Processed by: SB Summary file for RA123A deployed at 15N90E From 2014-12-06 11:30:00 [340] to 2016-03-05 11:08:00 [065] Tube 709, software version 4.07: Wind 101166 100% - note that the propeller shaft was found slighty bent, but still spinning freely AirT 59297 100% RH 59297 100% SWR 37087 ~100% (6 flags) Rain 1636 95.2% -numerous large spikes and noise had to be flagged and interpolated LWR 37711 ~100% (2 flags, and also 2 spikes were flagged) BP 104102 100% Notes: 2016-11-01 CWF tube released. Module summary: Temp 92.9% Cond 89.7% Pres 100% Modules: 1m SSC #15355 100% t, 17.9% c 5m TC #15779 100% tc 10m TC #14852 100% tc 13m TV #15613 0% - bad file, software vers 5.10 20m TC #13999 100% tc 40m TC #15818 100% tc 60m TC #15807 100% tc 80m T #15137 100% 100m TC #15563 100% tc 120m T #15325 100% 140m TC #14939 100% tc 180m T #15492 100% 300m TP #13338 100% tp 500m TP #13985 100% tp Notes: 2016-11-01 CWF request redump of TV15613. All other modules released. ra123.doc: 15N 90E RA-123 14/340 Deployed, IO314. OTN 112525 at 185m. Chipods 813 (15m, 814 (30m), P1741 815 (45m). CO2 sys 108, SBE16 and sami ph sensor on bridle legs. Met obs 340/1400z wdir 046 spd 8 kn, at 27.2 sst 28.5, rh 77. 15/056 SSC (cond) reads high (logged 15058) ??/??? V10 intermittent 16/005 rain failed, 0 accum, high % time raining (logged 16008) 16/065 Recovered, IO5-15. We found lots of fishing gear fouled on the bridle, but none on any of the subsurface sensors. About 5m of nilspin plastic jacket was stripped from the wire at about 600m, leaving exposed and rusty bare steel wire. We recovered all of the ATLAS surface MET sensors and subsurface sensors undamaged, except for the RMYoung wind sensor, which had its propeller shaft slighty bent, but still spinning freely. Notes: There weren't any obvious anomalies in the winds compared with climatology and previous/subsequent data (see plot). There are out of range pressures due to strong currents, but apparently no vandalism events. The 1m-5m temperature differences show an inversion of about -0.06 at the start to about -0.08 at the end of the mooring. As a comparison, a plot of 1m-5m ra140 temperatures shows an ordinary set of differences around 0. The 1m temperatures are likely too low, given that the 1m salinities are too high by the same amount (~0.06). Unfortunately we don't yet have post-cals, so there's no way of knowing how much the temperatures drifted. In the past, a few sets of temperatures (such as ra092a, pm452b, ke001a) have had their temperatures pre-post-cal interpolated to correct for drifts... In checking the 5m-10m temperature diffs, these fall above 0 by ~0.04 in an arch (nonlinear), and the 1m-10m support (after subtracting out the -0.06) the idea that the 10m temps are too low in the middle of the mooring. The 5m temps may also be a bit high. I've been plotting ra092b ke001a, and pm452b temps, and none of them show differences as far from 0 as the ones in ra123. Also, they only had a single problematic depth, whereas there may be 2 or 3 here. Unfortunately we don't have a precedent for adjusting temperatures the way we do with salinities. Should I wait (months!) for the post cal drifts, or try adjusting them based on the inter-depth differences? A plot of the 1m-20m temperatures matching up with the start of the ra140 mooring confirms the 1m temps are too low (inverted below the 5m-20m temps), but the 5-10m temps are as well...? Paul alerted me to the possibility that the temperature inversions are real, based upon this location, where large cold freshwater rivers dump into the warmer ocean and a temperature inversion could exist indefinitely (as long as there isn't also a density inversion). So this could be real! Also, he reminded me that it's highly unusual for a temperature module to drift nonlinearly; a much more like scenario for a bad temperature would be the use of a bad pre cal, or even a linear drift... Back to the original (pre-adjusted) densities: although there were numerous inversions, I'd adjusted them all on the basis of salinity inversions, assuming the salinities rather than the temps to be the cause of the density inversions (this is almost always the case, but still I shouldn't have assumed it, - I should have checked the temperatures to make sure). Anyhow, in looking again at the sal inversions, they match the density inversions quite well, except that for 5m-1m, the sal diffs were ~0.015 smaller than the dens diffs (the sal diffs sloped from -0.016 to -0.021 over the course of the mooring, while the dens diffs sloped from -0.032 to -0.036). I'd ended up adjusting the 1m sals downward by a much larger amount (-0.063) based on the matchup with ra106... Plots of the dens, sal and temp matchups with the previous mooring (ra106) show the ra123 1m sal and dens to be too high by ~the same amount (~0.1), and the ra123 1m temp to be too low (~0.1). I noted in my adjustment file that the 1m adjustments downward were based partly on their matchup with the end of ra106. Back to the temperatures... Since the salinity and density diffs and matchups correlate so well, I think that means the salinity inversions were responsible for the density inversions (with perhaps a small amount of temperature-inversion influence in the 5m-1m dens and sal inversions, indicated by those being ~0.015 apart). So the temperature inversions didn't create the density inversions, indicating they were indeed real temperature inversions. I should revisit this once post cals are available, in case the ~0.015 temperature component of the 5-1 density inversion can be explained by a 0.015 drift in the 1m data, or else a bad pre cal. For now, I'll leave it 'as is' because it's below the nominal module accuracy of 0.02 degrees C. Note that the freshening curve of the 5m-10m (inexplicably not seen in 10m-20m) spans April-November, which encompasses the monsoon season (May-Sept, with the monsoons starting over the BoB in late May and advancing across India). A profile plot of the average temperatures versus depth (with each Tavg averaged across the full deployment at a single depth) shows that there were inversions not only between 1m-5m (with 1 colder), but also between 10m-20m (with 10 colder). But these are all fairly small wavers along the downward part of the curve that starts to rapidly turn left (colder) below 40m. I plotted similar profiles for the subsequent (ra140) and previous data sets at this site, dating back to 11/2009 (pm863a, ra057ab, ra072a, ra093a, and ra106a) and among the 7 profiles, ra123 was unfortunately the only one that exhibited inversions at this location. This is unsettling, but I really need more info before assuming bad temps and adjusting them. So based on the above, I'll release the temperatures as they are and then revisit the data after temp and sal post cals have been performed... Here's some background info on freshwater in the Bay of Bengal: Major rivers such as the Ganges, Bramhaputra, Irrawaddy, Krishna and Godavari flow into the Bay of Bengal and create a fresh surface there, especially during monsoons, when the fresh layer deepens. In this online article (http://www.bbc.com/news/science-environment-33479507), researchers with the Ocean Mixing and Monsoon project have found in the BoB a "sharp separation between river water and seawater on scales ranging from 100m to 20km". Here's another article underscoring the importance of rainfall and advected river water on the SST and SSC in the BoB. http://drs.nio.org/drs/bitstream/handle/2264/3939/J_Geophys_Res_116_JC006804a.pdf?sequence=1 Here's another article discussing the near-surface (1-20m) temperature inversions near the mouths of the rivers flowing into the BoB. http://www.sciencedirect.com/science/article/pii/S0967063702000444 With huge (1.6-2.4 degrees C) inversions near fresh water inputs, It makes sense that farther away from these inputs there could be smaller temperature inversions, perhaps on the order of about -0.2 to -0.06 that occurred in ra123 near-surface depths. Here's another article (http://www.oalib.com/paper/1545445#.WMIWkfJCjaI) that talked about inversions at depths of 20-60m due to runoff and rainfall. Here's an overview by http://journals.ametsoc.org/doi/10.1175/JCLI-D-14-00553.1 : "The northern Bay of Bengal is characterized by distinct surface stratification caused by freshwater supply from the Ganges and Brahmaputra Rivers. Surface water is freshened from summer to fall by river runoff associated with heavy precipitation during the summertime Indian monsoon, resulting in sharp haline stratification and a thin mixed layer along the coast (Howden and Murtugudde 2001; Rao and Sivakumar 2003; Akhil et al. 2014). Surface circulation further advects low-salinity water southward along the western boundary of the bay in fall and winter (Han and McCreary 2001; Han et al. 2001; Akhil et al. 2014). In early spring, currents carry surface freshwater southward along the eastern boundary and also to the interior of the bay (Han and McCreary 2001; Vinayachandran et al. 2002; Thadathil et al. 2007)." Salinity Notes: Mike asked Ken to have me process and release the ra123 data before the post cals arrive (since that could be quite awhile yet). Here's part of his email from 11/4/2016: "Mike M. has asked to proceed with posting ra123 salinity data to the web without post-cal comparisons using the justification that if these were realtime data they would be up there anyway and replaced later once the RAM processing (and post-cal comparison) is complete." Curran and Dan weighed in, and the plan is for me to process and create the hry and davg files, then give those to Dan (without running RAMdbload) for Dai to post in place of the current realtime files (so we won't be releasing them as high res files; that way nobody will get confused about why we're releasing non-post-caled high res data). p.s. The salinities are being released now at a later date, despite having no post cals yet, because the data are needed and it was decided that the QC2 and 3 codes sufficiently encompassed the uncertainty created by a lack of post cals, given that buddy checks were thorough and that other types of data (ctds, subsequent moorings, adjacent depths for buddy checking, or other info needed for processing) are often lacking in other moorings without any special consideration (other than QC2 and QC3 codes). -SB 3/10/2017 processA2Mod ra123a /home/noor/salinity/moorings/ra123/ --cal=/home/noor/salinity/moorings/ra123/ra123a.cal flag_data ra123a . /home/noor/salinity/moorings/ra123/flag123a.txt /home/noor/salinity/salinity_code_localized/nxgonew --rclc ra123 . /home/noor/salinity/salinity_code_localized/ The spectacular 1m failure (seen in the dens123a.flg plot of all depths) was flagged from 2015056223000 to the end. For mooring ra123, hanning_qc is being run out of the /home/noor/salinity/salinity_code_localized/ directory. For this mooring, nominal depths will be used to calculate salinities and densities following the pre-postcal interpolation. Salinity drift (post cal data - pre cal data): ra123a From nxgonew --rclc at 2016-11-04 16:03:55 For the listed drift negative values indicate scouring; positive values indicate fouling; 1m 0.0000 5m 0.0000 10m 0.0000 20m 0.0000 40m 0.0000 60m 0.0000 100m 0.0000 140m 0.0000 Running /home/noor/salinity/salinity_code_localized/hanning_qc 123a dft /home/noor/salinity/moorings/ra123/ 13 1 1 ra /home/noor/nxram/ra123/ 0 Compare the ra123a.flg (or dft) file with the ra140a realtime data Curran emailed me from ketch (now located in the figures subfolder of my ra123 salinity processing directory), as well as the online plots from the TAO internal home page -> data monitoring files -> plot deployed realtime, for ptt#17667. Here are the first data points from the w17667.cnd file Curran sent: The comparisons (even comparing the ra140 1m to the ra123 5m since the 1m ended very early, on 2015056222000) are reasonable, with no drifts large enough to justify adjustments or cause any alarm. #YYYYDDD HH MM SS DDD HHMM NZ # Z Cond Saln Dens #-------------------------------------------------------------------------------- 2016066 04:38:17 066 0000 7 1.0 51.888 31.887 20.012 10.0 51.934 31.969 20.099 20.0 53.306 32.613 20.442 40.0 53.593 33.183 21.048 60.0 53.428 33.149 21.063 100.0 51.311 34.302 23.072 140.0 46.112 34.805 25.034 2016067 00:44:11 067 0000 7 1.0 44.853 27.122 16.456 10.0 51.796 31.895 20.055 20.0 53.203 32.580 20.434 40.0 53.715 33.209 21.040 60.0 53.684 33.327 21.198 100.0 51.997 34.472 23.068 140.0 46.125 34.775 24.999 2016068 01:21:52 068 0000 7 1.0 52.110 31.872 19.920 10.0 51.777 31.892 20.058 20.0 52.574 32.293 20.289 40.0 53.493 33.091 20.968 60.0 53.922 33.423 21.237 100.0 51.898 34.545 23.179 140.0 46.359 34.781 24.948 2016069 00:55:51 069 0000 7 1.0 52.230 31.953 19.980 10.0 52.042 32.001 20.104 20.0 52.568 32.331 20.338 40.0 53.489 33.072 20.946 60.0 53.800 33.425 21.279 100.0 51.882 34.621 23.271 140.0 46.522 34.810 24.938 2016070 00:29:12 070 0000 7 1.0 52.651 32.028 19.933 10.0 52.233 32.109 20.173 20.0 52.678 32.418 20.408 40.0 53.494 33.052 20.921 60.0 53.828 33.421 21.264 100.0 51.806 34.626 23.299 140.0 46.591 34.811 24.921 2016071 00:18:38 071 0000 7 1.0 52.926 32.246 20.110 10.0 52.683 32.386 20.366 20.0 53.001 32.589 20.512 40.0 53.541 33.078 20.937 60.0 53.908 33.463 21.290 100.0 51.917 34.664 23.310 140.0 46.339 34.822 24.996 Buddy checks: 1m: In buddy checks, the remaining small segment of 1m data from the start of the deployment is about ~0.06 too high (20-1, also 5-1) The plot of dens106a.adj with dens123a.flg also shows the 1m ra123 data looks a bit high from the very start, and could certainly be lowered by about 0.06 to match up better with the end of dens106a (in this case the ra123 data rather than the ra106 data needs to be adjusted because the 1m ra123 data is inverted in the buddy checks). 5m: the 5m may be a bit too high (~0.02) near the start (5-1). The 20-5 drifts down to ~-0.08, while the 40-5 drifts down to ~-0.2 and the 40-20 drifts down to ~-0.15, indicating that the 5m drifts high by ~0.08 (or 20m drifts low by -0.08, or some combination of both) while the 40m drifts low by ~0.12. Comparing with 20-10 and 10-5 indicates that the 20m drifts down by ~0.02 while the 5m drifts up by ~0.06. 10m: The 10m is too high (10-1,10-5, 20-10). The 10-5 and 20-10 diffs show a discontinuity in late July in the 10m data. The 20-10 goes down to -0.13, while 40-10 goes down to -0.23. Comparing with the 20-5 and 40-5, it appears that 10m drifts high by ~0.1 (20m down by .03), and the 40m drifts down by ~0.13 20m: The 20-5 plot shows the 5m data a bit high (~0.02) at the start, followed by upward drifting (or downward 20m drifting) to reach -0.08 by the end. Comparing with 20-10 indicates that the 20m drifts down by ~0.02 while the 5m drifts up by ~0.06. 40m: the 40-5, 40-10 and 40-20 plots show the 40m data drifting down (starting in June or July) to -0.2 and -0.15 respectively, by the end. Comparing with 20-5 indicates 40m drifts low by ~0.13. 60m: the 60-40m showed ~0.05 inversions in Jan-Feb 2016 during 2 small periods when the layers were mixed (so the 60-40 diffs should have been positive). So the 60m data were linearly adjusted upward from June to the end. /home/noor/salinity/salinity_code_localized/hanning_qc 123a adj /home/noor/salinity/moorings/ra123/ 13 1 1 ra /home/noor/salinity/salinity_code_localized/reformat_sd 123a Release the data for now (email Dan and Dai, but do not run RAMdbload); will have to reprocess later when post cals return... -SB 11/16/2016