New Computer Fund

Sunday, February 1, 2015

South China Sea Reconstruction Comparisons

As noted on the chart this is a comparison of SST reconstructions of the South China Sea.  The Herbert et al. 2010 is a UK37 recon which I "adjusted" to the Wei2007 Mg/Ca recon from a very close location.  The adjustment amounted to +0.34C which is in the uncertainty range of the reconstructions.  As you can see, either there is one wicked current edge around the rough 80 nautical miles that separate the two core locations or there are some issues with the "bug" turned into thermometers.  Dating is a likely issue with the cores, so the Herbert2010 might be time compressed between 150kyr bp( yep there is a typo on the chart) and 250kyraBP.  The temperature range though is more important for my purposes than the dating.  The Herbert2010 and Wei2007 locations are at the northern edge of the tropics and the Steinke2008 is in the heart of the tropics.  If I adjusted Herbert2010 up to that recon, we would have the 28C peaks during the inter-glacials.

Instrumental in that region has a large range from roughly 24C to 30C.

The 0 to 10 degree north SSTs have a tighter range of roughly 27.5C to 29.5C.  

Comparing just the overlap period of all three reconstructions the peak values of the Wei2007 more northern tropical South China Sea come close to the Steike2008 more equatorial recon.  So while it isn't a normal situation for paleo, I can scale the northern tropical recons to the equatorial for a rough reconstruction of peak SST.  

Now I can use the 2.5ma Herbert2010 South China Sea reconstruction as a South China Sea approximate Peak SST reconstruction with enough wiggle words.

Expanded to 1000kyr you can see why I will need to look up new wiggle words.  The Wei2007 and Herbert2010 don't agree all that well as you go back in time.  

So this is more like the range of temperatures one could expect.  That isn't bad really, +/-1.5 C "temperatures" based on the life and times of plankton.  So the easiest way to "adjust" Herbert2010 South China Sea would be a simple average of the actual recon temperature and estimated peak temperature with +/-1.5C stuck on the end.  It is pretty unlikely that UK'37 will underestimate sea surface temperatures.  

So if you see me referencing 2.1 million years of South China Sea summer sea surface temperatures this is what you get and how I got it.

Saturday, January 31, 2015

Laboratory versus the Real World

Since I suspect that the Herbert et al. tropical SST reconstruction data using UK'37 will require some re-calibration, I am becoming a Mg/Ca kinda guy, I looked for a few comparisons using the same core or region.  DuBois et al. used the Herbert method and LuDuc et a. did a comparison of UK'37 and Mg/Ca.

Nothing is simple though.  The lat/lon are the first two numbers in the legend with the core identifier and method.  From the read me with the DuBois 2009 data we have this:

Cores ME0005A-27JC, ME0005A-43JC and TR163-31P were analyzed for alkenone unsaturation at Dalhousie University following standard laboratory procedures detailed by Kienast et al. (2006), whereas cores TR163-19P and TR163-22P where analyzed at Brown University following similar laboratory procedures detailed by Herbert et al. (1998). 
The age models for these cores were adopted as published in earlier studies (ME0005A-27JC from Kienast et al. 2007, ME0005A-43JC from Benway et al. 2006, TR163-19P from Lea et al. 2000, TR163-22P from Lea et al. 2006, TR163-31P from Martin et al. 2002). 
The alkenone unsaturation index UK037 is calculated as UK'37 = (C37:2)/(C37:3 + C37:2), where (C37:2) and (C37:3) are concentrations of the diunsaturated and triunsaturated C37 methyl alkenones. For conversion into temperature estimates, we used the culture calibration of Prahl et al. (1988) (UK'37 = 0.034T + 0.039). Replicate analyses of selected samples indicate an analytical precision of about ±0.01 UK'37 units (0.3°C). 

So we have lab quality of about +/- 0.3C but real world temperature ranges of about +/- 3.0 C degrees.  The eastern tropical Pacific should be about the worst case.  There is a great deal of ENSO related upwelling and northward circulation.  The UK'37 method can be "calibrated" to regional temperatures but you may not be able to capture the variance.  So while the eastern Pacific is an interesting region, temperature reconstructions are going to be a beyatch.  So let's just say Eastern Tropical Pacific is going to be put on hold.

Thursday, January 29, 2015

Favorite Ocean Paleo Proxies

My favorite is becoming the Mg/Ca (g. ruber.) pretty quickly.  Instead of the standard spaghetti chart this is something a bit different.  This is just all the Marcott et al. selected reconstructions for the central tropical region combined in one big time series.   This gives you an eye for the variance in the individual proxy reconstructions versus a roughly 100 initial to 500 year smooth back near 13000 BP.   Since this doesn't include "caps" or the most recent data for the core locations, there is a "hockey stick" starting around 800 years before present.  The "caps" are all pretty much in the orange noise so the variance here is not a bad indication of the overall confidence interval.  If these were all thermometers, you could reduce the error range but paleo "bugs" can have biases so there is some question about how much you can reduce the error bars.

I like the Mg/Ca because they appear to produce more realistic temperatures more often.

The Uk'37 and TEX 86 are alkenone based "thermometers" seem to have issues figuring out if they are "surface" or "subsurface" temperature indicators.  While there where not many southern tropical reconstructions selected by Marcott et al. the ones selected have a much higher variation.  Near the Pacific coast that variation could be real with changing currents or it could be "bugs" going with the flow from colder waters with the currents.   That is useful, but I would think a serious ocean current model would be required to take advantage of these.  TEX 86 though seems to be mainly a fresh water thing.  The Teirney et al Lake Tanganyika reconstruction which is included in this group, so it not a "true" SST reconstruction, is responsible for the more noticeable peaks.

It has a number of spikes that are likely related to the precessional cycle.

There are more northern tropical reconstructions in the Marcott et al. selections and these tend to be in better agreement.  Not a huge difference between Glacial and Inter-Glacial in the northern tropics it would seem.

If I combine all the reconstructions, since the central equator has more recons it tends to dominate.  You get the same hockey stick shifted toward 300 years BP.  Since the UK37 recons tend to run a little on the cold side the variance is much larger.

As I said at the start this isn't your standard reconstruction but I think it does show potential uncertainty better than you will normally see.  btw, the last combines all the reconstructions listed on the first three charts.

Here is the complete list.

Number Location / Core Reference

1 GeoB5844-2 Arz et al., 2003 (4)
2 ODP-1019D Barron et al., 2003 (5)
3 SO136-GC11 Barrows et al., 2007(6)
4 JR51GC-35 Bendle and Rosell-Melé 2007 (7)
5 ME005A-43JC Benway et al.,2006 (9)
6 MD95-2043 Cacho et al., 2001 (10)
7 M39-008 Cacho et al., 2001 (10)
8 MD95-2011 Calvo et al., 2002 (11)
9 ODP 984 Came et al., 2007 (12)
10 GeoB 7702-3 Castañeda et al., 2010 (14)
11 Moose Lake Clegg et al., 2010 (15)
12 ODP 658C deMenocal et al., 2000 (16)
13 Composite: MD95-2011; HM79-4 Dolven et al., 2002 (17)
14 IOW225517 Emeis et al., 2003 (18)
15 IOW225514 Emeis et al., 2003 (18)
16 M25/4-KL11 Emeis et al., 2003 (18)
17 ODP 1084B Farmer et al., 2005 (20)
18 AD91-17 Giunta et al., 2001 (21)
19 74KL Huguet et al., 2006 (22)
20 74KL Huguet et al., 2006 (22)
21 NIOP-905 Huguet et al., 2006 (22)
22 NIOP-905 Huguet et al., 2006 (22)
23 Composite: MD01-2421; KR02-06 St.A GC; KR02-06 St.A MC Isono et al., 2009 (24)
24 GeoB 3910 Jaeschke et al., 2007 (25)
25 Dome C, Antarctica Jouzel et al., 2007 (26)
26 GeoB 7139-2 Kaiser et al., 2008 (27)
27 Dome F, Antarctica Kawamura et al., 2007 (28)
28 18287-3 Kienast et al., 2001 (29)
29 GeoB 1023-5 Kim et al., 2002 (30)
30 GeoB 5901-2 Kim et al., 2004 (31)
31 KY07‐04‐01 Kubota et al., 2010 (32)
32 Hanging Lake Kurek et al., 2009 (33)
33 GeoB 3313-1 Lamy et al., 2002 (34)
34 Lake 850 Larocque et al., 2004 (35)
35 Lake Nujulla Larocque et al., 2004 (35)
36 PL07-39PC Lea et al., 2003 (36)
37 MD02-2529 Leduc et al., 2007 (37)
38 MD98-2165 Levi et al., 2007 (39)
39 MD79-257 Levi et al., 2008 (39)
40 BJ8 13GGC Linsley et al., 2010 (40)
41 BJ8 70GGC Linsley et al., 2011 (40)
42 MD95-2015 Marchal et al., 2002 (41)
43 Homestead Scarp McGlone et al., 2010 (42)
44 Mount Honey McGlone et al., 2011 (42)
45 GeoB 10038-4 Mohtadi et al., 2010 (43)
46 TN05-17 Nielsen et al., 2004 (44)
47 MD97-2120 Pahnke and Sachs, 2005 (45)
48 MD97-2121 Pahnke and Sachs, 2006 (45)
49 17940 Pelejero et al., 1999 (46)
50 Vostok, Antarctica Petit et al., 1999 (47)
51 D13822 Rodriguez et al., 2009 (48)
52 M35003-4 Rühlemann et al., 1999 (49)
53 OCE326-GGC26 Sachs 2007 (50)
54 OCE326-GGC30 Sachs 2007 (50)
55 CH07-98-GGC19 Sachs 2007 (50)
56 GIK23258-2 Sarnthein et al., 2003 (51)
57 GeoB 6518-1 Schefuß et al., 2005 (52)
58 Flarken Lake Seppä and Birk, 2001; Seppä et al. 2005 (53, 54)
59 Tsuolbmajavri Lake Seppä and Birk, 2001; Seppä et al 1999 (54, 55)
60 MD01-2390 Steinke et al., 2008 (56)
61 EDML Stenni et al., 2010 (57)
62 MD98-2176 Stott et al., 2007 (58)
63 MD98-2181 Stott et al., 2007 (58)
64 A7 Sun et al., 2005 (59)
65 RAPID-12-1K Thornalley et al., 2009 (60)
66 NP04-KH3, -KH4 Tierney et al., 2008 (62)
68 GeoB6518-1 Weijers et al., 2007 (64)
69 MD03-2707 Weldeab et al., 2007 (65)
70 GeoB 3129 Weldeab et al., 2006 (66)
71 GeoB 4905 Weldeab et al., 2005 (67)
72 MD01-2378 Xu et al., 2008 (68)
73 MD02-2575 Ziegler et al., 2008 (69)
Update: Here is a semi-random mix of tropical recons.

This is mainly eastern tropical Pacific which I used with no particular plan in mind (semi-random).  Other than the "cap" recons there aren't that many that were missed by Marcott et al. 2013, other than some of the longer, low frequency recons.  Some of those have clusters of higher frequency data points around interesting events like the LGM and the Younger Dryas.  Including those will play hell with estimating confidence intervals.  This chart though has the average scale on the right (note I have the ledges reversed) and the average on the right.  I was going to trash this but figured I might as well post it so it is saved from my next round of playing.