Desktop To Grids - PowerPoint PPT Presentation

1 / 17
About This Presentation
Title:

Desktop To Grids

Description:

Environmental Science corresponds to the scientific principles, ... Then there's the trendy aspirations of climate modeling. GIS Geographic Information System ... – PowerPoint PPT presentation

Number of Views:79
Avg rating:3.0/5.0
Slides: 18
Provided by: hess2
Category:
Tags: desktop | grids | trendy

less

Transcript and Presenter's Notes

Title: Desktop To Grids


1
Desktop To Grids
  • Environmental Science

2
Environmental Science
  • Environmental Science corresponds to the
    scientific principles, concepts, and
    methodologies required to understand the
    interrelationships of the natural world, to
    identify and analyze environmental problems both
    natural and human-made, to evaluate the relative
    risks associated with these problems, and to
    examine alternative solutions for resolving
    and/or preventing them.

3
Why Environmental Science Matters
  • The environment drives a lot of our news, our
    culture, our society, and especially our energy
    use, our politics.
  • Environmental problems are hard
  • Energy conversions
  • Variables are extremely interconnected
  • Environmental problems have social and cultural
    ramifications
  • Time scales are on the orders of lifetimes

4
Types of Environmental Science Computing
  • The most natural example is weather forecasting
  • Then there's the trendy aspirations of climate
    modeling
  • GIS Geographic Information System
  • a system for capturing, storing, analyzing and
    managing data and associated attributes which are
    spatially referenced to the Earth.

5
Desktop Computing. Feasible?
  • Weather tools
  • The Beaufort Wind Scale
  • developed in 1805 by Sir Francis Beaufort of the
    Royal Navy. He wanted a consistent way to measure
    wind force on fighting ships. Today's version
    states the wind's effects on land as well as sea.
  • Doppler Radar
  • a type of radar that sees rain and snow and
    measures wind speed and direction. It is most
    useful for short-range forecasts on the day of
    the weather event.
  • Infrared Radiation
  • an invisible form of radiation given off by all
    objects both day and night. Satellites equipped
    with infrared sensors take images of clouds and
    of the Earth's surface.

6
Weather Models
  • NAM Models
  • The NAM model is a regional mesoscale model using
    enhanced terrain and improved parameterization of
    surface and precipitation processes. It produces
    forecasts out to 84 hours at 00, 06, 12 and
    18UTC.
  • Grid resolution is 40km, run over CONUS (which
    is the Continental US).

7
Weather Models
  • NGM Models
  • The Nested Grid model (NGM) - Its name comes from
    the technique of using a finer grid over North
    America and coarser grid over the oceans.
  • The NGM provides guidance out to 48 hours at 00
    and 12UTC.

8
Weather Models
  • GFS Models
  • The Global Forecast System (GFS) - The GFS is a
    global spectral model used primarily for aviation
    weather forecasts.
  • The GFS provides guidance out to 384 hours at 00,
    06, 12 and 18UTC.

9
Weather Models
  • RUC Models
  • The Rapid Update Cycle (RUC) -
  • The RUC provides guidance out to 12 hours at 23,
    00, 01 and 02UTC.

10
Weather Models
  • ADAS
  • ADAS is the AWOS (Automated Weather Observation
    System) Data Acquisition System
  • ADAS uses NAM data along with other data from
    radars, satellites, mesonets, wind profilers,
    aircrafts, and other surface data to a 10km grid
    (as opposed to a 40km grid in NAM).

11
Napkin-back calculations
  • What type of weather can you reasonably simulate?
  • Let's try ADAS over CONUS...
  • 10km spacing
  • Area of US is approximately 9,600,000
  • Say... 8x1M, or 8 x 220
  • Atmosphere 10 miles at the equator
  • Say 20 km, or two cells high
  • How many cells?
  • (8 x 220) x 2 224 cells

12
Napkin-back calculations
  • What type of weather can you reasonably simulate?
  • 224 cells. But how much data is being gathered
    at each cell?
  • Temp
  • Pressure
  • Wind speed
  • Wind vector
  • any others? (yes! quite a few!)
  • Say 8 variables per cell. That makes 227 total
    variables to keep track of
  • Each variable might be a double, which is 64
    bites or 8 bytes.
  • An even 230 bytes.

13
Napkin-back calculations
  • What type of weather can you reasonably simulate?
  • 230 bytes, or an even Gigabyte just to initialize
    the data.
  • That's certainly reasonable to do on a single
    computer these days.
  • Now, let's run the simulation
  • Forecasts out to 84 hours, dumping every 6
    hours.
  • 15 snapshots...15 GB.
  • Sounds rather do-able.

14
Simulations on a workstation
  • You can download the WRF model from NCAR (WRF
    Weather Research Forecast, NCAR National
    Center for Atmospheric Research).
  • Compiling the code is a pain.
  • Running the code is an exercise in patience and
    sacrifice.
  • Code runs, task complete.

15
Simulations on a Cluster
  • A lot of things happens in six hours.
  • Weather forecasters say that weather changes at
    minute resolutions.
  • Over 84 hours, that's 5041 data sets! 5 TB of
    data.
  • Still looking at regional resolutions, and
    non-decorated simulations

16
Simulations on a Grid
  • Larger resolution or more decorations require
    moving to a Grid
  • Even moving of the data around to visualize would
    require a high-performance filesystem
  • http//portal.leadproject.org
  • A TeraGrid Science Gateway.

17
Links
  • Lead Portal
  • https//portal.leadproject.org
  • Earth Systems Grid
  • http//www.earthsystemgrid.org/
  • Weather Forecasting Code
  • NCAR
  • http//www.ncar.ucar.edu/
  • Weather Research and Forecasting Code
  • http//www.mmm.ucar.edu/wrf/users/download/get_sou
    rce2.html
Write a Comment
User Comments (0)
About PowerShow.com