A review on systematic differences between Sippican's XBT and TSK's XBT Shoichi Kizu (Tohoku University, Japan) The 4 th XBT workshop November 2014, Beijing, China
Speed (kt) Range (m) since T561830’71.8- T615460’72.7- T715760’78.4- T7 (20kt)20760‘97.8- T ’79.1- Probe types (as of 2013) Speed (kt) Range (m) since T430460’65.6- T561830’71.6- Fast Deep201000’91.9- T615460’68.4- T715760’67.6- Deep Blue20760’81.4- T ’72.3- T116460’06.2- Rated accuracy: Temp.: +/- 0.2K, Depth: +/- 5m or 2% Common FRE T4(?) ? Sippican ( LMS) TSK US, Europe, Australia, etc.Japan, China, Korea*, etc. Notes 1) T4/T6 are with shorter wire in the probe (lighter nose weight) than T7/DB. 2) T4/DB are with longer wire in the canister than T6/T7 to allow higher ship speed. 3) They produce by their own, only sharing basic design and thermistor. * for civilian use only
Sippican T7TSK T7 A key difference: wire Same gauge, but different coating.
Some extreme “pull” test of wire TSK T7 technical limit in this measurement 3 ℃ 20 ℃ 10 ℃ TSK’s wire is very “resistant” to insulation break by wire stretch. Relative Stretch (%) Sample Number by courtesy of TSK
Sippican’s wire tends to lose insulation more easily in cold water. Sippican T7 3 ℃ 10 ℃ 20 ℃ Relative Stretch (%) Sample Number Some extreme “pull” test of wire by courtesy of TSK
Temperature ( o C) Case Temperature ( o C) Depth (m) Case Spiky and/or positively-biased profiles Found 1) Only with Sippican 2) Mostly in cold water Black: Sippican Red: TSK Blue: CTD Black: Sippican Red: TSK Blue: CTD - Examples from our side-by-side tests near Japan -
Facts known (or believed at the time of H95) about the two manufacturers’ XBT TSK Sippican Wire is different. TSK’s wire coating is thicker and heavier than Sippican’s. But the weight of their probe was believed to be the same. The difference in wire weight was believed to be cancelled by adjusting the metal nose weight by inside hollowing. Therefore, they were believed to fall at an identical rate. Actually, the difference in their fall-rates was marginal in H95.
(Cowley, 2010; personal communication) TSK T7 is heavier than Sippican T7. This difference comes from the difference in weight of wire. (Kizu et al., 2011) Difference in probe weight (in air) The Sippican’s probe shows larger variance (batch-to- batch difference). Manufacturer’s tolerance Sippican: +/- 5g TSK: +-/ 1-2g Probe weight (g) Sippican T7 (Tohoku) TSK T7 (Tohoku) Sippican T7 (CSIRO) Frequency ~10g (1.5%)
H95 A typical comparison between recent Sippican T7 and TSK T7 (Kizu et al. 2011) Manu spec. (2%) Depth (m) Depth error (m) Sippican (dashed) TSK (thin solid) Sippican: positive depth error TSK: negative depth error positive error = slower fall Sippican TSK CTD (thick solid) negative error = faster fall
Side-by-side comparisons (May 2008; East of Japan; K11) Depth (m) Sippican: Depth error is positive. True fall rate is >2% smaller than H95. TSK: Depth error is negative. True fall rate is <2% greater than H95. Manu spec. (2%) Depth error (m) Negative Error ↓ Faster fall TSK Positive Error ↓ Slower fall Sippican Depth error (m) Much bigger than the weight difference ~4%
Cowley et al. (2013) +4% -8% +4% -6% After GK07 - Suggested histories of XBT biases Depth Temperature Manu spec. (+/- 2%) H95 Strip-chart Sippican has slowed. TSK T7
TSK T7 shows different history of fall-rate bias from Sippican’s. (T6 does, too.) Mar 1998 (4) Manu spec. (+/- 2%) Mostly within +/- 2%, but with both signs. (not very systematic) sample size Aug 1998 (4)Apr 1999 (11)Aug 2000 (9)Oct 2000 (5) Jan 2001 (10)Jan 2002 (11)Feb 2003 (6)Dec 2011 (3)Jun 2012 (23) slow fast
SIP T6 SIP T7 TSK T6 TSK T7 Inter-manufacturer difference : Sippican vs TSK, and T6 vs T7 (late 2000s – early 2010s) Similar 1-1.5% different 2% different4% different Recent Sippican T6/T7 fall more slowly than H95. Recent TSK T7 falls more quickly than H95. opposite! Much bigger than the weight difference Also detected by H95
Effect of weight change circular ditch by courtesy of TSK Effect of weight change (10g and 20g) is tested by two methods: 1)Reducing the nose weight by making ditch inside 2)Reducing the wire in the probe
Fall rate: TSK > TSK(-10g) > TSK (-20g) > LMS Impact of weight difference TSK ( nose -10g ) TSK (normal) TSK ( nose -20g ) 11g lighter than TSK Test in 2011 (nose) 10 grams = 1-1.5% in fall-rates SIPPICAN 4g lighter than TSK TSK ( wire -10g ) TSK (normal)SIPPICAN TSK ( wire -20g ) Test in 2012 (wire) Manu spec. (+/- 2%) 0m 800m Depth 50m Depth error -50m 0m 800m
Structural differences between Sippican T7 and TSK T7 (K11) Sippican T7TSK T7 Not concentricConcentric Inside of nose weight: Very different According to the manufacturers, All TSK’s XBT are with concentric design. Sippican’s (now LMS) T7 and Deep Blue are non-concentric, but T4, T5, T6 and T10 are concentric. “Never changed.” Other differences Length of nose weight Girth of nose weight Diameter of nose hole Shape and thickness of tail fins Inner volume of afterbody Size of inner water pass
Sippican (T7/DB) Sippican (T4/T6) Seemingly variable TSK T7Sippican T7 The nose metal of Sippican T7 is not concentric for unknown reason. They weigh the same! (at the level of tolerance) Sippican samples were provided by CSIRO &AOML Batch-to-batch difference in the shape and density
Bias of manufacturers’ FRE for T5 The manufacturers’ FRE: worked well with Sippican T5, but it shows large positive bias for TSK T5. So we proposed a new FRE for TSK T5: Kizu and Hanawa (2005a) Sippican T5 (almost bias-free) positive depth error (slower fall) negative depth error (faster fall) ①②③④ ⑤ Rated depth error (2% or 5m) TSK’s T5 is heavier than Sippican’s T5! TSK T5 (shows very large positive bias) Manu FRE
Temperature-dependence of fall-rates (T5) North of KE front (cold water) South of KE front (warm water) Kizu and Hanawa (2005b) Temperature Depth (m) Cold water Warm water Sippican T5TSK T5 TSK T5 showed larger temperature dependence than Sippican T5.
Temperature-dependence of fall-rates (T7) A test off Honshu (143E) 35N (S03) 37N (S02) 39N (S01) 14 o C a b Faster fall Slower fall XBT “feels” change of water viscosity. The time change is not likely to be the whole story of the fall-rate bias. (Kizu and Hanawa, to be published) ~1%/10K
Summary 1) Sippican’s probes and TSK’s probes have different histories of the fall-rate bias. The two manufacturers’ XBTs should be separated in the analysis. 2) The change in probe weight can contribute to the fall-rate change. It’s impact is estimated to be about 1-1.5% / 10g. 3) Change of fall-rate by weight reduction is not proportional to the weight change, indicating that factors other than total weight (maybe structure) is also important. 4) The inter-manufacturer difference and the historical change of fall-rate cannot be explained by their weight difference alone. Again, structural difference/change should also be important. 5) Water temperature (viscosity) causes difference in the fall-rate. It’s impact is estimated to be about 1% for 10K.
Thank you. Shoichi Kizu
backup slides
Obtained FR coefficients a b K11 (TSK) K11 (LMS) H95 Sippican H95: Hanawa et al. (1995) K11: Kizu et al. (2011) Faster fall Slower fall
Obtained FR coefficients H95: Hanawa et al. (1995) Faster fall Slower fall
Obtained FR coefficients a b K11 (TSK) K11 (LMS) H95 Sippican H95: Hanawa et al. (1995) K11: Kizu et al. (2011) Faster fall Slower fall
a b
a b
Brief history of the instrument Developed in early 1960’s at the request of US Navy. Three manufacturers met in a competition in 1965, and Francis Associates, the predecessor of the present Lockheed Martin Sippican, won it. “The Navy is seeking a new BT that will be more suitable for use aboard ASW and other vessels and that will improve the sonar range prediction capability of the fleet.” (Arthur Little Inc., 1965) “A more specific object is to provide … a greater depth range … with ships underway at higher speed and higher sea state, while maintaining flexibility of ship maneuvers.” (Campbell et al., 1965; US patent 3,221,556) TSK joined XBT business in 1970s under a license agreement with Sippican. TSK produces several probe types basically of the same design with Sippican, but manufacture of the two companies is totally independent except that they use thermistors of the same brand, and their probes do have differences.
・ Sippican’s FRE (S65): a=6.472, b= ・ IGOSS’s FRE. (Hanawa et al, 1995; H95): a=6.691, b= commonly for Sippican and TSK T4/T6/T7/DB The fall-rate bias : early history (until 1990s) Many works suggested negative depth bias of Sippican’s FRE for T6/T7. 20 - 20 Sippican T7 Flierl and Robinson (1977) D(t) = at – bt % correction to depth Singer (1990) “XBT wave” Sippican T7
Anatomy -Detail inspection of expendable probes- We investigated weight of probe w/wo wire in air weight of probe w/wo wire in water position of the center of gravity shape and structure of probes line density of wire quality of wire coating (i.e. electric insulation) for some sample of T-5 from both TSK and Sippican T-7 from both TSK and Sippican And, we learned that weight was different, shape was different, structure was different, … though we had only known that wire was different. *Photos are taken during inspection of T-5.
There is a long story leading to Hanawa et al.(1995), though I was not involved during those years… Surface transient of Japanese XBT systems - Kizu and Hanawa (2002a; Deep-Sea Res.) Bowing problem in some Japanese acquisition systems - Kizu and Hanawa (2002b; J. Oceanogr.) Fall-rate of TSK T-5 - Kizu et al. (2005a; J. Oceanogr.) Temperature-dependency of T-5’s fall-rate - Kizu et al. (2005b; J. Oceanogr.) Fall-rate of XCTD-1, 2, and some developmental models - Kizu et al. (2008; Deep-Sea Res.) Fall-rate of T-7 (“revisit”) … - to be submitted soon Calibration of XCTD … - being tackled, but very difficult Some chronology (in Tohoku) after mid-1990s
The XBT Afterbody (plastic) Fin stabilizers (their end part is angled) Nose weight (metal) Front view Thermistor Rear view Center hole (for water and wire outlet) Center hole (for water inlet) A photo of TSK T-7 Probe wire
Cited from AOML’s web site. Canister wire Probe wire Canister Nose weight Probe Thermistor The XBT
Metadata for XBT Most important Manufacturer (Sippican, TSK, other) Probe type (T4, T5, T6, T7, DB, FD, etc.) Coefficients of FRE used Probe serial number (or Date of Manufacture) Deck unit (recorder) Auxiliary Country Institution Ship And, of course, time and position of measurement
Depth error estimation (Hanawa and Yasuda, 1992) 1) Assume CTD temperature profile as truth. 2) Find that gives the minimum of for each depth. 3) gives the depth error profile for that probe. *) Resistive to errors in temperature. *) Does not work well in areas with weak thermal stratification.
S65 H65 S65 H65 Fall rate changes with time Sippican T7/DB probes have an slowing of fall rate from the 1990 ’ s to now. Slower fall rate results in warm biases S65 = Sippican fall rate, 1965 H95 = Hanawa et al fall rate, 1995 Biases in historical XBT data | Rebecca Cowley 37 | Sippican T4/T6 profiles dominate the early records and have a slower fall rate in the mid-1970s. Slower, warmer Faster, cooler Slower, warmer Faster, cooler Modified Cowley et al. Sippican T4/T6 Sippican T7/DB Not always similar Newest scheme shows …
“pure” Temperature bias also changes with time The temperature bias after depth correction is highest in the pre ’ s data. and it’s been “positive” The higher bias may be due to the recorder types used at the time (analog strip chart recorders were used up till the mid-1980s) Biases in historical XBT data | Rebecca Cowley 38 | Modified Cowley et al. Positive T bias remains in XBTs with pressure switches (Goes et al., 2013), too. Sippican T4/T6 Sippican T7/DB
a b (x10 3 ) a, b coefficients Sippican : a=6.472, b= H95 : a=6.691, b= Faster fall Slower fall TSK T6 falls more slowly than TSK T7 by 1-1.5%. TSK T6 TSK T7
Sippican : a=6.472, b= H95 : a=6.691, b= a b (x10 3 ) a, b coefficients (TSK T6) Faster fall Slower fall Oct 1990 (11) Nov 1990 (11) Jul 1991 (7) Sep 1992 (22) Jun 1995 (11) Nov 1997 (9) Jun? 2012 (22)
Sippican : a=6.472, b= H95 : a=6.691, b= a, b coefficients (TSK T7) a b (x10 3 ) Faster fall Slower fall Mar 1998 Aug 1998 Apr 1999 Aug 2000 Oct 2000 Jan 2001 Jan 2002 Feb 2003 Dec 2011 Jun 2012
Some curious examples from our side-by-side tests Sippican T7 vs. Sippican T7 Gray: CTD # (734.2g) # (728.5g) Heavier probe fell more slowly. # (728.8g) # (734.2g) Almost no depth difference, but some temperature bias.
Structural differences (nose weight; K11) Length: TSK T7 is 1mm (1.7%) longer Diameter: TSK T7 is 0.2mm (0.4%) larger Sippican T-7 TSK T-7 Diameter of hole (water inlet): TSK T7 is 0.5 mm (4.6%) smaller
Structural differences (afterbody; K11) Angled part of the tail fins: Shape is different Three fins: TSK T-7 is thinner Inner volume of afterbody: TSK T-7 is smaller by about 5 cm 3.
Structural differences between Sippican T7 and TSK T7 (afterbody) Water pass (four holes around the central rod): TSK T7’s holes are smaller Four protuberances exist only in Sippican’s
Seemingly variable (T4/T6) TSK T6 Oct 1990 (120767) Sep 1992 (142796) Nov 1997 (197271) Apr 2011 (256930) Change in density of die cast (zinc-alloy) Change in the inside form of the nose weight (by an effort to keep their mass constant) They weigh the same! but with different size of the bore. Change in inside water flow? Change in moment of inertia? Provided by F. Reseghetti Could it be a cause of year-to-year fall-rate changes? Sippican (T4 ; 1992?) Change of outer shape is minor, but density has changed.