VOCALS2008_soundings_z_V4.0.readme.txt :: 2010 August 24 :: Simon de Szoeke README metadata for VOCALS2008_soundings_z_v4.0.nc Title: Rawinsonde data for VOCALS 2008 Version: 4.1 Creation date: 2010-08-23 Author: Simon de Szoeke Data are provided by Simon de Szoeke (Oregon State University), Sandra Yuter (North Carolina State University), and Chris Fairall (NOAA/ESRL/PSD) with funding from the NOAA Climate Program Office. Find this metadata at ftp://ftp.coas.oregonstate.edu/dist/sdeszoek/vocals/VOCALS2008_soundings_z_v4.0.readme.txt Data Dimensions 1 launch rawinsonde launch number (216, record dimension) 2 altitude altitude in 10-m increments (2501) Data Set Variables Name [units] Description TIME(launch) vectors 1 year [Gregorian year] always 2008 2 yday [day] Decimal year day of sonde launch, since 2008 January 1 00:00 UTC 3 month [month] month of year 4 day [day] day of month 5 hour [hour] hour of day 6 minute [minute] minute of hour 7 lat [degrees] north latitude 8 lon [degrees] east longitude 20 tqflag 8-bit quality control flag VERTICAL(altitude) vector 9 height [meters] bottom of 10-m vertical averaging interval DATA(launch, altitude) 10 pres [hPa] pressure 11 T [degree C] temperature 12 RH [percent] relative humidity 13 Td [degree C] dew point temperature (*) 14 wnddir [degree] meteorological convention (out-of) wind direction; degrees heading clockwise from north 15 wndspd [m/s] wind speed 16 u [m/s] zonal eastward wind velocity 17 v [m/s] meridional northward wind velocity 18 theta [K] potential temperature (+) 19 q [g/kg] specific humidity (*+) (*) derived from RH (+) derived from temperature In the netcdf(.nc) file, IEEE not-a-number (NaN) values denote missing values. Preprocessing Preprocessing of the raw digital rawinsonde signals is performed by three generations of the Vaisala DigiCora software. Most data is processed by DigiCora version 3.62. Three soundings could not be processed with DigiCora 3.62 and were processed with DigiCora 3.51 instead. The processed files eliminate glitches, but often interpolate across true observations in the vicinity of the sharp trade inversion. These interpolated layers were identified by a human operator and reverted to DigiCora Research-mode data, which does not interpolate. Interpolation to Standard Levels and Derived Variables Data provided here are vertically interpolated to standard 10-m height increments. For a nominal ascent rate of 5 m/s the 10 m height interval corresponds roughly to the 2 s time interval to which the data are interpolated by the Vaisala DigiCORA sonde processing software. The data decorrelation time scale is estimated at about 5 s from the sonde ascent rate and response time of the sensors, so the 10 m and 2 s intervals slightly over-sample the data. Thermodynamic variables pressure, temperature, and relative humidity are measured by sensors on the sonde boom. Other thermodynamic variables Td, theta, and q are computed from these in-situ measurements. Td is computed by the Vaisala software. Variables theta and q are computed in post-processing from T and RH. Winds are calculated from horizontal displacements of the rawinsonde measured by the GPS receiver on the sonde. Wind direction and speed are output at 2 s increments by the Vaisala sounding processing software. These are converted to vector eastward zonal and northward meridional components (u, v) and interpolated to standard 10-m intervals. Wind direction and speed (wnddir, wndspd) are computed from the interpolated wind components. Quality Control Hugely negative relative humidity was found for two sondes. All humidity variables (RH, Td, q) for these sondes are set to NaN, and the launch is flagged (tqflag bit 6) to indicate a humidity issue has been masked out. Three sondes had alternating cycles of RH values. This is probably due to a mismatch between the two hygrometers on the sonde, which are alternately sampled and heated to remove liquid water contamination. The humidity variables for these sondes are set to NaN, and they are flagged for humidity cycling (tqflag bit 8). Sonde observations from 20-40 m have been compared with observations from sensors on the ship's forward mast. The 20-40 m level average agrees best with the mast temperature (to within 0.1 C in the mean). The temperature from 15 sondes disagrees by more than 1C with the mast thermometer. Their temperature is flagged (tqflag bit 2), and the temperature data (T, q, theta) are adjusted to agree with the mast at 20-40 m. The adjusted temperature agrees better with neighboring sondes in the marine boundary layer and lower troposphere, but often it does not in the upper troposphere and stratosphere. We also flag adjusted data with a temperature caution above 5000 m (tqflag bit 1). Use care in interpreting the adjusted temperature in the upper atmosphere. If the disagreement between sonde and mast is more than 3.5 C the data are replaced with NaNs. Sondes sit on the deck of the ship (~7 m) before being launched and the sonde processing software adjusts the first few data points to match data entered on a terminal during launch preparation. Since the source of these lower-altitude data is different from the sensors on the rawinsonde itself, data from levels below 30 m are expunged from the rawinsonde data set. Wetting of the thermistor results in brief cooling right above cloud top. Temperature lapse rates more than 15 C/km near the inversion are flagged to caution of possible wetting (tqflag bit 1). Wind data that are zero for both wind speed and direction are assumed to represent missing wind speeds and are replaced with NaNs before computing wind vector components or interpolating. Quality Control Flags The variable 'tqflag' contains a single byte for each (launch,altitude) denoting data quality for that sounding and level. The bits of tqflag should be interpreted as follows: flag bit description 1 T caution 2 T adjusted 3 T replaced with NaN 4 not used 5 RH caution 6 RH adjusted (not used in v.4.0) 7 RH replaced with NaN 8 RH cycles Note bits are numbered in fortran notation, 8 is the highest order bit. Summary of sondes that have flagged data at 50 m height yearday yyyy-mm-dd hh flags QC notes 298.6403 2008-10-24 15 00000010 T adjusted 298.8167 2008-10-24 19 00000010 T adjusted 299.1431 2008-10-25 03 00000010 T adjusted 299.4695 2008-10-25 11 00000010 T adjusted 299.6403 2008-10-25 15 00000010 T adjusted 302.4889 2008-10-28 11 00000010 T adjusted 308.4750 2008-11-03 11 00010010 T adjusted, RH caution 315.9951 2008-11-10 23 00000010 T adjusted 316.6667 2008-11-11 16 00000100 T replaced with NaN 318.0160 2008-11-13 00 01010000 RH replaced with NaN 319.8326 2008-11-14 19 10000000 RH cycles 321.5111 2008-11-16 12 10000010 T adjusted, RH cycles 323.3430 2008-11-18 08 10000000 RH cycles 327.6604 2008-11-22 15 00000010 T adjusted 328.9972 2008-11-23 23 00000100 T replaced with NaN 329.8292 2008-11-24 19 00000010 T adjusted 331.6597 2008-11-26 15 00000010 T adjusted 332.3423 2008-11-27 08 00000010 T adjusted % Matlab code using snctools to read netcdf sounding file qcfile='VOCALS2008_soundings_z_v4.0.nc'; height =nc_varget(qcfile,'height'); yday =nc_varget(qcfile,'yday'); month =nc_varget(qcfile,'month'); day =nc_varget(qcfile,'day'); hour =nc_varget(qcfile,'hour'); lat =nc_varget(qcfile,'lat'); lon =nc_varget(qcfile,'lon'); pres =nc_varget(qcfile,'pres'); T =nc_varget(qcfile,'T'); theta =nc_varget(qcfile,'theta'); RH =nc_varget(qcfile,'RH'); q =nc_varget(qcfile,'q'); Td =nc_varget(qcfile,'Td'); u =nc_varget(qcfile,'u'); v =nc_varget(qcfile,'v'); wnddir =nc_varget(qcfile,'wnddir'); wndspd =nc_varget(qcfile,'wndspd'); tmp=int16(nc_varget(qcfile,'tqflag')); % convert tqflag to uint8 without losing data tqflag=uint8(tmp); tqflag(tmp<0)=uint8(tmp(tmp<0)+256); % interpret tqflag bits is_rhcycle=bitget(tqflag,8); is_rhzap= bitget(tqflag,7); is_rhadj= bitget(tqflag,6); is_rhnote= bitget(tqflag,5); % bit 4 not used is_tzap= bitget(tqflag,3); is_tadj= bitget(tqflag,2); is_tnote= bitget(tqflag,1);