Numerical forecasting is a term that refers to the usage of mathematical equations to forecast the weather and is based on a large number of very-high-speed calculations that only supercomputers are capable of performing. It is inferred that these equations are more multifaceted than those used in aerospace engineering.

On an average, National Weather Service (NWS) receives as high as 100 million weather observations every day. Data on a number of different variables such as wind speed, air temperature, barometric pressure and humidity is collected from many sources including land based observation points and ships moving around in the ocean. This information is then fed into a supercomputer, which usually has two sets of several fast processors capable of parallel processing. The first set performs weather-related calculations used in forecasting, while the other set is continuously looking for ways to improve the supercomputer’s software, thereby making weather predictions more accurate and meticulous.

Fig-4
Fig. 4: A 24-hour cloud forecast loop from LAPS using MM5 as the forecast model (Source: www.laps.fsl.noaa.gov)

TriNet is a five-year-old collaborative project focused on producing a better, more effective, real-time earthquake-information system for southern California, USA. A high-tech system developed by TriNet has the capability to provide instantaneous damage reports for emergency relief in certain areas of greater Los Angeles area (USA). At its core, Tri-Net system consists of a broad array of earthquake sensors linked to computers. When an earthquake occurs, seismographs automatically record and transmit data to a computer at California Institute of Technology. The computer synthesises a shake map from the entire ground-motion data, which immediately informs emergency managers about the worst-shaken location. Emergency relief can then first be provided to the most severely affected areas. The shake map also indicates where the ground did not shake, thus enabling emergency crews to establish relief shelters and hospitals in areas near destruction points.

IBM’s Deep Thunder
Deep Thunder is a research project by IBM that aspires to develop short-term local weather forecasting using high-performance computing. The project belongs to the same family as Deep Blue system that beat world chess champion Garry Kasparov in May 1997. It uses information gathered by NWS but focuses on much smaller geographic areas than NWS and that too in greater detail.

Fig-5
Fig. 5: India’s Param Yuva II supercomputer (Source: www.crazyengineers.com)

Deep Thunder takes data and puts it through a numerical model, which predicts the weather. It works on software called Local Analysis and Prediction System (LAPS) that can process up to a million separate pieces of information each day. This entire system consists of several hardware and software components in an integrated environment, namely, a high-performance computer system (IBM RS/6000 SP), a forecasting model (such as RAMS, MM5 or WRF), a data-assimilation package (like LAPS), visualisation software (Data Explorer) and associated peripherals.

LAPS software developed by Forecast Systems Laboratory (FSL) of NOAA is a data assimilation and analysis package, which takes in local, national as well as global data from various sources such as satellite, radar and aircraft. LAPS functions as a pre-processing assimilation step, therefore resultant grid data from LAPS is used as initial conditions for the model. It executes serially on a single processor and does more than just model initialisation.

LAPS provides a high-resolution view of the atmosphere in its current state and derived products (like icing, visibility and clouds) and variables (like heat index and buoyancy), which prove useful for a wide array of real-time applications. It produces surface analysis and three-dimensional (3D) wind, temperature, cloud and moisture analyses, and also incorporates facilities to assess data quality.

Deep Thunder’s power can be gauged by the fact that it can produce highly-accurate weather predictions within the narrow ranges of a single city. The system was used during 1996 Atlanta Olympics to successfully schedule weather-affected events like sailing and the closing ceremony. It is also proposed to be used for 2016 Summer Olympics in Rio de Janeiro, Brazil.

How India forecasts its weather
Param was the first mission taken up by C-DAC for the development of a high-performance parallel computer and was completed in July 1991. Later, Param Yuva II was launched in 2013, which was capable of performing at a peak of 524 teraflop per second and was used for research in weather forecasting and seismic data analysis.

SHARE YOUR THOUGHTS & COMMENTS

Please enter your comment!
Please enter your name here