of the knematc mount, the flter radometer can be moved n and out of the optcal path n a reproducble manner, allowng frequent montorng of the source used as a stmulus. The dstance to the source can be determned va a precson reference plane just below the entrance aperture, whch has a calbrated offset wth respect to the plane of the precson aperture of the DFU. The electroncs are ftted together nto one separate box, so that the whole setup can be transported and nstalled wth ease. The whole system can be controlled remotely from a laptop connected through an Ethernet cable. Fg. 3 shows a photograph of the complete realzed system. Fg. 2. The DFU carousel nsde the flter radometer housng, the cover of whch has been removed for easy vewng. In the pcture, only one DFU s mounted. In total the flter radometer can hold three DFUs, for UV-VIS, NIR and SWIR. Fg. 3. Photograph of the complete system, set up to measure the spectral rradance of a FEL lamp. From front to back: electroncs box wth controllng laptop, FEL lamp, external baffles to sheld for stray lght and flter radometer. Proc. of SPIE Vol Q-4 Downloaded From: on 8/21/2018 Terms of Use:

5 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 III. CALIBRATION The UV-VIS DFU was calbrated on the Absolute Cryogenc Radometer (ACR) faclty at VSL, whch s the prmary standard for radometry n the Netherlands [5,12]. Fg. 4 shows an overvew of ths faclty, the heart of whch s an ESCR. Performance of the ACR faclty has been well establshed n nternatonal comparsons [13,14]. The ACR features several lght sources, to cover the full wavelength range from UV to SWIR. The radaton of the selected lght source s sent nto a double-monochromator system (subtractve mode) to dsperse the radaton nto ts spectral components. The output of the second monochromator s sent through an optcal hgh-pass flter to remove any remanng hgher order spectral components. The radaton then passes through an optcal shutter, whch can be closed for dark measurements. Mrror optcs s then used to form a spot wth a dameter of approxmately 4.5 mm and an openng angle of f/8. Ths spot can drected onto the ESCR to determne the spectral radant flux, or onto a transfer detector that has been prevously calbrated aganst the ESCR. o QTH I O p a Xe-arc I, Secondary transfer unt Order sortng fl er Shutter I 1 u Double Monochromator 4 Gate valve and pumpng port ' Transfer.detector Cryogenc Radometer,,Ar-arc Fg. 4. Layout of the ACR faclty, whch s the prmary standard for radometry n the Netherlands. Recently a laser drven lght source (LDLS) was added to the avalable lght sources, whch s not shown n the pcture. A mrror (denoted M n fg. 4) can be used to drect the lght onto a secondary transfer unt. The transfer unt can be scanned n two dmensons (x and y) perpendcular to the ncomng beam. Our scannng spot method [15,16] allows us to determne the spectral rradance responsvty of an unkown detector by scannng a spot over the actve area of the detector, wthout requrng ether the spot or the detector senstvty to be spatally unform. Also, the area of the detector aperture need not be known beforehand. The spectral rradance responsvty of the detector S (λ) can be calculated accordng to Rst ( λ) ϕ d ( x, y) S( λ ) = dxdy (1) ϕ ( x, y) st where R ( λ) s the known spectral flux responsvty of a standard detector prevously calbrated aganst the st ESCR, ϕ d ( x, y) the sgnal measured for the unknown detector at scan poston ( x, y) and ϕ st ( x, y) that for the standard. In the case of the flter radometer, the unknown detector s the DFU, whch for each channel s a combnaton of a (trap) detector and a band-pass flter. It was decded to calbrate the combnaton rather than the separate elements, so that the nteractons between the elements and the way ther nhomogenetes add up are automatcally taken care of. The calbraton thus results n a set of spectral rradance responsvty curves S (λ), one for each flter n the DFU. The uncertanty budget for the DFU calbraton on the ACR s gven n table 2 for three representatve channels, wth wavelengths near the begnnng, mddle and end of the UV-VIS range. Detals on the uncertanty estmaton method are gven n [15]. Proc. of SPIE Vol Q-5 Downloaded From: on 8/21/2018 Terms of Use:

6 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 Contrbuton Standard uncertanty 280 nm 540 nm 900 nm Amplfcaton factor standard 0.002% 0.002% 0.002% Amplfcaton factor detector 0.000% 0.000% 0.000% Responsvty standard 0.503% 0.058% 0.076% Step sze 0.416% 0.410% 0.410% Numercal approxmaton 0.000% 0.000% 0.000% Non-orthogonalty of x and y axes 0.061% 0.061% 0.061% Varaton n rradance due to msalgnment of scan plane and detector plane 0.005% 0.005% 0.005% Stray lght 0.010% 0.010% 0.010% Wavelength error 0.435% 0.057% 0.013% Angular algnment of detector 0.333% 0.333% 0.333% Combned 0.854% 0.538% 0.537% Table 2. Uncertanty budget for the calbraton of the DFU on the ACR faclty, for three representatve channels ndcated by ther respectve nomnal wavelengths. IV. FITTING OF IRRADIANCE SPECTRA If we use the flter radometer to measure a broad-band lght source, we wll get a dark-subtracted measurement sgnal ϕ for each flter. For deal flters, wth nfntesmal small band wdth λ, the source spectral rradance E( λ ) at the channel wavelength λ can then be constructed from the measured sgnal va the smple relaton E( λ ) = ϕ ( S ( λ ) λ ). Subsequently, a of the rradance dstrbuton E ( λ, p) contanng a number of parameters p can be ftted through the measured spectral rradances to obtan the spectral rradance at ntermedate wavelengths. However, the actual flters n the flter radometer have fnte bandwdths, as do all practcal flters, so that the spectral rradance E (λ) may not be treated as constant wthn the flter bandwdth. The measured sgnal n volts s bult up from contrbutons over the full wavelength range over whch the combnaton of flter and detector has a non-zero responsvty: ϕ = ( λ) E( λ) dλ S. (2) In order to fnd the spectral rradance dstrbuton from the measured sgnals, E (λ) s replaced by the dstrbuton E ( λ, p) n (2) to obtan a led sgnal ϕ ( p). The absolute dfference,, ϕ ( p) ϕ ( p) s then mnmzed as a functon of p over all channels n a Levenberg-Marquardt ft procedure. Note that n each teraton of the Levenberg-Marquardt algorthm, the ntegral (2) s evaluated. From the optmzed parameters p ~, the spectral rradance at each wavelength can be obtaned by the substtuton E ( λ) = E (, ~ λ p ). For fttng the spectral rradance of FEL lamps, a modfed Planckan spectrum was chosen [7]: E n 1,, 0,, ) = = b n nλ λ T b K bm. (3) λ [exp( c λt ) 1] FEL ( 5 Here T s the temperature of the Planckan radator and c = mk a constant. The polynomal n the numerator expresses the fact that the emssvty s not constant wth wavelength. The choce of polynomal order Proc. of SPIE Vol Q-6 Downloaded From: on 8/21/2018 Terms of Use: m

7 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 s a compromse between the closeness of ft at the poston of the flter channels and the stablty of the ftted curve n between channels. Wth the flters currently mplemented n the UV-VIS DFU, t was found that a thrd-order polynomal gave the best results. V. PERFORMANCE VERIFICATION To verfy the performance of the flter radometer n UV-VIS, verfcaton measurements usng FEL lamps wth known spectral rradance were performed. The FEL lamps have been calbrated for spectral rradance on the Spectral Irradance faclty (SIRF) of VSL [17]. The verfcaton setup s depcted schematcally n fg. 5. The 1000 W FEL lamps were placed at a dstance of 50 cm from the flter radometer aperture, wth two baffles wth crcular apertures n between to sheld stray lght and a thrd baffle under 45º behnd to reduce back reflectons. Algnment of the flter radometer and the apertures s done wth the ad of a laser, and the dstance s measured wth a measurng rod. Each verfcaton measurement conssts of a dark measurement and 30 measurements per selected flter. In between flters there s a 3 s transton tme for mechancal changeover of the flter and stablzaton of the sgnal. obsuuâz Fo wqncs 2;1.9a 1Ia14 Fg. 5. Setup used for the performance verfcaton measurements. The results of the verfcaton experment were evaluated by comparng the ftted spectral rradance based on the flter radometer measurements to the known spectral rradance of the FEL lamps. In fg. 6, the relatve dfference n spectral rradance for one lamp s plotted. The ftted rradance spectrum E ( λ, p ~ ), based on a thrd order polynomal accordng to (3), s evaluated at the nomnal channel wavelengths and then compared to the known spectral rradance of the calbrated FEL lamp Eref ( λ ). A combned expanded uncertanty U ( E (, ~ λ p ) Eref ( λ )) at 95% level of confdence was calculated for the dfference E ~ ( λ, p ) Eref ( λ ), and s ndcated n fg. 6. Ths uncertanty nvolves four contrbutons: the uncertanty of the flter radometer from ts calbraton, the uncertanty of the reference spectrum, the uncertanty comng from the verfcaton experment tself (algnment, dstance, stray lght and reproducblty) and the uncertanty connected to the fttng procedure. The rato E n, E = U ( E mod el mod el ( λ ~, p) E ( λ, p ~ ) E ref λ ( λ ). (4) ( λ )) ref Proc. of SPIE Vol Q-7 Downloaded From: on 8/21/2018 Terms of Use:

8 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 ndcates whether the observed dfference between measurement and reference s consstent wth the estmated uncertanty. A successful verfcaton result (to the stated level of confdence) requres that E n, 1. Ths s ndeed the case for all channels n fg. 6. 8% Ftted mnus reference spectral rradance 6% 4% 2% 0% -2% -4% -6% -8% Flter nomnal wavelength / nm Fg. 6. Relatve dfference between ftted and reference spectral rradance. The error bars denote the uncertanty at 95% level of confdence (k = 2). VI. CONCLUSION AND OUTLOOK We have desgned a flter radometer as a portable traceablty soluton for the on-ste calbraton of lght sources used n the ground-based calbraton of earth observaton nstruments. The module for the UV-VIS subrange was realzed and calbrated aganst the prmary standard of radometry n the Netherlands. Performance verfcaton aganst known FEL lamps showed devatons consstent wth the estmated uncertantes. The flter radometer tself has a standard uncertanty of % dependng on the channel. The ntenton s to extend the project wth the realzaton of the NIR and SWIR modules. ACKNOWLEDGEMENT The authors wsh to acknowledge the fnancal support from the Netherlands Space Offce (NSO). We thank Gerard Otter of TNO for valuable nput from a user perspectve. REFERENCES [1] G. Ohrng (ed.), Achevng Satellte Instrument Calbraton for Clmate Change (ASIC3), 2007 [2] Bureau Internatonal des Pods et Mesures, The Internatonal System of Unts (SI), 8th ed., 2008 [3] J. E. Martn, N. P. Fox, and P. J. Key, A cryogenc radometer for absolute radometrc measurements, Metrologa, 21, pp , 1985 [4] T. R. Gentle, J. M. Houston, J. E. Hards, C. L. Cromer, and A. C. Parr, Natonal Insttute of Standards and Technology hgh-accuracy cryogenc radometer, Appl. Opt., 35, pp , 1996 [5] C. A. Schrama, R. Bosma, K. Gbb, H. Rejn and P. Bloembergen, Comparson of monochromator-based and laser-based cryogenc radometry, Metrologa, 35, pp , 1998 [6] N. P. Fox, Trap Detectors and ther Propertes, Metrologa, 28, pp , 1991 [7] P. Karha, P. Tovanen F. Manoocher and E. Ikonen, Development of a detector-based absolute spectral rradance scale n the 380nm to 900nm spectral range, Appl. Opt., 36, pp , 1997 [8] P. Tovanen, F. Manoochehr, P. Karha, E. Ikkonen, A. Lassla, Method for characterzaton of flter radometers, Appl. Opt., 38, pp , 1999 Proc. of SPIE Vol Q-8 Downloaded From: on 8/21/2018 Terms of Use:

9 Tenerfe, Canary Islands, Span Internatonal Conference on Space Optcs 7-10 October 2014 [9] T. Kubarsepp, P. Karha, F. Manoocher, S. Nevas, L. Ylanttla and E. Ikonen, Spectral rradance measurements of tungsten lamps wth flter radometers n the spectral range 290 nm to 900 nm, Metrologa, 37, pp , 2000 [10] M. Durak, F. samadov, Realzaton of a flter radometer-based rradance scale wth hgh accuracy n the regon from 286 nm to 901 nm, Metrologa, 41, pp , 2004 [11] Y. J. Lu, G. Xu, M. Ojanen, E. Ikonen, Spectral rradance comparson usng a mult-wavelength flter radometer, Metrologa, 46, pp , 2009 [12] C. A. Schrama, P. Bloembergen and E. W. M. Van der Ham, Monochromator-based cryogenc radometry between 1 µm and 20 µm, Metrologa, 37, , (2000) [13] S. Brown, T. Larason, and Y. Ohno, Report on the Key Comparson CCPR-K2.a-2003 Spectral Responsvty n the Range of 900 nm to 1600 nm, BIPM, 2003 (avalable from kcdb.bpm.org) [14] R. Goebel, M. Stock, Report on the key comparson CCPR-K2.b of spectral responsvty measurements n the wavelength range 300 nm to 1000 nm, BIPM, 2004 (avalable from kcdb.bpm.org) [15] C.A. Schrama, H. Rejn, Novel calbraton method for flter radometers, Metrologa, 36, pp , 1999 [16] C. A. Schrama, E. W. M. Van der Ham, Samplng perod crteron n a scannng-beam technque, Appl. Opt., 39, pp , 2000 [17] E.W.M. van der Ham, H. C. D. Bos and C. A. Schrama, Prmary realzaton of a spectral rradance scale employng monochromator-based cryogenc radometry between 200 nm and 20µm, Metrologa, 40, pp , 2003 Proc. of SPIE Vol Q-9 Downloaded From: on 8/21/2018 Terms of Use: