*********************************** *** FluxNet REVISION HISTORY *** *********************************** Version 4.0, 1 October 2001: Significant changes: - Changed profile input from pressure (mb) to sigma levels - Replaced the input cloud pressure variable with cloud temperature - Added 200+ cold and high elevation profiles from the Arctic and Antarctic - Based the network on Streamer v3.0 Minor changes: - Changed the range of cloud particle effective radius for both liquid (lower maximum) and ice (lower minimum) clouds - Constrained water clouds such that they cannot exist in profiles with maximum temperatures less than 253 K. Previously water clouds could have temperatures lower than 253 K but wouldn't if a profile had one or more temperatures above 253. Version 3.1, 1 June 2001: - Changed input profile levels from pressure to sigma levels, so input files that use profiles are not compatible with previous versions. - Retrained the network with Streamer v3.0 - Added a normal random number generator for better sampling of some variables in training. - Added another sample application. Version 3.0a, 7 Feb 2000: - Added an initialization of 'j' in snns/mkdat*.c programs. This could be a small problem on some systems, but will NOT affect FluxNet itself. It's only an issue in sampling for training data. (No new version number) Version 3.0, 15 July 1999: - Added TOA fluxes to output; added specification of surface type to input - Fixed bug in cloud water content - Expanded the range of cloud droplet size and ice cloud temperature - Increased the percentage of clear and completely overcast cases - Set cloud properties to fixed values in clear cases - Added an HTML version of the user's guide. Version 2.1.1, 5 Apr 1999: - Modified manual to reflect "bug" in specification of cloud water content. Important! (No change to code.) Version 2.1, 9 Sep 1997: - Retrained with greater number of dark and clear sky cases - streamlined 'run.f'. Version 2.01, 22 May 1997: - Fixed bug concerning incorrect cloud top temperature in Streamer run. Version 2.0, 20 Apr 1997: - One network outputs shorwave and longwave fluxes - Redesigned network for use with global TOVS profiles - Simplified steps for redesign and implementation Version 1.0, 28 Aug 1996: - Version 1.0 release. Expanded the albedo range for snow revised the documentation. Version 1.0B, 16 Jul 1996: - First Beta test release.