GUISDAP 1.65 is available to process data from both the new, upgraded, systems and the previous ND-based systems. However, the mechanisms are not yet too stable and are likely to change as the upgraded systems come fully into operation. This document represents the state of the art on .
The current versions of GUISDAP 1.65 are installed and maintained on the main process computer, puny. This is a multi-processor machine and can run multiple copies of the analysis without loss of performance.
The best available analysis for the current systems can be found in /opt/guisdap/TonyTromso, and for the pre-2000 systems can be found in /opt/guisdap/TonyOldTromso. In each case, cd to the specified directory, start MATLAB and execute the scipt 'analysis.m'.
Data from the new systems is integrated with an earlier version of Nigel Wade's integration program (as it is on Svalbard) and this requires that /opt/guisdap/uk/lib be available in the load path (use LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/guisdap/uk/lib in the k-shell for example) before starting MATLAB.
analysis.m has now been replaced by newNamesAnalysis.m which implements the new experiment naming convention and can process the following codes:
| Programe | Comments |
| cp1lt | Processes both ac and lp data, but does not combine them |
| cp4bv | Allows selection of which beam to process |
| tau1_singel_v | Normally select whole antenna |
newNamesAnalysis.m will ask a series of questions to determine the program to be processed, the beam selected (where appropriate), the start and end times required, the integration period and whether the analysis is to check for new real-time data (this carries some overhead and should not be used when processing completed datasets - and anyway doesn't work correctly at the moment). Note that some responses are required in quotes or square brackets.
Two graphics windows will be created, one displaying the real part of the data dump and the second the fitted results. If not required (for example to speed up the processing once it has been determined that the analysis is proceeding correctly), these can be closed using the window controls (top left button) and each will disappear when the window is next addressed.
analysis.m can process the following codes:
| Programe | Comments |
| CP1K | |
| CP2E | |
| CP3F | |
| CP4B | Processes both beams sequentially |
analysis.m will ask a series of questions to determine the program to be processed, (both the program and the pulse scheme) the start and end times required, the integration period and whether the analysis is to check for new real-time data (this carries some overhead and should not be used when processing completed datasets - and anyway doesn't work correctly at the moment). Note that some responses are required in quotes or square brackets.
Two graphics windows will be created, one displaying the real part of the data dump and the second the fitted results. If not required (for example to speed up the processing once it has been determined that the analysis is proceeding correctly), these can be closed using the window controls (top left button) and each will disappear when the window is next addressed.
The analysed results will be written to the analysed data area /data0/analysis. MATLAB files will be written to a directory whose name is constructed as <YYYY>-<MM>-<DD>_<program>_<code>_<integration time>_<antenna>. For example, 2001-04-03_cp1lt_aclp_240_uhf contains Tromsø UHF data form 3 April 2001 taken with the cp1lt program, both alternating code and long pulse data with four minutes integration, 2001-03-21_tau1_singel_v_ac_60_vhf0 contains Tromsø VHF data from the whole antenna taken with the tau1_singel_v program (altenating code only) with 60 seconds integration and 1993-04-20_CP1K_aclp_60_tro contains Tromsø UHF data form 20 April 1993 taken with the cp1k program, both alternating code and long pulse data with 60 seconds integration.
Additional files are constructed in /data0/analysis containing the same data in NCAR format and special velocity data (also in NCAR format) as used by the SuperDARN radars (though there is some problem with these data at present). The latest NCAR records of both types are also posted to the web server and can be checked via the links on the monitors page (disabled at present).
The progress of the analysis can be checked using the vizu program. cd to /opt/guisdap/TonyTromso/vizu, start MATLAB and run 'vizu.m'. The available analysed data directories in /data0/analysis will be listed, select the appropriate directory and add a comment if appropriate. The plot can be updated using the 'update' button; the update/edit button gives access to some further options; always select whole hour boundaries for the start and end points of the plot. Typing vizu('save') at the prompt will cause a PostScript file to be created in /data0/analysis with the same name as the data directory and of type '.ps'.
Quality-checked analysed data sets can be loaded into the MADRIGAL database where appropriate. Please contact Tony if you would like to make your data available in this way.
Updated by Tony,