One of the main purposes of Aspen is to apply quality control algorithms to a sounding. These algorithms attempt to systematically detect data points that are likely to be incorrect, and remove them from the QC data set.

The processing begins with a copy of the original (raw) observations. In most cases, all parameters are considered separately. The main exception to this is the handling of the wind observations. In this case, the wind speed and direction are separated into u and v components, and the QC tests are individually applied. However, if one of the two components fails a test, then both components as well as the speed and direction are removed from the QC data.

There are other QC tests that depend upon the relationship between two variables; in some algorithms, the processing of a given parameter requires input from another parameter. For instance, the GPS-measured velocity is used as a discriminator for the horizontal wind observations. In another case, the dynamic adjustment procedure requires the time constant of the sensor making the measurement. This time constant is a function of density and ventilation rate, which are obtained from the pressure and vertical velocity.

The Algorithms

The following table lists the QC processing steps, in the order that Aspen applies them. Also listed are the sounding direction and variables to which each step is applied. Following the table is a more detailed description of each step.

Procedure Downward Upward P T RH* Winds GPS Position
Temperature-dependent dry bias correction†          
Launch parameter overrides
Copy raw data set to QC data set
End of drop override
Ambient equilibration    
Post-splash data check          
Apply fixed offsets    
RH thresholding        
Limit check
Satellite check      
Wind error check        
Buddy check
Outlier check  
Filter check
Pressure smoothing        
Monotonic pressure check        
Compute vertical velocity (1)          
Temperature dynamic adjustment        
RH limiting (1)        
Final smoothing    
RH limiting (2)        
Pressure adjustment          
Compute vertical velocity (2)          
Vertical velocity check          
Compute vertical winds            
Create surface observation      
Winds dynamic adjustment        
Fast fall check            
Compute altitude          
Compute position          

* Aspen assumes humidity is given as RH w.r.t. water, which is the meteorological standard, but does not check this. Refer to documentation for the sensor you are using for more information.

† Only applied to specific range of RS94 dropsondes that need the correction.

More details about each processing step follows. Some algorithms are linked to their pages in the QC Algorithms section of AspenDocs. If the step can be adjusted via a configuration parameter, the name of the configuration parameter in the config set and how to access the parameter value in Aspen are listed. See the Configuration section for details on managing Aspen configurations.

Temperature-dependent dry bias correction

A correction is applied to the RH series if a) the sonde is in the specific range of RD94 sondes that need the correction and b) if the correction is not already present in the input data. If the correction is present in the input data, any RH values of 0.1 (signifying very low but nonzero RH) are flagged and removed.

Launch parameter overrides

Settable: Main tab

The user-specified launch parameter overrides, if any, are substituted for the ones given in the data file.

Copy raw data set to QC data set

Parameters: DiscardBadCrcData

Settable: Options > Processing > Discard frames with CRC errors

The QC data set is initialized with a copy of the original observation. If the option is chosen to discard frames with CRC errors, these frames are not copied to the QC data set, except for the very last frame in the original observation (applies only to AVAPS dropsonde soundings).

End of drop override

Settable: Main tab

If the user has specified an end of drop time, all data beyond this time is ignored.

Ambient equilibration

Parameters: WindEquilTime

Settable: Options > QC Parameters > Dropsonde equilibration time override

See Ambient Point.

Post-splash data check

See Post-splash check.

Apply fixed offsets

Parameters: PresOffset, TdryOffset, RHOffset

Settable: Options > QC Parameters > Offset to add

See Offset Adjustment.

RH thresholding

Parameters: RHThreshold

Settable: Options > QC Parameters > Thresholding

Optionally thresholds RH values outside the 0-100% range that still may contain valid data. Thresholding sets RH values between -5% and 0% to 0% and between 100% and 110% to 100%. This thresholds values before quality control takes place, unlike the separate RH limiting that gets applied after processing and is not user-configurable.

Limit check

See Limit Check.

Satellite check

Parameters: WindDisableSatsCheck, WindSats

Settable: Options > QC Parameters > Disable sats check or set minimum number of satellites

See Satellite Check.

Wind error check

Parameters: WindDisableErrorCheck, WindErrorHigh, WindErrorLow

Settable: Options > QC Parameters > Disable wind error check or set limits in upper and lower atmosphere

See Wind Error Check.

Buddy check

Parameters: PresBuddySlope, TdryBuddySlope, RHBuddySlope, WindBuddySlope, GPSPosBuddySlope, GPSAltBuddySlope

Settable: Options > QC Parameters > Buddy check slope

See Buddy Check.

Outlier check

Parameters: PresDisableOutlierCheck, PresOutlier, TdryDisableOutlierCheck, TdryOutlier, RHDisableOutlierCheck, RHOutlier, WindDisableOutlierCheck, WindOutlier

Settable: Options > QC Parameters > Outlier check or disable outlier check

See Outlier Check.

Filter check

Parameters: PresQCWL, PresQCDev, TdryQCWL, TdryQCDev, RHQCWL, RHQCDev, WindQCWL, WindQCDev, WindDisableQCFilter, GPSPosQCWL, GPSPosQCDev, GPSAltQCWL, GPSAltQCDev

Settable: Options > QC Parameters > QC filter wavelength, QC filter deviation limit, disable QC filter

See Filter Check.

Pressure smoothing

Parameters: PresSmoothWL

Settable: Options > QC Parameters > Final smoothing wavelength

The smoothing is applied to the pressure data. The pressure needs to be smoothed before the pressure monotonic check. If the final smoothing wavelength is set to 0, this is not applied. A small amount of final smoothing should almost always be applied to pressure, so that reverse trends in pressure caused by noise will not cause a data point to be discarded. See Smoothing for more details.

Monotonic pressure check

Parameters: PresMonoCheck

Settable: Options > QC Parameters > Pressure monotonic check

The pressure trace must change monotonically in order for pressure-based searching to be performed. The pressure time series is scanned; when a point is found which follows an incorrect trend (i.e. decreasing for a dropsonde, increasing otherwise), it is removed from the series.

Compute vertical velocity

See Vertical Velocity.

Temperature dynamic adjustment

Parameters: TdryDynCor, TdryDynCorWL

Settable: Options > QC Parameters > Dynamic correction and dynamic correction wavelength

See Dynamic Adjustment.

RH limiting

Incoming RH values can be outside the 0-100% range, but Aspen will not output QC values outside of the range. After other QC processing has been done, make sure RH values are still within what’s physically possible. RH values less than 0.1% are set to 0.1% and RH values greater than 100% are set to 100%. This is performed a second time after final smoothing, since that can adjust RH values to outside of the allowable range.

Final Smoothing

Parameters: TdrySmoothWL, RHSmoothWL, WindSmoothWL, BlendLength

Settable: Options > QC Parameters > Final smoothing wavelength

The series are smoothed using the final smoothing wavelength. If the final smoothing wavelength is set to 0, this is not applied. See Smoothing for more details.

Pressure Adjustment

Settable: Main tab

See Scalar Adjustment.

Vertical velocity check

Parameters: WindVVDelta, WindVVPresWL, BlendLength, ChuteArea, DragCoef, DropSondeMass

Settable: Options > QC Parameters > Vertical velocity difference limit and pressure smoothing wavelength, Options > Processing > Dropsonde parameters

The GPS-measured fall velocity is a good discriminator for the quality of the GPS-derived horizontal wind. If the difference between the GPS-measured vertical velocity and the hydrostatically determined fall velocity is greater than the specified limit, and the same comparison passes for the theoretical fall velocity, the GPS measured horizontal wind value is discarded. The theoretical fall velocity is based on a model of the sonde/parachute aerodynamic properties. Note that a separate pressure series is used for the hydrostatically derived fall velocity, so that it can have a different smoothing wavelength than the standard pressure series. This allows it to be tuned in order to catch the expected variance in the GPS-measured vertical velocity.

Compute vertical winds

Parameters: WindDisableW, ChuteArea, DragCoef, DropSondeMass

Settable: Options > QC Paremeters > Disable Vertical Winds Calculation, Options > Processing > Dropsonde Parameters

See Vertical Wind.

Create surface observation

See Create Surface Observation.

Winds dynamic adjustment

Parameters: WindDynCor, WindDynCorWL

Settable: Options > QC Parameters > Dynamic correction and dynamic correction wavelength

See Dynamic Wind Adjustment.

Fast fall check

The hydrostatically computed vertical velocity is compared with the theoretical value. If more than 50% of the values are greater than 1.35 times larger than the theoretical value, the observation is flagged as a “fast fall” sounding. See Fast Fall Check.

Compute altitude

See Altitude and the Post-Processing Adjustment section for a description of height computations.

Compute Position

Parameters: ReportObservedPos, PosInterpSpan

Settable: Options > Processing > Report observed position instead of integrated winds position, position interpolation interval

See Position.

Guidelines for Adjusting QC Parameters

The supplied default values for the Aspen QC parameters in general produce excellent results. However, there may be times when a group of soundings have peculiar error characteristics, and the Q/C parameters need to be modified in order to produce better discrimination against outliers. Some tips for adjusting the configuration parameters are given here. The user should carefully read and understand the descriptions of the QC algorithms given in the preceding section; this will provide the best guidance on modifying the QC parameters to match the error characteristics of the sounding data.

  • Toggle the display of raw data off and on in order to see which data points are being rejected by the algorithms. Toggle the “mark points” option, and use the zoom and pan capabilities freely to investigate the effectiveness of the procedures.
  • If the sounding system does not report the number of GPS satellites used in the winds calculation, set the GPS satellite threshold to 0, so that winds are not discarded due to a lack of satellites.
  • The most frequently adjusted parameters are the the QC filter wavelength and the QC filter deviation, which are used in the filter check and are applied together. In this algorithm, a smoothed curve is calculated from the data series. If the filter wavelength is small, the curve will closely follow the data; if large the curve will be a very smooth representation of the data. The deviation is the amount that a data point is allowed to deviate from the smoothed curve. If the difference between the curve and the raw data point is larger than the deviation, the data point is discarded. The interplay between the amount of smoothing and the allowed deviation allow for wide latitude in tuning the filter check algorithm.
  • The dynamic adjustments for temperature and wind speed can significantly increase the variance of the data series. If the QC results appear noisier than the raw data, it will be due to the dynamic adjustments. Use of the dynamic adjustments is the prerogative of the user, and can be deactivated in the configuration. Note that the tendency of the series, used in the adjustment, is calculated from a smoothed version of the data. The dynamic correction smoothing wavelength is used for the smoothing value during the dynamic adjustment procedure. If a very small smoothing wavelength is specified, the variance increase in the QC data can become significant.
Edit me