The Effects of Grid Nudging on Polar WRF Forecasts in Antarctica Daniel F. Steinhoff 1 and David H. Bromwich 1 1 Polar Meteorology Group, Byrd Polar Research Center, and Atmospheric Sciences Program, Department of Geography, The Ohio State University, Columbus, Ohio
19
Embed
The Effects of Grid Nudging on Polar WRF Forecasts in Antarctica Daniel F. Steinhoff 1 and David H. Bromwich 1 1 Polar Meteorology Group, Byrd Polar Research.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The Effects of Grid Nudging on Polar WRF Forecasts in
AntarcticaDaniel F. Steinhoff1 and David H. Bromwich1
1Polar Meteorology Group, Byrd Polar Research Center, and Atmospheric Sciences Program, Department of Geography, The Ohio State University, Columbus, Ohio
“You’re the Inspiration”
• During my visits to the McMurdo forecasting office, it was clear that AMPS forecasts beyond 72 hours are not well trusted– 72-120 hours: Events generally “pushed back” in
time in subsequent forecasts– Only taken seriously inside of 72 hours
Last Year
• Aaron Wilson from our group found that grid nudging of Polar WRF simulations in the Arctic improves extended simulations compared to non-nudged simulations.
• Are similar improvements possible in the Antarctic? Can AMPS forecasts out to 120 hours be more trustworthy?
Grid Nudging
• Grid nudging involves an extra time tendency term in the forcing equation for a given variable
• This extra term “nudges” the tendency term towards the driving analysis/forecast, weighted by a representative time scale parameter
• Originally designed for dynamic initialization of forecasts, but also useful during simulations
Testing Strategy
• Testing Simulations– Forced by NCEP GFS 1.0° forecasts– NUDGE: 120 hour simulation, nudged for top 5 model
vertical levels only– NONUDGE: Same except no nudging
• Validation Simulation– Forced by NCEP FNL 1.0° analyses– 48 hour simulations, reinitialized every 24 hours
• So we compare the GFS-forced NUDGE and NONUDGE runs (24-120 hours) with FNL-forced “validation runs” (24-48 hours) at corresponding times
Testing Strategy (Cont’d)
• September – November 2009• Testing simulations initialized at 0000 UTC on
the 5th, 10th, 15th, 20th, 25th, and 30th of each month (Total of 18 cases)
• Average verification statistics for each forecast hour from 24-120 hours computed (first 24 hours of each simulation discarded for model “spin-up”
• Surface pressure: Scattered differences over ocean, but large expanse of lower RMSE over Antarctica for NUDGE
BLUE: NUDGE BETTERRED: NUDGE WORSE
RMSE Differences – Z500
• 500 hPa geopotential height: Scattered differences over ocean, but large expanse of lower RMSE over Antarctica for NUDGE
BLUE: NUDGE BETTERRED: NUDGE WORSE
RMSE Differences – T2
• 2 m air temperature: No discernible pattern
BLUE: NUDGE BETTERRED: NUDGE WORSE
RMSE Differences – q850
• 850 hPa mixing ratio: No discernible pattern
BLUE: NUDGE BETTERRED: NUDGE WORSE
Antarctic Land Only - PSFC
• Taking only the land points:– Bias and RMSE differences larger in a relative
sense than for entire grid
Ocean Only - PSFC
• Taking only the sea points:– Bias smaller, RMSE larger than land– Biases remain constrained longer (almost 96 hours)– Smaller RMSE differences than land
Conclusions
• Grid nudging improves verification statistics presented here– Small improvements on grid average, but greater
improvements over Antarctic land– Grid nudging can help/hurt over certain ocean areas
(likely placement of synoptic-scale cyclones)• More work needed: – Determine causality of changes to forecasts by
nudging over both Antarctica and Southern Ocean– Verification against station and AWS observations