P.  Poli  et  al.:  SVP-BRST:  genesis,  design,  and  initial  results

203

www.ocean-sci.net/15/199/2019/

Ocean  Sci.,  15,199-214,  2019

Table  2.  Similar  to  Table  1  but  for  HRSST-2  SVP-BS  buoys  (each  buoy  was  also  fitted  with  a  CT  probe).

WMO  identifier

Deployment  basin

HRSST  sensor  model  and  S/N

Start  date

End  date

4100736

North  Atlantic

Digital  YSI46000  10014

14/02/2012

26/01/2013

6200513

North  Atlantic

Digital  YSI46000  10011

18/03/2012

17/01/2013

6200505

North  Atlantic

Digital  YSI  46000  10017

25/03/2012

10/04/2013

6200501

North  Atlantic

Digital  YSI  46000  10019

29/06/2012

10/12/2012

6100788

Mediterranean  Sea

Digital  YSI  46000  10020

04/09/2012

16/02/2013

3100739

North  Atlantic

Digital  YSI  46000  10016

30/11/2012

06/07/2013

3100740

North  Atlantic

Digital  YSI  46000  10044

01/12/2012

06/03/2013

6100530

Mediterranean  Sea

Digital  YSI  46000  10013

30/01/2013

19/05/2013

6100525

North  Atlantic

Digital  YSI  46000  10042

22/02/2013

16/08/2013

6100524

North  Atlantic

Digital  YSI  46000  10049

22/02/2013

05/05/2013

6200504

North  Atlantic

Digital  YSI  46000  10045

24/05/2013

27/11/2014

1300899

Tropical  Atlantic

Digital  YSI  46000  10043

26/05/2013

10/12/2013

6200509

North  Atlantic

Digital  YSI  46000  10062

27/05/2013

15/10/2013

2300587

Indian  Ocean

Digital  YSI  46000  10071

09/06/2013

07/09/2013

2300588

Indian  Ocean

Digital  YSI  46000  10053

09/06/2013

07/09/2013

4100737

North  Atlantic

Digital  YSI  46000  10059

06/12/2013

10/03/2015

4100800

North  Atlantic

Digital  YSI  46000  10058

06/12/2013

16/01/2015

6200500

North  Atlantic

Digital  YSI  46000  10054

12/06/2014

18/02/2016

6500511

North  Atlantic

Digital  YSI  46000  10056

17/06/2014

25/06/2014

3100719

Tropical  Atlantic*

Digital  YSI  46000  10020

11/04/2015

20/06/2015

during  daytime  (the  hull  sensor  being  located  closer  to  the
surface).  The  differences  are  smaller  at  night  and  when  the
Sun  is  more  than  30°  below  the  horizon.  The  large  departures ­
  observed  sometimes  during  daytime  suggest  that  one
or  other  of  the  two  SST  sensors  may  have  been  differentially
affected  by  direct  solar  radiation,  or  by  the  buoy  heating  up
the  sensor  through  heat  conduction.
Unlike  promising  new  developments  with  wave  drifters
(Centurioni  et  ah,  2016),  the  HRSST-2  drifters  did  not  provide ­
  any  information  about  sea  state.  In  past  SST  studies,
wind  speed  is  generally  used  to  describe  sea-state  mixing
(e.g.,  Donlon  et  al.,  2002;  Morak-Bozzo  et  al.,  2016).  In
this  study,  we  also  consider  significant  wave  height.  Information ­
  about  both  parameters  can  be  obtained  by  co-locating
with  the  ERA5  reanalysis  (Hersbach  and  Dee,  2016;  C3S,
2017).  The  ERA5  reanalysis  data  are  interpolated  in  space
from  their  original  resolution  (spectral  truncation  T639)  to
the  buoy  locations,  using  the  nearest-in-time  hourly  reanalysis ­
  map.  Figure  2b  and  c  show  (respectively)  that  the  largemagnitude
  SST  difference  mostly  arise  when  the  wind  speed
is  up  to  moderate  (under  8-10ms -1 )  and  when  the  wave
heights  are  up  to  moderate  (under  2-3  m).  The  agreement
between  the  sensors  increases  when  there  is  more  wave  activity, ­
  probably  because  of  greater  mixing.  When  such  is  the
case,  almost  all  SST  differences  are  found  in  the  range  from
—0.1  to  0.0  K.  Sea-state  mixing  caused  by  waves  cannot  be
controlled  or  mitigated  by  a  platform  as  small  as  a  40  cm  diameter ­
  drifter.  However,  the  role  of  the  waves,  probably  via
mixing,  is  suggested  here  to  be  quite  important  when  using
the  SST  data  collected  by  drifting  buoys.  A  knowledge  of

the  local  SST  dynamics,  as  the  buoy  is  following  a  pendulum ­
  movement  and  senses  the  temperature  surface  at  various
depths  within  the  top  few  meters  of  the  ocean,  would  help
better  understand  the  distribution  of  SST  that  is  measured,
and  how  it  corresponds  to  satellite  measurements,  or  how  it
should  be  considered  in  the  cal/val  process.
The  differences  between  the  probes  can  also  be  inspected
as  a  function  of  mean  solar  local  time  (MSLT)  for  each  buoy.
For  this,  we  only  retain  the  buoys  that  reported  at  least  for
250  days,  without  issue.  For  the  subsequent  data  analysis,
we  filter  out  12  cases  when  differences  are  larger  than  20  K
(visible  in  Fig.  1),  likely  to  be  erroneous.  Figure  3  shows
that  the  mean  differences  feature  a  diurnal  cycle,  with  the
maximum  positive  differences  around  12:00  MSFT.  This  is
consistent  with  the  depth  difference  of  the  two  probes  in  the
context  of  diurnal  vertical  stratification  of  the  surface  temperature. ­
  Diurnal  stratification  tends  to  peak  around  14  h  (e.g.,
Reverdin  et  al.,  2013;  Morak-Bozzo  et  al.,  2016),  and  temperature ­
  stratification  larger  than  0.1  K  within  the  upper  0.5  m
would  tend  to  occur  only  at  the  lowest  wind  speeds.  However, ­
  this  daily  cycle  in  difference  may  also  be  partially  explained ­
  by  the  hull  sensor  being  heated  by  the  surrounding
buoy  and/or  by  direct  solar  radiation  (an  effect  which  might
tend  to  peak  more  around  12h  MSFT).  These  latter  effects
are  not  related  to  the  environment  and  should  be  avoided.
2.3  Recovered  buoys
Three  HRSST-2  buoys  manufactured  in  2012,  deployed  in
2014,  ran  ashore  in  2016  in  Great  Britain  and  Brittany.  They
were  recovered  and  offered  together  a  unique  opportunity  to