================================================================================ SVN $Id: ChangeLog 35698 2012-03-22 23:59:57Z kauff $ SVN $URL: https://svn-ccsm-models.cgd.ucar.edu/tools/mapping/trunk_tags/mapping_121113b/gen_mapping_files/runoff_to_ocn/ChangeLog $ ================================================================================ Originator: kauff Date: Thu Mar 22 2012 Model: mapping/gen_runoff_ocn_maps Version: One-line summary: exact copy of gen_runoffmap trunk ================================================================================ ================================================================================ Originator: kauff Date: Thu Mar 22 2012 Model: gen_runoffmap Version: gen_runoffmap_120322 One-line summary: move ref1br branch code onto main trunk ================================================================================ ================================================================================ Originator: kauff Date: Thu Mar 22 2012 Model: gen_runoffmap Version: ref1br04_gen_runoffmap_080619 One-line summary: modify size estimates for smoothing matrix & mat-mat multiply map_mod.F90 - bump up size estimate for result of matrix-matrix multiple - size is 20% greater, as required by some paleo application, circa 2011 (ask nanr) smooth.F90 - set size estimate for smoothing matrix to max size that it can be given that it is indexed with a 4-byte integer (ie. close to 2 ^ 31 -1) - add diagnostics wrt the min, max, avg, bin'd number of cells that each src cell is smooth into ================================================================================ Originator: kauff Date: Wed Mar 18 14:29:34 MDT 2009 Model: gen_runoffmap Version: ref1br03_gen_runoffmap_080619 One-line summary: fix bug in nn algorithm Changes made: - add support for this runoff data on this domain... /fis/cgd/cseg/csm/inputdata/lnd/dlnd7/RX1/runoff.daitren.annual.090225.nc which is observational data an a regular 1x1 deg lat/lon grid - replace optimize nn map generation algorithm with a brute-force approach because the optimized version doesn't work correctly. May need to debug and reintroduce the optimized version for use with tx0.1 grids - add support for bluefire, remove support for blueice ================================================================================ Originator: tcraig, jdennis Date: Fri Jun 27 14:58:02 MDT 2008 Model: gen_runoffmap Version: ref1br01_gen_runoffmap_080619 One-line summary: update code Changes made: - update smoothing algorithm from jdennis - generate nearest neighbor initial mapping on the fly - read in raw rdirc file instead of rdirc scrip grid file - refactor mat_mat_mult to take advantage of nearest neighbor implementation -- input map MUST BE nearest neighbor - clean up some memory use issues ================================================================================ Originator: tcraig Date: Model: gen_runoffmap Version: ref1br00_gen_runoffmap_080619 One-line summary: create branch ? Changes made: ================================================================================ ================================================================================ Originator: kauff Date: Thu Jun 19 17:04:19 EDT 2008 Model: gen_runoffmap Version: gen_runoffmap_080619 One-line summary: remove dead code, add recur_setdist Changes made: - remove unused POP_* routines from smooth_mod.F90 - add recur_setdist routine ================================================================================ Originator: kauff Date: Tue Jun 10 16:01:27 MDT 2008 Model: gen_runoffmap Version: gen_runoffmap_080610 One-line summary: performance enhancements motivated by r05_tx0.1v2 mapping Changes made: - memory & efficiency mods required by hi-rez (r05_tx0.1v2) mapping (thanks to John Dennis) - is hard-coded to apply smoothing only to points that are implicated in the destination domain of the original (unsmoothed) map (which is reasonable and also a lot less work). "Implicated" means mapped to -- there may be many unmasked destination grid cells that are not "implicated." Implicated points are presumably only very near to coasts. - used to create: map_r05_to_tx0.1v2_e1000r50_080605.nc, which apparently works OK (?) - NOTE: has been validated as b4b with previous tag (by recreating map_r05_to_tx1v1_e1000r500_080505.nc) WITH THE SRC MOD of *not* restricting smoothing to implicated ocn cells (smooths all unmasked ocn cells) Timing info: o input : file_orig = 'map_r05_to_tx1v1.nc ' o output: file_new = 'map_r05_to_tx1v1_e1000r500_yymmdd.nc o very hi-rez tx0.1v2 ocn grid o restricted sources (only smooth implicted ocn cells) o only a 50km smoothing radius o basically: step2 = 4 hours, step3 = 4.75 hours o on one blueice node using 16 threads (sorting & mat-mult are threaded) ================================================================================ Originator: kauff Date: Wed May 28 10:57:36 MDT 2008 Model: gen_runoffmap Version: gen_runoffmap_080528 One-line summary: tag code cleanup circa Jan 2007 Changes made: - code cleanup circa Jan 2007 - new eFold & rMax nml vars - time stamp progress reports when making smoothing map (May 2008) - validated against gen_runoffmap_070104 by recreating map_r05_to_tx1v1_e1000r500_080505.nc (there are roundoff differences) ================================================================================ Originator: kauff Date: Thu Jan 4 11:52:13 MST 2007 Model: gen_runoffmap Version: gen_runoffmap_070104 One-line summary: new correction & sorting algorithms: faster, more accurate Changes made: - this code worked on 2007 Jan 2 for S. Yeager for making map_r05_to_gx1v5_e1000r500_070102.nc - all steps/phases: < 6 hrs on bluesky, threaded on 24 processors - new correction algorithm more accurately smooths runoff data - new sort algorithm is much faster - also works on bluevista, but only 8 processors are available on a node which is slower than 24 processors on bluesky ================================================================================ Originator: kauff Date: Tue Dec 19 15:50:32 MST 2006 Model: gen_runoffmap Version: gen_runoffmap_061219 One-line summary: initial import of gen_runoffmap code Changes made: - this code apparently worked in Sep 2006 for B. Briegleb for making the kx3v1 paleo grid runoff map. ================================================================================