Navigation Links

 

Products Guide

Gauge Quality Control

Gauge Quality Control

Short Description

Algorithm that flags the hourly gauge data and identifies reports that most likely contain errors and may degrade the quality of gauge-derived products.

Subproducts

None.

Primary Users

None. This algorithm is not a stand-alone product, but rather is primarily incorporated into other products, such as the QPE – Gauge Only, QPE – Mountain Mapper and QPE – Radar w/ Gauge Bias Correction.

Input Sources

Hydrometeorological Automated Data System (HADS) gauge network; Radar Quality Index; 1-hr QPE – Radar Only; hourly model Surface Wet-Bulb Temperature

Resolution

Spatial resolution: n/a

Temporal resolution: 60 minutes (with latencies due to gauge collection)

Product Creation

The Gauge Quality Control (QC) Algorithm goes through a series of steps to determine if a gauge report falls into one of the categories that could cause it to be omitted in gauge-based product generations. The QC flags gauges that:

  • Report accumulations that are more than 5 minutes outside of the valid time
  • Missing data
  • May be frozen due to winter precipitation
  • May be suspect due to poor radar quality that makes comparison impossible
  • Report “false zero” as compared to the QPE - Radar Only
  • Report “false precipitation” as compared to the QPE - Radar Only
  • Are considered “outliers high” based on predefined maxima from power curves
  • Are considered “outliers low” based on predefined minima from power curves

Technical Details

The decision tree is shown in Figure 1, and described here:

Gauges that are flagged as “Pass” or “Unchecked” are used in MRMS product generation (green in Figure 1). All other gauge QC flags (in red) cause the gauge to be omitted in MRMS product generation.

  1.  Determine if the hourly gauge observation is within ± five minutes from the top of the hour.

a.  If beyond ± five minutes from the top of the hour, then flag the observation as “Outside of Time Window.”

  2.  Determine if the hourly gauge observation is a negative value, which suggests a missing value. Flag as “Missing” if value is negative.

  3.  Determine the model surface wet bulb temperature at the gauge location.

If the model surface Twb is ≤ 0.00°C, winter precipitation is assumed to be the precipitation type that could occur at the gauge site. The following logic is then applied:

  4.  If the gauge value is 0.00 mm, and the RQI is < 0.10, then the gauge observation is flagged as “Unchecked.”

  5.  If the gauge value is 0.00 mm, and the RQI is ≥ 0.10, then compare the gauge to the radar-only QPE.

a.  If the radar-only QPE value is also 0.00 mm, then the gauge is marked as “Pass.”

b.  If the radar-only QPE value is > 0.00 mm, then gauge is flagged as “Frozen” due to the assumption that the gauge is either clogged or not capable of measuring the winter precipitation.

  6.  If the gauge value is > 0.00 mm, and the RQI is < 0.10, then the gauge is compared to a threshold value.

a.  If the gauge value is ≥ 6.35 mm (0.25 in.), then the gauge is flagged as “Suspect.” This is to remove any erroneous values that are likely due to technical or reporting problems. The value is derived from historical hourly snow accumulations and snow-water equivalent ratios.

b.  If the gauge value is < 6.35 mm, then gauge is marked as “Unchecked.”

  7.  If the gauge value is > 0.00 mm, and the RQI is ≥ 0.10, then the gauge is then subjected to the following logic:

a.  If the gauge value is ≥ 6.35 mm (0.25 in.), then the gauge is flagged as “Suspect.”

b.  If the gauge value is < 6.35 mm, then the gauge value is compared to radar QPE.

i.  If the radar-only QPE value is 0.00 mm, then the gauge is flagged as reporting “False Precipitation.”

ii.  If the radar-only QPE value is > 0.00 mm, then the following checks are conducted:

1.  If the gauge value is greater than the winter Rmax value, then the observations is flagged as “Outlier High.”

2.  If the gauge value is less than the winter Rmin value, then the observation is flagged as “Outlier Low.”

3.  If the gauge observation lies between Rmax and Rmin, then the observation is marked as “Pass.”

If the model surface Twb is > 0.00°C, non-winter precipitation is assumed to be the precipitation type that could occur at the gauge site. The following logic is then applied:

  8.  If the gauge value is 0.00 mm, and the RQI is < 0.10, then the gauge observation is flagged as “Unchecked.”

  9.  If the gauge value is 0.00 mm, and the RQI is ≥ 0.10, then compare the gauge to the radar-only QPE.

a.  If the radar-only QPE value is also 0.00 mm, then the gauge is marked as “Pass.”

b.  If the radar-only QPE value is > 0.00 mm, then gauge is flagged as a “False Zero” observation.

  10.  If the gauge value is > 0.00 mm, and the RQI is < 0.10, then the gauge is compared to a threshold value.

a.  If the gauge value is ≥ 50.8 mm (2.00 in.), then the gauge is flagged as “Suspect.” This is to remove any erroneous values that are likely due to technical or reporting problems. The value is derived from internal analysis showing that obvious erroneous hourly gauge observations in poor radar coverage will be properly removed through this check.

b.  If the gauge value is < 50.8 mm, then gauge is marked as “Unchecked.”

  11.  If the gauge value is > 0.00 mm, and the RQI is ≥ 0.10, then the gauge is compared to radar QPE and subjected to the following logic:

a.  If the radar-only QPE value is 0.00 mm, then the gauge is flagged as reporting “False Precipitation.”

b.  If the radar-only QPE value is > 0.00 mm, then the following checks are conducted:

i.  If the gauge value is greater than Rmax value, then the observation is flagged as “Outlier High.”

ii.  If the gauge value is less than Rmin value, then the observation is flagged as “Outlier Low.”

iii.  If the gauge observation lies between Rmax and Rmin, then the observation is marked as “Pass.”

Figure 1: QC algorithm decision tree for hourly rain gauges. Gauges flagged with a green label is used in the hourly gauge analysis. Gauges flagged with a red label are omitted from the analyses.

Figure 2: Graph depicting the range of maximum and minimum hourly QPE potential (Rmax and Rmin) based on a derived set of power curves. Gauges that passed the QC algorithm are represented by blue dots between the two curves. Gauges that were flagged as outliers are represented by black dots.

Figure 3: Similar to the previous graph, but for winter preciptation. Graph depicting the range of maximum and minimum hourly QPE potential (Rmax and Rmin) based on a derived set of power curves. Gauges that passed the QC algorithm are represented by blue dots between the two curves. Gauges that were flagged as outliers are represented by black dots.

References

* Zhang, J., K. Howard, S. Vasiloff, C. Langston, et al., 2011: National Mosaic and multi-sensor QPE (NMQ) system: description, results and future plans. Bull. Amer. Met. Soc., 92, 1321-1338.

* Zhang, J., K. Howard, S. Vasiloff, C. Langston, B. Kaney, Y. Qi, L. Tang, H. Grams, D. Kitzmiller, J. Levit, 2014: Initial Operating Capabilities of Quantitative Precipitation Estimation in the Multi-Radar Multi-Sensor System. 28th Conf. on Hydrology, Amer. Meteor. Soc.

Qi, Y., S. Martinaitis, J. Zhang, and S. Cocks, 2016: A real-time automated quality control of hourly rain gauge data based on multiple sensors in MRMS system. J. Hydrometeor., 17, 1675–1691.

Other MRMS product documentation: RQI, QPE - Radar Only, QPE - Radar w/ Gauge Bias Correction, QPE - Gauge Only, QPE - Mountain Mapper

* Changes were applied to the QC algorithm after the completion of both of these references.