Meteorology 485 Long Range Forecasting Friday, February 13, 2004
Long Range Guidance n CDC - Climate Diagnostic Center (Boulder) n Blended model - both dynamic and statistical n n DYNAMIC: Four AGCM - NCEP MRF 9; GFDL-R 30; NCAR-CCM 3 and IRI-ECHAM 3 using forced SST’s from 1950 -99 climatology STATISTICAL: Multi-variate regression model trained on the relationship between tropical SST’s and US Seasonal T, P training period 1950 -94. Tested from 1995 -2003 for skill scores based on hindcasts http: //www. cdc. noaa. gov/seasonalfcsts/
Models n n Climate Diagnostic Center Precipitation Temperature
Long Range Guidance n Statistical Models n CA=Constructed Analog/ Climate Prediction Center n Huug van den Dool A linear combination of past observed anomaly patterns such that the combination is as close as desired to the initial state. n A forecast is obtained by persisting the weights assigned to each year in the historical record and linearly combining the states following the initial time in the historical years. n See: ftp: //ftpprd. ncep. noaa. gov/pub/cpc/wd 51 hd/sst/200312/cahgt_anom. 1. gif
Models n CA Outlook for Feb-Mar-Apr, 2004
Constructed Analog n February 2004 Departures (to date) http: //www. cpc. ncep. noaa. gov/soilmst/index_jh. html
Constructed Analog n February 2004 Departures: Analog-Mapper http: //www. cdc. noaa. gov/USclimate/USclimdivs. html
Constructed Analog n CA based on Feb 1 -11 Anomalies for March, 2004 http: //www. cdc. noaa. gov/USclimate/USclimdivs. html
Constructed Analog n Checks and Balances n n n Do the Temp and Precip fields make sense? Are the fields dominated by 1 -2 highly anomalous years? Do other techniques produce similar analog years? n n n 500 mb flow pattern analogs (NA or NH) Can the historical contributions be weighted? Can GFS 10 day surface anomaly fields be used to extend adjust the constructed analog Is an ‘ensemble’ of analogs useful, or even a lagaveraged analog What is the reliability of downscaling? Can this be automated?
Constructed Analog n Downscaled - example for western Pennsylvania
Constructed Analog n Downscaled - example for a western U. S. city
Long Range Guidance n n Statistical Models Markov – Climate Prediction Center n n A future variable is determined by the present variable, but is independent of the way in which the present state arose from its predecessors. Examples: n n Weather conditions deduced from seaweed (ie, wet seaweed means it is raining, dry means sunny, damp is indeterminate) State of the weather not restricted to state of seaweed, but most recent past conditions combined with current state can yield a credible future state. . . See: ftp: //ftpprd. ncep. noaa. gov/pub/cpc/wd 52 yx/web/MKmodel_ellfb_clim 71 -00/fig 1. gif
Models n Markov Outlook for SST
Long Range Guidance n Soil Moisture - Constructed Analog System Database too limited (not enough actual observations) -must generate estimated soil moisture using simple hydrologic model (Huang, et al 1996) d. W/d. T = Precip - Evaporation - Runoff + Transpiration n Challenge: P varies by 2 or 3 times annual E n Data calculated for 344 Climate Divisions since 1932 n Analog constructed based on soil moisture anomalies with each year contributing a weight for the best fit n Projection based on same weights, but for next month or two or season. n Most effective tool in April-July when d. W/d. T changes http: //www. cpc. ncep. noaa. gov/soilmst/index_jh. html n n
Models n Soil Moisture
Next Few Weeks n Commercial Long Range Forecast n n Accuweather, WSI, Dynamicpredictables, Wx. Risk, Wx Data International Forecast Centers n UKMET, ECMWF, CMC, Brazil, South Africa, Australia, Japan and Korea
Скачать презентацию Meteorology 485 Long Range Forecasting Friday February 13
742130fa272a84c0e275d585456627fe.ppt