Operational mesoscale atmospheric dispersion prediction using a parallel computing cluster |
| |
Authors: | C V Srinivas R Venkatesan N V Muralidharan Someshwar Das Hari Dass P Eswara Kumar |
| |
Institution: | (1) Radiological Safety Division, SE Group, Indira Gandhi Centre for Atomic Research, 603 102 Kalpakkam, India;(2) National Centre for Medium Range Weather Forecasting, 110 003 New Delhi, India;(3) Institute of Mathematical Sciences, 600 113 Chennai, India |
| |
Abstract: | An operational atmospheric dispersion prediction system is implemented on a cluster supercomputer for Online Emergency Response
at the Kalpakkam nuclear site. This numerical system constitutes a parallel version of a nested grid meso-scale meteorological
model MM5 coupled to a random walk particle dispersion model FLEXPART. The system provides 48-hour forecast of the local weather
and radioactive plume dispersion due to hypothetical airborne releases in a range of 100 km around the site. The parallel
code was implemented on different cluster configurations like distributed and shared memory systems. A 16-node dual Xeon distributed
memory gigabit ethernet cluster has been found sufficient for operational applications. The runtime of a triple nested domain
MM5 is about 4 h for a 24 h forecast. The system had been operated continuously for a few months and results were ported on
the IMSc home page.
Initial and periodic boundary condition data for MM5 are provided by NCMRWF, New Delhi. An alternative source is found to
be NCEP, USA. These two sources provide the input data to the operational models at different spatial and temporal resolutions
using different assimilation methods. A comparative study on the results of forecast is presented using these two data sources
for present operational use. Improvement is noticed in rainfall forecasts that used NCEP data, probably because of its high
spatial and temporal resolution |
| |
Keywords: | Operational mesoscale dispersion model cluster computer MM5 RSL FLIC |
本文献已被 SpringerLink 等数据库收录! |
|