Regional Weather Modeling and Grid Computing at CDAC
































- Slides: 32

Regional Weather Modeling and Grid Computing at C-DAC SECG/CAS/e. Science-2007 Computational Atmospheric Sciences Team, Centre for Development of Advanced Computing, Pune

Scope • Background • Mausam Grid • Case Study – Regional Coupled Climate Model on Grid SECG/CAS/e. Science-2007 – Grid-based Meteorology-Air Quality Modeling 2

Computational Atmospheric Sciences Activities • Computational Research • Workflow Environment Development • Technology Development • Parallel Programming • Model Porting, Optimisation & SECG/CAS/e. Science-2007 Simulations • Grid Computing Joint Collaborative Research Turnkey solutions Contract Projects Consultancy 3

SECG/CAS/e. Science-2007 C-DAC’s Vision – Mausam Grid 4

Mausam Grid • • • SECG/CAS/e. Science-2007 • • A portal for Atmospheric & Environmental Sciences Community Multimodel, Multidata, Query based Problem Solving Environment Secured Access Based on Garuda Infrastructure Regional Models – Reg. CM, WRF, RSM Ocean Models – ROMS, HYCOM Air Quality Models – AERMOD, WRF-CHEM 5

Resources SECG/CAS/e. Science-2007 • Specifies Information regarding the available resources on Garuda • Status of resources, such as Cluster Name, availability of Processors and nodes are displayed • User can take decision regarding no. of processors to specify based on machine availability. 6

Job Submission • • • SECG/CAS/e. Science-2007 • • • Specify the model from the drop down list. (Users can run various atmospheric models ported by the team of experts across Garuda Infrastructure. ) User has to specify the type of resource required. (e. g Linux/aix) Easy to remember Job ID. (Jobs are recognized by their name) Selection of resources is automatic, based on the availability of resources of specified type. Can provide multiple input & output files Can specify command line & environmental variables. 7

CASE STUDY SECG/CAS/e. Science-2007 • Reg. CM 3 – ROMS • WRF - AERMOD Reference: http: //nctr-people. pmel. noaa. gov/cmoore/wrf-roms/wrf_roms 1. 2. tar. gz 8

Regional Atmosphere-Ocean Coupled Model EUIndia Project: Development of regional coupled model on Grid SECG/CAS/e. Science-2007 Regional Climate Model (Reg. CM 3): Atmospheric Reg. CM 3 Regional Ocean Modeling System (ROMS): Ocean Development of online coupler : Reg. CM 3 and ROMS Coupled System tested on EGEE grid ROMS Long term runs and Analysis in progress Coupled System used for Climate simulations 9

SECG/CAS/e. Science-2007 WRF - ROMS Coupling • The Models coupled using Model Coupling Toolkit • MCT consists of a set of distributed-memory routines based on MPI • WRF Domain 161 by 161 lateral points and 30 levels • ROMS domain is 200 by 200 horizontal points and 30 vertical levels • ROMS supplies the SST to WRF • WRF provides wind stresses and heat to ROMS • Vortex generation is modeled by initiating a tropical storm like vertical wind field in gradient wind balance. • The source code for this can be downloaded from • http: //nctr-people. pmel. noaa. gov/cmoore/wrf-roms/wrf_roms -1. 2. tar. gz 10

Meteorology Air Quality Coupled System Meteorological data critical input for Air Quality models High resolution NWP models used for AQ model initialization SECG/CAS/e. Science-2007 C-DAC: Development of Coupler for WRF and AERMOD Simulation time/day- WRF: 30 min, AERMOD: 15 min Simulation time – 6 day (serial flow) : 270 min 11

Heterogeneous Grid Infrastructure • Exploit the various Windows & Linux based resources available in the research lab • Middleware used: SECG/CAS/e. Science-2007 – Globus Toolkit – Alchemi (Window based) 12

WRF - AERMOD Grid - Flow SECG/CAS/e. Science-2007 Schematic Presentation of Meteorology-Air Quality Grid System Contact: cas@cdac. in 13

Seismic Tomographic Inversion on EU-India Grid Seismic Data Processing Team Centre for Development of Advanced Computing SECG/CAS/e. Science-2007 Pune University Campus, Pune - 411 007 (India) http: //www. cdac. in/html/secg/sdp. asp 14

What is Tomography ? The word "tomography" derived from the Greek tomos (slice) and graphia (describing). The method used in: • Medicine • Astronomy SECG/CAS/e. Science-2007 • Geology • Materials Science 15

Seismic Tomographic Inversion • Seismic inversion provides method for direct estimation of velocity distribution in the subsurface of the earth using initial earth model as guess. • It generates a cross-sectional picture of an object by utilizing SECG/CAS/e. Science-2007 the object’s response to external energy source. • Seismic applications like mining, dam foundation building, ground water probing etc. 16

Seismic Tomographic Survey Ray diagram for different positions of sources and receivers SECG/CAS/e. Science-2007 Cross-hole test method – field setup P-wave velocity contour along the vertical Velocity distribution in different pixels plane between source and receiver borehole 17

Seismic Tomographic Inversion • The inversion scheme is Real-coded Genetic Algorithm. SECG/CAS/e. Science-2007 • Hybrid Island model (Chipperfield and Fleming, 1996) is adopted for the parallel implementation of the genetic algorithms 18

EU-INDIA GRID SECG/CAS/e. Science-2007 DEMONSTRATION

Files required for job submission: • gatom 4 grid_crossplot_N (executable) • grid_job. param (job card) • init_vel_L. mod (initial model file) SECG/CAS/e. Science-2007 • Traveltime_L. dat (field data file) • Tomo. Output. tar (application specific output file) • Output. Directory. sh 20

Tomo. jdl [ Executable = “gatom 4 grid_crossplot_N “; Job. Type = “MPICH”; Node. Number = 4; Std. Output = “std. out”; Std. Error = “std. out”; SECG/CAS/e. Science-2007 Input. Sandbox = {“gatom 4 grid_crossplot_N “, “grid_job. param”, “vel_L. mod”, ”Traveltime. dat”, ”init_vel_L. mod”, ”Output. Directory. sh”}; Output. Sandbox = {“std. out”, ”std. err”, ”Tomo. Output. tar”}; ] 21

EUIndia Grid Submission Proxy Certification SECG/CAS/e. Science-2007 jvratnam@egrid-4: ~/SEISMIC$ voms-proxy-init -voms euindia Cannot find file or dir: /home/jvratnam/. glite/vomses Your identity: /C=TW/O=AP/OU=GRID/CN=Akshara Kaginalkar/Email=akshara@cdac. in Enter GRID pass phrase: Creating temporary proxy. . . Done Contacting voms 2. cnaf. infn. it: 15010 [/C=IT/O=INFN/OU=Host/L=CNAF/CN=voms 2. cnaf. infn. it] "euindia" Done Creating proxy. . . . Done Your proxy is valid until Sat Sep 22 01: 52: 29 2007 22

Job Submission jvratnam@egrid-4: ~/SEISMIC$ glite-wms-job-submit –a -o id Tomo. jdl Selected Virtual Organisation name (from proxy certificate extension): euindia Connecting to host eu-india-02. pd. infn. it, port 7772 Logging to host eu-india-02. pd. infn. it, port 9002 ====== glite-job-submit Success ============ The job has been successfully submitted to the Network Server. Use glite-job-status command to check job current status. Your job identifier is: SECG/CAS/e. Science-2007 - https: //eu-india-02. pd. infn. it: 9000/Qc. CFm. I 6 l-IRXXc. VT 2 f. IF 5 A The job identifier has been saved in the following file: /home/jvratnam/SEISMIC/id 23

Job Status jvratnam@egrid-4: ~/SEISMIC$ glite-wms-job-status -i id ******************************* BOOKKEEPING INFORMATION: SECG/CAS/e. Science-2007 Status info for the Job : https: //eu-india 02. pd. infn. it: 9000/Qc. CFm. I 6 l-IRXXc. VT 2 f. IF 5 A Current Status: Scheduled Status Reason: Job successfully submitted to Globus Destination: ictpgrid-ce-1. ictp. it: 2119/jobmanager-pbs-euindia Submitted: Fri Sep 21 13: 53: 35 2007 CEST ******************************* 24

Job Status jvratnam@egrid-4: ~/SEISMIC$ glite-wms-job-status –i id ****************************** BOOKKEEPING INFORMATION: SECG/CAS/e. Science-2007 Status info for the Job : https: //eu-india-02. pd. infn. it: 9000/Qc. CFm. I 6 l -IRXXc. VT 2 f. IF 5 A Current Status: Running Status Reason: Job successfully submitted to Globus Destination: ictpgrid-ce-1. ictp. it: 2119/jobmanager-pbs-euindia Submitted: Fri Sep 21 13: 53: 35 2007 CEST ******************************* 25

Job Status jvratnam@egrid-4: ~/SEISMIC$ glite-wms-job-status –i id ****************************** BOOKKEEPING INFORMATION: SECG/CAS/e. Science-2007 Status info for the Job : https: //eu-india-02. pd. infn. it: 9000/Qc. CFm. I 6 l. IRXXc. VT 2 f. IF 5 A Current Status: Done (Success) Exit code: 0 Status Reason: Job terminated successfully Destination: ictpgrid-ce-1. ictp. it: 2119/jobmanager-pbs-euindia Submitted: Fri Sep 21 13: 53: 35 2007 CEST ******************************* 26

Retrieving of the Output SECG/CAS/e. Science-2007 jvratnam@egrid-4: ~/SEISMIC$ glite-wms-job-output --dir output –i id Retrieving files from host: eu-india-02. pd. infn. it ( for https: //euindia-02. pd. infn. it: 9000/e_l 5 k. Abe 0 rkeg 2_Jp. VEmag ) ************************************ JOB GET OUTPUT OUTCOME Output sandbox files for the job: https: //eu-india-02. pd. infn. it: 9000/e_l 5 k. Abe 0 rkeg 2_Jp. VEmag have been successfully retrieved and stored in the directory: /output ************************************ 27

jvratnam@egrid-4: ~/SEISMIC/output$ tar -xvf Tomo. Output. tar SECG/CAS/e. Science-2007 Tomo. Output/. /job. status Tomo. Output/. /output. dump Tomo. Output/. /error. crv Tomo. Output/. /fitness. crv Tomo. Output/. /gen_err. out Tomo. Output/. /vel_mod_L_iter_1. GRD Tomo. Output/. /pre_rectime_L_1. dat Tomo. Output/. /obs_pre_rectime_L_1. dat Tomo. Output/. /vel_mod_L_iter_2. GRD Tomo. Output/. /pre_rectime_L_2. dat Tomo. Output/. /obs_pre_rectime_L_2. dat 28

SECG/CAS/e. Science-2007 RESULTS

SECG/CAS/e. Science-2007 Output Data. . . : : : : : : : : : : : : : : : : : : : : : : : : : PARALLEL GENETIC ALGORITHM : : : : For Seismic Traveltime Tomography : : : : (2 D Elastic Impedance Modelling) : : : : ============== : : : : (c) Seismic Data Processing Group : : : : C-DAC, Pune : : : : All rights reserved. : : : : : : : : : : : : : : : : : : : : : : : GA GA Type : � No. of Runs : 1 No. of Generations : 10 No. of Populations : 10 Probability Crossover : 0. 900000 Probability Mutation : 0. 000100 Bits per Gene : 0 Tournament Size : 2 Starting seed : 0. 200000 Crossover type : b eta and mutation : 0. 000000 Open. Mod 30

Result of L shape Model (a) (b) (c) SECG/CAS/e. Science-2007 (d) (a) initial velocity model (b) reconstructed model (c) original model (d) error vs generation graph for a Lshaped model (velocity of 2. 5 km/sec and 3. 5 km/sec) with the background velocity of 3. 0 km/sec. 31

SECG/CAS/e. Science-2007 Thank you e-mail : sdp@cdac. in