Loading...
Please wait, while we are loading the content...
Similar Documents
epl draft Coping with Dating Errors in Causality Estimation
| Content Provider | Semantic Scholar |
|---|---|
| Author | Smirnov, Dmitry A. Marwan, Nurfakhzan Breitenbach, Sebastian F. M. Lechleitner, Franziska A. Kurths, Juergen |
| Copyright Year | 2017 |
| Abstract | We consider the problem of estimating causal influences between observed processes from time series possibly corrupted by errors in the time variable (dating errors) which are typical in palaeoclimatology, planetary science and astrophysics. “Causality ratio” based on the Wiener – Granger causality is proposed and studied for a paradigmatic class of model systems to reveal conditions under which it correctly indicates directionality of unidirectional coupling. It is argued that in case of a priori known directionality, the causality ratio allows a characterization of dating errors and observational noise. Finally, we apply the developed approach to palaeoclimatic data and quantify the influence of solar activity on tropical Atlantic climate dynamics over the last two millennia. A stronger solar influence in the first millennium A.D. is inferred. The results also suggest a dating error of about 20 years in the solar proxy time series over the same period. Introduction. – Revealing cause-and-effect relationships between observed processes at various time scales is an important step in understanding many physical, biological, physiological and geophysical systems [1–8]. Frequently, this issue must be addressed with rather limited knowledge about the systems under study, amounts of observational data, and dating accuracy. A general approach to detect and quantify causal couplings, i.e., to find out “who drives whom”, is the Wiener – Granger (WG) causality [9, 10]. In its simplest version, the idea is to check whether a present value of one process (X) can be predicted more accurately using the past of a second process (Y ) in comparison with predictions based solely on the past of X. In fact, this concept generalizes a conditional (partial) cross-correlation [11] and has been followed by a number of elaborations such as information-theoretic measures [3,12–15] and various nonlinear approximations [16]. Despite some limitations and obstacles [17–20], the WG causality appears quite useful in practice, allowing meaningful dynamical interpretations [21,22] and becoming increasingly widely used in different fields, such as biomedicine [1,5,8] and geophysics [6]. Causal coupling estimation is also of great value in climate science, where temporal changes of climatically sensitive proxies [23] are the main source of information about past climate dynamics over long time intervals. The stalagmite YOK-I from the Yok Balum Cave in Southern Belize is especially well dated [24] and provides a highresolution reconstruction of low-latitudinal Atlantic moisture variations [25]. Making use of solar irradiance reconstructions (e.g. [26]), one can ask “How do variations in solar activity affect regional Atlantic climate?”. Answering this question helps further delineating the timevariant processes that drive climate variations. However, this question leads directly to the main difficulty with such data: dating accuracy of the reconstructions used. Uncertainties inherent to sampling and dating methods limit our knowledge of the time instant of each proxy observation, so that temporal ordering of the observations from the two time series may be distorted uniformly or irregularly in the course of time. This makes questionable any application of the WG causality approach, which essentially requires a clear distinction between the future and the past. In this Letter, we propose a solution with an appropriate specification of the problem setting and adaptation of the WG causality characteristics. We consider a situation |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://quest.pik-potsdam.de/wp-content/uploads/2017/03/DatingErrors_v7.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
