1: \begin{abstract}

2: 

3: The estimation of covariance operators of spatio-temporal data 

4: is in many applications  only computationally feasible under simplifying assumptions, such  as 

5: separability of the covariance into strictly temporal and  

6: spatial factors. 

7: Powerful tests for  this  assumption  have  been proposed 

8: in the literature.

9:  However, as real world systems, 

10: such as climate data are  notoriously inseparable, validating this assumption by statistical  tests, seems inherently questionable.  

11: In this paper we present an alternative approach: By virtue of 

12: separability measures, we quantify how strongly the data's

13: covariance operator diverges from  a  separable approximation. Confidence intervals  localize these measures with statistical guarantees. This method provides users with a 

14: flexible tool, to  weigh the computational gains of a separable model against the associated increase in bias.

15: As separable approximations we consider the established methods of partial traces and partial products, and develop  weak convergence 

16: principles for the corresponding estimators.  Moreover, we also prove such results for

17:   estimators of optimal, separable approximations, which are arguably of most interest in applications. In particular we

18:    present for the first time  statistical inference for this object, which has been  confined to estimation previously. 

19: 

20: Besides confidence intervals, our results encompass  tests for approximate separability. 

21: All methods proposed in this paper  are free of nuisance parameters and do neither  require

22: computationally expensive resampling procedures nor the estimation of  nuisance parameters. 

23: A simulation study underlines the advantages of our approach and its applicability is demonstrated by 

24:  the investigation of German annual temperature data.

25: 

26: 

27: 

28: 

29: \end{abstract}

30: