1. CTD/XBT Comparison, Quality of JJYY Data and XBT Data Analysis of the Mixed Layer Depthby
LT Mike Roth
OC3570
20MAR01

2. Purpose XBT/CTD temperature comparison
Value lost though the use of JJYY data points
Determine impact of mechanical and thermal forcing on the MLD

3. XBT-CTD Comparison
Insruments
Sea Bird CTD
T-7 Sippican XBTs

4. XBT-CTD Comparison Location of collocated XBTs and CTDs
1-7 cruise one
8 –10 cruise three

5. XBT-CTD Comparison Data Processing
Converted CTD dbars to m (Saunders, 1981)
Data deeper than deepest depth of collocated instrument discarded
XBT had smaller sampling interval (0.7 m: 1.98 m)
Linear interpolation of XBT to standard CTD levels
383 levels / 3,326 data pts each/ total=6,652

6. XBT-CTD Comparison Quality Control of the data Causes
Spikes due to XBT copper wire hitting hull
Bad XBTs
Upper 4 m (16.28 %)
3 Stage Process
0.2° C criteria to flag points
Visual
Final re-run

7. XBT-CTD Comparison
Quality Control of the data
Example of a bad XBT

8. XBT-CTD Comparison Quality Control of the data Results of QC
Reduced sample size to 9 CTD/XBTs
None of CTD data flagged
Removed: 501 pts (1,002)
2,825 pts (5,650); or 85% remained for statistical analysis

9. XBT-CTD Comparison Data Analysis
XBT subtracted from CTD temp at each depth for each collocated XBT/CTD

10. XBT-CTD Comparison
Data Analysis
Composite Plot

11. XBT-CTD Comparison Findings XBT generally > CTD
Greatest T-diff in upper 100 m
60-80 m was largest T-diff ( ºC at 69.5 m) with max variability as well ( ºC at 67.5 m)
CTD>XBT (+ T-diff) between m (max ºC at 91.3 m)
105 to 760 m mean T-diff = ºC and smaller std

12. XBT-CTD Comparison Comparison to Similar Studies NPS R/V Point Sur
NPS R/V Point Sur
February 2001
July-August 2000
Schmeiser (2000)
Heinmiller et al. (1983)
CTD – XBT Temp. Diff.
Depth (m)
Mean ( C)
Std ( C)
25-125
0.1779
0.3598
-0.17
0.08
0.0903
0.2194
-0.10
0.10
0.0951
0.1975
0.11
0.0960
0.1981
-0.13
0.16
Mean
0.1047
-.1549
0.2151

13. JJYY-XBT/CTD Comparison
Results of QC
Much easier to perform
XBT-6 salvaged for upper 100 m
Only had 7 collocated JJYY-XBT/CTD
Started with 2,247 pts (4,494) after interpolation
Ended up with 2,245 pts (4,490); 99% remained for statistical analysis

14. JJYY-XBT/CTD Comparison
Results of Data Analysis
NAVO SVPG software with user interface performs a pre-Quality control of the data and
With a linear interpolation of the data points does perform well as representing the full XBT profile ….

16. Analysis of MLD using XBTs

17. Analysis of MLD using XBTs

18. Analysis of MLD using XBTs

19. Conclusions Good Quality Control is a necessity, start below 4m
Sometimes you are limited to small sample sizes which can have a significant impact on final results
XBT/CTD comparison did reveal similar results as past studies (ie. XBT>CTD bias, upper 125 m)
JJYY-XBT/CTD comparison showed SVPG is a useful tool in representing the original XBT profile and makes QC much easier
Ungridded XBT data is not ideal for MLD analysis, but does show effects of thermal and mechanical forcing