2018/02/21 ---------- make new processing dir on laptop Jordan: c:\projects\healy10_ta\aw_process_0910 previous processing took place in projects\nopp10\aw_pp copy data dir projects\nopp10\aw_pp\bs3_aw to new processing dir: c:\projects\healy10_ta\aw_process_0910\bs3_aw start with call_awproc.m: - update aw_params.m - set to run f_all_awfiles.m only, first time through 2018/02/22 ---------- running - prf 152: spike at 26m oddly enough, the run only seemed to have 152 profiles. save the run at bs3rbr200910_aw_run1.mat 2018/02/23 ---------- finished call_awproc.m run with do_proc_raw option. Lags were mostly -2 (193 times), but a lot of -1 (117 times) and some 0 (10 times). ======================================== continue during HLY1902: 2019/09/25 ---------- look at calibration. In contrast to the last year, we DO have a microcat at the bottom of the Arctic Winch. Its done all by one script, "knitted" together: inv_awcal.m (for investigate AW calibration): look at deepest good AW bins, selected "manually" as lastgoodbins = [77 78 79]; %its nominal depths are: %77.0000 38.2500 %78.0000 38.7500 %79.0000 39.2500 Calculate mean conductivity over those three bins ("mean deep cond") Calculate vertical conductivity gradient over those three bins: simply difference bin(79)-bin(77). They happen to be 1m apart. The depth difference between the microcat and the mean of bins 77:79 is 1.5m, so add 1.5 times the cond gradient to the mean deep cond. Fit mc cond to AW times; estimate, roughly by eye, a mean difference, use that as base to exclude outliers (this replaces Paula's iterative scheme, which I found could be mislead by times with large cond changes, both in time as well as vertically) calculate linear fit to the non-outlier cond diff: has no slope, and 0.0827 mS/cm offset: mc-(AW+grad) positive, => AW cond too low Leave this for now; will apply that small correction in "make_final_aw.m" ------------------------------------ Continue with average profile check and possibly editing: p_awstag.m Aha: but I need input file _aw_grd.mat: that's made by cont_proc_aw.m Edit that for 2009-10 (simple input / output file name change); also drop the "first look at cal" done for 2008-09 using CMP So first: run call_awproc.m with flags set to only run cont_proc_aw.m 2019/09/26 To save the original version of the input file, I copy it first (once!) !copy bs3rbr200910_aw_grd.mat bs3rbr200910_aw_grd_noedits.mat ran through p_awstag.m a few times, catching quite a few blibs early in the record when TS was wild and there was lots of large signals. After the first 20% of the records ,there were just a couple isolated blibs. ------ run make_final_aw.m ================================================================= ========== previous AW 2008-09 processing notes ============= ================================================================= 2018/02/07 ------------ look again at Arctic Winch data, to finish off entries for the AON web page. For 2008-09, there was confusion on the mooring names, and BS3 was called "BS4" for this deployment. There also was a BS3, in slightly deeper water, which we are not considering here. --------------------- a lot of processing was done previously, including an initial plot entry to the web page as well as a pressure time series for Rick (he edits the ULS data for contaminations from the rising Arctic winch float). After looking through my previous processing, I decided to re-do things once more. Start in c:\projects\nopp09\arctic_winch\bs4_recover\arctic_winch_1\mfiles which is where I constructed the file for Rick, and take files from there to build a new processing dir as c:\projects\healy10_ta\aw_process\ start with ascpprof_1.m make that ascpprof_2.m; also change t_salts4.m to t_salts5.m 2018/02/08 ---------- got something going here. Looking through files: lag for dir1 file 10 (F07-2008 F2312356.08) : lag=-2 same for dir 1 file 12 note: 2008 is a leap year: Feb 29 exists. ------- re-hash processing: - start with calling program for overall processing: call_awproc.m it starts with a parameter file that sets all kinds of things like input directories etc: aw_params.m sorting the input files by time: f_all_awfiles.m for "find all Arctic Winch files" prf 34: top hook 2018/02/09 prf 174: use lag 0 to avoid hook near buoy but only for that one! At 173, go back to lag=-2 Unfortunately, software kept lag=0 and I did not notice until 190. Fixed. Other than that one profile, I kept lag at -2 for the whole dataset. finished full run of call_awproc.m it works by: - set params with aw_params.m - organizes raw files by time into structure "aw" with f_all_awfiles.m - calls the workhorse ascpprof_3.m, which in turn calls: -- auto_raw_edits.m: automatic edits by different criteria -- raw_plots.m simple plots -- set_figs_0.m just set plot sizes -- t_salts5.m biggy; selects indeces for different lags, finds down cast --- depth_bin_aw.m vertically bins down cast more plots -- asks whether to accept edits and lags, then adds current profile to structure "aw" and saves it ----------------------- next: look at cal Ah: I find that I need conductivity in the aw structure, which I forgot. Another run of call_awproc.m But this time, I don't want to do all that interactive stuff. So I come up with a few modifications, as shown in the top of call_awproc.m; a few differences in script calls (adding "auto" to names) and introducing the flag "do_auto" 2018/02/15 continue with C cal save output of cont_proc_aw.m : the gridded AW data: file bs4rbr200809_aw_grd look for vertical conductivity gradient of MMP: f_mmp_cgrad.m uses bs3_cmp.mat (holds Paula's processed MMP data ) calculates vertical conductivity gradient for the top-most 4 bins. extrapolates to depth of lowest (2) AW bins looks at difference in conductivity between those bins and extrapolated MMP conductivity manually determine time periods of apparently different strucure in AW and MMP record; specify those as extensive "find" command calculate a fit of [MMP_cond - AW_cond] as function of time: dC = 0.0004 * 0.0646*dday (in mS/cm). will apply that to AW C one more editing: staggered plots of AW: p_aw_stag.m (based on p_uphyst_ts3.m from ITP processing) Makes a FLAG mask with NaNs for bad data. Did that, saved edited output to bs4rbr200809_aw_grd_edt.m apply cal: Bit of a conundrum: The simplest thing would be to adjust the matrix C. then re-calculate S and be done. But C contains basic conductivity, not LAGGED conductivity. A better approach would be to start with the structure aw, the apply: - raw flags - sensor lags - cal - grid - gridded flags - convert A/D voltages Make script make_final_aw.m to do that.