[SEM18-07] Application of the Tsunakawa-Shaw method on “aged” thermoremanent magnetizations in laboratory
Santos and Tauxe (2019) examined the behavior of natural basaltic and trachytic samples during absolute paleointensity (API) experiments on both the original and laboratory-acquired thermoremanent magnetizations (TRMs). They separated the samples into “straight” (single-domain-like) and “curved” (multidomain-like) groups based on the API plots obtained from the original TRMs. When they gave a “fresh” laboratory TRMs to the samples in a 70 uT field and conducted an IZZI-Thellier (Yu et al., 2004) API experiments on the samples, the straight-group-samples recovered the laboratory field with high precision (70.5+/-1.5 uT, N=12) while the curved-group-samples resulted in much more scattered results (71.9+/-5.2 uT, N=12). Because the API plots on the fresh laboratory TRMs were also classified into “straight” and “curved”, the sample sets were finally categorized into four types of “straight-straight” (SS), “straight-curved” (SC), “curved-straight” (CS), and “curved-curved” (CC) (original TRM behavior – laboratory TRM behavior).
Sister specimens from each sample of Santos and Tauxe (2019), which were given “fresh” laboratory TRMs in a 70 uT field, were further aged in laboratory for two years also in a 70 uT field but in perpendicular direction. Tauxe et al. (2019 AGU Fall Meeting) reported the API results of the IZZI-Thellier experiments on the “aged” specimens: 70.6+/-4.8 uT for the SS type specimens (k’=0.031+/-0.106; N=6); 71.4+/-5.1 uT (k’=0.191+/-0.184; N=6) for the CC type specimens (N=6). It is noted that k’ is the curvature statistic defined in Paterson et al. (2014), which is the value of curvature k (Paterson, 2011) for the measurements actually used in the API calculation.
In the present study, we have applied the Tsunakawa-Shaw (TS) method (Tsunakawa and Shaw, 1994; Yamamoto et al., 2003) to the “aged” specimens of the four types. We adopt usual selection criteria (e.g. Yamamoto et al., 2010). Measurement results are analyzed by a python code specially developed to analyze a series of remanence data obtained by the TS method. The analytical procedure is as follows: (1) calculate API statistics for all possible coercivity intervals; (2) discard the statistics not satisfying the selection criteria; (3) sort the statistics by a value of dAPI (relative difference from the expected API) and select the best 10 statistics; (4) sort the statistics by a fraction of NRM (frac_n) and select the best one.
The TS method resulted in APIs of 71.4+/-2.9 uT (N=5), 70.1+/-0.1 uT (N=4), 69.8+/-0.6 uT (N=7) and 68.3+/-10.0 uT (N=6) for the SS, SC, CS and CC types, respectively. Percentage fractions of anhysteretic remanent magnetization (ARM) erased by low-temperature demagnetization (LTD) are 5.4+/-2.1 % (N=6), 5.4+/-1.9 % (N=4), 7.4+/-3.1 % (N=7) and 11.6+/-5.1 % (N=6) for the SS, SC, CS and CC types, respectively. It is suggested that multi-domain (MD) like components are efficiently erased in the TS method, resulting in the good API estimates even for the CS and CC types. In the NRM-TRM1 diagrams, there are less curvatures for the SS type (k=0.09+/-0.09) and the SC type (k=0.11+/-0.07) than for the CS type (k=0.19+/-0.13) and the CC type (k=0.43+/-0.34). The same is true for the ARM0-ARM1 diagrams, namely, the curvatures are less in the SS type (k=0.04+/-0.04) and the SC type (k=0.03+/-0.04) than in the CS type (k=0.19+/-0.20) and the CC type (k=0.39+/-0.43). Because the curvature values are almost the same between the NRM-TRM1 diagrams and the ARM0-ARM1 diagrams for each type, ARM corrections are thought to yield linear NRM-TRM1* diagrams even for the CS and CC types.
Sister specimens from each sample of Santos and Tauxe (2019), which were given “fresh” laboratory TRMs in a 70 uT field, were further aged in laboratory for two years also in a 70 uT field but in perpendicular direction. Tauxe et al. (2019 AGU Fall Meeting) reported the API results of the IZZI-Thellier experiments on the “aged” specimens: 70.6+/-4.8 uT for the SS type specimens (k’=0.031+/-0.106; N=6); 71.4+/-5.1 uT (k’=0.191+/-0.184; N=6) for the CC type specimens (N=6). It is noted that k’ is the curvature statistic defined in Paterson et al. (2014), which is the value of curvature k (Paterson, 2011) for the measurements actually used in the API calculation.
In the present study, we have applied the Tsunakawa-Shaw (TS) method (Tsunakawa and Shaw, 1994; Yamamoto et al., 2003) to the “aged” specimens of the four types. We adopt usual selection criteria (e.g. Yamamoto et al., 2010). Measurement results are analyzed by a python code specially developed to analyze a series of remanence data obtained by the TS method. The analytical procedure is as follows: (1) calculate API statistics for all possible coercivity intervals; (2) discard the statistics not satisfying the selection criteria; (3) sort the statistics by a value of dAPI (relative difference from the expected API) and select the best 10 statistics; (4) sort the statistics by a fraction of NRM (frac_n) and select the best one.
The TS method resulted in APIs of 71.4+/-2.9 uT (N=5), 70.1+/-0.1 uT (N=4), 69.8+/-0.6 uT (N=7) and 68.3+/-10.0 uT (N=6) for the SS, SC, CS and CC types, respectively. Percentage fractions of anhysteretic remanent magnetization (ARM) erased by low-temperature demagnetization (LTD) are 5.4+/-2.1 % (N=6), 5.4+/-1.9 % (N=4), 7.4+/-3.1 % (N=7) and 11.6+/-5.1 % (N=6) for the SS, SC, CS and CC types, respectively. It is suggested that multi-domain (MD) like components are efficiently erased in the TS method, resulting in the good API estimates even for the CS and CC types. In the NRM-TRM1 diagrams, there are less curvatures for the SS type (k=0.09+/-0.09) and the SC type (k=0.11+/-0.07) than for the CS type (k=0.19+/-0.13) and the CC type (k=0.43+/-0.34). The same is true for the ARM0-ARM1 diagrams, namely, the curvatures are less in the SS type (k=0.04+/-0.04) and the SC type (k=0.03+/-0.04) than in the CS type (k=0.19+/-0.20) and the CC type (k=0.39+/-0.43). Because the curvature values are almost the same between the NRM-TRM1 diagrams and the ARM0-ARM1 diagrams for each type, ARM corrections are thought to yield linear NRM-TRM1* diagrams even for the CS and CC types.