Progress towards developing the observing system for air-sea fluxes proposed at OceanObs99 is reviewed. The main elements of the system are surface flux buoys, improved Voluntary Observing Ship (VOS) observations and automated flux packages on research ships. Surface flux buoys are being deployed as part of a broader ocean time series program although in limited numbers at present. The VOSClim project is producing a high quality subset of VOS observations with associated metadata that that can potentially be used to reduce uncertainty in model based fluxes. Automated flux packages are not yet routinely deployed but prototypes have been developed and used on a number of research cruises.

Marine air temperature reports from ships can contain significant biases due to the solar heating of the instruments and their surroundings. However, there have been very few attempts to derive corrections. The biases can reverse the sign of the measured air–sea temperature differences and cause significant errors in the sea surface latent and sensible heat flux estimates. In this paper a new correction for the radiative heating errors is presented. The correction is based on the analytical solution of the heat budget for an idealized ship, using empirical coefficients to represent the physical parameters. For the first time heat storage is included in the correction model. The heating errors are estimated for the Ocean Weather Ship Cumulus and the coefficients determined. When the correction is applied to the Cumulus data the average estimated error is reduced from 0.32° to 0.04°C and the diurnal cycle in the error is removed. The rms error is reduced by 30%. The correction technique, although not the coefficients derived here that are specific to the Cumulus, can be applied to air temperature data from any type of ship, or to data from groups of ships such as the Voluntary Observing Ships.

In this paper we shall consider the accuracy of in situ measurements of wind and wind stress over the ocean, and also the contrasting characteristics of different wind stress parameterisations. There is much scatter in the drag coefficient or roughness length measurements for winds below 10 m/s. While this scatter may be caused by sampling limitations and other measurement errors, there is increasing evidence that swell waves may modify the effective surface roughness. However, at these lower wind speeds, the resulting uncertainty in the wind stress is very small, only a few percent of the magnitude of the wind stress at 20 m/s. At wind speeds between 10 to 20 m/s the measurements from the open ocean are less scattered with both eddy correlation and inertial dissipation wind stress estimates giving similar values. At these higher wind speeds the instrumentation on meteorological buoys is relatively low compared to the height of the dominant waves. However we shall present data for wind velocity fluctuations and buoy motion which demonstrate that meteorological buoys can be used to adequately determine the wind velocity, and hence the wind stress, even in high wave conditions. At wind speeds above 20 m/s, different parameterizations predict significantly different wind stress values. At these higher wind speeds we suggest that, compared to the behaviour at lower wind speeds, the sea surface roughness will increase less rapidly with increasing wind speed. Unfortunately the available wind stress data are few, particularly for winds above 25 m/s, and insufficient to test this prediction.

[1] The accuracy of two empirical formulae used in recent climatological studies to estimate the atmospheric longwave flux at the ocean surface from ship meteorological reports has been evaluated using research cruise measurements from the northeast Atlantic. The measurements were obtained with a pyrgeometer and corrected for differential heating of the pyrgeometer dome and shortwave transmission through the dome. The formulae tested were from Clark et al. [1974] and Bignami et al. [1995]; neither was capable of providing consistently reliable estimates of the longwave flux. Clark overestimated the mean measured longwave of 341.1 Wm(-2) by 11.7 Wm(-2), while Bignami underestimated by 12.1 Wm(-2). A new formula is developed that expresses the effects of cloud cover and other parameters on the longwave through an adjustment to the measured air temperature. The air temperature is adjusted by the amount necessary to obtain the effective temperature of a blackbody with a radiative flux equivalent to that from the atmosphere. A simple parameterization of the adjustment in terms of the total cloud amount gives longwave estimates that have an improved mean bias error with respect to the measurements of -1.3 Wm(-2). The new formula is still biased under overcast, low cloud base conditions. However, by including a dependence on dew point depression in the formula, this bias is resolved, and the mean error reduced to 0.2 Wm(-2). The new formula has been tested using measurements made on two subsequent cruises and found to agree to within 2 Wm(-2) in the mean at middle-high latitudes.

bridge University Press. IPCC (2001) Climate Change 2001: The Scientific Basis, Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change, Houghton JT, Ding Y, Griggs DJ, et al. (eds) Cambridge: Cambridge University Press. Kettle AJ and AndreaeMO (2000) Flux of dimethylsulphide from the oceans: a comparison of updated data sets and flux models. Journal of Geophysical Research 105: 26793–26808. Kettle AJ, AndreaeMO,AmourouxD, et al. (1999) A global database of sea surface dimethylsulfide (DMS) measurements and a procedure to predict sea surface DMS as a function of latitude, longitude, and month. Global Biogeochemical Cycles 13: 399–444. Kettle AJ, Rhee TS, von Hobe M, Poulton A, Aiken J and Andreae MO (2001) Assessing the flux of different volatile sulfur gases from the ocean to the atmosphere. Journal of Geophysical Research 106: 12193–12209. KingDB, Butler JH,Montzka SA,Yvon-Lewis SAandElkins JW (2000) Implications of methyl bromide supersaturations in the temperate North Atlantic Ocean. Journal of Geophysical Research 105: 19763–19769. LawCS, Sjoberg TN and Ling RD (2002) Carbonmonoxide cycling and flux in a seasonal study of the Scheldt Estuary. Biogeochemistry 59: 69–94. Liss PS and Merlivat L (1986) Air–sea gas exchange rates: introduction and synthesis, In Buat-Menard P (ed.) The Role of Air–Sea Gas Exchange in Geochemical Cycling, pp. 113–127. Norwell, MA: D. Reidel. Liss PS and Slater PG (1974) Flux of gas across the air–sea interface. Nature 247: 181–184. Lovelock JE (1987) Gaia, A New Look at Life on Earth. Oxford: Oxford University Press. Moore RM and Groszko Z (1999) Methyl iodide distribution in the ocean and fluxes to the atmosphere. Journal of Geophysical Research 104: 11163–11171. Nightingale PD, Liss PS and Schlosser P (2000) Measurements of air–sea gas transfer during an open ocean algal bloom. Geophysical Research Letters 27: 2117–2120. Takahashi T, Feely RA, Weiss RF, et al. (1997), Global air–sea flux of CO2: an estimate based on measurements of sea-air pCO2 difference. Proceedings of the National Academy of Sciences of the USA 94: 8292–8299. Watts SF (2000) The mass budgets of carbonyl sulphide, dimethyl sulphide, carbon disulfide, and hydrogen sulphide. Atmospheric Environment 33: 3503–3514. Wanninkhof R (1992) Relationship betweenwind speed and gas exchange over the ocean. Journal of Geophysical Research 97: 7373–7382. WanninkhofRandMcGillisWR (1999)A cubic relationship between air–sea exchange and wind speed. Geophysical Research Letters 26: 1889–1892. Woolf DK (1997) Bubbles and their role in gas exchange. In: Liss PS, Duce RA (eds) The Sea Surface and Global Change, pp. 173–205.Cambridge:CambridgeUniversity Press. Xie HX and Moore RM (1999) Carbon disulfide in the North Atlantic and Pacific Oceans. Journal of Geophysical Research 104: 5393–5402.

The accuracy of two empirical formulae used in recent climatological studies to estimate the atmospheric longwave flux at the ocean surface from ship meteorological reports has been evaluated using research cruise measurements from the northeast Atlantic. The measurements were obtained with a pyrgeometer and corrected for differential heating of the pyrgeometer dome and shortwave transmission through the dome. The formulae tested were from Clark et al. [1974] and Bignami et al. [1995]; neither was capable of providing consistently reliable estimates of the longwave flux. Clark overestimated the mean measured longwave of 341.1 Wm−2 by 11.7 Wm−2, while Bignami underestimated by 12.1 Wm−2. A new formula is developed that expresses the effects of cloud cover and other parameters on the longwave through an adjustment to the measured air temperature. The air temperature is adjusted by the amount necessary to obtain the effective temperature of a blackbody with a radiative flux equivalent to that from the atmosphere. A simple parameterization of the adjustment in terms of the total cloud amount gives longwave estimates that have an improved mean bias error with respect to the measurements of −1.3 Wm−2. The new formula is still biased under overcast, low cloud base conditions. However, by including a dependence on dew point depression in the formula, this bias is resolved, and the mean error reduced to 0.2 Wm−2. The new formula has been tested using measurements made on two subsequent cruises and found to agree to within 2 Wm−2 in the mean at middle‐high latitudes.

It is proposed that the sea surface roughness zo can be predicted from the height and steepness of the waves, zo/Hs = A(Hs/Lp)B, where Hs and Lp are the significant wave height and peak wavelength for the combined sea and swell spectrum; best estimates for the coefficients are A = 1200, B = 4.5. The proposed formula is shown to predict well the magnitude and behavior of the drag coefficient as observed in wave tanks, lakes, and the open ocean, thus reconciling observations that previously had appeared disparate. Indeed, the formula suggests that changes in roughness due to limited duration or fetch are of order 10% or less. Thus all deep water, pure windseas, regardless of fetch or duration, extract momentum from the air at a rate similar to that predicted for a fully developed sea. This is confirmed using published field data for a wide range of conditions over lakes and coastal seas. Only for field data corresponding to extremely young waves (U10/cp > 3) were there appreciable differences between the predicted and observed roughness values, the latter being larger on average. Significant changes in roughness may be caused by shoaling or by swell. A large increase in roughness is predicted for shoaling waves if the depth is less than about 0.2Lp. The presence of swell in the open ocean acts, on average, to significantly decrease the effective wave steepness and hence the mean roughness compared to that for a pure windsea. Thus the predicted open ocean roughness is, at most wind speeds, significantly less than is observed for pure wind waves on lakes. Only at high wind speeds, such that the windsea dominates the swell, do the mean open ocean values reach those for a fully developed sea.

The effects on a dataset of smoothing by successive correction have been investigated. The resulting spatial resolution is estimated using a distribution of ship reports from a sample month. Although the smoothing uses the same characteristic radii over the whole globe, the resulting resolution is spatially variable and, in data-sparse regions, will show large month-to-month variability with changes in the distribution of the ship tracks. The climatological dataset, which is gridded at 1 degrees, is shown to have a typical resolution of 3 degrees. In some regions the resolution is much coarser.Using sea surface temperature as an example, it is shown that the successive correction procedure as used, for example, in a recent climatological dataset, is not successful in removing all of the noise in data-sparse regions. Additionally, the well-defined intermonthly variability in the main shipping lanes, where there are many observations, is degraded by the influence of poorer-quality data in the surrounding regions. This typically increases the intermonthly variability estimates in the shipping lanes by a factor of 2. Further, the reduction of intermonthly variability, by up to a factor of 6, in highly variable regions such as the Gulf Stream, is greater than can be accounted for by noise in the individual ship reports. This reduction is due to the removal of small-scale variability by the smoothing process. Removal of coherent and persistent small-scale variability has an effect on the temporal and spatial characteristics of the data. It is suggested that smoothing by successive correction, although commonly used, is poorly suited to such spatially inhomogenous data as those from the merchant ships.However, the effect of successive correction on variability analysis using empirical orthogonal functions (EOFs) is shown to be small for the most significant modes of variability identified in the Gulf Stream region. This is because the EOF analysis picks out the large-scale variability in the highest-order modes. However, too large a fraction of the total variance explained is ascribed to the large-scale modes of variability. Variability with small spatial scales is more likely to be significant if raw data are used in the EOF analysis. Little significance should be given to EOF modes with spatial scales similar to the size of gaps between shipping lanes; this varies from region to region.

A synthetic dataset is used to show that apparent variations between different stability classes in the mean drag coefficient, C-D10n, to wind speed relationship can be explained by random errors in determining the friction velocity u(*). Where the latter has been obtained by the inertial dissipation method, the variations in C-D10n have previously been ascribed to an imbalance between production and dissipation in the turbulent kinetic energy budget. It follows that the application of "imbalance corrections" when calculating u(*) is incorrect and will cause a positive bias in C-D10n, by about 10(-4).With no imbalance correction. random errors in u(*) result in scatter in the C-D10n values, but for most wind speeds, there is no mean bias. However, in light winds under unstable conditions random errors in u(*) act to positively bias the calculated C-D10n values. This is because the stability related effects are nonlinear and also because for some records for which C-D10n would be decreased, the iteration scheme does not converge. The threshold wind speed is typically 7 m s(-1), less for cleaner datasets. The biased C-D10n values can be avoided by using a u(*) value calculated from a mean C-D10n-U-10n relationship to determine the stability. The choice of the particular relationship is not critical. Recalculating previously published C-D10n values without imbalance correction, but with anemometer response correction, results in a decrease of C-D10n but only by about 0.05 x 10(-3).In addition to removing a previous cause of scatter and uncertainty in inertial dissipation data, the results suggest that spurious stability effects and low windspeed biases may be present in C-D10n estimates obtained by other methods.

We read with interest Tolman's (1998, hereafter T98) analysis of the effect of observation errors on validation of marine winds. His demonstration of the importance of the correct treatment of the errors in comparative analysis of wind-speed land other types of) data is welcome, as examples of inaccurate comparisons are common, particularly for satellite data validation which is often published in non-refereed reports. Kent et ad. (1998) show that apparent trends of the order reported can be directly attributed to errors in both datasets and that the trend appears smaller and in the opposite sense if the satellite wind speed rather than the ship wind speed is used as the independent variable for plotting.T98 states that 'in special cases, where the ratio of... errors can be estimated, more advanced regression techniques can be used'. Our aim in this note is to advocate a simple, established method of data analysis which can lead to reliable comparisons of pairs of nearly co-located and simultaneous observations from two sources, both containing random errors. This method enables the use of these advanced regression techniques since the ratio of random errors in the datasets to be compared is estimated. The error estimates can then be verified by using the effects of errors on bin-averaged analyses highlighted by T98. Following T98 and Kent et al. (1998), we shall discuss satellite and in situ wind-speed data comparisons, but again expect the results to be more widely applicable.

Quality controlled wind speed observations from merchant ships have been compared with ERS-1 scatterometer data. The ship and satellite wind speed pairs were well correlated at 120km separation in the open ocean, reducing to 40km in the North Sea. The maximum allowed separation of ship and scatterometer wind speed pairs had to be further reduced to 20km in the North Sea to avoid matching of coastal ship wind speed data with scatterometer data from more exposed regions. Spurious biases in the comparisons were caused by the error variability of the scatterometer data (0.5 m s(-1)) being significantly less than that for the ships (2.0 m s(-1)). The unbiased regression was:U-10n(ship)= 1.025 x U-10n(satellite)+ 0.255No significant enhancement of the scatterometer wind speeds occurred in the coastal region.

A large dataset of wind stress estimates, covering a wide range of wind speed and stability conditions, was obtained during three cruises of the RRS Discovery in the Southern Ocean. These data were used by Yelland and Taylor to determine the relationship between 10-m height, neutral stability values for the drag coefficient, and the wind speed, and to devise a new formulation for the nondimensional dissipation function under diabatic conditions. These results have been reevaluated allowing for the airflow distortion caused by the ship. The acceleration and vertical displacement of the flow have been modeled in three dimensions using computational fluid dynamics (CFD). The CFD modeling was tested, first by comparison with.wind tunnel measurements on models of two Canadian research ships and second, by analysis of data from four anemometers on the foremast of the RRS Charles Darwin. Originally, the four anemometers gave drag coefficient values that differed by up to 20% from one to another and were all unexpectedly high.The CFD results showed that the airflow had been decelerated by 4%-14% and displaced vertically by about 1 m. These effects caused the original drag coefficient results to be overestimated by up to 60%. After correcting for flow distortion effects, the results from the different anemometers became consistent, which gave confidence in the quantitative CFD-derived corrections.The CFD modeling showed that the anemometer position on the RRS Discovery was much less affected by airflow distortion. For a given wind speed the CFD corrections reduced the drag coefficient by about 6%. The resulting mean drag coefficient to wind speed relationship confirmed that suggested by Smith from a more limited set of open ocean data.The effects of how distortion are sensitive to changes in the relative wind direction. It is shown that much of the scatter in drag coefficient estimates may be due to variations in airflow distortion rather than to the effect of changing sea states. The Discovery wind stress data is examined for evidence of a sea-state dependence: none is found. It is concluded that a wave-age-dependent wind stress formulation is not applicable to open ocean conditions.

An automatic inertial dissipation system was used during three cruises of the RRS Discovery in the Southern Ocean to obtain a large dataset of open ocean wind stress estimates. The wind speed varied from near calm to 26 m s(-1), and the sea-air temperature differences ranged from -15 degrees to +7 degrees C. The data showed that the assumption of a balance between local production and dissipation of turbulent kinetic energy is false and that the sign and magnitude of the imbalance depends critically on both stability and wind speed. The wide range of stability conditions allowed a new formulation for the nondimensional dissipation function under diabatic conditions.A minimum in the 10-m neutral value of the drag coefficient occurred at 6 m s(-1). At lower wind speeds the data were fitted by the relationship1000C(D10n) = 0.29 + 3.1/U-10n + 7.7/U-10n(2) (3 less than or equal to U-10n less than or equal to 6 m s(-1)),where U-10n is the 10-m neutral wind speed (m s(-1)). At higher wind speeds1000C(D10n) = 0.60 + 0.070*U-10n (6 less than or equal to U-10n less than or equal to 26 m s(-1)),which gives drag coefficients that are about 10% higher than those from previous open ocean studies (which assumed a balance between production and dissipation). Wave measurement suggested that the sea state was not, on average, fully developed at wind speeds above 15 m s(-1). However, contrary to findings from other studies, no large anomalies in the drag coefficient were detected despite the range of conditions and sea states encountered. It is believed that the ideal conditions (such as the absence of swell) needed to detect the effects of sea state on the wind stress may occur rather infrequently over the open ocean.

The effect of incoming solar radiation on merchant ships' observations of air temperature was assessed as part of the Voluntary Observing ShiPs' Special Observing Project for the North Atlantic (VSOP-NA). The ships' reports were compared with interpolated output from a numerical weather model. Differences between the ship values and the model values for air temperature (DELTAT(a)) were found, in the mean, to be independent of instrument type, ship size, and, except for very badly exposed sensors, exposure. The differences were related to the relative wind speed over the ship (V) and the incoming shortwave radiation (R). The formula derived for the radiative heating error deltaT was deltaT = 2.7 x 10(-3) R - 3.2 X 10(-5) RV, where deltat has units of degrees Celsius, R is in watts per square meter, and V is in knots.After correcting the DELTAT(a) values, an approximately constant bias remained with the ship reports on average 0.4-degrees-C lower than the model air temperatures. This offset probably represents a mean bias in the model estimates; however, a residual bias in the ship observations is also a possibility. There was also evidence that heat generated by the ship caused a temperature overestimate of about 0.4-degrees-C at zero relative wind, decreasing to a negligible level at a relative wind speed of 20 kt.For the North Atlantic dataset used, the correction reduced daytime marine air temperature reports by 0.63-degrees-C on average. Applying the correction to the VSOP-NA air temperature data was found to significantly change estimates of sensible and latent heat fluxes.

A new commercial slide latex particle agglutination test for rapid (2 min) diagnosis of vaginal candidosis was evaluated and compared with conventional methods. Of the 263 women studied, 63 (23.9%) had yeasts in the vagina. Clinical signs of vulvitis or vaginitis were seen in 23 women (8.8%) and 40(15.2%) were harbouring yeasts without clinical signs. Yeast counts were generally higher in women with clinical signs of vaginal candidosis than in those without. The test was positive in 15 of the 23 women (65.2%) with clinical signs, the incidence of a positive test increasing in direct proportion to the amount of yeasts isolated. The test's sensitivity, specificity and predictive values were comparable to those of microscopy and culture. Being both rapid and simple to perform, this new test offers a useful alternative to conventional methods for the diagnosis of vaginal candidosis.