IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 52, NO. 12, DECEMBER 2014 7775

A Microbolometer Airborne Calibrated InfraredRadiometer: The Ball Experimental Sea

Surface Temperature (BESST) RadiometerWilliam J. Emery, Fellow, IEEE, William S. Good, William Tandy, Jr., Miguel Angel Izaguirre, and

Peter J. Minnett, Member, IEEE

Abstract—A calibrated radiometer has been developed to en-able the collection of accurate infrared measurements of sea sur-face temperature (SST) from unmanned aerial vehicles (UAVs).A key feature of this instrument is that in situ calibration isachieved with two built-in blackbodies (BBs). The instrumentis designed so that the 2-D microbolometer array produces in-frared images incremented as the aircraft travels, resulting in awell-calibrated strip of SST. Designed to be carried by medium-class UAVs, the Ball Experimental SST (BESST) instrumenthas been also successfully flown on manned aircraft. A re-cent intercalibration of BESST was carried out at the Univer-sity of Miami using their National Institute of Standards andTechnology traceable water-bath BB and a Fourier transform in-terferometer, the Marine-Atmospheric Emitted Radiance Interfer-ometer (M-AERI). The characterization of the BESST instrumentwith the Miami BB demonstrates the linearity and precision ofthe response of the microbolometer-based radiometer. Coincidentmeasurements of SST from a nearby pier clearly demonstratedthe excellent performance of the BESST instrument with a meanSST equal to that of the M-AERI and an RMS of 0.14 K veryclose to the microbolometer’s advertised precision of 0.1 K. Coldcalibration was not possible in Miami due to condensation, but aBall BB was characterized relative to the Miami water-bath BB,and calibrations were made in Boulder at lower temperatures thanwere possible in Miami. The BESST instrument’s performance re-mained linear, and the mean and RMS values did not change. UAVflights were conducted in summer/fall of 2013 over the AlaskanArctic.

Index Terms—Microbolometer, radiometer, sea surface temper-ature (SST), thermal infrared.

I. INTRODUCTION

SATELLITE infrared instruments measure the emitted ra-diation from the skin layer of the ocean, which is related

in a rather complex fashion to the subsurface temperaturesmeasured by buoys and ships [1]–[4]. To further advance ourunderstanding, it is important that we improve our ability to

Manuscript received October 11, 2013; revised January 30, 2014; acceptedMarch 8, 2014.

W. J. Emery is with the University of Colorado, Boulder, CO 80309 USA(e-mail: William.Emery@colorado.edu).

W. S. Good and W. Tandy, Jr. are with Ball Aerospace and Technologies,Boulder, CO 80301 USA (e-mail: wgood@ball.com; wtandy@ball.com).

M. A. Izaguirre is with the Rosenstiel School of Marine and AtmosphericScience, University of Miami, Miami, FL 33149 USA (e-mail: mizaguirre@rsmas.miami.edu).

P. J. Minnett is with the Meteorology and Physical Oceanography, RosenstielSchool of Marine and Atmospheric Science, University of Miami, Miami,FL 33149 USA (e-mail: pminnett@rsmas.miami.edu).

Digital Object Identifier 10.1109/TGRS.2014.2318683

collect well-calibrated coincident measurements of skin seasurface temperatures (SSTs). Collecting more accurate data ataltitudes close to the sea surface will also be of benefit sincethere are fewer atmospheric error terms to consider. The datafrom these infrared instruments could then be used to validateinfrared measurements from polar and geostationary orbitingsatellites. Despite these potential benefits, these validation mea-surements can presently only be made from either researchvessels or merchant ships. Unfortunately, these ships cannotcollect a sample over an area as large as even a single 1-kmpixel of one of the satellite sensors. In addition, the rather slowspeed of ships means that it is not possible to rapidly collectcontinuous in situ skin infrared emitted radiation from shipsover time. The result is that the intervals between data sets arelikely to be longer than the time scales of the processes thatcontrol the variability of skin SST.

An attractive alternative would be to fly an infrared radiome-ter on a plane or an unmanned aerial vehicle (UAV) that couldmeasure 2-D strip maps of skin-emitted thermal radiation. Sucha radiometer would have to be much smaller and lighter thanthose that make measurements from a ship, but they wouldstill have to conform to the standard for making accurate SSTmeasurements by having two blackbodies (BBs) to providea frequent two-temperature calibration of the sensed infraredradiation. To build such an instrument, Ball Aerospace designedBall Experimental SST (BESST) around a microbolometer asthe sensor array, which both simplified the instrument and madeit light and small in size. To provide the BB calibration, twosmall BBs were built and installed. Since the measurement ofSST includes a component reflected off the sea surface fromdirectly incoming thermal radiation from the atmosphere, theBESST instrument also makes a sky observation to correctfor reflected sky infrared radiation in the derivation of theSST. Finally, to provide continuity across these measurements,BESST uses the same microbolometer for viewing the sea, thereflected sky radiation, and the two reference BBs.

These requirements dictated the size and shape of the BESSTradiometer. Thus, the design includes a mirror that changes theview of the microbolometer between the view ports and theBBs. The temperature of one BB is allowed to “float” withthe ambient temperature, whereas the second BB is heated toat least 12 K above the temperature of the ambient BB. Thesetemperatures are continuously monitored and recorded fromvarious locations on the BESST instrument. The small size ofthese reference BBs might have been a cause for concern, but

0196-2892 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

7776 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 52, NO. 12, DECEMBER 2014

the results section will show that this concern turned out to beunjustified. In order to use the same microbolometer for bothtarget and calibration sensing, a small mirror on a shaft attachedto a step motor was installed, which rotates for each calibrationsequence. The initial sampling scheme was to collect three SSTsamples per second for 53 s and then perform a 7-s calibrationsequence. The calibration sequence requires a brief loss of databy the BESST instrument.

The BESST instrument was taken to the University ofMiami’s Rosenstiel School of Marine and Atmospheric Science(RSMAS) for characterization with the National Institute ofStandards and Technology (NIST) traceable water-bath BB[5], [6] and for a comparison with the Marine-AtmosphericEmitted Radiance Interferometer (M-AERI) [4]. Against theNIST traceable BB, the BESST instrument was found to behavelinearly, and these measurements were used in subsequent com-parisons to calibrate the BESST measurements. Measurementsof SST from a nearby pier showed the BESST instrument tohave a mean SST exactly the same as the M-AERI and astandard deviation of only 0.14 K, which was only slightlylarger than the 0.1 K precision claimed for the microbolometer.These results are also confirmed by earlier laboratory and fieldcalibration experiments made in Colorado [7]. This value ofthe standard deviation also includes contributions from theM-AERI measurements, which are also estimated to be ∼0.1 K.

II. INSTRUMENT DESIGN

The radiometer was conceived of as a “pushbroom” imagingradiometer that would produce strips of infrared data as theaircraft flew forward. A 324 × 256 pixel thermal imaginguncooled microbolometer could be equipped with filters (in the8–12 μm range) that include the wavelength ranges primarilyused for SST observations from satellites. The goal of thisdesign would be to have calibrated SST measurements accurateto ±0.1 K. This would be achieved by frequent calibrationwith two onboard BBs and with a correction for reflected skyradiation in the thermal infrared.

In its final configuration, the BESST radiometer’s field ofview is 18◦, covering 200 pixels cross-track, resulting in aswath width of about 1/3 of the flight altitude above groundlevel. We do not use pixels on the edge of the detector array(due to optical distortion), taking only the center 200 pixels tooptimize performance. Thus, the image resolution depends onthe altitude of the aircraft that carries the BESST instrument.For altitudes of about 600 m, the surface resolution is about 1 mwith a swath width of 200 m. An airborne system with thisspatial resolution is able to resolve the SST variability withinthe 1-km pixels typical of spaceborne infrared radiometers.

From the start, BESST was a modular design with a cen-tral cube to which target baffles, BBs, mirror motor, andmicrobolometer focal plane array (FPA) would be attached(Fig. 1). This design has evolved only slightly, and the BESSTinstrument now appears as in Fig. 2. This image includes amounting bracket for a particular UAV.

A single germanium lens looks through the downward-looking baffle to collect incident radiation from the target scene,focusing it onto a 2-D uncooled microbolometer detector with

Fig. 1. Modular design of the BESST radiometer.

Fig. 2. BESST #1 in the operational configuration.

a pixel size of 38 μm. The microbolometer is a FLIR PhotonThermal Imaging Camera Core (www.flir.com). The opticalsystem is F/2 with a 25-mm focal length. The detector ispartitioned into three spectral bands by the use of individualinterference filters positioned in front of the detector array.The channels selected are fairly narrow channels centered at10.8 μm and another centered at 12 μm. These channels areselected to correspond to the primary infrared channels usedby satellite radiometers to compute SST. The third channel isbroad, covering 8–12 μm, and provides a more sensitive SSTchannel by integrating a larger amount of thermal radiation. Inprocessing, these channels can be individually treated, or theycan be combined to generate an SST product. The filters areplaced along the flight path viewing the same ground scene witha latency of 0.3 s or less, depending on the platform’s speed overthe ground. An antireflective coating on the germanium windowinto the FPA has a passband from 8 to 12 μm and that providesadditional out-of-band blocking into the microbolometer.

The basic design principle is to have the motor/scene—mirror opposite the detector array as shown in Fig. 3. The motorrotation axis is directly inline with the center of the detectorarray. In this way, the optical path reflects off the fold mirrorand can be directed by rotating the fold mirror. The optical pathalways reflects off the mirror so that the only difference to thedetector is the actual scene. In addition, the optical stop is builtinto the turret; thus, the stop rotates with the mirror. Thermistors

EMERY et al.: MICROBOLOMETER AIRBORNE CALIBRATED INFRARED RADIOMETER 7777

Fig. 3. Detector, mirror, and drive motor configuration.

Fig. 4. Calibration modules.

continually monitor the temperatures of these elements forcorrection in processing the microbolometer data.

Onboard calibration is done by alternating the observedocean scene and the two temperature-monitored BB references,thus making it possible to construct a standard two-point cal-ibration curve. A dark uniform temperature flag (right abovethe filters and near the detector) provides a common referencefor all pixels and is flipped in as part of the once-per-minutecalibration routine. For calibration, the mechanical design hasthree distinct advantages. First, the tight fit between the mirrorhousing and the lens housing keeps stray light off of thecalibration sources. Second, the angle of incidence of the mirroris equal for the calibration sources and the scene. Third, the areaof the mirror is the same for both the calibration source andthe target scene. The sky view is also part of the corrections tocalculate SST in that it corrects for the sky radiation reflectedinto the downward-looking radiometer.

A. Calibration Modules

The calibration targets consist of an aluminum tube with aconical end that forms the BB calibration source. The insideis painted with black absorptive paint, so that when the detec-tor views the BB, it sees an accurately defined and uniformtemperature. The calibration module shown in Fig. 4 can bemade in two different configurations depending on the cali-bration temperature required. For a warm calibration source,a fiberglass-mounting flange is used so that the BB target isthermally isolated from the rest of the system. A strip heater,attached to the outside of the aluminum tube, heats the tube to apredetermined temperature (about 12 K above ambient), and thethermistor attached to the module reports its temperature. For

TABLE IMASS SUMMARY

the ambient module, no heater is installed, and the temperatureis monitored to insure that its temperature is close to the temper-ature of the sensor-mounting cube. The BBs and temperaturesensors are covered by Mylar insulation to protect them fromradiative heating and to trap the heat in the calibration module.The design of the BBs also protects them from contaminationby rain while the insulation reflects sunlight.

B. Mass

The total mass of the BESST instrument is presented here inTable I and does not include the electrical cables, computer, orbattery required to operate the system. The heaviest componenton the list is the motor. The motor also includes a gear head andcontrol electronics. A much simpler and lighter motor couldbe procured, but the selected unit proved adequate for the taskas it was easy to interface with and easy to control. Even withthis motor, the total weight is very light, making this an idealinstrument for flying on even rather small UAVs.

The detector array captures images of each scene, and theBB readings are used to provide highly accurate calibrationsof the scene measurements. The sequence is automated by themini-PC software, which also collects the data over the regionof interest. For this data collection, the frame rate is 3 Hz.A delay of 333 ms between frames and the inherent mini-PCcomputer latency makes this overall frame rate slightly less than3 Hz. A total of 130 frames are collected during a data frame,which takes about 53 s, and a calibration cycle takes only 7 s.Again, computer latency can affect these time periods, but timestamping of images with Global Positioning System time froma receiver allows accurate knowledge of image collection time.

C. Electronics

A block diagram of the BESST electronics is shown herein Fig. 5. The airborne computer is a dedicated mini-PC thatcontrols the mirror position and records the data from themicrobolometer. The motor is commanded to rotate the centralfold mirror to view the four scenes: water, sky, warm BB,and ambient BB. The calibration cycle itself takes about 7 s,resulting in about 21 frames of target data lost during a calibra-tion cycle. Thus, for a 25-m/s ground speed of the UAV, BESSTwould collect data for 1.3 km and then undergo a calibrationcycle for 0.2 km, which would then be repeated over the flightof the UAV.

Overall, the BESST instrument is collecting target data 87%of the time.

7778 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 52, NO. 12, DECEMBER 2014

Fig. 5. BESST electronics schematic.

III. ABSOLUTE CALIBRATION

Initial laboratory calibrations at Ball with a reference BBindicated the absolute BESST accuracy to be ±0.3 K. This ref-erence BB was not NIST traceable, however, and this absoluteaccuracy is better than the repeat performance over the GreatSalt Lake. To better establish the absolute temperature accuracyof the BESST, we carried out an intensive intercomparisonexperiment at the University of Miami’s RSMAS in June 2012.RSMAS has a NIST traceable BB to use as a reference andalso operates the M-AERI for SST measurements from ships.The plan for this intercomparison was to first characterizeBESST against the RSMAS water-bath BB and then make anindependent set of BB comparisons to test the absolute accu-racy of the measurements made in the laboratory. Then bothM-AERI and BESST would be taken to a nearby pier to makecoincident measurements of SST over a period of time. Thispier intercomparison was cut short due to the arrival of after-noon storms, but the two instruments were coincidentally oper-ated for at least 2 h. In addition, laboratory measurements weremade to characterize the Ball BB in terms of the Miami NISTtraceable BB. This turned out to be a future advantage sincecondensation due to the humid Miami atmosphere restrictedcomparison with the NIST BB at lower temperatures, wherewe hoped to operate BESST in the future. These referencemeasurements were later made back in Boulder with the BallBB, which had now been characterized against the Miami NISTtraceable water-bath BB, as will be described below.

The characterization measurements of BESST against theMiami water-bath BB are presented here in Fig. 6. An averagevalue of the reported temperatures within the area of the water-bath BB in the BESST 2-D image is plotted. The BESST rawtemperature are found to be linear relative to the water-bathBB temperatures, indicating a predictable performance of themicrobolometer when calibrated with its internal BBs. Thisintercomparison uses the best linear fit between the BESSTonboard warm and cold BBs to interpolate the target bathtemperature.

From the plot, we can determine the slope and offset forthe line that corrects the measured BESST value to the actualtarget scene temperature. These constants can then be appliedto correct future temperature measurements. When this is donewith the same data set we used to derive the values, we get

Fig. 6. Comparison of NIST calibrated BB target temperatures to raw BESSTmeasurements under ambient conditions.

Fig. 7. Independent measurement applying linear correction to raw BESSTdata.

temperature errors that range between −0.1 and +0.06 K(within the ±0.1 K precision of the microbolometer). Anotherindependent data set was collected to estimate this scatter. Theresults are presented in Fig. 7.

The scatter of this small data set ranges from 0.027 ◦C to±0.157 ◦C. All of the differences are less than 0.16 ◦C, whichis close to the absolute error of 0.1 ◦C suggested by the earliercharacterization measurements.

As aforementioned, the characterization of the BESST in-struments’ performance was not extended below 10 ◦C for theMiami testing due to condensation on the NIST and Ball BBsin the relatively humid atmosphere. In order to extend the rangeof calibrated performance of the instrument, BESST was in-stalled in a dry-nitrogen-filled thermal chamber with the newlycharacterized Ball BB, where the temperature could be loweredwithout condensation. The Ball BB was cross calibrated withthe NIST BB during the Miami testing using the MAERI as atransfer radiometer. The resulting linear fit to the error of theBall BB setpoint versus actual radiant temperature is shown inFig. 8. This slope and offset was applied to correct later datasets with the Ball BB, making it possible to extend the BESSTcharacterization to colder temperatures than were possible inthe Miami characterization study.

Having calibrated the Ball BB versus the NIST traceablewater-bath Miami BB, measurements were collected under coldambient conditions over a range of Ball BB target temperatures(Fig. 9). The measurements show linear performance of the

EMERY et al.: MICROBOLOMETER AIRBORNE CALIBRATED INFRARED RADIOMETER 7779

Fig. 8. Linear fit to characterize Ball BB versus the NIST BB.

Fig. 9. Characterization at varying ambient temperatures.

BESST instrument for various ambient temperature conditions.Measurements of ambient temperature are collected by thethermistors mounted on the BESST internal ambient BB andon the instrument itself and updated for every data sample at1.5 Hz. There is also another thermistor installed on the “hot”BB. The changing ambient temperature due to field conditions,including altitude changes, is used to determine the linear fitto apply to the BESST data in order to accurately find thetarget scene temperature for each image. A lookup table of in-terpolated slopes and offsets provides the necessary correctionfor changing ambient temperatures. This ambient temperaturecorrection is required due to nonlinearities in the sensitivityof the microbolometer array to external temperature changes.The microbolometers are designed to nominally operate at25 ◦C, and their sensitivity to incident thermal radiation varieswhen the temperature varies from this nominal condition. Timestamps synchronize the temperature data with the image datato allow for postprocessing of all of the BESST data with theambient temperature correction slopes and offsets.

Next, both the M-AERI and the BESST instruments weretaken to a nearby pier to make coincident measurements ofSST for an extended period. The BESST (Fig. 10, top reddashed box) was mounted on top of the much larger M-AERI(bottom yellow dashed box) to approximately view the samewater surface.

The view angle for the M-AERI is 55◦, and the height fromthe pier to the water surface was approximately 2 m. The

Fig. 10. (Top red) BESST and (bottom yellow) M-AERI mounted on the pier.

Fig. 11. SST from the Miami pier.

M-AERI was about 0.61 m above the pier surface, whereasBESST was about 1.22 m above the pier. M-AERI was viewingabout 3.05 m away from the pier, whereas BESST was nadirmounted and viewed the water about 0.46 m in front of theM-AERI aperture. The local current and the shallow (5 m)depth resulted in the observed surface waters being well mixed,which would result in a rather uniform SST. Two hours ofcoincident data were collected. The results are shown in Fig. 11,which is a time-series plot of both the BESST and M-AERImeasurements of the SST. For these measurements, a wa-ter emissivity of 0.9876 was used for the BESST 8–13 μmband, where the characterization values in Fig. 9 were usedto calibrate BESST for these measurements. An average ofthe temperatures from the central area of the BESST imageis shown in the figure. The M-AERI values are averages ofwavelengths near 7.7 μm [4]. As shown in Fig. 11, the BESSTvalues showed a transient over the first 30 min of the time series.The M-AERI transient values dominated the entire first hour for

7780 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 52, NO. 12, DECEMBER 2014

Fig. 12. BESST SST example (left) collected along the flight track (right).

some unknown reason (which may be due to a problem withthe hot BB heater) and have been eliminated from this plot.While M-AERI has a measurement every 10 min, the BESSTinstrument is putting out three measurements per second duringthese observations.

Interestingly, both the BESST and the M-AERI recordedthe same mean SST of 30.18 K over the entire 2-h period.The BESST measurements included the upward-looking skycorrection for reflected radiation, and the M-AERI data werecorrected for both reflected sky radiation and for atmospherichumidity. For the intercomparison, the first hour of M-AERImeasurements were neglected, and only the last four M-AERISST measurements are included in Fig. 12. Over this shorterperiod, M-AERI gave a mean SST of 30.04 K, whereas theBESST data for this period yielded a mean value of 30.01 K.The M-AERI SST standard deviation for this period was0.54 K, whereas that for the BESST was 0.14 K. The largevariability for the M-AERI is clearly due largely to the singlelow value at about 21.7 h. The BESST data also show a slightSST decrease at this point in time but do not go as low as theM-AERI value. Taken over the entire series, the SST standarddeviation for BESST is only 0.14 K, which includes the tran-sient start-up period, which is consistent with the claimed 0.1 Kprecision of the microbolometer.

BESST measurements exhibit a relatively small variabilityand a nearly constant SST after about the point labeled as21.5 h on the plot. The time axis refers to the Greenwich MeanTime as collected by the instruments during the measurements.

IV. PAST AND FUTURE FIELD EXPERIMENTS

BESST was also flown on a Twin Otter aircraft over theDeepwater Horizon oil spill in the Gulf of Mexico in thesummer of 2010 [8]. The goal of this deployment was todetermine how well the BESST instrument would represent theoil slicks in conjunction with coincident measurements madeby visible and radar instruments. Unfortunately, the radar didnot operate properly, and the visible instrument collected onlya very limited data set. BESST, however, performed well andprovided some very interesting images of the SST associatedwith surface oil. In Fig. 12, we present an image (left) of theSST along a BESST strip located over the oil as indicated by

Fig. 13. BESST in ScanEagle payload bay.

the map in the center panel collected along the track shown inthe right panel. These data were collected during the daytimewhen the oil is typically warmer than the surrounding water,as is suggested by the red and yellow patches in the left panelin Fig. 12. The temperature range is 24 ◦C–36 ◦C. The orangestripe corresponds to the flight path region blown up on the rightpanel and represents the location of the BESST SSTs collected.The loop represents the area where the plane turned around togo back on the same dashed line. The SST panel on the leftis not completely rectangular, reflecting the changes in planelocation due to local winds. The interesting thing about theinfrared sensing of oil is that, during the daytime, the oil appearswarmer than the surrounding water, whereas at night, the oilcools faster than the water, and an infrared image should thenshow the oil as cool relative to the water. The real benefit hereis the rapid collection of this SST information and its ability tocapture the fine detail that is not possible with satellite imagery.

The first UAV deployments were made in summer/fall of2013 on the ScanEagle UAV, which is a medium-class UAVplatform frequently used for video reconnaissance by themilitary.

The BESST instrument is integrated into a payload bay forthis aircraft, as shown here in Fig. 13.

The big advantage of operating an accurate SST radiometerfrom a small UAV is the ability of these aircraft to fly low andslow over the water for an extended period of time. Thus, theywill operate below the major atmospheric attenuation layer andwill be able to directly measure the complex spatial structure ofthe SST.

We will be able to actually measure the SST wavenumberspectrum of a 1-km2 pixel of the typical infrared satellite SSTsensor. This will help to resolve the long-standing question ofhow much subpixel spatial variability contributes to the uncer-tainty budget in the satellite SST retrievals when this is derivedusing comparisons with point measurements from buoys.

EMERY et al.: MICROBOLOMETER AIRBORNE CALIBRATED INFRARED RADIOMETER 7781

V. DISCUSSION AND CONCLUSION

The BESST radiometer has been developed to be a small,accurate, compact, and low-cost thermal infrared radiometerfor UAV deployments. Both laboratory and field measurementshave demonstrated that this instrument is capable of makingSST measurements accurate to ±0.14 ◦C with no bias dif-ference and with accurate reference temperatures. Based ona microbolometer, the BESST instrument does not requirecooling, and its small size makes it possible to be deployed ineven the smaller classes of UAVs. Laboratory measurementsagainst a water-bath BB demonstrate the linear behavior ofthe BESST instrument. Coincident pier measurements with aninterferometer (M-AERI) demonstrate the comparable behaviorof the BESST instrument and recommend it as a small andlow-cost alternative to measuring in situ skin SST. AlternativeBESST designs were considered, but the upward and downwardview ports are the most logical for UAV applications. In thefuture, a seagoing version may be needed, and the positionsof the baffle tubes and the BBs may be changed. Such aseagoing instrument would also need rain and sea spray pro-tection mechanisms. Fortunately, a primary advantage of theBESST’s modular design is the flexibility to enable such al-ternative configurations.

ACKNOWLEDGMENT

The authors would like to acknowledge the continued supportof Ball Aerospace in developing BESST and funding the testflights and the Miami intercomparison tests. They would alsolike to thank M. Kuester, who began development of BESSTback when she worked at Ball; E. Baldwin-Stevens, who, asan Intern at Ball, carried out a lot of the initial characterizationof BESST; and R. Warden, P. Kaptchen, and T. Finch for theircontributions to the development and testing of BESST. Theywould also like to acknowledge NASA support of this instru-ment as part of MIZOPEX (http://ccar.colorado.edu/mizopex/index.html).

REFERENCES

[1] W. J. Emery, S. Castro, G. A. Wick, P. Schluessel, and C. Donlon, “Estimat-ing sea surface temperature from infrared satellite and in situ temperaturedata,” Bull. Amer. Meteor. Soc., vol. 82, no. 12, pp. 2773–2785, Dec. 2001.

[2] C. J. Merchant and A. R. Harris, “Toward the estimation of bias in satelliteretrievals of sea surface temperature 2: Comparison with in situ measure-ments,” J. Geophys. Res.-Oceans, vol. 104, pp. 23 579–23 590, 1999.

[3] C. J. Donlon et al., “Towards operational validation of satellite sea surfaceskin temperature measurements for climate research,” J. Climate, vol. 15,no. 4, pp. 353–369, Feb. 2002.

[4] P. J. Minnett et al., “The Marine-Atmospheric Emitted Radiance Interfer-ometer (M-AERI), a high-accuracy, sea-going infrared spectroradiometer,”J. Atmos. Ocean. Technol., vol. 18, no. 6, pp. 994–1013, Jun. 2001.

[5] J. B. Fowler, “A third generation water-bath-based BB source,” J. Res. Natl.Inst. Std. Technol., vol. 100, no. 5, pp. 591–599, Sep. 1995.

[6] J. P. Rice et al., “The Miami 2001 infrared radiometer calibration andintercomparison: 1. Laboratory characterization of BB targets,” J. Atmos.Ocean. Technol., vol. 21, no. 2, pp. 258–267, Feb. 2004.

[7] W. Good, R. Warden, P. Kaptchen, T. Finch, and W. Emery, “Absolute ther-mal radiometry from a UAS,” presented at the AIAA Infotech@AerospaceConference, St. Louis, MO, USA, 2011.

[8] W. S. Good et al., “Absolute airborne thermal SST measurements and satel-lite data analysis from the deepwater horizon oil spill,” in Monitoring andModeling the Deepwater Horizon Oil Spill: A Record-Breaking Enterprise.Washington, DC, USA: Amer. Geophys. Union, 2011, ser. GeophysicalMonograph Series 195.

William (Bill) J. Emery (F’03) received the Ph.D.degree in physical oceanography from the Universityof Hawaii, Honolulu, HI, USA, in 1975.

After working at Texas A&M University, CollegeStation, TX, USA, he moved to the University ofBritish Columbia, Vancouver, BC, Canada, in 1978,where he created a Satellite Oceanography facility/education/research program. He was appointed Pro-fessor of aerospace engineering sciences at the Uni-versity of Colorado, Boulder Campus, Boulder, CO,USA, in 1987. He is an Adjunct Professor of in-

formatics at Tor Vergata University, Rome, Italy. He has authored over 177refereed publications on both ocean and land remote sensing and two textbooks.

Prof. Emery is a Fellow of the American Meteorological Society (2010),the American Astronautical Society (2011), and the American GeophysicalUnion (2012). He is the Vice President for Publications of the IEEE Geoscienceand Remote Sensing Society (GRSS) and a member of the IEEE PeriodicalsCommittee. He was the recipient of the GRSS Educational Award in 2004 andits Outstanding Service Award in 2009.

William (Bill) S. Good received the M.S. degree inaerospace engineering sciences from the Universityof Colorado, Boulder, CO, USA, in 2003 and theM.S. degree in condensed matter physics from IowaState University, Ames, IA, USA, in 2002.

He has worked at Ball Aerospace and Technolo-gies, Boulder, since 2003, developing new instru-ments for earth remote sensing, including UV-LWIRspectrometers and IR radiometers. Since 2008, hehas led the Ball Civil and Operational Space Air-borne Initiative.

Mr. Good has been a member of the American Geophysical Union since2009.

William (Bill) Tandy, Jr. received the M.S. degreein aerospace engineering from the University ofTexas at Austin, Austin, TX, USA, in 2006.

He has worked at Ball Aerospace and Technolo-gies, Boulder, CO, USA, since 2006 as a StructuralAnalyst and a Systems Engineer. His key projectsinclude the JWST aft optics bench, the MOIREspacecraft, and the OMPS ozone monitoring instru-ment. He has been involved with the BESST systemas a programmer, developer, and operator since 2010.

Miguel Angel Izaguirre received the bachelors de-gree from the Universidad Metropolitana-Iztapalapa,Mexico City, Mexico, in hydrobiology in 1981 and aM.A. degree in liberal studies from the University ofMiami, Coral Gables, FL, USA, in 2014.

He is a Research Associate with the RosenstielSchool of Marine and Atmospheric Science, Uni-versity of Miami, Miami, FL, USA. He has partic-ipated in numerous research cruises throughout theAtlantic, Pacific, and Indian Oceans, the CaribbeanSeas, and the Gulf of Mexico, as well as many field

campaigns in the Canary Islands, Barbados, Puerto Rico, and the Maldives. Hisfocus of interest is in atmospheric physics and chemistry, and instrumentation.

Peter J. Minnett (M’09) received the B.A. de-gree in physics from the University of Oxford,Oxford, U.K., and the M.Sc. and Ph.D. degrees inoceanography from the University of Southampton,Southampton, U.K.

Since 1995, he has been with the RosenstielSchool of Marine and Atmospheric Science, Uni-versity of Miami, Miami, FL, USA, where he iscurrently the Chair of Meteorology and PhysicalOceanography. His research interests are in derivingaccurate measurements of ocean variables, primarily

sea surface temperature, by remote sensing from satellites and ships. He servesas the Science Team Chair of the Group for High Resolution Sea-SurfaceTemperature.