Statistical Quality Control of High-Resolution Winds of Different Radiosonde Types for Climatology Analysis

K Houchi, A Stoffelen, GJ Marseille, J de Kloe

Quality control (QC) is among the most important steps in any data processing. These steps are elaborated for high-vertical-resolution radiosonde datasets that were gathered and analyzed to study atmospheric winds. The database is composed of different radiosonde wind-finding systems (WFSs), including radio theodolite, Loran C, and GPS. Inspection of this database, particularly for wind, wind shear, and ascent height increments (dz), showed a nonnegligible amount of outliers in radio theodolite data as compared to the two other WFSs,
thus denoting quality differences between the various systems. An effective statistical QC (SQC) is then developed to isolate and eliminate outliers from the more realistic observations. Improving the accuracy of the radio theodolite WFS is critical to the derivation of the vertical motion and the vertical gradients of the horizontal wind--that is, wind shear--mainly because of the direct dependence of these quantities on dz.
Based on the climatological distribution of the quality-controlled dz, a new approach is suggested to estimate these wind quantities for radio theodolite data. The approach is validated with the high-quality modern WFSs (Loran C and GPS). Although initially of reduced quality, applying SQC and using the climatological mean dz of 12-s smoothed radio theodolite profiles shows very good improvement in the climatological wind analyses
of radio theodolite WFSs. Notably, the climatologies of ascent rate, vertical motion, horizontal wind, and vertical shear now look comparable for the various WFSs. Thus, the SQC processing steps prove essential and may be extended to other variables and measurement systems.

Bibliografische gegevens

K Houchi, A Stoffelen, GJ Marseille, J de Kloe. Statistical Quality Control of High-Resolution Winds of Different Radiosonde Types for Climatology Analysis
published, J. Atm. Oceanic Technol., 2015, 32