1. Introduction
The ocean’s surface temperature has been the subject of increased interest in the past decade because of its importance to climate change, weather prediction, and air–sea interaction. These problems require sea surface temperature (SST) accuracy of 0.3°C (Emery et al. 2001) and have emphasized the difference between the “skin” and “bulk” temperatures. The skin is the 10-μm surface layer seen by satellite IR sensors whereas the bulk layer is between 1 and 5 m deep and is sampled by ships and buoys. The skin SST differs from the bulk SST because of daytime solar heating of the top few meters of the ocean and evaporative and IR cooling of the top 10 cm (Fairall et al. 1996a). The former effect can be as large as several degrees Celsius, while the latter is typically 0.2°–0.5°C.
The structure and temperature of the ocean’s skin is important to the exchange of heat, momentum, and trace chemicals such as carbon dioxide. These same processes are also relevant for the coastal zone and inland lakes. Inland lakes have a wide range of temperatures and atmospheric conditions and can easily be instrumented for satellite calibration. They also do not have the swell and large-amplitude waves found in the ocean. On the other hand, the complicated and variable boundary layers over small lakes increase the uncertainty in the calculated heat and momentum fluxes (Mahrt et al. 1998).
Most oceanic data have been obtained from extended campaigns from research ships that have measured the surface water temperature to an accuracy of 0.1°C (Wick et al. 1996). These studies have provided an abundance of data but are expensive to conduct. Moreover, the ship itself has a significant effect on the atmospheric flow and on temperature measurements in the top few centimeters of water. Blanc (1987) found that ship-induced errors in the calculated heat and momentum fluxes were 20%–30%.
Drifting and moored buoys routinely measure the bulk water temperature and have much less influence on airflow than ships. The sensor accuracy is believed to be ∼0.15°C, but sensor error plus natural variability is ∼0.4°C (Emery et al. 2001). Satellite sea surface temperature measurements are being compared with buoy and ship data to improve the algorithms that relate the bulk temperature to the skin temperature (Fairall et al. 1996a, b) and instrument development has been directed toward accurate but rugged sensors to measure the skin sea surface temperature for extended periods (Emery et al. 2001).
This paper describes a lightweight, portable apparatus to measure the skin temperature and related meteorological variables. The apparatus is relatively transparent to the wind and current and has sensors positioned within 1 m of the air–water interface to minimize the effects of advection and internal boundary layers. It is attached to existing buoys and can collect data for several days. Data from a small lake are presented to illustrate the performance of the apparatus and to determine the heat and momentum fluxes and the skin temperature depression with bulk aerodynamic relationships.
2. Apparatus
Since the skin is only a few microns thick, its temperature must be measured remotely with an IR radiometer. However, the IR skin temperature is less than its true value because of reflection of (colder) sky radiation. This effect can be calculated if the emissivity of water is known as a function of wave state along with the sky radiance. This calculation is complicated, however, and requires accurate measurement of water and sky radiance and sea surface shape (Donlon and Nightingale 2000). Corrections for the sky radiation can be minimized if the skin temperature is found by subtracting the ocean’s surface IR temperature from that of a stirred bath of constantly replenished ocean water (Schluessel et al. 1990). This technique reduces sky IR reflection and instrument errors because both measurements should be similarly affected. However, the emissivity of water is dependent on the viewing angle, and the shape of the sea surface may differ from that of the stirred bath. Furthermore, shading from the ship’s hull may result in a different sky radiation field for the stirred bath than for the ocean’s surface. In addition, contaminants floating on the water’s surface could alter the ocean’s emissivity compared to the bath water. Finally, the bath water is collected from a depth of 0–2 m where a daytime warm layer may exist. Wick et al. (1996) eliminated the latter problem by restricting the analysis to nighttime data.
A variation on the stirred-bath method is to pump water to the sea surface periodically and compare the temperature of the pumped water with that of the undisturbed surface. This technique will lead to comparable radiative environments for the pumped (bulk) and natural (skin) water surfaces and thus an accurate temperature difference that can be added to the bulk water temperature to obtain an accurate skin temperature. This apparatus is simple to design and does not require a correction for sky radiation. However, care is required to ensure that the jet of pumped water is reliably inserted onto the water’s surface in the radiometer’s field of view for all wind and water conditions.
All three methods employ periodic calibration of the radiometer with a reference standard—the first method uses a separate blackbody, whereas the stirred and pumped-water methods use seawater, with accurately measured temperature, as the calibration target.
The apparatus described in this paper is a pumped-water design and derives the skin temperature from the difference between IR measurements of the undisturbed surface water and of water pumped from a depth of 10 cm every few minutes to a point in the radiometer’s field of view. The pumped-water temperature is obtained at the point where the jet breaks the surface, before the skin can form. In practice, a radiometer offset is calculated by comparing the pump-on temperature with the 10-cm thermocouple temperature and then applied to all the radiometer data to determine the skin temperature. The difference between the IR temperature with the pump on and off includes both the skin depression and any gradient in temperature between the surface and 10 cm. To eliminate the temperature gradient, the skin temperature depression was found by subtracting the corrected radiometer temperature from the 1-cm water temperature instead of the 10-cm value.
Figure 1 shows the experimental apparatus, and Table 1 lists the sensor properties. The apparatus frame is 1.3 m2 with the corner legs held together with stainless steel rods. The apparatus is attached to a buoy with a 10-m tether and orients itself with the front (sonic anemometer and IR radiometer) facing into the wind. It has been deployed in winds of up to 12 m s−1.
The IR radiometer is a chopped pyroelectric instrument with a range of 8–14 μm and a field of view of 3 cm at a distance of 0.7 m. It is mounted on an upstream leg of the apparatus and sighted at 20° from the zenith. Every few minutes, the pump mounted under the IR radiometer pumps water to the surface in the radiometer’s field of view for 10 s. Since data are sampled every 2 s, five measurements of the pumped water are obtained for each period.
Other variables required to test skin temperature parameterizations were measured with the apparatus (Table 1). A two-dimensional sonic anemometer, mounted on one of the front legs of the apparatus, collects u and υ winds 3 times a second and outputs a 1-s average value. The temperature/RH sensor and electronic compass are mounted on downwind legs of the apparatus. The electronic compass was used to calculate the absolute wind direction and the orientation of the apparatus. Data were logged and the pump was controlled with a Campbell CR10 logger.
Thermocouple temperature probes were mounted on a lightweight float assembly in the center of the apparatus at depths of 1, 10, and 50 cm. The assembly floats independently of the main apparatus to minimize bouncing of the 1-cm probe out of the water from wave action. A reference thermistor is located in the logger connector in the insulated logger/battery container at the base of the rear leg.
3. Calibration
Since the IR skin measurements are “calibrated” by comparison with water periodically pumped from a depth of 10 cm, thermocouple accuracy and the radiometer’s precision and repeatability, rather than absolute accuracy, are most important. Nevertheless, the radiometer and thermocouples were calibrated prior to each use in a stirred blackbody water bath. The bath is 38 cm deep and 15 cm in diameter with an embedded, cylindrical blackbody cavity near its base that is 5 cm in diameter and 13.5 cm long. The last 6 cm of the cavity is wedge shaped to increase its effective emmissivity. Temperatures were monitored with a precision glass thermometer and Omega HH41 recording digital thermometer calibrated with a National Institute of Standards and Technology (NIST)-traceable standard to 0.01°C accuracy. For a temperature range of 1° to 30°C, a constant offset of 0.1°C was sufficient to correct the thermocouples to a standard deviation of 0.04°C with respect to the HH41. A linear correction to the radiometer values for this temperature range yielded a standard deviation of 0.06°C with respect to the HH41. The sensors drifted by less than 0.1°C in 1 yr. Uncertainty due to the blackbody itself, for example, an emissivity of less than 1.0, is not considered.
The standard deviation of 0.06°C includes contributions from instrument accuracy, inherent precision, and resolution. The radiometer resolution is 0.1°C (Table 1) and will cause errors of 0.0° to a maximum of 0.05°C, with an equivalent standard deviation of 0.029°C. If we assume that the accuracy, precision, and resolution are uncorrelated; remove the contribution to the total variance from the resolution; and assume that the accuracy and inherent precision contribute equally, we obtain values of 0.035°C for each, that is, 0.06°C = √
A second test investigated the dependence of the measured temperature on the instrument’s internal temperature. Figure 2 shows the effect of placing the radiometer in an environmental chamber whose temperature was stabilized at 10°C and then gradually raised to 25°C during a 45-min period while the instrument was pointing at a blackbody with constant temperature of 20.2°C. The measured temperature decreased by 0.6°C during the first 15 min and then increased gradually to within 0.05°C of the original reading. Since a 15°C change in 45 min is much greater than expected in normal operation, transients observed in the field should be less than 0.6°C. During deployment, a reflective sleeve is placed over the instrument to minimize solar heating and temperature fluctuations.
Instrument performance was also evaluated in the field. Since water pumped from a depth of 10 cm, monitored with a thermocouple, functions as a blackbody reference standard, the uncertainty in the skin temperature measurement has four components: 1) thermocouple uncertainty; 2) radiometer precision, repeatability, and resolution over time scales of minutes; 3) possible changes to the water pumped between 10 cm and the surface; and 4) possible differences in the reflected sky radiation between the natural and pumped-water surface.
The thermocouple accuracy determined in the laboratory (0.04°C; Table 1) should be achievable in the field. Similarly, the radiometer’s precision and repeatability in the field should be comparable to the laboratory, at least for short time periods. Uncertainty 3 above includes cases when the water does not reach the surface, is mixed with surface water, or is displaced away from the radiometer’s field of view. In normal operation, the pumping rate was adjusted to produce a mound of water ∼2–3 cm high and ∼10 cm across. This diameter is much larger than the radiometer’s field of view at the surface (3 cm), and thus displacement of the pumped jet is not likely to shift the mound completely from the field of view.
One way to examine uncertainties 3 and 4 is to compare variability of the IR measurements of the natural and pumped-water surfaces with the 10-cm thermocouple variability. If the variability of the pumped-water temperature (IR) is comparable to that of the 10-cm (thermocouple) water and less than that of the natural surface (IR), then we can assume that the pumped water has not mixed with surface water. As a test, we computed the mean 10-cm thermocouple and IR temperatures during pump operation, the mean IR temperature for the 10 s prior to pump operation (natural surface), and the corresponding standard deviations from the means, for a 4-h dataset, on 4 May 2000. The means were 22.53° (10-cm thermocouple), 22.30°, (IR pumped water) and 21.89°C (IR natural surface), and the standard deviations were 0.015°, 0.040°, and 0.062°C, respectively. Since the standard deviations are substantially less than the differences in the means, disruption to the pumped water was not a significant source of error. The precisions are also comparable to laboratory values (Table 1).
The pump performance was also investigated with an IR camera positioned above the apparatus. During light winds, the pumped water was observed to spread out over the surface to a diameter of 1 m or more, while the skin took as long as 30 s to reform after pump shutoff. For this reason, IR data collected within 30 s of pump shutoff were not included in the averages.
Mixing of pumped water with surface water was more apparent in higher winds (>7 m s−1), when pitching of the apparatus caused regular interruptions in the water stream reaching the surface and rapid reformation of the skin. Infrared camera images showed a regular cycle of warm and cold spots under the radiometer with the rocking frequency of the apparatus. This behavior was also evidenced by greater variability in the 2-s IR data, which implies greater uncertainty, and also a cold bias, in the pumped-water IR temperature. To reduce the bias for these situations, the mean of the warmest (50%) of the pumped-water data was chosen as the IR temperature of the pumped water. In this study, strong wind days were of less interest than light wind days because skin temperature depressions are small (0.2°–0.3°C), and the upper surface of water is well mixed and statistically stationary. Greater accuracy on windy days could be achieved with longer periods of pumped water, for example, >30 s, to obtain more significant statistical samples. However, for extended operation in strong winds, a useful modification to the apparatus to ensure uninterrupted flow of 10-cm water to the surface would be the addition of a curved tube from the submerged pump that directs water downward from the side of the radiometer to the viewing point on the surface.
The uncertainty of the measured skin temperature can now be determined from the data in Table 1. Since the skin temperature is equal to the 10-cm thermocouple temperature plus the natural IR temperature minus the IR temperature of the pumped water, the total uncertainty, assuming uncorrelated errors, is the root-sum-square of 0.04°, 0.04°, and 0.04°, or 0.07°C. As noted above, the uncertainty is greater for strong winds. These results are comparable to those reported in a radiometer intercomparison test reported by Barton et al. (2004).
Field tests were also performed to evaluate the radiometer and thermocouple operation and the effect of sky reflection. Figure 3 shows data at midday in April in nearly calm winds (<1.5 m s−1), with and without a reflective aluminum cone in place over the radiometer target area. The cone obscured the zenith to ∼45°. The figures show a difference of 0.7°C between the IR temperature and the 10-cm thermocouple without the reflective cone in place that decreased to 0.45°C when the pump was operating [1208:00 to 1208:08 eastern daylight time (EDT)]. This difference implies a cool-sky radiative effect of −0.45°C for this day. The lower panel in Fig. 3 shows that the sky reflective effect is reduced to 0.2°C with the cone in place (1154:02 to 1154:08 EDT). Of particular interest is that with the cone in place, the IR temperature just after the pump turns off is within 0.05°C of the 10-cm thermocouple temperature. The reason for this is that the pump jet creates a slight mound of water under the radiometer that reflects cooler-sky radiation from zenith angles greater than 45° into the radiometer. Immediately after the pump shuts off, and before a cool skin forms, the water surface becomes nearly flat, and only radiation from the recently pumped 10-cm water is reflected back to the radiometer, instead of sky radiation. The same behavior was seen at 1156, 1158, and 1200 EDT, when the IR temperature just after pump shutoff exceeded the 10-cm thermocouple temperature by +0.09°, +0.02°, and +0.07°C, respectively. These differences combine with the −0.05°C difference at 1154 EDT to yield an average of +0.03°C, which is as close as can be expected considering other sources of variability.
Figure 3 also shows constant temperature readings when the pump was operating that indicate a short-term instrumental precision of <0.05°C and variability before and after the pump operation that is real and is not instrumental noise. The figure also shows the reformation of the cool skin in the 10 s after the pump was turned off.
4. Observations
As mentioned in the introduction, most studies of the skin temperature have been on the ocean, where temporal and spatial variability is small and the fetch is large. On lakes, however, temporal variation, internal boundary layers, and short fetches are much more important. These problems do not affect the measurement of the skin temperature but do influence the variables required to evaluate skin temperature parameterizations, for example, heat and momentum fluxes. This variability means that similarity relationships and the flux measurements on which they are based are subject to more uncertainty than over the ocean (Mahrt et al. 1998). On the other hand, the variability over inland lakes provides a greater range of conditions to test skin temperature theory than over the ocean.
Although fluxes of momentum, heat, and moisture can be measured directly with the covariance method, long averaging times are required to remove the effects of mesoscale eddies. However, this approach is not realistic for inland lakes because of rapidly changing air and water conditions. This problem can be reduced by placing sensors as close to the water’s surface as possible where the energy is contained in smaller (high frequency) eddies, but above the wave boundary layer where the fluxes depend on wave amplitude (Mahrt et al. 1998). Sensor placement near the surface of small lakes is possible because swell and large-amplitude waves common to the ocean are absent. Surface layer cospectra of vertical heat and momentum flux show that when the ratio of the height (z) to the Monin–Obukhov length (L) is small, the normalized frequency n = fz/U = 0.1, where f is the frequency and U is the wind speed (see, e.g., Fig. 2.18 in Kaimal and Finnigan 1994). For a wind speed of 3 m s−1 and z equal to 1 m, the maximum covariance corresponds to an eddy period of ∼3 s. This implies that reasonably accurate bulk fluxes can be obtained for averaging intervals as short as 1 min.
The applicability of the apparatus to skin temperature studies was investigated with 4 h of data collected on 4 May 2000 from the center of a lake 1 km wide by 6 km long. In Fig. 4 1-min-averaged data for the entire period are shown, and 2-s data for the period 1440–1459 EDT are shown in Fig. 5. During the 4-h period, the air temperature increased by 4°C, the relative humidity decreased by 30%, and the water temperature increased by 0.5°C. The 1- and 10-cm water temperatures were within 0.05°C of each other and were 0.2°C greater than the 50-cm temperature.
The 2-s radiometer data in Fig. 5 clearly show the periods of pump operation at 2-min intervals. Note that the water temperature measured with the pump on is 0.1°–0.2°C less than the 10-cm (thermocouple) temperature because of reflected cold-sky radiation. This effect was much less on this day than on the April day shown in Fig. 3 because of partly cloudy skies and warmer, more humid conditions. Figure 5 also shows a wind speed range of 0.5–7 m s−1 during this 20-min period, which was typical of the entire 4-h period. Also apparent in this figure is a negative correlation between the IR temperature and the wind speed (positive correlation between the skin temperature depression and the wind speed; see, e.g., the period between 1451 and 1455 EDT in Fig. 5).
The data collected with the skin temperature apparatus can be used to calculate the skin temperature depression with the bulk aerodynamic method (Wick et al. 1996). In contrast to previous studies, however, much shorter time intervals were employed in the present study (1 and 5 min) rather than 1 h. The shorter averaging interval was selected to explain the correlation between the variables on scales of minutes (Fig. 5) and because of rapid temporal variations observed over the lake.
The significance of bulk aerodynamic fluxes for varying time intervals can be demonstrated in several ways. For example, the flux calculated at two different elevations above the surface should be comparable if advection is negligible and averaging times are appropriate. A more direct method is to compare the measured water temperature near the surface with that calculated from the time-integrated bulk heat flux. The assumption here is that if the computed surface heat flux is correct (an appropriate time average is used) and corresponds to energy flow into and out of the water, it should be most apparent in the topmost layer of the water.
The results for the 5-min averages are shown in Figs. 6 and 7. The integrated heat flux is the numerator of Eq. (3), with coefficients b and c adjusted to produce the same trend as seen in the 1-cm temperature. This is justified because we are interested in the short-term fluctuations in fluxes. The flat section of the curves between 1200 and 1317 EDT is caused by a decrease in the solar flux of −100 W m−2 due to clouds.
Figure 7 shows the measured and calculated skin temperature depressions for the 4-h period. The “measured” value is found by first calculating the effect of sky reflection by subtracting the IR temperature with the pump operating from the corresponding 10-cm thermocouple temperature. This correction (∼0.1°C; see Fig. 5) was added to the IR measurements (pump off), which were then subtracted from the l-cm thermocouple temperatures to determine the skin depression with respect to the 1-cm temperature. This eliminates any temperature gradient between 10 and l cm (see, e.g., the period from 1452 to 1455 EDT in Fig. 5).
Figure 6 shows good correlation between the integrated heat flux and 1-cm temperature and implies that the 5-min averaging period is long enough to include most of the heat-transporting eddies. The calculated and measured skin temperature depressions are in good agreement (Fig. 7) and average 0.45°C. However, the correlation between the calculated and measured skin temperature depression is not as good as found for the temperature and heat flux. This is surprising since the skin is only a few microns thick and should approach equilibrium much faster than the l-cm temperature.
Figures 8 and 9 show the same results as Figs. 6 and 7, but for 1-min averages. The 1-min integrated heat fluxes are noisier than the 5-min results, the correlation between the calculated and measured skin depression is worse than for the 5-min data, and the variance of the latter is considerably greater than the former.
The rms difference between the measured and calculated skin depression for the 5-min data was 0.05°C, and the correlation coefficient was 0.52. These values can be compared, respectively, with 0.04°C and 0.64, obtained by Wick et al. (1996, their Table 2), which were obtained with Fairall et al.’s model and the Central Equatorial Pacific Experiment (CEPEX) data. [Wick et al.’s results with the Fairall et al. model and the METEOR data were worse than the CEPEX dataset but are not considered here because of the much wider range of skin depression (0.1°–0.5°C) than in the CEPEX data or the present study.] These results suggest that the 5-min averaging time is sufficiently long to determine heat and momentum fluxes to the same degree of accuracy as is possible with longer time averages.
The rms difference and correlation coefficient for the 1-min averages were 0.07°C and 0.35, respectively. As suggested above, these numbers are worse than for the 5-min fluxes, but this is partly due to greater variability in the 1-cm temperature, and not to the calculated heat and momentum fluxes.
The skin depression relationships are based on the momentum as well as the heat flux across the air–water interface. Although we have not assessed the accuracy of the bulk momentum flux, it is reasonable to assume that it is comparable to that of the heat flux. If this is true, then the computed skin temperature depression should correlate as well with the measured value as the 1-cm temperature does with the integrated heat flux. The fact that the correlation is significantly worse suggests that some quantities are not in steady state or that there is a more fundamental problem with the theory. One variable probably not in equilibrium with the surface wind is the surface water current. Our calculations assumed a zero current velocity. In reality, of course, the wind-driven current will be adjusting constantly to variable winds and inaccurate fluxes may result. These errors will affect the heat flux calculations as well as the skin depressions, but the latter may be more sensitive to them.
5. Conclusions
A portable, lightweight apparatus has been described that measures the skin temperature and related variables required to evaluate theoretical relations for the skin temperature depression. The apparatus has a minimal effect on the air and water flow and measures wind, temperature, and humidity within 1 m of the air–water interface.
Since the apparatus is intended for use mainly on small lakes, which have greater variability than the open ocean, its performance was evaluated with bulk aerodynamic relationships using 1- and 5-min averages. The time-integrated heat flux correlated well with the 1-cm water temperature for both the 1- and 5-min averages, and the measured skin temperature depression (0.45°C) also agreed well with the theoretical result. The rms difference and correlation coefficient between the measured and calculated skin temperature for 5-min averages were comparable to 1-h-average results obtained over the open ocean. However, the measured and calculated skin temperature depressions were not as well correlated in time as were the heat flux and 1-cm water temperature. This is believed to be due to variability on time scales longer than minutes in quantities such as the surface current or to possible deficiencies in the theoretical relations.
The apparatus collects high-resolution data closer to the air–sea interface than previous studies and can be deployed on lakes and coastal locations for periods of several days in light or moderate winds. However, it would have to be modified for longer, unattended operation in heavy seas.
Acknowledgments
This work was supported by the Department of Energy’s Multispectral Thermal Imager project. The authors thank Allen Weber for a critical review of the manuscript. Reviewers’ suggestions are also appreciated.
REFERENCES
Barton, I. J., Minnett P. J. , Maillet K. A. , Donlon C. J. , Hook S. J. , Jessup A. T. , and Nightingale T. J. , 2004: The Miami 2001 infrared radiometer calibration and intercomparison. Part II: Shipboard results. J. Atmos. Oceanic Technol., 21 , 268–283.
Blanc, T. V., 1987: Accuracy of bulk method determined flux, stability, and sea surface roughness. J. Geophys. Res., 92 , 3867–3876.
Donlon, C. J., and Nightingale T. J. , 2000: Effect of atmospheric radiance errors in radiometric sea-surface skin temperature measurements. Appl. Opt., 39 , 2387–2392.
Emery, W. J., Castro S. , Wick G. A. , Schlussel P. , and Donlon C. , 2001: Estimating sea surface temperature from infrared satellite and in situ temperature data. Bull. Amer. Meteor. Soc., 82 , 2773–2785.
Fairall, C. W., Bradley E. F. , Godfrey J. S. , Wick G. A. , Edson J. B. , and Young G. S. , 1996a: Cool-skin and warm-layer effects on sea surface temperature. J. Geophys. Res., 101 , 1295–1308.
Fairall, C. W., Bradley E. F. , Rogers D. P. , Edson J. B. , and Young G. S. , 1996b: Bulk parameterization of air-sea fluxes for Tropical Ocean-Global Atmosphere Coupled-Ocean Atmosphere Response Experiment. J. Geophys. Res., 101 , 3747–3764.
Kaimal, J. C., and Finnigan J. J. , 1994: Atmospheric Boundary Layer Flows. Oxford University Press, 289 pp.
Mahrt, L., Vickers D. , Edson J. , Sun J. , Hojstrup J. , Hare I. , and Wilczak J. , 1998: Heat flux in the coastal zone. Bound.-Layer Meteor., 86 , 421–446.
Schluessel, P., Emery W. J. , Grassi H. , and Mammen T. , 1990: On the bulk-skin temperature difference and its impact on sea surface temperature. J. Geophys. Res., 95 , 13341–13356.
Wick, G. A., Emery W. J. , Kantha L. H. , and Schlussel P. , 1996: The behavior of the bulk − skin sea surface temperature difference under varying wind speed and heat flux. J. Phys. Oceanogr., 26 , 1969–1988.
The skin temperature apparatus on a lake at the Savannah River site.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Heitronics KT 15.82 measurement of a blackbody held at a constant temperature of 20.2°C while the KT 15.82 was in an environmental chamber whose temperature was raised from 10° to 25°C during a 45-min period. Also shown is the internal KT 15.82 temperature.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Water temperature at a depth of 10 cm and the IR surface water temperature (top) without and (bottom) with a reflective aluminum cone in place over the radiometer. The heavy line labeled ON shows the time when the pump was operating, i.e., water from a depth of 10 cm was being pumped to the surface under the radiometer.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
One-minute averages of the air temperature, relative humidity, water temperatures, and IR (surface) temperature on 4 May 2000.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Data collected every 2 s from 1440 to 1449 EDT on 4 May 2000. The spikes in the IR data are periods when the pump was operating.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Numerator of Eq. (3) (integrated heat flux) and 1-cm temperature (5-min averages).
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Calculated and measured skin temperature depression for 5-min averages. For improved visualization, 0.1°C has been added to the measured values.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Same as in Fig. 6, but for 1-min averages.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Same as in Fig. 7, but for 1-min averages.
Citation: Journal of Atmospheric and Oceanic Technology 22, 9; 10.1175/JTECH1780.1
Sensor properties. The laboratory temperature accuracy is based on a comparison with a stirred water bath with embedded blackbody cavity over a temperature range of 1°–30°C after linear correction (Heitronics) and constant offset (Vaisala and thermocouples). Blackbody errors are not included. The radiometer field precision is derived from measurements during the pump operation cycle and includes both precision and resolution.
