Evaluating GEM-LAM precipitable water vapour output using the southern Alberta GPS network during UNSTABLE 2008 • Craig D. Smith • Climate Research Division, Environment Canada, Saskatoon, SK • Jason Milbrandt • Atmospheric Numerical Prediction Research Section, Meteorological Research • Division, Environment Canada, Dorval, QC • Susan Skone • Dept. of Geomatics Engineering, University of Calgary, AB CMOS Congress 2009, May 31 -June 4, Halifax, NS
Background: GPS Derived PWV • GPS signal is refracted or “delayed” • • • by the atmosphere Through post-processing techniques, the signal delay due to Precipitable Water Vapour (PWV) can be isolated Requires precise P 0 PWV at temporal resolution of 30 to 60 minutes no costly expendables all weather conditions Accuracies of ± 1 -2 mm PWV Page 2
Background: GPS Derived PWV • • Alberta GPS Atmospheric Moisture Evaluation • Page 3 July 2003 and 2004 ~110 radiosonde flights Olds, Airdrie, and Sundre July 2004: 10 GPS point intercomparisons with the 15 km GEM REG Oct-2008 A-O
A-GAME Year r 2 Bias* SD mm mm n 2003. 76 -1. 7 1. 6 24 2004. 82 -0. 6 1. 4 14 2003. 84 -0. 9 1. 2 17 2004. 81 1. 4 2. 3 10 Sundre 2003. 84 -1. 0 1. 4 14 Airdrie Olds *Bias = (GPS – RS) (Smith et al, 2008) Page 4
Elapsed Forecast Time r 2 Bias mm SD mm n All . 63 0. 5 2. 9 870 0 . 71 0. 2 2. 5 209 3 . 71 0. 4 2. 5 224 6 . 59 0. 5 3. 0 219 9 . 52 0. 9 3. 2 218 (Smith et al, 2008) Page 5
UNSTABLE 2008 • 8 GPS receivers, 2 fixed radiosondes • 1 km GEM LAM – – – Same physics package as 15 km GEM REG Enhanced (1 km) topography and land-cover July 7 -23 (IOP) Initialized daily at 15 Z, 12 1 -hour prognosis including PWV During IOP: severe weather forecasting and mission guidance Post IOP: generated output for examining pre-storm processes and storm evolution – Requires validation – No vertical or horizontal interpolation • Preliminary results Page 6
UNSTABLE 2008 Limestone Mt Olds Cremona EA 3 WVX Cochrane Uof. C Strathmore Priddis Vulcan Page 7
UNSTABLE 2008: GPS vs. Radiosonde PWV r 2 Bias* mm SD mm n . 92 0. 9 50 *Bias = (GPS – RS) Page 8
UNSTABLE 2008: 1 km GEM LAM vs. GPS PWV r 2 Bias* mm SD mm n . 52 1. 0 2. 9 1198 *Bias = (GEM – GPS) (Includes Limestone) Page 9
UNSTABLE 2008: 1 km GEM LAM vs. GPS PWV Page 10
UNSTABLE 2008: 1 km GEM LAM vs. GPS PWV Page 11
Page 12
Page 13
Summary and Conclusions • • • GPS vs. RS intercomparison of PWV at EA 3/Olds very good Overall, 1 km GEM LAM is too wet (average bias of 1. 0 mm) Bias starts low at initialization and peaks between 21 Z and 23 Z Correlations start high but drop towards 19 Z before increasing to a peak towards the end of the model run - ? ? GEM vs. GPS correlations unaffected by elevation but the bias tends to increase with elevation (can’t all be explained by Delta-Z) Page 14
Further Work • • • Figure out what is wrong with GEM LAM data at Limestone and fix Add some missing GPS data (Rocky Mt. House and Okotoks) More spatial and temporal trend analysis Include more of the mobile RS data Add 2. 5 km GEM LAM and revisit the 15 km GEM REG for UNSTABLE 2008 Page 15
Thank you! Page 16
Скачать презентацию Evaluating GEM-LAM precipitable water vapour output using the
4bb73c4ff8a5843f9abf0bc10d2f3970.ppt