Clog QA Methods

• Summary of clog quality checks
  • What is clogging and how can we tell?
    • How a gauge get’s blocked
    • Identifying a clog
    • Unclogging
    • How does this affect the data?
  • How big an impact does this have?
    • Pre-cleaning
    • Clogs
  • Quality Checks/Rules
    • Pre-cleaning
      • Flag double delayed precip
      • Flag repeating values
      • Flagging Estimates of empty tanks
    • Cross Probe Quality Checks
      • Building Cross Comparison Tables
      • Comparing Two Probes
      • Deciding on Final Flagging
      • Examples
• Exploring Clogs
  • Let’s get some Data
    • Clear up 10/1 repeats
    • Track down missing values
  • GCE flagging during clogs
    • J flag
    • F flag
    • Do GCE flags ID clog periods
    • Is precip always 0 when M
    • Flags need to propagate
    • Remove duplicate values
    • 9/6/22 ‘EM’ & “0 tank” issue
      • Find empty (0) tank
      • Find EM
  • Comparing paired sensors during clogs
    • Precip Minus PairedPrecip
    • Looking for optimal timestep
    • What metric is consistent for entire clog?
• Capturing clogs from ACC ratio
  • Create clean ACC data
  • Find the clog
  • Fine Tune parameters
    • threshold for below normal ratio
    • running average
  • Check false positives
  • Early season lag
  • WY19 test
    • Weird exceptions
      • Oct 13-16, 2018
      • Oct 29-31 2018
      • Dec 18-25, 2018
      • Feb 12-21, 2019
      • Apr. 7-16, 2019
• Flagging Clogs: Finding U or C
  • Get Some Data structured
  • CENT 02 (Shelter gauge)
    • Fix overflagging issue
    • Undercatch or Cumulative: WY 2019
      • 10/29/18 example
      • Test shifting clog on other examples
• Using multiple probes to catch clogs
  • First Look
    • Pros
    • Cons
    • Strategy
  • Finding Clogs Using CS2MET
    • Setting levels
      • Lowest Normal Level
      • Minimum accumulation
      • Window Precision
      • Rolling Window Size
  • Weighting Clogs From Other Sites
    • How heavily should within site gauges be weighted.
    • Testing U and C flag weights
  • Reverse Relationship to Identify Known CS2MET Clogs
    • Over flagging
    • Delayed Accumulation Could Signal a Clog in Reverse
      • Can Weighting Fix This?
      • Weighting + Manual Unflagging
      • Rolling Avg at Second Scale
• Reverse Clogs- When the Pair has a Clog
  • Distinguishing Between Clog and Delayed Precip in a Pair
    • Finding the Shift in Ratio Slope
    • Multiple Running Averages
      • Centered Windows
    • Ratio Slope Compared to Ratio Dropping
      • Smooth
      • Smoothing with SUM
  • Applying to Test Clog Detection
    • Check False Positives
      • Dec 2019 example
      • Other WY19 Examples
      • Reduce Sensitivity
  • Restart by Smoothing the Differential
    • ratio.diff().rolling(window=’16D’)
    • ratio.rolling(window=’8D’).diff().rolling(window=’16D’)
    • Broader Clog Check Finds Early Season Issues
  • Need to Reduce Impact of Low ACC Spikes
    • Reset Min-Accum
  • Final Check for False Positives
    • 11/14/17
    • 10/29/18
    • 11/25/19
    • Nov 2020
  • Do True Positives Still Work?