NMQ Verification System (QVS) Tutorial
)
Figure 20. Gauge main
display.
)
Figure 20a. Gauge Circles display.
)
Figure 20b. Gauge Circles display.
)
Figure 21. Gauge Circles display.
)
Figure 22. Scattergram display.
)
Figure 23. Bias Histogram display.
)
Figure 24. Contingency Table display.
Tabs: Gauge
This tab allows the use of rain gauge data to verify precipitation products (Figure 20a at right). Hourly gauge data are FTPed from NCEP every 10 min. Just after the top of the hour, the number of gauges is small. All gauges are in after 6 hours. Data from the Oklahoma Mesonet are also available beginning 1 March 2007 (Figure 20b at right).
Note that the upper left buttons can be used to select the gauges and products to be compared.
Display Type: By default (see Figure 20, at right), the circle size indicates gauge amounts and the color represents bias. The QPE C. button switches the circle size to QPE amount. The size can be reset to gauge amount by using the Gauge C. button.
Scatterplot: Button creates a plot of gauge data versus user-selected products. Summary statistics are also shown (more on this below).
Analysis Tools: Allows more flexibility for diagnosis with tabs at the top as follows:
- Gauge circles (default window) (Figure 21)
- Scattergrams (Figure 22)
- Creates the same scatter plot as mentioned above.
Data Options - click the Table Output button and Submit to retrieve a pop-up window with gauge QPE data in a data table.- User's web browser must be set to allow pop-ups
- The data can be cut and pasted into other applications (e.g., spreadsheet)
- Bias Histogram tab (Figure 23)
- Shows a scatterplot with summary statistics replaced by a histogram of the
bias values.
Yes/No Threshold: change minimum amounts to be displayed in scatter plot and histogram. - Contingency Table tab (Figure 24)
- Scatter plot: Shows a plot of "predictor" values vs. "truth" values
In a 2 x 2 contingency table:- hit = b
- miss = d
- false alarm = a
Contingency data: 2 x 2 contingency table shows data values for the following:- Hit = QPE > threshold and gauge > threshold;
- Miss = QPE < threshold and gauge > threshold;
- False alarm = QPE > threshold and gauge < threshold
- Probability of detection POD = hits / (hits + misses)
- False alarm ratio FAR = hits / (hits + false alarms)
- Threat score TS = hits / (hits + misses + false alarms)
- Heidke skill score HSS = 2 (ad - bc) / [ (a + c)(c + d) + (a + b)(b + d) ]
- Bias = QPE > threshold / gauge > threshold
- Bias
- Root mean square error RMSE
- Correlation coefficient CC
){kind=link}