13th Symposium on Global Change and Climate Variations
18th International Conference on IIPS

J2.6

A suite of web pages for analyzing climate signals in large datasets

Catherine A. Smith, NOAA/CIRES, Univ. of Colorado, Boulder, CO; and J. A. Collins and R. H. Schweitzer

The ability to readily search for and quantify climate relationships is obviously of enormous importance in climate studies of all types. In practice it can take a long time to perform the analysis as various datasets must be downloaded and read, code must be written to analyze the relationships and the results plotted. Time and sufficient computer resources are necessary and even then, up-to-date data and new datasets must be constantly added. At CDC, we have enormously simplified the work needed to be done by exploratory researchers by providing a suite of webpages that allow users to analyze many of our large monthly mean datasets including model analysis, climate model output and other datasets like the Reynolds SST. Specifically, the user can interactively examine relationships between monthly(or seasonal) time series and the variables from various gridded climate datasets. The user can seek linear relationships by examining correlations between an index time series and different variables or they can look at possible nonlinear relationships by compositing variables using specified averaging criteria. The timeseries used for correlations or composites can be pregenerated, user supplied or generated from our webpages.

The pages can be found at http://www.cdc.noaa.gov/PublicData/web_tools.html

Our suite of pages consists of three basic parts. The first page creates monthly mean timeseries from different datasets including the US climate division dataset and the NCEP/NCAR Reanalysis. The user selects the dataset, the averaging region and the variable and a monthly mean timeseries is returned. This timeseries can be saved by the user or stored in a location that can be easily used by CDC's climate analysis webpages. The 2nd part, the correlation webpage, can use this timeseries, a pregenerated one stored at CDC (e.g. PNA), or a user supplied one and correlate it with variables from different datasets. The timeseries can be lead/lagged relative to the correlating dataset among other options and the results are both plotted and available as text. As an example, a user can seek SST precursors for springtime precipitation in the southwestern US.

The third analysis page calculates composites from different datasets. Composites can be averaged over time (by entering separate years, a year range or a set of years subtracted from another) or they can be made for years that fit a specified criteria from a specified timeseries. For example, the user can specify that composites of omega pressure over the US be made for years when precipitation exceeds a threshold of 2sigma. Other types of critera allowed are values (e.g. temperature below -10C) or percentages (e.g. geopotential height at or below the 33rd percentile).

This web interface has also been applied to output from the NCAR CCM3 model. We anticipate that it will be similarly used to ease diagnosis of AMIP-type climate model run output. We may use new technology like Java Servlets, Java Server Pages and Java Beans to assist the integration of the webpages. We will also be updating the archived datasets, adding more datasets if warranted and adding to our selection of pregenerated timeseries.

extended abstract  Extended Abstract (68K)

Joint Session 2, Climate Model Diagnostics: Tools (Joint with the 18th Conference IIPS and 13th Symposium on Global Change and Climate Variations)
Monday, 14 January 2002, 3:30 PM-4:59 PM

Previous paper  

Browse or search entire meeting

AMS Home Page