1. University at Albany Dept. of Atmospheric & Environmental Sciences
Using GAZPACHO to create forecast snowfall bias/error maps stratified by flow regime
Joe Villani
NWS Albany, NY
Mike Main
University at Albany Dept. of Atmospheric & Environmental Sciences
NROW – November 1-2, 2017

2. What is GAZPACHO? Not talking about cold
soup here…
GAZPACHO = Gridded Automated Zonal Precipitation And Complete Hi-res Output
GAZPACHO is an automated program that was created to assist WFOs with snowfall verification (rainfall added)

3. What is GAZPACHO? Creates maps of:
Observed precipitation (rain, snow or both)
Zone-average rain/snow
NDFD forecast rain/snow (from NOMADS)
Difference (or error) maps of forecast minus observed rain/snow (inches and %)
Spreadsheet table of zone-average statistics

4. What is GAZPACHO?
Created by Danny Gant (GSP), Vasil Koleci (ALY) & Joe Villani (ALY)
Uses ArcGIS software and Python scripts
GAZPACHO is run on a PC (with ArcGIS installed) via a simple GUI or command line

5. What is GAZPACHO?
Three main input options (along with start/end time of precipitation event:
A QC’d list of snow (or rain) reports from home WFO (preferably with a few from surrounding WFOs too) via a PNS with metadata (LAT/LON)
NOHRSC snowfall or AHPS rainfall analyses as the input source
blend of NOHRSC/AHPS and PNS data

6. Method for Bias/MAE maps
Run GAZPACHO using archived snowfall reports at ALY (55 total events from )
Use difference maps for each event in database:
Sum difference maps (using raster calc in ArcGIS) to compute bias
Sum of absolute value of error maps divided by # of events to compute mean absolute error (MAE)

7. Method for flow regime stratification
Stratify by wind direction & speed at 850/925 mb:
4 quadrants for wind direction
3 categories for speed
= 12 combined categories 0°
270°
90°
180°
Speed category 1
0-19 kt 2
20-39 kt 3
40+ kt

8. Method for flow regime stratification
Use Albany, NY sounding data to determine wind at 925 & 850 mb
Sounding time (0000 UTC or UTC) closest to midpoint of each event

9. Method for flow regime stratification
Stratify events into the 12 categories using a spreadsheet

10. Bias/MAE maps by category
Compute bias/MAE (using ArcGIS) for each category using 925 & 850 mb wind
Speed Category 1
0-19 kt 2
20-39 kt 3
40+ kt

11. Terrain influence on precip
Terrain influence (upslope/down slope) on precip likely influenced more by 925 mb winds than 850 mb due to ALY’s mountains all below 850 mb
Focus on 925 mb flow regimes here

12. Notable 925 mb categories Category ii-2 (90-180°, 20-39 kt) 5 events
Negative bias NW half of area
Positive bias SE of Albany
Possible terrain influence: under- forecast upslope southern Adirondacks, Saratoga, southern Greens

13. Notable 925 mb categories Category iii-2 (180-270°, 20-39 kt) 8 events
Positive bias leeward side of southern Adirondacks
Negative bias SW upslope areas
Fairly good upslope & downslope signatures noted

14. Notable 925 mb categories
Category iv-2
( °, kt)
11 events
Lake Effect event position error
Significant negative bias terrain influence for W/NW flow events: Catskills, Taconics, Berkshires, Litchfield Hills
Lake Effect contribution
Includes 2016 Nov extreme upslope event

15. Notable 925 mb categories Category iv-2 (270-360°, 20-39 kt) 11 events
MAE good(+/- 0 to 1”) in Hudson Valley
Greater error in higher terrain

16. 2016 Nov Upslope Event
Extreme snowfall differences in short distances Weighted NW flow category bias/MAE

17. Use as a Forecast Application
Use bias/MAE maps in operations
Envision using maps as a forecast tool
Look at forecast 850/925 mb wind direction/speed at ALB around the midpoint of upcoming event
Reference bias/MAE maps based on category
Bias
Speed category
925 mb 1
0-19 kt 2
20-39 kt 3
40+ kt
MAE

18. Future Work Expand database with each additional winter season
Ideas for improving limitation of using single point (ALY sounding) for wind stratification
Stratify events based storm track?

19. Questions/Comments?