Struggling with an age-depth model that covers the last 200 ky

[Antonio García-Alix]

Hi everyone,

 

I’m running an age-depth model in OxCal of a 42 m-long sediment core that covers the last 200 ky. We have 85 radiocarbon dates for sediments younger than 48 ky as well as 7 dates coming from amino acid racemization and paleomagnetism for older sediments. I used the P-Sequence tool and I got the command P_Sequence("",1,0.03,U(-2,2)). I also used a general outlier model. AAR and paleomagnetic dates were entered like this:

 

Date("Iceland Basin start",N(calBP(191700),1800/2))

   {

    Outlier(0.5);

    z=4011;

   };

 

I ran the model yesterday afternoon and it is still computing. I only got a very small blue line in the “done” progression bar. I’m using the downloaded OxCal distribution. Is this normal due to the number of dates and the length of the sediment core? Would be faster “OxCal online”? 

 

I have another question. The two oldest 14C dates we got were reported by the lab as >52 ky and > 55 ky without any associated error. These depth-dates could make sense. Is there any way to enter them in the model?

 

Thank you very much in advance,

 

Antonio

[Erik Marsh]

Hey Antonio – 

I am not sure which version of OxCal runs faster, but it seems like it's taking too long of a model with 92 dates. The P_Sequence tool is the right approach here. The k-value of 0.03 looks lowish, but could be right for this sequence. Make sure you include all the dates when using the tool (not just the 14C dates) and double check the depth units.

For your outliers, you probably want 0.05 (not 0.5).

For very old dates, the best way to input them is with the percentage of modern carbon. For example, this date from Australia is older than 45,000 years with 0.0031±0.001% modern carbon, from Macken et al. 2013:Table 1. In OxCal you input it as:

R_F14C("WeC16e", 0.0031, 0.001);

If you don't have that, you can use the Before command.

Hope this helps, Erik

[Antonio García-Alix]

Hi Erik, thank you very much for your quick reply!

 

I tried to run the model several times. Since I suspected that the main problems were some 14C outliers, I’m only running the first ca. 9 meters that cover all the radiocarbon ages to see if I can obtain the modelled ages. In this part of the record I have clear outliers related to plant roots (e.g., 14C ages of 23 ky BP when they should be around 45 ky BP). I set the outlier probability to 100% in these cases of clear outliers, as in the paper from Macken et al. (2013), but I got errors like:

Warning! Cannot resolve order
Warning! Cannot determine value - U [3]=0.698388
Warning! Cannot determine value - P_Sequence [4]=0.844719

 

However, when I remove these clear outliers from the model, it works. Is this fine?

 

Additionally, we are trying to get the F14C values of the oldest 14C dates from the labs where we analyzed the samples. If we don’t get them eventually, we will use the Before command. Since we do not have the error of these 14C ages, should we enter only the 14C ages? like this:

 

Before()

  {

   R_Date("old date", 52000);

  };

 

Thank you very much in advance!