Abstract

Extensive comparisons are made of the infrared sea surface skin temperature Tskin measured by the Calibrated Infrared In situ Measurement System (CIRIMS) and the Marine-Atmospheric Emitted Radiance Interferometer (M-AERI). Data were collected from four separate deployments on the NOAA research vessel (R/V) Ronald H. Brown and the U.S. Coast Guard (USCG) Polar Sea over a wide range of latitudes and environmental conditions. The deployment time totaled roughly 6 months over a 4-yr period and resulted in over 7000 comparison values. The mean offset between the two instruments showed that CIRIMS consistently measured a lower temperature than the M-AERI, but by less than 0.10°C. This mean offset was found to be dependent upon sky condition, wind speed, and ship roll, which implies the offset is likely due to uncertainty in the emissivity. The CIRIMS Tskin was recomputed using two alterative emissivity values, one based on emissivity measured by the M-AERI and the other based on a wind-speed-dependent model. In both cases, the recomputation of the CIRIMS Tskin significantly reduced the mean offset. The overall standard deviation between the M-AERI and CIRIMS Tskin was 0.16°C, did not significantly depend on environmental conditions, and was within the expected values of instrument and comparison uncertainties. These comparisons demonstrate the success of CIRIMS in achieving good agreement with the M-AERI over a wide range of conditions. The results also highlight the importance of the sea surface emissivity when measuring the ocean surface skin temperature.

1. Introduction

The importance of validating satellite-derived sea surface temperature (SST) using in situ measurements has led to the development of a variety of sea-going sensor systems (Barton et al. 2004). The Calibrated Infrared In situ Measurement System (CIRIMS) is an autonomous radiometer system for ship-based validation that has been extensively deployed. Jessup and Branch (2008) summarized the design and operation of the CIRIMS and presented laboratory and field testing results. Here we present extensive at-sea comparisons of skin temperature Tskin measured by the CIRIMS and the Marine-Atmospheric Emitted Radiance Interferometer (M-AERI; Minnett et al. 2001).

The CIRIMS measures infrared (IR) radiation from the sea and sky to calculate Tskin. It has one sea-viewing and one sky-viewing radiometer that both operate in the 9.6–11.5-μm wavelength range. Calibration is achieved with a National Institute of Standards and Technology (NIST)-traceable internal blackbody cavity, which the sea-viewing radiometer rotates to view. The blackbody is used to perform a two-point dynamic calibration procedure. The measurement cycle typically consists of four 6-min sea views and one 6-min blackbody view. The sky-viewing radiometer continuously measures sky radiation. The uncertainty in the CIRIMS Tskin was estimated from laboratory and field tests to be 0.081°C (Jessup and Branch 2008).

The M-AERI is an accurate, self-calibrating, Fourier transform IR spectroradiometer that measures emission spectra from the sea and atmosphere (Minnett et al. 2001). It uses two IR detectors to achieve a wide spectral range, and these are cooled to ∼78 K (i.e., close to the boiling point of liquid nitrogen) by a Stirling cycle mechanical cooler to reduce the noise equivalent temperature difference to levels well below 0.1°C. The M-AERI includes two NIST-traceable internal blackbody cavities for accurate real-time calibration. A scan mirror directs the field of view from the interferometer to either of the blackbody calibration targets or to the environment from nadir to zenith. The mirror is programmed to step through a preselected range of angles. The measurements are integrated over a preselected time interval, usually a few tens of seconds, to obtain a satisfactory signal-to-noise ratio. A typical cycle of measurements includes two view angles to the atmosphere, one to the ocean and one for calibration measurements, and takes from 5 to 10 min. The environmental variables derived from the spectra include Tskin with an absolute uncertainty of <0.1°C. This level of uncertainty is based in large part on the results of a side-by-side field comparison of two M-AERIs reported by Minnett et al. (2001). They found that the mean ± standard deviation of the difference between Tskin from the side-by-side instruments was 0.01° ± 0.079°C for 890 values obtained over a 10-day period. The M-AERI Tskin is calculated from radiation measured at a wavelength of 7.7 μm.

2. Overall Tskin comparison

The four major deployments with simultaneous CIRIMS and M-AERI measurements are listed in Table 1. During the 2001 deployment of the National Oceanic and Atmospheric Administration (NOAA) research vessel (R/V) Ronald H. Brown, the simultaneous measurements totaled roughly 85 days over a 3-month period and covered three cruises designated GasEx01 (Gas Exchange 2001), ACE-Asia (Asian Pacific Regional Aerosol Characterization Experiment), and FOCI (Fisheries–Oceanography Coordinated Investigations). The dataset that resulted from the 2001 deployment of U.S. Coast Guard (USCG) Polar Sea was relatively short, covering only about one week. The largest comparison dataset is from the 2004 Aerosol and Ocean Science Expedition (AEROSE) and Windward Passage (WP) cruises on the Brown. This deployment provided the opportunity to reassess the comparison after nearly three years. In late 2004, the CIRIMS and M-AERI were deployed on the Brown for the Tropical Atmosphere Ocean (TAO) cruise with the IR SST Autonomous Radiometer (ISAR; Donlon et al. 2008). Preliminary results of this three-way comparison (Jessup and Branch 2008) indicated good agreement, and a detailed analysis of this unique and valuable intercomparison will be the subject of a future publication.

Table 1.

Simultaneous CIRIMS and M-AERI deployments.

Simultaneous CIRIMS and M-AERI deployments.
Simultaneous CIRIMS and M-AERI deployments.

The overall comparisons for each ship deployment are summarized in Table 2 as the RMS difference, the mean difference , and the standard deviation σMC for the quantity ΔTMC = TM-AERITCIRIMS. The Brown comparisons are considered the most significant since the number of data points on both the 2001 and 2004 cruises were an order of magnitude greater than on the 2001 Polar Sea cruise. As explained in the next section, for most of the comparisons on the Brown the CIRIMS was operated without the optional IR transparent window, which can be used to protect the optics and blackbody when needed (Jessup and Branch 2008).

Table 2.

Difference statistics of TM-AERITCIRIMS from three cruises.

Difference statistics of TM-AERI − TCIRIMS from three cruises.
Difference statistics of TM-AERI − TCIRIMS from three cruises.

The histograms of ΔTMC from the 2001 and 2004 Brown comparisons in Fig. 1 show that TCIRIMS is consistently lower than TM-AERI, but by less than 0.10°C. The RMS difference and σMC were between 0.12° and 0.16°C, and were less in 2004 than in 2001. The simultaneous time series of TM-AERI and TCIRIMS in Figs. 2 and 3 further illustrate the characteristics of the comparison. The 8-day period from GasEx01 shown in Fig. 2 includes CIRIMS measurements made with and without the IR transparent window, which are discussed below. The example in Fig. 3 is from the FOCI cruise and covers a much wider range of temperatures than Fig. 2. The time series comparisons show that ΔTMC varies with time and that the scatter in the CIRIMS data is comparable to that of the M-AERI.

Fig. 1.

Histograms of TM-AERITCIRIMS for cruises on the R/V Brown from (a) 2001 and (b) 2004. The mean TCIRIMS is consistently lower than the mean TM-AERI by less than 0.10°C. The RMS difference and the std dev were between 0.12° and 0.16°C, and were lower in 2004 than in 2001.

Fig. 1.

Histograms of TM-AERITCIRIMS for cruises on the R/V Brown from (a) 2001 and (b) 2004. The mean TCIRIMS is consistently lower than the mean TM-AERI by less than 0.10°C. The RMS difference and the std dev were between 0.12° and 0.16°C, and were lower in 2004 than in 2001.

Fig. 2.

Time series of Tskin measured by M-AERI and CIRIMS over an 8-day period on the GasEx01 cruise showing results with and without the IR transparent window in use. The CIRIMS measurement with the window is consistently greater than that without the window.

Fig. 2.

Time series of Tskin measured by M-AERI and CIRIMS over an 8-day period on the GasEx01 cruise showing results with and without the IR transparent window in use. The CIRIMS measurement with the window is consistently greater than that without the window.

Fig. 3.

Time series of Tskin measured by M-AERI and CIRIMS over a 4-day period during the FOCI cruise on the R/V Brown in 2001. The data cover a wide range of temperatures and show that the scatter in both measurements is comparable.

Fig. 3.

Time series of Tskin measured by M-AERI and CIRIMS over a 4-day period during the FOCI cruise on the R/V Brown in 2001. The data cover a wide range of temperatures and show that the scatter in both measurements is comparable.

The three separate cruises on the Brown in 2001 covered a wide range of environmental conditions and geographic locations. From February through March 2001, the CIRIMS and M-AERI operated continuously together during the GasEx01 cruise from Florida to Hawaii. After GasEx01, the Brown traveled from Hawaii to Japan during the ACE-Asia cruise and then on to Alaska during FOCI. The range of temperatures and wind speeds encountered during the two latter cruises was much greater than during GasEx01. The statistics of the comparison for the three different 2001 Brown cruises along with the wind speed and sea temperature ranges are listed in Table 3.

Table 3.

Difference statistics of TM-AERITCIRIMS from 2001 Brown cruises, including with the window (GasEx01 – window) and without the window (all other entries).

Difference statistics of TM-AERI − TCIRIMS from 2001 Brown cruises, including with the window (GasEx01 – window) and without the window (all other entries).
Difference statistics of TM-AERI − TCIRIMS from 2001 Brown cruises, including with the window (GasEx01 – window) and without the window (all other entries).

a. Comparison using the IR transparent window

During the GasEx01 cruise, CIRIMS was operated alternately with and without the IR transparent window to evaluate the performance of the window correction described in Jessup and Branch (2008). The statistics for the comparison with and without the window are listed in Table 3. The RMS and standard deviation were 0.05°C higher for the measurements made with the window compared to those without the window. The mean difference changed from 0.04° to −0.07°C when the window was in place, indicating that the CIRIMS was measuring higher temperatures than the M-AERI. The degree to which the measurements agreed is illustrated in the time series plot in Fig. 2, which shows the M-AERI data plotted with the CIRIMS data with and without the window. The changes in both the mean and standard deviation are consistent with the self-comparison of measurements with and without the window described in Jessup and Branch (2008). In deployments of the CIRIMS subsequent to GasEx01, the window was used only when the risk of contamination was great enough to warrant a 0.05°C increase in the standard deviation. If the risk of contamination was minimal, then the CIRIMS was operated without the window. This was the case during the deployments on the Brown, since the CIRIMS was mounted high enough to avoid significant sea spray (Jessup and Branch 2008). The integrity of the sensor housing was periodically monitored when operating without the window on the Brown and showed no contamination by spray. This decision made it possible to investigate differences between the CIRIMS and the M-AERI without the complication of the window correction as an additional source of error.

b. Sky correction effects

Radiometric determination of Tskin is based on inversion of the equation for the sea surface radiance L(T) measured with an IR radiometer. If the distance to the surface is small enough to neglect atmospheric effects, the radiance measured by a radiometer operating in the wavelength range λ1λλ2 and viewing the sea surface at an incidence angle θ is the sum of the emitted and reflected radiation, given by

 
formula

where Lλ,b(λ, T) is the spectral radiance at temperature T given by Planck’s function, R(λ) is the instrument responsivity, ɛ(λ, θ) is the spectral emissivity, and ρ(λ, θ) is the spectral reflectivity. In practice, (1) is solved for Tskin by combining separate measurements of the upwelling radiance for the sea and the downwelling radiance from the sky:

 
formula

For uniformly cloudy conditions with clouds that have a low base, the sky radiance L(Tsky) is close to the sea radiance L(Tskin) and so (2) is relatively insensitive to the accuracy of the sky correction. For uniformly clear conditions, the sky radiance is much less than the sea radiance and thus the correction for reflected radiation from the sky is critical. For partly cloudy conditions, the situation is complicated by both spatial and temporal variations in the sky radiance.

Sorting the CIRIMS and M-AERI comparison statistics by sky conditions provides evidence that the sky correction is the main source of the mean difference. The statistics ΔTMC and σMC as a function of sky conditions are given in Table 4 and shown in the category plots in Fig. 4 for each individual cruise and the combined data from all cruises. The statistic is lowest for cloudy conditions, intermediate for partly cloudy conditions, and highest for clear skies for the GasEx01 and ACE-Asia cruises as well as for the combined data. This consistent increase in as the sky conditions vary from cloudy to clear suggests that the sky correction is the main cause of changes in this quantity. On the other hand, σMC shows no consistent trend with sky conditions when each cruise is considered separately and no significant difference with sky conditions for the combined data. This suggests that the cause of the standard deviation is not a function of the sky correction.

Table 4.

Difference statistics of TM-AERITCIRIMS for 2001 Brown cruises by sky condition.

Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises by sky condition.
Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises by sky condition.
Fig. 4.

Category plot of (left) mean and (right) std dev of for individual cruises (FOCI, GasEx01, and ACE-Asia) and combined cruises as a function of sky conditions. The mean difference shows a dependence on sky conditions while the std dev does not.

Fig. 4.

Category plot of (left) mean and (right) std dev of for individual cruises (FOCI, GasEx01, and ACE-Asia) and combined cruises as a function of sky conditions. The mean difference shows a dependence on sky conditions while the std dev does not.

To determine if the sky correction was the primary source of ΔTMC, raw sea-viewing brightness temperatures were compared. The data were acquired when the M-AERI was operated during GasEx01 at the same incidence angle as the CIRIMS. The spectral capabilities of the M-AERI allowed for a calculation of the brightness temperature at the CIRIMS wavelengths. The mean difference in brightness temperatures was zero and the standard deviation was 0.10°C. The zero mean difference in brightness temperature demonstrates that the mean offset in Tskin is not due to calibration of the sea-viewing radiometer but rather to the sky reflection correction. The sky reflection correction could be incorrect due to either the value of the sea surface emissivity used in (2) or inaccuracy of the sky measurements. If the mean difference is due to the emissivity, then it should vary with the environmental conditions that affect emissivity. If the mean difference is due to inaccurate sky measurements, then it would not be expected to vary with environmental conditions.

Rough sea conditions can significantly affect the sea surface emissivity due to increased surface roughness and variation in local incidence angle due to the ship’s roll and wave slope. A 1% error in the emissivity for the CIRIMS wavelength band could cause up to a 0.66°C error in Tskin (Hanafin and Minnett 2005). The emissivity used in the CIRIMS sky correction did not take into account roughness or changes in local incidence angle. Since surface roughness increases with wind speed, an indication of its effect is given by sorting the comparison statistics from the three 2001 Brown cruises by wind speed range as shown in Table 5. The statistic ΔTMC increases from 0.01°C for low winds to 0.08°C for high winds, while σMC and the RMS show no clear pattern.

Table 5.

Difference statistics of TM-AERITCIRIMS for 2001 Brown cruises by wind speed.

Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises by wind speed.
Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises by wind speed.

An example of the effect of increased ship’s roll on ΔTMC is illustrated in Fig. 5. The upper trace shows the RMS value for the ship’s roll, which increased significantly a few hours before midnight UTC on 4 April 2001. The lower plot of the time series of Tskin from both instruments shows a significant increase in the difference over the period of increase in the ship’s roll. The emissivity will be dependent upon both the mean and standard deviation of the roll, which change with wind conditions, wave conditions, and ballast shifting. For instance, Hanafin and Minnett (2005) found that the distribution of the effective incidence angle, which is a result of ship’s roll and wave conditions, was symmetric at low winds but more highly spread and negatively skewed at high winds. The increase in ΔTMC with both wind speed and roll provides convincing evidence that ΔTMC is due to sea surface emissivity.

Fig. 5.

Time series of skin SST for M-AERI and CIRIMS on the R/V Brown from Hawaii to Alaska in 2001. The period between 1200 UTC 3 Apr 2001 and 1200 UTC 4 Apr 2001 shows that an increased difference is correlated with increased ship roll due to rough conditions.

Fig. 5.

Time series of skin SST for M-AERI and CIRIMS on the R/V Brown from Hawaii to Alaska in 2001. The period between 1200 UTC 3 Apr 2001 and 1200 UTC 4 Apr 2001 shows that an increased difference is correlated with increased ship roll due to rough conditions.

3. Discussion of CIRIMS and M-AERI comparisons

The extensive comparisons between the M-AERI and CIRIMS over the three 2001 Brown cruises show that the mean offset is related to uncertainty in the emissivity, which is a function of incidence angle and surface roughness. Sea surface emissivity has been theoretically derived to decrease as wind speed increases for the range of incidence angles used by the CIRIMS and M-AERI (Masuda et al. 1988). However, the magnitude of the effect has been shown to be less than predicted (Watts et al. 1996). Furthermore, recent field measurements by Hanafin and Minnett (2005) indicate that emissivity can be independent of or even increase with wind speed, depending on incidence angle. For wind speeds of 3 to 13 m s−1, they reported that the mean emissivity was constant at an incidence angle of 40° and increased by 0.004 at a 55° incidence angle. Since the M-AERI measurements were made at 55° and those of the CIRIMS at 40° on the Brown, one might suspect that the reason for the dependence of the difference on sky conditions is an increase in emissivity with wind speed affecting the M-AERI measurements. If the M-AERI measurements were sensitive to changes in emissivity, an underestimate of the emissivity at high wind speeds would cause TM-AERI to be higher than TCIRIMS, which is consistent with the overall comparison statistics.

However, the sensitivity of the M-AERI to changes in local incidence angle should be an order of magnitude less than CIRIMS because of a fundamental difference in sky correction methods (Minnett et al. 2001). The CIRIMS radiometers operate in the 9.6–11.5-μm range, which is a relatively transparent atmospheric window. Therefore, the sky radiance measured by CIRIMS under clear-sky conditions is much less than the sea radiance and the sky correction is sensitive to uncertainty in the emissivity. The M-AERI uses a narrow wavelength band centered on 7.7 μm, for which the atmospheric pathlength is much shorter than in the 9.6–11.5-μm range. Thus, the difference between the sea and sky radiances measured by the M-AERI is much smaller under clear-sky conditions than that measured by the CIRIMS. Therefore, Tskin retrieved by the M-AERI should be much less sensitive to changes in emissivity. The effects of the atmosphere between the sensor and the sea surface are much greater in the M-AERI wavelength range than in that of the CIRIMS. A correction for the greater atmospheric absorption is made in the M-AERI algorithm using a parameterization based on the air temperature, instrument height, and incidence angle (Minnett et al. 2001). Although the effective optical depth for the two instruments differs by about 10 μm, the gradient in the skin layer is such that the difference in measured temperature due to this effect is O(0.01) K, which is much less than the accuracy of O(0.1) K (McKeown et al. 1995).

An alternative explanation for the mean offset is that the CIRIMS measurement is too low because the emissivity is overestimated. The effective emissivity used to compute Tskin from the CIRIMS data was based on tabulated values for zero wind speed by Shaw and Marston (2000), who used the complex refractive index values for pure water by Hale and Querry (1973). Hanafin and Minnett (2005) reported values for emissivity measured in the field at 40° incidence angle of 0.9836 for 9 μm and 0.9911 for 11 μm. These wavelengths roughly bracket the wavelength range of the CIRIMS (9.6–11.5 μm). Therefore, a reasonable estimate of the effective emissivity for the CIRIMS is the average of the emissivity at 9 and 11 μm reported by Hanafin and Minnett (2005). This average of 0.9873 is significantly less than the value of 0.9894 used for the 2001 Brown measurements taken at 40° incidence. If the actual emissivity for the CIRIMS is indeed lower than the value used in the computation, then this would explain some of the mean offset.

The CIRIMS Tskin values for all the 2001 Brown data were recomputed using the emissivity value of 0.9873 derived from the M-AERI data for the GasEx01 cruise. The resulting statistics are shown in Table 6 and should be compared to those in Table 3. The mean offset for the GasEx01 data was reduced to zero, which implies 0.9873 was a good estimate for the sea surface emissivity during that cruise. Although the mean offsets for the ACE-Asia and FOCI cruises decreased in the sense that they went from positive to negative values, their absolute values remained comparable, implying the sea surface emissivity during those cruises was between the value of 0.9873 based on the Hanafin and Minnett (2005) results from GasEx01 and the value of 0.9894 used in the original CIRIMS algorithm.

Table 6.

Difference statistics of TM-AERITCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity of 0.9873.

Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity of 0.9873.
Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity of 0.9873.

To further investigate the variability of the sea surface emissivity during the three 2001 cruises on the Brown, an average emissivity was calculated based on the average wind speed during each cruise. The average emissivity was computed by linearly interpolating the Shaw and Marston (2000) values of emissivity at different wind speeds to the average cruise wind speed. The CIRIMS Tskin was then recomputed for each cruise with the appropriate emissivity. The results are shown for each cruise and the combination of cruises in Table 7 and should be compared to the original values given in Table 3. The absolute value of the mean offset for each cruise was reduced to 0.02°C and that for the combination of cruises was reduced to zero. Although a more exact result might be achieved by computing an emissivity value for each datum rather than using an average for the entire cruise, the potential gain is likely minimal considering that the mean offset resulting from the approach used here already is very small. These results suggest that the variability in the mean offset between the CIRIMS and M-AERI may be due to the variability in the emissivity due to wind speed. They also emphasize the importance of wind speed in determining a sea surface emissivity.

Table 7.

Difference statistics of TM-AERITCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity determined by the average cruise wind speed.

Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity determined by the average cruise wind speed.
Difference statistics of TM-AERI − TCIRIMS for 2001 Brown cruises when TCIRIMS is calculated with a sea surface emissivity determined by the average cruise wind speed.

The standard deviation between the CIRIMS and M-AERI did not significantly depend on sky conditions or wind speed. For the combined data from the 2001 Brown cruises for all sky conditions, σMC was 0.16°C. This value is a combination of the instrumentation uncertainties for each instrument and measurement uncertainties such as different fields of view and different measurement times. Jessup and Branch (2008) estimated the overall uncertainty of the CIRIMS to be 0.081°C for variable sky conditions. A corresponding overall measurement uncertainty for the M-AERI is given by the standard deviation of 0.079°C for the side-by-side comparison of two M-AERIs noted above (Minnett et al. 2001). The root of the sum of the squares of these CIRIMS and M-AERI uncertainties is 0.11°C, which is comparable to values of 0.12° and 0.16°C for σMC during the 2001 and 2004 Brown cruises, respectively, for all sky conditions. Thus, the standard deviation of the difference between the CIRIMS and M-AERI measurements is consistent with our expectations of the individual instrument uncertainties.

4. Conclusions

Over 7000 skin temperature measurements from the CIRIMS were compared to those from the M-AERI, for multiple cruises between 2001 and 2004 spanning a wide range of environmental conditions. From the combined datasets, the CIRIMS Tskin was consistently lower than that from the M-AERI. The data collected on the R/V Brown in 2001 were analyzed in detail in order to determine the source of the mean offset. The mean difference between the M-AERI and CIRIMS Tskin ranged from 0.00°C for cloudy conditions to 0.05°C for clear conditions. It also varied from 0.01°C for low wind conditions to 0.08°C for high wind conditions. The increase in the mean offset with changing sky and wind conditions indicates that the offset is likely due to uncertainty in the emissivity. The emissivity used in the original CIRIMS Tskin calculation was based on published theoretical values and did not include wind speed dependence. The emissivity was changed to a value measured by the M-AERI during the GasEx01 cruise and the recomputed comparison resulted in a zero mean offset. However, use of the M-AERI-derived emissivity did not significantly improve the comparison for deployments other than the GasEx01 cruise. The comparisons then were recalculated using an emissivity value that was dependent upon the wind speed conditions for each cruise. The absolute value of the offset was reduced to 0.02°C for each cruise and the mean offset for the three cruises combined was reduced to zero. The reduction in the mean offset for both recalculations is compelling evidence that the mean difference between the CIRIMS and M-AERI measurements is primarily due to the emissivity used in the sky correction. The standard deviation of the difference was comparable to the sum of the independently determined overall measurement uncertainty of the M-AERI of 0.079°C (Minnett et al. 2001) and that of the CIRIMS of 0.081°C (Jessup and Branch 2008). These comparisons have demonstrated the reliability of the CIRIMS measurements relative to the M-AERI over long time periods, large distances, and a wide range of environmental conditions. They also emphasize that the environmental dependence of emissivity should be included in radiometric measurements of ocean surface skin temperature.

Acknowledgments

We thank T. Litchendorf of the Applied Physics Laboratory, University of Washington, for assistance in deployment and data analysis for the CIRIMS. We thank J. Hanafin, K. Maillet, and G. Szczodrak of the Rosenstiel School of Marine and Atmospheric Science, University of Miami, for assistance in deployment and data analysis for the M-AERI. We are grateful to the captains and crews of the NOAA R/V Brown and the USCG Polar Sea.

REFERENCES

REFERENCES
Barton
,
I. J.
,
P. J.
Minnett
,
K. A.
Maillet
,
C. J.
Donlon
,
S. J.
Hook
,
A. T.
Jessup
, and
T. J.
Nightingale
,
2004
:
The Miami2001 infrared radiometer calibration and intercomparison. Part II: Shipboard results.
J. Atmos. Oceanic Technol.
,
21
,
268
283
.
Donlon
,
C. J.
,
I. S.
Robinson
,
W.
Wimmer
,
G.
Fisher
,
M.
Reynolds
,
R.
Edwards
, and
T. J.
Nightingale
,
2008
:
The Infrared Sea Surface Temperature Autonomous Radiometer (ISAR) for deployment aboard Volunteer Observing Ships (VOS).
J. Atmos. Oceanic Technol.
,
25
,
93
113
.
Hale
,
G. M.
, and
M. R.
Querry
,
1973
:
Optical constants of water in the 200-nm to 200-micrometer wavelength region.
Appl. Opt.
,
12
,
555
563
.
Hanafin
,
J. A.
, and
P. J.
Minnett
,
2005
:
Measurements of the infrared emissivity of a wind-roughened sea surface.
Appl. Opt.
,
44
,
398
411
.
Jessup
,
A. T.
, and
R.
Branch
,
2008
:
Integrated ocean skin and bulk temperature measurements using the Calibrated Infrared In Situ Measurement System (CIRIMS) and through-hull ports.
J. Atmos. Oceanic Technol.
,
25
,
579
597
.
Masuda
,
K.
,
T.
Takashima
, and
Y.
Takayama
,
1988
:
Emissivity of pure and sea waters for the model sea surface in the infrared window regions.
Remote Sens. Environ.
,
24
,
313
329
.
McKeown
,
W.
,
F.
Bretheron
,
H. L.
Huang
,
W. L.
Smith
, and
H. L.
Revercomb
,
1995
:
Sounding the skin of water: Sensing air–water interface temperature gradients with interferometry.
J. Atmos. Oceanic Technol.
,
12
,
1313
1327
.
Minnett
,
P. J.
,
R. O.
Knuteson
,
F. A.
Best
,
B. J.
Osborne
,
J. A.
Hanafin
, and
O. B.
Brown
,
2001
:
The Marine-Atmospheric Emitted Radiance Interferometer: A high-accuracy, sea-going infrared spectroradiometer.
J. Atmos. Oceanic Technol.
,
18
,
994
1013
.
Shaw
,
J. A.
, and
C.
Marston
,
2000
:
Polarized infrared emissivity for a rough water surface.
Opt. Express
,
7
,
375
380
.
Watts
,
P. D.
,
M. R.
Allen
, and
T. J.
Nightingale
,
1996
:
Wind speed effects on sea surface emission and reflection for the Along Track Scanning Radiometer.
J. Atmos. Oceanic Technol.
,
13
,
126
141
.

Footnotes

Corresponding author address: Andrew T. Jessup, Applied Physics Laboratory, University of Washington, 1013 NE 40th St., Seattle, WA 98105-6698. Email: jessup@apl.washington.edu