Selection of prior distributions and multiscale analysis in Bayesian temperature reconstructions based on fossil assemblages
Citations
283 citations
Cites background or methods from "Selection of prior distributions an..."
...…development and refinement of reconstruction methods (e.g. Vasko et al. 2000; Toivonen et al. 2001; Korhola et al. 2002; Gersonde et al. 2005; Erästö & Holmström 2006; Haslett et al. 2006; Holden et al. 2008; Hübener et al. 2008; Goring et al. 2009; Velle et al. 2011), we are probably close…...
[...]
...…potential in palaeoclimatology because they can assess which features seen in a range of smoothed data are statistically significant and thus may be environmentally significant (see Korhola et al. 2000; Erästö & Holmström 2005, 2006; Weckström et al. 2006 for palaeoecological applications)....
[...]
82 citations
65 citations
47 citations
Cites background from "Selection of prior distributions an..."
...In a sensitivity analysis on the selection of prior distributions in Bayesian paleoclimate temperature reconstructions Erästö and Holmström (2006) demonstrate how crucial the selection of a prior probability distribution for the past values can be, particularly when the data are noisy and the…...
[...]
...Highly informative prior distributions, for example modern temperatures in the prior parameters of past temperatures, as in Toivonen et al. (2001) and Korhola et al. (2002), can to a large extent prescribe the reconstructed temperature (Erästö and Holmström, 2006)....
[...]
45 citations
Cites background or methods from "Selection of prior distributions an..."
...…that it is actually more meaningful to explore several different smooths of the whole posterior distribution since then one can make inferences about the salient features of past temperature variation and their statistical significance in many different time scales (cf. Erästö and Holmström, 2006)....
[...]
...More recently Haslett et al. (2006), Erästö and Holmström (2006), and Holden et al. (2008) have also used a Bayesian approach....
[...]
...We describe here only the main idea; for further details and additional discussion we refer to the original paper, the application described in Korhola et al. (2002), as well as to the related work discussed in Erästö and Holmström (2006) and Haslett et al. (2006)....
[...]
...Both in Korhola et al. (2002) and Erästö and Holmström (2006) the prior distribution for the x*is had the largest influence on the reconstructions but the choice of prior distributions for αk, βk and γk was not that important....
[...]
References
10,258 citations
1,900 citations
1,331 citations
"Selection of prior distributions an..." refers background in this paper
...It can be shown that Sj0 indeed acts like a smoother and that it is in fact a discrete analogue of cubic spline smoothing often used for continuous data (Erästö and Holmström 2005a; Green and Silverman 1994)....
[...]
...We have also added in the maps two curves whose horizontal distance shows the effective smoothing window width when the smoother Sj is viewed as an approximate kernel smoother (Green and Silverman 1994)....
[...]
622 citations
620 citations
"Selection of prior distributions an..." refers background or methods in this paper
...Another idea put forward in this paper is the integration of the reconstruction model with a multiscale feature analysis technique known as SiZer, or SIgnificant ZERo crossings of derivatives (Chaudhuri and Marron 1999)....
[...]
...More discussion on this particular prior can be found in Erästö and Holmström (2005a) and Green and Silverman (1994) and its earliest use in estimation can in fact be traced to Whittaker (1923) (see also de Boor 1978)....
[...]
...which features are statistically significant? An interesting way to look at this question is offered by SiZer methodology (Chaudhuri and Marron 1999; Erästö and Holmström 2005a; Holmström and Erästö 2002; Godtliebsen et al. 2003)....
[...]