Presentation is loading. Please wait.

Presentation is loading. Please wait.

Do the NAM and GFS have displacement biases in their MCS forecasts? Charles Yost Russ Schumacher Department of Atmospheric Sciences Texas A&M University.

Similar presentations


Presentation on theme: "Do the NAM and GFS have displacement biases in their MCS forecasts? Charles Yost Russ Schumacher Department of Atmospheric Sciences Texas A&M University."— Presentation transcript:

1 Do the NAM and GFS have displacement biases in their MCS forecasts? Charles Yost Russ Schumacher Department of Atmospheric Sciences Texas A&M University Research supported by COMET grant #Z10-83387

2 Outline Brief Background Data and Methodology Results Case Studies Future Work

3 Background Mesoscale Convective Systems (MCSs) are responsible for a large percentage of rain during the warm season Researchers and forecasters noticed the NAM and GFS consistently predicted these events too far north HPC and Texas A&M University collaborated to investigate

4

5

6

7

8

9

10 Cases Searched April through August of 2009 and 2010 using – Radar to identify MCSs – Stage IV to analyze amounts 29 unique 6 hour intervals – Ranging from April 13 to August 18 – Several cases outside of initial time frame

11 Data Stage IV – 6 hourly multi-sensor precipitation analyses North American Mesoscale Model – 0Z and 12Z model runs – 6 hourly precipitation forecast Global Forecast System Model – 0Z and 12 Z model runs – 6 hourly precipitation forecast

12 Methodology “Eyeball” Test Method for Object-Based Diagnostic Evaluation (MODE)

13 Note on terminology 1 st Forecast: most recent model forecast 2 nd Forecast: second most recent forecast 3 rd Forecast: third most recent forecast Example: 6Z to 12Z – 1 st Fore: 0Z – 6 to 12hr – 2 nd Fore: 12Z (previous day) – 18 to 24hr – 3 rd Fore: 0Z (previous day) – 30 to 36hr Note: 0Z and 12Z are 12 hour forecasts, 6Z and 18Z are 6 hour forecasts

14 August 18, 2009 – 12Z

15

16

17 Method for Object-Based Diagnostic Evaluation Tool (MODE) Resolves objects in observed and forecasted fields Provides detailed information about the objects – Centroid location, object area, length, width – Axis angle, aspect ratio, curvature, intensity Can pair observed and forecasted objects

18 Process for Resolving Objects Davis, C., B. Brown, and R. Bullock, 2006a: Object-based verifica- tion of precipitation forecasts. Part I: Methods and application to mesoscale rain areas. Mon. Wea. Rev., 134, 1772–1784.

19 MODE Tool Settings ModelRadiusThreshold GFS4≥ 7.5 NAM6≥ 10 GFS was re-gridded to the 212 grid. NAM remained at the 218 grid Stage IV was regridded to the corresponding forecast’s grid Radii and thresholds were selected to match what a human would draw

20 MODE Tool Output Fields used: – Centroid (center of mass) – Area – Length – Width Determine forecast error: “Forecast – Observed”

21

22

23 Results

24

25 “Eyeball” Test

26 GFS Forecast Errors QuadrantNumber of PointsPercentage 1 (NE)4248% 2 (NW)2124% 3 (SW)911% 4 (SE)1517% TOTAL87100% MeanMedianStand. Dev. Distance (km)266.67216.80183.74

27

28 “Eyeball” Test

29 NAM Forecast Errors QuadrantNumber of PointsPercentage 1 (NE)2946% 2 (NW)2438% 3 (SW)610% 4 (SE)46% TOTAL63100% MeanMedianStand. Dev. Distance (km)249.19228.49134.45

30 Forecasting Questions Is there a correlation between forecast error (distance) and forecast area? Is there a correlation between forecast error (distance) and forecast width? Is there a correlation between forecast error (distance) and forecast length?

31

32

33

34

35

36

37

38

39

40 Conclusions “Eyeball” test and MODE test are consistent with each other Clear northern bias in the NAM 84% of cases No temporal bias GFS northern bias present, not as strong 72% of cases Tends to move system through early (65%) No clear bias with area, width, or length

41 Future Work Expand the time period to include more years and cases Does this bias exist in higher resolutions? – NSSL WRF What are the causes of this bias?

42

43

44


Download ppt "Do the NAM and GFS have displacement biases in their MCS forecasts? Charles Yost Russ Schumacher Department of Atmospheric Sciences Texas A&M University."

Similar presentations


Ads by Google