1	Data QA/QC (Quality Assurance/Quality Control) Introduction to RDI's Data Quality Indicators
2	Having 4 beams enables a unique screening of raw data, producing higher quality results Evaluates how well all beams see the same flow field => affects noise, bias in data Provides a quantitative basis for QA/QC at each depth layer of each ping Enables superior screening of raw data, allows greater certainty of quality of results
3	Data Quality Philosophy: Quality vs. quantity An 80/20 rule often applies to Content of a data series Time consumed by data reduction Taking advantage of RDI's unique data quality indicators permits more efficient data reduction Identifying both poor and high quality data segments Permitting faster production of results and reports.
4	What is unique to RDI? 4 beams permit immediate ping-by-ping error detection via redundant information (cf. Packet-by-packet checking in digital messages in telecoms) BroadBand signaling takes multiple samples within a single ping => statistical indicators of single-ping quality Advantage: QA/QC on a ping-by-ping basis maximizes the volume of high quality signal recorded
5	RDI's Data Quality Indicators Quality of processed data: Error Velocity Variability of velocity data Informational content in signal: Correlation Velocity Signal Strength in received echo Strength of received signal : Echo Intensity Changes in Received Signal Strength Consistency between beams
6	Using Error Velocity Focus: Quality of processed data Measures variability of velocity data Provides a far more sensitive screen for data quality than can be achieved by inspecting echo intensity Screens each ping for unacceptable noise in the data ( e.g., due to fish, turbulence, or eddy variability), maximizes the volume of high quality signal recorded Detects consistent obstructions from solid scatterers => causing bias in the data
7	Using Correlation Focus: Informational content in signal Measures velocity signal strength in echo Causes of poor correlation Low signal-to-noise ratio in returned echo Too much variability in the velocity signal returned from depth cell Rapidly varying scatterers Diverse Doppler shifts
8	Decorrelation: Rapidly Varying Scatterers Causes Turbulence and Boat Heave: scatterers move rapidly between bins (or along beam) High boat speed, Pitch and Roll: scatterers move rapidly across beam (short residence time) Options to improve correlation Slow down.. Scatterers move across beam slower Use larger bins.. fractional change in scatterers is less Increase ambiguity velocity Find another measurement section
9	Decorrelation: Diverse Doppler shifts Causes High Shear: velocity changes too much across the depth of the bin (especially the ambiguity resolving bin) Beam divergence: differences between Doppler shifts measured at edges of beams become greater at higher speed Options to improve correlation Smaller bins.. Reduce range of velocities in cell Slow down.. Reduce range between Doppler shifts measured at beam edges
10	Using Echo Intensity Focus: Strength of received signal Strength of backscattering reveals Amount and patchiness of scatterers Acoustic energy into the water Bottom depth Consistency between beams Anomalies in backscattering permit detection of potential Doppler contamination Fish "2nd bounce" off hard bottom
11	Data Display and Review Refining Data Quality: typically an Iterative cycle Control: Data screening Objective checking of numbers => machine Assurance: Data review Subjective checking of patterns => person 2-D contour plots, 1-D Displays Reveal patterns of variation with both depth & time/distance Vertical profiles, Tables (see variation with depth) Time series (see variation vs. time--for one depth cell) Ship-track (see variation vs. distance traveled)
12	Checking Data Quality Times At ADCP installation Anticipate sources of data contamination Acquisition Confirm data quality Playback Refine data quality via review & reprocess
13	Data Acquisition Emphasis: Avoid bad data, Confirm data quality Configurable Thresholds for Screening Error Velocity, Correlation, Fish detection Difficult Environments for Quality Data Highly dynamic ADCP motions Bottom tracking over soft bottom Water tracking in highly variable water conditions 2-d, 3-d turbulent motions, extreme shear Rocky bottom, schools of fish
14	Data Playback Emphasis: Review data and retain high quality Sequence for shooting-trouble: Error Velocity => Correlation => Echo Intensity Using Data Quality Indicators Set values: Error Velocity, Vertical Velocity, Fish detection Review Patterns: Echo Intensity, Correlation, Error Velocity Re-processing: Option for 3-beam processing where data drop-outs occur on one beam