Outline for NATS 101 with Professor Castro at UA
Outline for NATS 101 with Professor Castro at UA
Popular in Course
Popular in Department
This 46 page Class Notes was uploaded by an elite notetaker on Friday February 6, 2015. The Class Notes belongs to a course at University of Arizona taught by a professor in Fall. Since its upload, it has received 20 views.
Reviews for Outline for NATS 101 with Professor Castro at UA
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/06/15
Use of Emerging Applications of the Weather Research and Forecasting Model to Investigate the North American Monsoon Christopher L Castro Francina Dominguez and Stephen Bieda Ill 1 M Department of Atmospheric Sciences and SAHRA r I ringi Sustainability n1 semiArid Hydrology and Riparian Areas UASCIENCE Presentation Outline What is a regional atmospheric model and why do we use it for dynamical downscaling What is WRF and how is it currently being used operationally here at UA Recently funded projects which are using WRF Shortterm monsoon forecasting and adjoint sensitivity Bieda RCM downscaling of NCEP GCM and IPCC data Dominguez Possible connections to hydrologic applications Some slightly modified figures that I present in my NATS 101 and graduate modeling courses Objective Analysis Data must be interpolated to some kind of grid so we can run the numerical weather prediction model this is called the initial analysis For a regional model these are equally spaced points Grid spacing 35 km Structure of atmospheric models Dynamical Core Mathematical expressions of Conservation of motion ie Newton s 2nd law F ma Conservation of mass Conservation of energy Conservation of water These must be discretized to solve on a grid at given time interval starting from the initial conditions analysis Parameterizations One dimensional column models which represent processes that cannot be resolved on the grid Called the model physics but it is essentially engineering code Equations to represent in dynamic core MUST SOLVE AT EVERY GRID POINT MASS CONSERVATION apat v m7 ENERGY CONSERVATION 6661 I7 V0 S9 CONSERVATION OF MOTION aVat 47 VV lpr g1 25 x 7 CONSERVATION OF MOISTURE a 1lt9t 7 an 54 n 1 2 3 PieIke 2002 Whv is just doinq this REALLY REALLY HARD Have discretize the equations so they can be solved on a grid Equations are nonlinear We haven t even accounted for parameterizations yet Precipitation processes Radiation Turbulent diffusion I Dynamic core Discretized dynamical equations i Boundary conditions Land surface energy balance Boundary layer Dynamical downscaling Definition Use some kind of numerical model to generate finer resqution information from courser resolution information For the atmosphere this is a limited area model Implicit assumption A finer resolution andor improved model physics parameterizations gives a better representation of weather and climate than the driving coarser resolution model ie GCM Better may more fidelity with observations andor improved representation of physical processes If this is not satisfied you re wasting money in terms of computer time to generate simulations and labor to analyze the results Dynamical Downscaling Types Castro et al 2005 Fairly TYPE 1 Shortterm numerical weather prediction out to certain in results 12 weeks and constrained TYPE 2 Retrospective smuation of past climate by downscaling a atmospheric reanalysis perfect latera boundary forcing TYPE 3 Downscale a atmospheric general circulation model forced with fixed surface boundary condition eg SST from some observed inita state 9 Seasonal forecast mode TYPE 4 Downscale a completely coupled atmosphere V ocean general circulation model for integrated for many Very years unconstrained Climate prolectlon mode and uncertain m THE WEATHER RESEARCH amp FORECASTlNG MODEL We Klingons are not just warriors we develop numerical weather prediction models too as you humans THE WEATHER RESEARCH E1 FORECASTING MODEL Based largely on the MM5 model originally developed at Penn State Two dynamical cores NMM NCEP and ARW NCAR The latter is what is used for most research applications and what we use Numerous parameterization options for physical processes Though most heavily used for shortterm weather prediction designed for a broad range of scales and applications Some advantages to WRF Model use and development occurring at numerous institutions user community is large spinup time is relatively quick via online tutorials or NCAR tutorials and runs on wide variety of computer platforms Real time UA forecasting in the Department of Atmospheric Sciences during monsoon WRF rainfall Radar estimated rainfall 13 km grid spacing HZHAL HSK U 24H DSIUEIZUUB IZUBZ r 39 r I 391 Preclpltatlnn In 315ml V I I l y I I12 50W 10374W NDH E MSEIHH 001 111 025 USU E175 M 15 2 2 4U 1D 5 0 Bil 80 1D 35UN Courtesy Mike Leuthold Some consistent problems in NWP monsoon forecasts for Arizona Poor or missing initialization of smallerscale features like Gulf surges outflow boundaries or clouds Model produces thunderstorms but they occur in the wrong place andor the intensity is off Different GCM forcing data different model simulation result Data to initialize the models is completely missing in Mexico What parameterizations to use Use a trial and error approach to figure out what works best operationally Severe weather events that affect urban areas are very difficult to simulate skillfully eg Phoenix Some recently funded WRFrelated projects in my group Use of Regional Atmospheric Modeling to Improve Short and Long term Forecasting Capability of the North American Monsoon System Pls C Castro F Dominguez Sponsoring agency NSF Using Regional Atmospheric Modeling to Investigate Heavy Monsoon Rainfall Events in Arizona and Socioeconomic Implications Pls C Castro S GrossmanClarke ASU Sponsoring agency Science Foundation Arizona Processes Linking Easterly Waves and the North American Monsoon System Pls Y Serra C Castro E Ritchie Sponsoring agency NSF Short term monsoon forecasting and adjoint sensitivity August 2 2005 Severe Weather Event in Phoenix Metro Area A Rim Shot Had typical ingredients 1 Upperlevel inverted trough 2 Lowlevel surge of moisture from the Gulf of California Net result Vertical wind shear high He in low levels upper level divergence and relatively high CAPE 7 xii 39 1 Water vapor imagery on Aug 2 2005 at 15 Terraininduced convection can organize into MCSs west of Mogollon Rim Corresponding NEXRAD radar imagery Severe thunderstorm in Phoenix area Approx 62 3 Aug 2005 Produced Major dust storm Golfball size hail Damaging winds Urban flooding Close to an inch or two or rain in isolated locations Accumulated Precipitation kcm Composite Mean 38N NOAAESRL Physical Sciences Division M 375N 37M 36395N 36M 35395N 35M 345N 34N SEEN iiSW HEW iiZW iiiW HOW iO9W iUBW 20050803 052 3h accumulated rainfall 32 to 62 3 Aug 2005 NARR product NOAA ESRL WRF V3 NWP Simulations of Aug 2005 Event 24 h simulation starting at 12 Z Aug 2 48 h simulation starting at 12 Z Aug 1 Western US domain 2127 km grid spacing on coarsest grid GFS model analysis lateral boundary forcing Standard WRF parameterizations Faama ilmi cag t i j Augugi g 2mg lm Tatal PreCIP calor39mm Accumulated Precipitation kgm Composite Mean 38N NOAAESRL Physical Sciences Division 38M 38M 34M szu JON 25M ZSN 24N iiZW iiiW HOW i09W 108W ii7IIII15UIi15Wli4Wl13U 12Mll lliou og ONIOMOSMOSMONI 20050803 052 ZEN 3h accumulaled rainfall 32 lo 62 3 Aug 2005 3h accumulaled rainfall 32 lo 62 3 Aug 2005 WRF Model V30 NARR product NOAA ESRL Framia ilmiiag iratbm g Total Precfp colormm b Accumulated Precipitation kgm Composite Mean 3M NOAAESRL Physical Sciences Division liZW lllW HOW l09W 108W ZEN iHUME 151MMll lws39l12MillllliOUIOQUIIOSUIIONIIOBWOEMO4W 20050803 052 3h accumulaled rainfall 32 lo 62 3 Aug 2005 3h accumulaled rainfall 32 lo 62 3 Aug 2005 WRF Model V30 42 hour forecasl NARR product NOAA ESRL A Lack of Observations in Mexico There have been lt 5 64 43 virtually no upper air Qm v observations in 3 73 w y 4 V northern MeXIco I since the end of f l quot NAME A Also no data along the Gulf of California to track gulf surges l 1 39 I 39 r V K T A consistent I Q m a 7 5 Pmb39em quotMed by I m Tucson and Phoenix 39 WSFOs during the monsoon 300mb Winds and streamlines Brief Overview of Adjoint Modeling Technique to determine the sensitivity of a NWP forecast for a selected target region to specification of initial conditions within the model domain High sensitivity regions and atmospheric parameters in which small perturbations can produce large effects on forecast features that can be identified Adjoint model is the transpose of the tangent linear operator of the given NWP model An estimate of a differentiable model forecast state response function R defined at a given forecast verification time t can be produced through a modifiable initial state X0 Adjoint Sensitivity of a Simple Response Function R defined at verification time 1 R VILY u v Horizontal winds X0 Model initial state 5R If 5R 6X0 aXf Xf Model final state Gradient of G d t f response function Adlomt ra 39en 0 response function at forecast at start of model mOde39 verification time integration adjoint sensivitity Response Function R Trajectory Basic State Figure 7 Schematic outlining the ow chart of adjoint sensitivity calculation Xiao et al 2008 Antarctic Windstorm Case First demonstration with WRFVAR Adjoint Sensitivity to Adjoint Sensitivity low level u to low level v RESP FUNCE39I ONV DEFI E HERE Units m s1 Xiao et al 2008 Adjoint model caveats for monsoon convection 1 Does the linearity assumption hold 2 Parameterized processes are not accounted for in the adjoint model yet Sensitivity only to dry dynamics Response Function R Rzggkunggfi Defined in a 10 x 10 grid point box over central Arizona for 27 km grid spacing Verification time is 62 Aug 3 Simulation hour 18 WRF model ow level winds ZEN 2m mum16wI5w1um13M12wquotImamoevnuamomosvnommw 10 Adjoint sensitivity to initial conditions Model level 5 Water Vapor Meridional Wind 125 425 o5 o7s 415 vim IGWI 5Wi um um I2WI Ilwi mmmomomosmosmom quotTimismsmkua39m12wquotwmovnomoamomcsvnasmom Units m2 5392 kg kg391 Units m S391 Adjoint sensitivity to initial conditions Model level 20 Water Vapor 3 WI 15M 15m er lam izvn I quotI tomsvnaawmmasmnsmmw Units m2 5392 kg kg391 Meridional Wind yimwm mm mm IJWI iZ MHM mmemoamomoawunsmmw Units m 5391 125 Ongoing work Higher resolution simulations comparable to current UA WRF monsoon forecasts Testing of additional forecast aspects more directly tied to the development of convection using the adjoint sensitivity method eg CAPE moisture flux convergence Incorporation of adjoints of parameterizations eg convection microphysics Simulation of intensive observing periods IOPs during the North American Monsoon Experiment NAME in 2004 and corresponding adjoint sensitivity experiment IOPs corresponded with development of organized convection like the Aug 2005 case Assimilate NAME field campaign data into the aforementioned NAME IOP simulations Expected outcome Identify hot spots of forecast sensitivity that will lead to a permanent longterm monsoon observing system for US and Mexico Regional climate modeling Definition A numerical weather prediction model integrated for a period longer than about two weeks so that the sensitivity to initial conditions is lost Successful representation of the monsoon in a retrospective sense Castro et al 2007 Regional model response Summer teleconnection Observed 500mh height anomalies m in early July associated with one of the dominant modes of Paci c SST variability Corresponding difference in diurnal moisture flux convergence as simulated hya regional model RAMS downscaling an atmospheric reanalysis Well that s great but can the same be done for seasonal climate forecasts and climate change projections Answer Yes with two caveats on the driving GCM 1 Does it have a reasonable climatology 2 Summer teleconnections captured If the answer to either is no wasting computer time My opinion on how to proceed with RCM climate forecasting 1 Downscaling of seasonal forecasts 2 Downscaling for climate change projection purposes ie IPCC simulations Comments I know we REALLY want to get 2 but must do 1 first Must assess value added in a seasonal forecast sense before proceeding to climate change projection which has more degrees of freedom Use consistent methodology for both Additionally the disconnects between the research communities that do weather and shortterm climate forecasting vs climate change projection don t help 2008 Official Climate Prediction Center Forecast for this past summer Temperature forecasts are becoming more dominated by longterm trends probably due to climate change Equal chances for monsoon precipitation in the Southwest htt wwwcdcnoaa ov Here s what happened So was the CPC forecast is good or not 60 days ending ZUOEJUIZS 5 ei l ilquot a lquotquot 5 gr V 3 39 391 in t it uh 111 quota nil 139 I39l Hr air If 6 all no u U up 1m 91 Dry Wot Climate Diagnostics Center Precipitation percent above or below normal for past 60 days Generally wet in the Southwest and dry in the Great Plains Note Northern Mexico also experienced 2nd wettest July on record with only 1955 being wetter according to Art Douglas Retrospective CFS Seasonal Forecast downscaling Use a similar domain as Castro et al 2007 RAMS simulations that covers the contiguous US and Mexico Simulate retrospective period 19822007 Downscale 5 ensemble members per year from the date of the May 1 forecast Simulation period through at least the end of August to capture the monsoon Will eventually employ a spectral nudging technique Expected outcome Improved representation of the monsoon in the regional model that will lead to a more accurate seasonal forecast Monsoon precipitation mm per day from WRF downscaled CFS ensemble member vs original GCM data Year 1993 dry monsoon JuneSep average Note An obvious problem in the spatial distribution of rainfall for this year in this particular memberbut the rainfall magnitudes are comparable to what happened with respect to central US flood event WRFCFS downscaled precipitation IPCC Simulated Rainfall During Control Period Precipitation mmday Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Obsewations bccicml bccrbcm20 cccmacgcm31 cccmagcgcm3 17t63 csiromk30 gdecm2D gfdlcm21 91557a o m gissmodeler iapjgoalslioig inmcrn30 ipslicm4 miroc32hires miroc3727medres miub echOAQ mple ch a m5 mrch cm2 32a nca rrcrsm30 nca ripcml ukmohadcm3 Preci itation mmday U39i Lu J Observations 39 39 bccrbcm20 cccmacgcm31 cnrmcm3 csiromk30 gdecmZJ gfdcm21 gissmodeler inmcm30 ipscrn4 miroc342medres miubiechoig mpiecham5 mricgcm232a ncariccsm30 ncarpcm1 ukmoihadcms Jan Feb Mar Apr May Jun 1 Jul v Aug Sep 1 l Oct Nov Dec ukmojadgeml 100an 42V 39 mi skataan minim quotI ampI Z r 3 A 39 R ltn39uri z Mummy W DetroltA ind 39 wk J Cic c39 nd 39 g39 i 239 Chicago Plushmis 3 V i 95me i F s 39 K 3m aCfal1 Ego Iscu fk g 753312 2 m lt v Kansas City fir At 3 hams y harlcslon i O i Allialloy 39 1 mm Xquot fMEntphiS f f Cll WW I A o i i Jr mamaquot Dallas R lacksgnwllel I V gt L H 1 I s New DILWR J A tgllliiio lm g L h E i i I 39 4 GlEHUFMEXEL39D quotinlamorus La g 39m o 3 A Mazali kl 9 K 39v V z 2 a i Tamplco M s V gt E 77 39 L v M id x Gu da aim wquot T gm 1 Eff q v J legw u n7 1231 ms W BELIZE ul ngwmcmruo V I puma Itselmup a Cd LNUJEQ ya 0633 2 I y Mum Mapulcoquot V w Gl cmaIa GUA I EMA 39 Average historical model runs sres20c3m 19702000 Courtesy Francina Dominguez Spectral Nudging in WRF Gonzalo Miguez Macho Universidaa e de Sam iago a e Composfea Eaicia Spain Spectral nudging f Model variables fObser39va l39ionS reanalySiS d 7 LQ Kx y Q Q0 DCIVies Udglng d T m0 6 Opera or Relaxa hon coeffICIenT d m x x S ec rr39al Z Le W LQ Z 219 ltan QWequot39 W npd in lnlgNlmlgM df InlgNlml lI HJ U 9 9 D mn Km spec rr39al nudging coefficient may depend on heigh r To nudge Iongwaves make if nonzero ONLY for39 small m and n RAMS mean June 2000 precipitation mmday 30W Spectral nudging Conventional nudging entire domain 500 mb Killer239 energy specfra Control no nudging 500mb Kinetic Energy control exp Spectral nudging 500mb Kinetic Energy spectral nudging exp Conventional nudging 500mb Kinetic Energy conventional nudging exp Hen days iio day Ha any liltrl any loglO variance loglO variance Hm days iin day 7 20 any keen day iogw variance u Hesn day i 5 to Is 20 25 en 15 40 5 SD 55 so 55 70 wave number Due to higher resolution the model generates small scales not present at t0 i 5 wave numbcr m i5 20 25 an 35 4o es 50 55 so 55 70 The model generates small scale structure as in the control i wl2025M3540455D 5505570 wave number No more structure than in driving fields reanalysis Concluding thoughts A regional model is potentially a very powerful tool to investigate the monsoon in Arizona both in a shortterm NWP sense and climate forecast and projection sense Generating a good result with WRF is by no means simple Sensitivities to the specification of domain size grid spacing model parameterizations length of model simulation For RCM simulations some means to control loss of largescale variability becomes an issue Emerging applications of WRF can very pressing problems with repect to the monsoon 1 How can we develop a longterm monsoon observing system 2 How can we improve summer seasonal climate forecasts How can this work tie to hydrologic applications Regional model simulations are approaching the scale at which they can be used as input to hydrologic models Direct input Additional statistical downscaling to finer resolution Moves away from the idea of stochastic forcing to hydrologic models which is typically used now Possible applications in Arizona and beyond Flash Flood forecasting Longterm streamflow projections Soil moisture forecasts Do you all have ideas I d like to know
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'