GrADS:
A Handy Tool for Data Access, Analysis, and Visualization
Jennifer M. Adams and Brian Doty
IGES/COLA
What is GrADS?
GrADS is an interactive tool that integrates data access, analysis, and visualization
it handles gridded and station data
it has a programmable interface for scripting
it runs on Unix, MS Windows, and Apple Mac
it has a busy users forum with 1800+ subscribers
it has been under active development since 1989
and it is open source software under GPL
What Types of Data Can GrADS Read?
GrADS handles the following data formats: Flat Binary NetCDF GRIB (versions 1 and 2) HDF (versions 4 and 5) BUFR (for station data)
Data Formats and Metadata
Data are arrays of numbers
Metadata provides information about the data so we know what the numbers mean Some data formats are self-describing: data and metadata are contained in the same file Accurate metadata is absolutely essential
Metadata and GrADS
Metadata is communicated to GrADS with a separate Data Descriptor File (a.k.a. Control File)
Gives the user more flexibility and control
Aggregates numerous data files into a single data set
Supplements or overrides the metadata in the data file
Descriptor file may not be necessary if the metadata in data file is good enough
GrADS Gridded Data Model
Internal grid structure supports 5 dimensions:
X Y Z T E ~ ~ ~ ~ ~ Longitude Latitude Level Time Ensemble
T and E axes must be linear X, Y, and Z axes may be non-linear Grid coordinate values must be 1-dimensional
GrADS Station Data Model
Used for in situ observational data
Individual reports may be located anywhere in space and time
Required metadata for each report:
Station Identifier (8-character string) Longitude Latitude Level (for level-dependent variables) Time
Linear time axis is used to organize reports
Visualization in GrADS
Display options include: Contour (lines or shaded) Grid fill (colors or numbers) Bar graph (with error bars) Scatter plot Winds: Vectors, Barbs, or Streamlines Output options include: Images Vector Graphics (e.g. Postscript) ASCII Binary or NetCDF file GIS-Compatible Formats
Data Analysis in GrADS
GrADS expressions are formulas that contain:
Operands (variables, functions, or constants) Operators (add, subtract, multiply, divide) Parentheses (to control the order of operation)
Expression handling is recursive, so an expression may be embedded within another. Some examples:
temp*9/5+32 aave((temp*9/5+32),global) hgt(lev=500)-hgt(lev=1000) ave(uwnd,lon=0,lon=360) sum(precip,t+0,t+24)
GOES VIS Image with Radar, 500mb Heights, and SLP
18z 4 June 2009
Another Example of GrADS Graphics Output
QuikSCAT Winds (HDF), Surface Obs (BUFR), and ETA Model SLP (GRIB)
00z 6 Dec 2003
Analysis of Surface METAR Observations with Radar
18z 29 May 2009
GrADS Output in Google Earth
Palmer Drought Severity Index July 2009
Ensemble Forecast Time Series
(Longitude, Latitude, and Level are fixed)
Forecast Time --->
Ensemble Forecast Grid
(Longitude, Latitude, and Level are fixed)
Ensemble Member
Forecast Time --->
Ten Ensemble Forecasts
(Longitude, Latitude, and Level are fixed)
Ensemble Member
Forecast Time --->
Ensemble Forecast Time Series
(Longitude, Latitude, and Level are fixed)
Forecast Time --->
Ensemble Mean, Ensemble Min/Max, +/- StdDev of Ensemble Mean
Forecast Time --->
What is OPeNDAP?
OPeNDAP (formerly known as DODS) is an Open-source Project for a Network Data Access Protocol OPeNDAP servers make local data files accessible from remote locations
OPeNDAP clients (such as GrADS) can access remote data files
Open a data set with http://server/filename instead of /disk/filename
OPeNDAP Server-Client Interaction
Data
Observational data Model-based analyses and forecasts
Server
Handles Metadata Requests
Returns Data Subsets
Evaluates Analysis Expressions
Analysis Result Cache
Internet
Client (Users Program)
Local or Remote Data Analysis Visualization Web-Based Interface
What is the GrADS Data Server?
The GrADS Data Server (GDS) is a type of OPeNDAP server it can serve any GrADS-readable data set it translates all data formats into NetCDF it offers subsetting and server-side analysis it has a user-friendly browser interface and it is running operationally at NCEP, NCDC, NASA, and many other locations
Why is Server-Side Analysis Important?
Reading data over the internet is slower than reading data from a local file
Some analysis tasks need a lot of data
Compute the monthly mean annual cycle from a 100-year climate model run
Analysis expressions are evaluated at the server, where the data reside on disk
Only the analysis result is delivered to the client Server-side analysis saves a lot of time when: (size of result) << (size of data operated on)
THORPEX Interactive Grand Global Ensemble
TIGGE is an international collaboration between operational centers and universities for the development of ensemble prediction
Ten participating centers provide their ensemble forecasts in near real time:
BoM (Australia) CMC (Canada) ECMWF (Europe) KMA (Korea) NCEP (USA) CMA (China) CPTEC (Brazil) JMA (Japan) MeteoFrance (France) UKMO (England)
TIGGE data portals are at ECMWF, NCAR, and CMA
TIGGE Data Behind GDS at NCAR
+ + + + + Data neatly handled by GrADS Boost to usage of TIGGE data Forecasts organized by date and by provider Time series of initialization analyses Server-side analysis capability
- 48-hour data embargo - Only a 3-week window of data is online - Intl agreement requires password protection
TIGGE Multi-Member Multi-Model Ensemble
500mb Geopotential Height valid August 30, 2008
7-day Lead 5-day Lead
3-day Lead
1-day Lead
Washington Post Weather Page
Extreme Makeover: Weather Graphics Edition
Step 1: Get the Daily Normals and Extremes
Check the web site of the local Weather Service Forecast Office for local climate data: http://www.weather.gov/climate/local_data.php?wfo=lwx
Grab ASCII files, which will require some manual editing:
DY NMX NMN NPCP NS RMX /YEAR LOMX/YEAR RMN /YEAR HIMN/YEAR MXPCP YEAR MXSN YEAR 01 43 29 0.10 0.1 69 /2005 17 /1918 -14*/1881 51 /1876 1.52 2003 4.5 1899 02 43 28 0.10 0.2 71 /1876 15 /1918 -1 /1899 52 /1876 1.44 1979 4.5 1925 03 43 28 0.10 0.2 68 /2004+ 10 /1879 -3 /1877 52 /2000 1.87 1914 2.4 1988 04 43 28 0.10 0.2 73 /1997 16 /1879 -3 /1877 60*/1950 2.28 1886 2.1 1980 05 43 28 0.10 0.2 71 /1997 18 /1896 -3 /1877 56 /1950 1.44 1949 4.8 1980
Convert ASCII to GrADS station data format (txt2stn.c)
Daily Normal and Extreme Temperatures Observed at Reagan National Airport
Step 2: Get Forecast Data for Next 10 Days
Browse through available TIGGE forecasts and select which initial date and model to use: http://vetscomm.ucar.edu:9090/dods/tigge/ http://vetscomm.ucar.edu:9090/dods/tigge
Run the GrADS script: ga-> get_fcst yyyymmddhh model The script creates a local data set (binary data and a descriptor file) containing daily min and max 2-meter temperatures for the first 10 days of the given forecast
Step 3: Get Observed Data for Past 10 Days
Recent hourly observations (past 15 days) are here: http://monsoondata.org:9090/dods/stn/metar/past360
Run the GrADS script (note date format is different):
ga-> get_past get_past ddMONyyyy
The script creates a local data set (binary data and a descriptor file) containing daily min and max surface temperatures observed over the past 10 days
Combine Climate, Forecast, and Observed Data in a New and Improved Graphic
Some URLs to Remember http://iges.org/grads
GrADS Home Page
http://iges.org/jma/AMS2010
This Presentation (plus all scripts and programs)
http://vetscomm.ucar.edu:9090
TIGGE GDS
