CSU Quality Control (QC) procedure for Version 1 sonde products ****************************************************************************** 1) Data is put into a common 5 mb ASCII format. Temperature and dew point are interpolated to 5 mb level. Winds are averaged in 5 mb layers. A 1-2-1 filter is applied to the 5 mb wind data to remove unrealistic high-frequency vertical noise. 2) Time mean and standard deviation (sigma) of each field is computed at each 5 mb level. These statistics are computed a second time not using any data that are more than 4*sigma from the mean. These stats are saved to file and used in gross limit checks (see step 3). 3) Objective QC: Perform a gross limit check flagging all data as questionable that is more that 3*sigma from the mean and as bad if data is more than 6*sigma from the mean. Hydrostatic (pressure decreasing with height, height increasing with decreasing pressure) and vertical consistency checks (excessive lapse rates or vertical shears) are performed and flagged accordingly. 4) Subjective QC: Visually QC all sondes at a given site from the surface up to 100 mb. By viewing consecutive skew-Ts in time, large temporal changes in a field can be noted and flagged if suspicious looking. 5) Linearly interpolate (in log pressure) missing data in the vertical if the layer of missing data is 200 mb or less. Interpolated data is flagged accordingly. ****************************************************************************** The idea of using QC flags is to never change a data value, but rather to provide data flags to quide the user. In this approach, it is up to the data user to decide whether or not to use a given data value. QC FLAGS flag meaning 1 parameter good 2 parameter questionable 3 parameter "visually" questionable 4 parameter bad 5 parameter "visually" bad 6 parameter interpolated 7 parameter estimated 8 parameter unchecked 9 parameter missing Value used for missing data is -999.0 Notes: Sonde data from the ISS sites were provided in 1 sec format. The raw soundings are first run through the ATD APSPEN processing system which performs smoothing and removes suspect data points. The soundings are then visually evaluated by ATD for outliers, or any other obvious problems. There is no record of the changes made to the data by ATD. The bottom point in the sondes at the ISS sites is from surface met station. Surface met station at Loreto failed so bottom point is from sonde measurement and human estimated winds. During its first cruise the R/V Altair did not report a surface wind. Any problems with the data should be reported to Paul Ciesielski (paulc@atmos.colostate.edu).