Variables for FASTEX experiments : date = year-month-day hour = decimal hour lat = latitude North ( - = South ) lon = longitude East ( - = West ) ws_s = true wind speed , ETL sonic ws_a = true wind speed, imet provane anemometer seaT_s = sea snake temperature, ETL 0.05 m depth seaT_t = tsg water temperature, 5 m depth airT_e = air temperature, ETL airT_i = imet air temperature qa_e = air specific humidity, ETL qa_i = imet air specific humidity swr = downward solar flux, ETL lwr = downward IR flux, ETL wdir_e = true wind direction, ETL sonic wdir_i = true wind direction, clockwise rel north, imet rain = rainrate, ETL STI optical rain gauge, uncorrected shf_c = sensible heat flux, covariance, ETL sonic anemometer shf_i = sensible heat flux, ID, ETL sonic anemometer shf_b = bulk sensible heat flux lhf_c = latent heat flux, covariance lhf_i = latent heat flux, ID lhf_b = bulk latent heat flux, includes Webb et al. correction hlwebb = correction to measured latent heat flux, Webb et al. head = ship heading, deg. clockwise rel north, SCS laser ring gyro reldir = relative wind direction, clockwise rel ship's bow, ETL sonic urel = relative wind speed, ETL sog = speed over ground, SCS gps cog = course over ground, SCS gps qse = sea surface specific humidity, from snake qs_tsg = imet bulk water specific humidity taucx = covariance streamwise stress, ETL sonic anemometer taucy = covariance cross-stream stress, ETL sonic anemometer tauib = ID streamwise stress, ETL sonic anemometer taub = bulk wind stress along mean wind sigoph = standard deviation of ophir fast hygrometer clear channel tiltx = flow tilt at ETL sonic anemometer, earth frame ushp = doppler log, SCS J = ship plume contamination index Jm = ship maneuver index hrain = rain heat flux, Gosnell et al. ct = ct^2 cq = cq^2 cu = cu^2 cw = cw^2 zu_etl = height of mean wind sensor, 17.7 m zt_etl = height of mean air temperature sensor, 15.5 m zq_etl = height of mean air humidity sensor, 15.5 m ETL Fastex Cruise Fluxes: Release #1.0 The first release of turbulent and bulk fluxes for Fastex cruise. This document is the Readme for fast_hr.txt and fas_10.txt files. The _hr refers to hourly averages; the _10 to 10-minute averages. crus Name Jul. Day Dates Hours Nom. Lat/Lon fast fastex 357-395 12/23-1/24 730 45 N 35 W The data files fast_hr.txt and contain measurements of turbulent and radiative fluxes plus bulk meteorological variables from the R/V Knorr in late December 1996 through January 1997. The files also contain bulk estimates of the turbulent fluxes computed using a recently updated version (2.6) of the COARE flux algorithm (see ftp://ftp.etl.noaa.gov/et7/users/cfairall/bulkalg/ for documentation). Most quantities have been subjected to several rounds of intercomparison/calibration scrutiny. Both direct (covariance) and inertial-dissipation (ID) turbulent flux calculations are included in this present data. The files are 47 columns; lengths depend on the length of the data sets (hours column in table above gives length for the _hr data). Following the decimal julian date, the next 10 columns are mean variables from the ETL system; the last 8 columns are similar data from the ships sensors. Columns 12-21 are turbulent fluxes (covariance, ID, and bulk); columns 22-23 the radiative fluxes and 24 the rain rate. Columns 25-28 are turbulence data quality indicators; 29-32 turbulent structure function parameters (indices of small-scale turbulence in the inertial subrange). Columns 33-34 are the minor (rain and Webb) heat flux components; 35-36 the latitude and longitude; 37-39 the heights of the ETL wind, temperature, and humidity mean sensors. The data columns are presently labeled and numbered at the top (first two rows). Delete the first two rows and restore as a .txt file so they can be directly acquired with a MATLAB 'load' statement. x=load('your_local_directory\crus_hr.txt');%read file with hr-average data; set your local directory The data in this file comes from three sources: The ETL motion-correct flux package [sonic anemometer acquired at 20.83 Hz, fast hygrometer acquired at 20 Hz, and 6-component motion measurements acquired at 10 Hz), the ships SCS system (acquired at 2 sec intervals), and the ETL mean measurement systems (sampled at 10 sec and averaged to 1 min). The sonic is 5 channels of data; the SCS file is 13 channels, and the ETL mean system is 16 channels. A series of programs are run that read these data files, decode them, compute covariance, variances, spectra, etc. at 10-min time resolution. A set of 10-min process files are written for each cruise. One particular file, crus_da.dat consists of 164 columns of 10-min data including the turbulent and mean variables used here. A final program reads the crus_da.dat files, applies various corrections, computes the ID and bulk fluxes and the data quality indices, and writes the crus_10.txt files. This program also contains routines that average the data to fixed one-hr time intervals, computes the ID and bulk fluxes from one-hr means, and writes the crus_hr.txt files. Note, that in the case of the turbulent variables (covariances, variances, structure function parameters) only 10-min values that pass the data indicator criteria are used in the one-hr average. Further experimental details are as follows: True wind speed is computed from the sonic anemometer using the ship's Laser ring gyro and the GPSsog/cog; thus, it is interpreted as the speed relative to the fixed earth. Air temperature and humidity from the ETL (aspirated Vaisala HMP-35) and the ship's IMET T/RH sensor were carefully compared with the handheld psychrometer values obtained during the experiment. Based on these intercomparisons, we decided to use the IMET values. At various times during some cruises the seasnake was removed from the water, in which case the ship's thermosalinograph, corrected for any warm layer effects, was substituted. Sea surface temperature measurements often showed contamination by the ship during stops and maneuvers at stations. The sensor was airborne a lot when the ship was underway in the stronger wind/rougher sea days and the accuracy under those conditions is uncertain. In many cases, the ships thermosalinograph was not properly flushed as the pump tended to quit when the intake came out of the water. Longwave flux was obtained from 2 Eppley PIR units, logged and computed as per Fairall et al. Jtech, 1998. Shortwave flux was obtained from 2 Eppley PSP units. The rainrate was obtained from an NCAR raingauge. However, data were only available in the second half of the experiment. Prior to that period, values from the Kiel gauge are used. Bulk estimates of air sea fluxes were computed using the COARE bulk algorithm version 2.6, which differs from the published version in only minor details. Turbulent fluxes were computed from motion-corrected time series of fast sensors. Covariance latent heat fluxes were obtained by cross-correlating the motion-corrected vertical velocity with fast humidity fluctuations from an OPHIR IR hygrometer. The turbulent fluctuations from the OPHIR were scaled by the mean ratio of the OPHIR humidity to the IMET humidity (i.e., increased about 9%) for application in computations of the w-q covariance and the standard deviation of humidity fluctuations. An additional scaling factor (1.0% increase) for the w-q covariance was used to account for the physical separation of the OPHIR and the sonic. Using a right-handed coordinate system with x boward, y to the port, and z up, the displacement vector from the OPHIR to the sonic is (-0.05, 0.0, 1.57) m. We used Kristensen et al. (J. Atmos. Oceanic Tech., 14, 814-821, 1997) to estimate the correlation loss. We normally use the OPHIR clear channel counts as an index of clean optics. Past experience has shown that the absolute calibration of the OPHIR is degraded as the optics become contaminated with salt and/or water. With clean optics the mean clear channel counts are around 2800-3000 and the standard deviation is between 2-15. In experiments with predominantly 'nice' weather we reject humidity flux values when the mean counts are below 2600 and/or the standard deviation is above 20. FASTEX had predominantly bad weather, so those criteria reject the majority of the data. To estimate the contamination effects on the fluxes, we did a linear regression of Flux/Flux_bulk vs mean clear channel counts for latent heat flux. For covariance, there is a positive slope and the ratio is 1.0 for mean counts greater than 2800; for ID fluxes, the slope is negative and the ratio is 1.05 for mean counts greater than 2800. Therefore, we have used these slopes to adjust the covariance and ID latent heat fluxes for mean counts <2800. At a mean clear channel count of 2000, this amounts to a 15% increase in covariance and a 19% decrease in ID latent heat fluxes. While an adjustment of this type violates our usual policy of never adjusting or rejecting direct flux values based on their agreement with the bulk model, we have accepted this compromise because it has essentially no effect of the average of the covariance and ID fluxes (i.e., they are adjusted in opposite directions). Purists who find this adjustment overly objectionable can restrict their analysis to data with the standard deviation of clear channel counts less than 20. An unadjusted version of the data set is available on request. Because the IR hygrometers detect water vapor mass concentration (rho_v in kg/m^3), their water vapor - velocity correlations must be corrected as per Webb et al Hlatent = Le + hl_webb The values given for covariance and ID latent heat fluxes in the file are Le. Values for hl_webb are included in column 34. This should be applied to the covariance and ID values. It is already included in the bulk values given here. Sensible heat flux was computed from vertical velocity - sonic temperature covariance. The humidity contribution to sonic temperature was removed using the bulk latent heat flux. Simple data quality indicators have been use to edit the turbulence data. At each 10-min time step, values for ID turbulence variables were computed if the following criteria were met: jj=find((reldir<90 | reldir>270) & sig_h<8 &sig_u<2 &sp2<2 & org<5 & sqrt(ww)./ugw<.55+.002*U.^2 & sqrt(vv)./ugu<1.1); reldir is the relative wind direction sig_h the standard deviation of ship heading (deg) sig_u the standard deviation of the ship speed sp2 the standard deviation of cross-ship motion corrections org the rainrate ww the vertical velocity variance vv the cross stream velocity variance Otherwise, the ID variables were set to NaN (not a number). In processing the 10-min data to one-hr averages, only the jj rows were used in averaging the turbulence variables. If there were no valid values in the 1-hr interval, the turbulence variables were set to NaN. Note that these criteria are rather weak; the turbulence data have not been strictly edited for outliers and some still remain. The criteria given above were subdivided to be approximately compatible with indices used in the past: J=ones(length(jdy),1); ii=find(sqrt(ww)./ugw<.55+.002*U.^2 & sqrt(vv)./ugu<1.1); J(ii)=0; jm=3*ones(length(jdy),1); ij=find(sig_h<8 & sig_u<2 & sp2<2); jm(ij)=sig_u(ij); *A value of J=0 implies no ship contamination. *A value of jm<3 implies no significant maneuver during the average. *Sigoph is an index of salt or rain contamination on the OPHIR fast hygrometer optics (see discussion above). Values for fluxes begin to be affected when sigoph exceeds 20 although a threshold of 50 gives acceptable data. *Turbulent fluxes are computed by converting the anemometer 3-component velocities to fixed earth coordinates, correcting the fast time series for ship motion, and re-setting the coordinate system normal to the 10-min mean flow through one rotation about the original vertical and one tilt. The variable tiltx gives the tilt used for the computation. Experience shows that tilts greater than less than about 5 or greater than about 15 deg give questionable fluxes. --------------72A15EC742B767E4BBC76EA9--