ARCS/IRI Regional Consortium J. Roads, S. Chen, J.
ARCS/IRI Regional Consortium J. Roads, S. Chen, J. Chen ECPC D. Lettenmaier, E. Salathe, E. Miles UW H. Juang, J. Han, S. Lord NCEP S. Cocke, T. Larow FSU A. Robertson, J. -H. Qian, S. Zebiak IRI Previous Work Background (current work) Examples (Roads) ECPC fireweather and model development FSU crops NCEP regional ensembles IRI statistical downscaling and model development Examples (Lettenmaier) Landsurface hydrology Ecology
2003: ARCS/IRI ARCs/IRI S. A. Reg. Mod. Domain Roads et al. 2003, J. Geophys. Res. (in press) (1) Scripps Experimental Climate Prediction Center regional spectral model (RSM) (2) Florida State Univ. nested regional spectral model (FSUNRSM) (3) Goddard Institute for Space Studies regional climate model (RCM) (4) IRI regional climate model (RegCm2) 2003: ARCS/IRI
IRI/ARCs Regional Model Comparison Model Reanalysis Scripps RSM FSUNRSM GISS RCM IRI RegCM2 Levels 28 28 27 16 14 Type spectral spectral spectral
grid grid Davies SW. Rad. Lacis and Chou and Lee CCM 3.6 (Kiehl CCM 3.3 (1982) Hansen (1996) et al. 1998) (Kiehl et (1974) al. 1998) Harshvardhan CCM 3.6 LW Rad. GFDL GFDL CCM 3.6 (Kiehl and Corsetti (Schwartzkopf (Schwartzkopf et al. 1998) (Kiehl et (1984) et al. 1994)
et al. 1994) al. 1998) Convective SAS (Kalnay SAS* (Hong Zhang-McFarlane modified Kuo Grell (Krishnamurti (1993) Param. et al. 1996) and Pan 1996) (1995) et al., 1990) Krishnamurti Holtslag et PBL Louis et al. Hong and Pan Holtslag and et al. (1990) (1982) (1996) Boville (1993) al. 1990 Land NOAA NOAA* (see FSU (Cocke and Fulakeza et BATS al. (2002)
Surface (Kalnay et al. Chen and LaRow 2000) (Dickinson 1996) Roads 2002) et al. 1993) 2003: ARCS/IRI Most regional models were able to adequately simulate the new Xie and Arkin .5 deg land precipitation climatology and interannual variability, although they added little skill to the driving R1. The regional model ensemble mean systematic errors were somewhat smaller than the driving NCEP/NCAR reanalysis systematic error However, the ensemble reduction in systematic error did not increase the correlations Threat scores also indicated that the regional model ensemble was not noticeably better than the driving reanalysis.
Part of the problem may be due to a bias in high rainfall events. All regional models apear to have this bias, which is only partially ameliorated in the ensemble forecast. In short, regional models provide many advantages, including better control over local conditions, which have yet to be taken advantage of, but at the same time they are still noticeably influenced by large-scale physical parameterizations, which need to be improved in future regional models. New Regional Application Work Basic regional model development is still taking place over US, Asia, Brazil However, as part of the IRI/ARCs regional consortium, we were subsequently requested to work with the applications community to better apply regional forecasting methodologies.
Basic applications we are now investigating include seasonal forecasts of: Hydrology, Fire Danger, Crops, Ecology Global IRI,NCEP,ECPC,FSU Regional IRI, NCEP, ECPC, FSU Hydrology IRI, ECPC, UW Fire ECPC Crop FSU Ecological UW 2003: ARCS/IRI ECPC Regional
Consortium Work ECPC is developing and evaluating the Regional Spectral Model, which is being used to drive various application models. Regional RSM forecasts (CA, SW, US, BZ) are being evaluated. Various improvements to the RSM are being implemented. ECPC has acquired the National Fire Danger Code, and is attempting to develop useful long range forecasts over the US and elsewhere. This fire danger code supplements its current simplified fire weather index, which is only influenced by wind speed, relative humidity, and temperature. ECPC has transferred its regional modeling methodology to Taiwan, Hong Kong, and Helinjong, and is currently driving their regional forecasts with ECPC global forecasts (4 month forecasts once a week). Taiwan and Hong Kong are concerned with hydrologic applications and Helinjong is concerned with fire danger forecasts. ECPC has recently acquired the VIC hydrologic model to develop land surface
forecasts, including streamflow for the US. Comparisons will be made to current Noah model We have also used a river network model, and have since found that for certain heavily managed basins that we will need to also include an engineering model. 2003: ARCS/IRI National Fire Danger Rating System Structure Input Site Description Fuel Model Slope Class Live Fuel Types Climate Class Latitude Output Calculated
Average Annual Precipitation 1300 LST Observation Relative Humidity Temperature Cloudiness 24-Hour Observations Precipitation Amount Maximum Temperature Wind Speed Fuel Stick Moisture Carryover Fuel Moistures (FM)
Max/Min RH 100-Hour Min Temp 1000-Hour Periodic Measurements Precipitation Duration Season Code & Greenness Factor Live Woody FM (88) KBDI (88)
Drought Fuel Ignition Component IC Optional pathway 1-hr FM (88) 10-hr FM 100-hr FM 1000-hr FM Live FM Contribution of dead FM to SC Contribution of dead FM to ERC
Spread Component SC Burning Index BI September 19,2000 Energy Release Component ERC 2003: ARCS/IRI Summer validation correlations with AC (cf. A. Westerling): (a) FWI; (b) IC;
FSU is assessing the skill of both global (T63L17) and regional (20km) models for driving crop models. One of the crop models being used to simulate maize yield is the CERES-Maize simulation model (Ritchie et al., 1998). This model is a dynamic process based crop model that simulates plant response to soil, weather, water stress and management practices. The model calculates development, growth and partitioning processes on a daily basis, beginning with planting and ending at harvest maturity. Input into the crop model are the Regional Models daily values of: Max. Temperatures Min. Temperatures Precipitation Surface Solar Radiation Crop model appears sensitive to all four input parameters. One measure of the skill will be the crop yields determined by the crop models and verified against the observed yields for selected locations in Florida and Georgia. Preliminary results for selected locations in Florida are encouraging and show greater skill of the FSU regional model to predict Maize yields when compared to the FSU global model (Jagtap et al 2002). 2003: ARCS/IRI
Spring/Summer growing season (MAMJJA) 180 days from 1987 to 1999 NCEP NCEP currently makes an ensemble of 20 7-month global forecasts, which are initialized every 12 hours 5 days before the start of the month and continuing to 5 days after. These forecasts are supplemented by hindcasts of 10 hindcasts for the same month but for each year for the previous 21 years. In all there are 230 7-month forecasts made every month. This strategy allows the model to be changed monthly if needed. NCEP is now developing a corresponding ensemble of 5 4-month regional US forecasts, initialized every 24 hours 2 days prior and continuing to 2 days after the start of the month. These forecasts are being supplemented by hindcasts, made the previous month, of 1 4-month hindcast for the same month but for each year of the previous 21 years. In all there will be 26 forecasts made each month. Depending upon computer time, these hindcasts will be
supplemented by up to 4 additional ensemble members, started every 24 hours 2 days prior and 2 days after the start of the month. Additional regions, like Brazil may could also be added (depending on computer resources). 2003: ARCS/IRI http:// wwwt.emc.ncep.govv.gov/ mmb/RSM Last years experiments: http://nomad2.ncep.noaa.gov/ cgi-bin/web_rsm.sh Download monthly mean climate average, and ensemble forecast average Plot variables of climate average and ensemble forecast average from
web by users 4 month GSM forecast R2 Analysis 4 month RSM forecast Future Experiments GSM: NCEP GFS 28LT62 Parallel RSM (MPI version): new GFS physics Forecast range: 4 month forecast Domain: 30km lon: 228.908 -295.832 lat: 20.823 - 50.994 3 member ensemble hindcasts, 5 members ensemble forecasts Data to be stored: flx, sig, sfc, r_sig, r_sfc, monthly mean of flx, pgb, r_pgb 2003: ARCS/IRI Statistical downscaling of daily rainfall occurrence at
IRI Construct a statistical transformation of atmospheric GCM predictions from the IRI two-tier system Predict local daily rainfall characteristics (e.g., occurrence frequency, dry-spell frequency) over Cear up to several seasons in advance Train on observed station data Application to IRI seasonal forecast Comparison with dynamical downscaling 2003: ARCS/IRI Statistical downscaling of daily rainfall occurrence over NE Brazil from a GCM with a Hidden Markov Model ECHAM4.5 run with observed SSTs 19752002 2003: ARCS/IRI
Interannual Changes in NHMM Rainfall Frequency (10-station average) 2003: ARCS/IRI Precipitation and lowlevel winds Feb 1999 (a) NCEP/NCAR reanalysis (b) SG1: uniform high resolution (50km grid) SG3: stretched grid from 150km to 50km (d) SG5: stretched grid from 250km to 50km 2003: ARCS/IRI ARCS/IRI Regional Model Consortium Summary (Part I)
The IRI/ARCS regional model consortium has now moved beyond simply comparing regional simulations (and forecasts) to connecting these regional models to the application community Firedanger, Crops, Hydrology, Ecology In fact, the application community has already connected directly to the global modeling community, in part because this community has larger forecast ensembles available. For example, the IRI is currently attempting to statistically downscale from its current multi-model global model ensemble rather than develop corresponding multi-model regional model ensembles. To provide additional regional model ensembles, NCEP has begun to develop ensembles of regional model forecasts for the US, which are freely available to interested researchers The regional consortium is also still attempting to improve and further develop regional models ECPC RSM CVS with pressure diffusion will soon replace current RSM96/97 versions and thus be more compatible with the ECPC SFM NCEP is replacing RSM97 physics with new GFS physics
Variable resolution global models are being investigated by the IRI 2003: ARCS/IRI ARCS/IRI Regional Applications Project 2. Hydrology and water resources projects Dennis P. Lettenmaier with contributions from John Roads (Scripps Institution of Oceanography), and Eric Salathe and Ed Miles (University of Washington) Climate Diagnostics and Prediction Workshop Sparks, NV October 23, 2003 2003: ARCS/IRI 1. Downscaling climate predictions for the Pacific Northwest using RCM (Regional Climate Model) Project leads: Eric Salathe and Ed Miles Climate Impacts Group University of Washington Background:
Mesoscale models must resolve below 20 km in order to simulate meteorology of the Puget Sound region The interaction of mesoscale processes with climate change and variability may yield a different response than interpolating from coarse resolution simulations MM5 will be applied as a Regional Climate Model forced by Global Climate Model simulations (PCM CCSM) 2003: ARCS/IRI RCM Details 10-year MM5 model runs nested in PCM. (1990-2000; 2045-2055; 2090-2100). MM5 nests at 108km, 36km, and 12 km horizontal resolution. WSU-EPA air pollution study over Chicago 4th nest at 12km. To be determined: 1. Nudging: on interior? How frequently? 2. Re-initializing MM5: How frequently? How much time overlap?
2003: ARCS/IRI Applications of MM5 Results for Climate Change 1. Water resources in small basins West of the Cascades Seattle hydropower and water supply 2. Air quality -- collaboration with EPA-STAR project at Washington State Biogenic and forest fire emissions CMAQ air quality model Puget Sound and Northern Midwest 2003: ARCS/IRI 2003: ARCS/IRI 2. COUPLING RSM with VIC land surface
hydrology Project lead: John Roads, SIO Initial Conditions reanalysis data RSM Regional Spectral Model Precipitation Pressure Radiation (Shortwaver, Longwaver) Wind Humidity Air temperature Boundary Conditions climatology reanalysis data RSM and VIC Physical processes VIC energy balance mode
simulates snow process simulates surface energy fluxes Sensible heat flux Latent heat fluxes Momentum Flux RSM models precipitation Process Subgrids represents land use/cover heterogeneity Variable infiltration capacity models the spatial and temporal variabilities of runoff-prediction VIC Land Surface Nonlinear baseflow based on the third layer soil moisture VIC simulates surface runoff and baseflow 2003: ARCS/IRI
Process of Initializing the RSM and VIC System RSM sea/land mask and Orography from the fixed fields RSM climatology data for initializing sea/land parameters and variables VIC land soil and vegetation fixed fields Initializing analysis fields using climatology data then, reading RSM analysis fields for updating sea/land variables Initializing forecast sea/land fields using analysis fields then, reading RSM forecast fields using provided base field Initializing VIC parameters and variables using VIC land soil and vegetation fixed
fields and RSM forecast fields Merging the analysis fields and forecast fields Check the consistence of RSM land surface mask with VIC land grids Output of VIC and RSM surface fields 2003: ARCS/IRI Preliminary Results using VIC and RSM 2003: ARCS/IRI 3. S/I Hydrologic Forecasting Project Project lead: Dennis Lettenmaier, University of Washington www.hydro.washington.edu/Lettenmaier/Projects/fcst/index.htm 2003: ARCS/IRI
Project Goals 1. Produce real-time seasonal ensemble hydrologic forecasts: based on experimental climate model forecast products based on established methods (such as ESP) of streamflow, for selected large river basins (primarily in the West) of snowpack / soil moisture anomalies 2. Assess experimental product skill relative to established forecast products 3. Evaluate relative hydrologic prediction skill due to ICs initial land surface conditions (soil moisture, snow) and due to climate forecast skill. 2003: ARCS/IRI Forecasting Approach General forecast ensemble meteorological sequences downscaling process *
local scale weather inputs hydrologic simulation streamflow, soil moisture, snowpack, runoff * for climate model forecasts Experimental forecast applications NASA NSIPP-I Forecasts 2-2.5 degree resolution monthly total P, avg T VIC Hydrology Model 1/8 degree resolution
daily P, Tmin, Tmax NCEP GSM Forecasts 1.9 degree resolution monthly total P, avg T 2003: ARCS/IRI Overview: VIC Simulations start of month 0 1-2 years back end of mon 6-12 forecast ensemble(s) VIC model spin-up climatology ensemble data sources NCDC met. station obs. up to 2-4
months from current LDAS/other real-time met. forcings for remaining spin-up climate forecast information snow state information Forecast Products streamflow soil moisture runoff snowpack derived products
2003: ARCS/IRI Overview: Spin-up approach, Index Stn Method 1. interpolate monthly percentiles from sparse index stations to 1/8 degree grid 2. find percentiles matching amounts in the dense station-derived climatology 2003: ARCS/IRI Current Columbia River basin forecasts 2003: ARCS/IRI Current Forecasts: Initial Soil Moisture CPC Percent of Normal Precip September 25, 2003
Jun-Jul-Aug % Sep 2003: ARCS/IRI Current forecasts: Snake River locations Upper Snake Storage Forecast 2003: ARCS/IRI Current forecasts: Reservoir system storage System Storage Forecast from SnakeSim includes: Jackson Lake Palisades Island Park Ririe
American Falls Lake Walcott Unconditional ESP Full Pool ensemble mean 2003: ARCS/IRI
Example. 1,000 people in the city. 1 is a terrorists, and we have their pictures. Thus the . base rate. of terrorists is 1/1000. Suppose we have a new terrorist facial recognition system that is 99% accurate.
OH MY GODS!! AN INTRODUCTION TO GREEK MYTHOLOGY * * * * * * * * * * * ARTEMIS TITLE-Goddess of hunting, wilderness and wild animals. She was also a goddess of childbirth, and the protectress of the girl...
Dash board. Cup Holder. arithmetic average by sites. arithmetic average by sight & type of vehicle. 2% of vehicles had MRSA. 36% of vehicles had S. aureus. Ascending. Asending. Staph isolation according to sight percent
Medical Term Elements Video Back to Directory Medical Terms Are Built from Word Parts Word Part Word root Prefix Suffix Combining form Example (Meaning) cardiogram (record of the heart) pericardium (around the heart) carditis (inflammation of heart) cardiomyopathy (heart muscle...
As is true with real lenses, every critical "lens" will make the work "look" a little bit different. Every critical theory has a set of assumptions about what is . most important . to look at in a work of...
Limited Budgets for Merit Pay - May result in zero sum game, in which some "good" performers are labeled "average" or "poor" in a ranking or forced distribution scheme to ration pay raises. Merit pay increases may be viewed as...
The limbic/emotional brain generally refers to amygdala, hippocampal formation, hypothalamus, thalamus, and nearby "paralimbic" cortex, such as the anterior cingulate cortex, orbitofrontal cortex insula, and temporal poles Given the role of aberrant, intrusive, emotional memory in PTSD symptomatology, the limbic...