Parallel Computing In Climate Research And Model
Parallel computing is the process of breaking down complex problems into smaller, independent, and frequently related sections that can be executed concurrently by several processors communicating via shared memory, with the results combined as part of an overall algorithm. The main goal of parallel processing is to increase available computing capacity for faster application processing and problem-solving.
Benefits of parallel computing
- It uses parallel computing models to solve real-world problems
- Saves Time
- Saves Money
- Solves more complex and larger problems
- It leverages remote resources
Motive Behind Climate Models
Climate models are created for various unique reasons. Climate models as nearly complete as feasible may be used to simulate the weather system. The model may be run for a long term and its annual cycle and variability approximately about that annual cycle can be assessed. Once the model has established itself on this simulation model, simulations may be carried out at the weather model, which definitely couldn’t be carried out at the climate. For example, the climate model may be used to simulate the reaction of the weather to outside changes, which include the solar constant or to volcanic eruptions.
In India: PRAM 10000
Predicting weather and climate involves solving a system of coupled nonlinear partial differential equations under appropriate boundary conditions. Predictions can range from short (up to 3 days) to long (up to 1 year) Climate modeling goes far beyond this limit.
In India Centre for Development of Advanced Computing is a national initiative in supercomputing under the support and guidance from the Department of Electronics, GoI. In order to research and tackle such tasks, CDAC developed a 100 GFLOPS High-Performance Computing System, PARAM 10000.
Long-range numerical weather forecasting, which allows routine weather predictions to be made a week or more in advance, is one of Param’s primary applications. The new Param model, for example, takes advantage of the fast processing speed of the widely used UltraSparc-II chip, which clocks in at 300 Mhz at all 48 parallel nodes. The weather models T21L05 and T42L15 had been ported on pram those models have a grid length of 600 km and 300 km respectively in the horizontal directions. Using the data parallelism approach, the models had been operational on the PARAM machine and a fairly good speedup was obtained.
Around The World
PCCM2.1
PCCM2.1 is a parallel model of the NCAR Community Climate Model, CCM2, carried out for MIMD vastly parallel computer systems by implementing a message-passing programming algorithm. Developed for Intel Paragons with 1024 processors and IBM SP2s with 128 processors. The code may be effortlessly ported to different multiprocessors helping a message-passing programming algorithm or run on machines dispensed throughout a network.
Parallelization breaks down the problem area into geographical patches and assigns to every processor the computation pertaining exclusively to a subset of those patches. With this decomposition, the physics calculations contain the simplest grid factors and data local to a processor and are executed in parallel. Using parallel algorithms created for the semi-Lagrangian transport, the short Fourier transform, and the Legendre transform, each physics, and dynamics are computed in parallel with minimum data motion and modest extrude to the authentic NCAR CCM2 source code.
POP(Parallel Ocean Program) CM-200
The Arctic Ocean is the sea from which maximum new insights applicable to climate change are probable to be received when its flow and connections to different oceans are effectively modeled. Different predictions could result from climate models if the polar areas are simulated more realistically
The Parallel Ocean Program (POP) was created for the worldwide ocean (with the Arctic Ocean excluded) at the Los Alamos National Laboratory on the CM-200. The team for researchers transformed and optimized the code to run on the Cray T3D in a joint attempt with Los Alamos, the National Center for Atmospheric Research, and Cray after which tailored it for the Arctic Ocean with a horizontal resolution of 1/6 degree and 30 vertical ranges to more realistically constitute blending and water mass formation withinside the Arctic Ocean.
The results were as follows, Several key features of flows and current interactions within the Arctic Ocean were resolved by the latest 200-year integration. The currents of the Arctic Ocean are topologically controlled and are typically 100 km wide. The current model simulates eddies approximately 100 km wide. The thickness of the ice and the concentration of ice depends on the oceanic and atmospheric fields as well as on the behavior of eddies. There have been good correlations between simulations of ice thickness and concentration and actual events. Several simulations have been run on both active and passive tracers to investigate Pacific water entry and circulation, nuclear contamination dispersion, and overall Arctic Ocean freshwater circulation.
Using the model, we can include advanced submodels of sea ice and a mixed layer in the model to represent physics realistically. In order to allow full-physics integration in future climate studies, this model has been coupled to the global Semtner Chervin model 1/4 degree to include interactions between the Arctic and the world’s oceans. The ARSC is currently developing a fully global ice-ocean model, including the Arctic Ocean, with a horizontal resolution of 1/3-degree.
Conclusion
Several kinds of climate models used for research, such as General Climate Models, Earth System Models, and Parallel Ocean Models must be run on supercomputers. These computers are invariably parallel at both the board and enclosure levels, using techniques such as multiple cores accessing shared memory. They provide great insights into our current change in climate and gives prediction based on simulations of what the future could look like.
Author: Rushikesh Shelke
Guide: Prof. Dhiraj Jadhav
I hope you found this blog interesting, feel free to drop your queries in the comments below. Stay tuned for more!
THANK YOU….
References:
[1]https://en.wikipedia.org/wiki/Climate_model
[2]WEATHER RESEARCH AND FORECASTING MODEL By NCAR UCAR
[3]Improving the Effectiveness of U.S. Climate Modeling (2001)National Research Council; Commission on Geosciences, Environment, and Resources; Board on Atmospheric Sciences and Climate; Panel on Improving the Effectiveness of U.S. Climate Modeling
[4]Climate, the Ocean, and Parallel Computing Robert C. Malone, Richard D. Smith, and John K. Dukowicz
[5]https://www.ipcc.ch/site/assets/uploads/2018/02/WG1AR5_Chapter09_FINAL.pdf
