The determination of the age of trees is referred to as dendrochronology. If a tree is felled or a drill core is taken from the trunk, the individual growth phases can be identified from the annual rings. Climate-growth correlations can be derived from the linkage of climate data with the annual ring chronologies. As violent volcanic eruptions can lower the global average temperature by up to 0.5 °C for several years, such eruptions can be read from the annual rings.
An appropriate drill core is first taken (from the living tree) using an increment borer.
Laboratory and analysis methods
For the annual ring width data, which form the basis of all further investigations and meth-ods, the drill cores are prepared as follows.
In the laboratory the “fresh” drill cores must dry for a certain time and adjust themselves to the room climate to avoid cracks or other problems after gluing. The dried samples are then glued onto wooden sample holders. It must be ensured that the tracheids are cut at right angles to their horizontal direction (Hannrup et al., 2001) to avoid later measur¬ing errors and problems with the wood anatomy measurements.
The glued samples are inserted into a microtome after a short drying period (Gärtner & Nievergelt, 2010), well moistened with water and their surface cut with a knife similar to a razorblade (Fujii, 2003) at a predefined angle, to make the annual rings and the wood structure visible.
The advantage of the microtome is that the samples can be cut over the whole of their length and their surface better prepared than with the method hitherto used of abrading the samples with a bandsaw. However, recourse is made to the bandsaw method depending on the condition and preparability of the individual samples.
This result is particularly significant if image analysis systems such as WinDENDRO (Regent Instruments, Canada) or TSAPWin are used for wood sample dating. If the surface of every sample is well-prepared, its cut side is scanned with a customary scanner.
Dating and cross-dating
During processing with WinDENDRO, a path is laid from the pith to the bark through the drill core. The annual ring boundaries are marked on this path. As oaks have a very irregular light/dark pattern for the computer due to the ring pores, the rings cannot be automatically identified by WinDENDRO and the annual ring boundaries must be set manually. To check and tune the correct settings of the ring boundaries on the scan in WinDENDRO, the drill cores are additionally observed using a binocular viewer.
WinDENDRO automatically produces a text file from the set ring boundaries. The wood samples can be dated with WinDENDRO, i.e. a calendar year is assigned to every annual ring. However, the dating is not considered to be binding as long as the annual ring width curves have not been synchronised (Cherubini et al., 2004). This method is referred to as ”cross-dating”.
According to Douglass (1941), cross-dating is the detection of the same annual ring patterns in different trees, so that the actual growth year of a ring is the same everywhere in this pattern. This means in particular that it should be possible to find characteristic maxima and minima in the curve progression with all individual series in the same calendar year, as they very probably reflect certain climate events which impact the location (Müller-Stoll, 1951).
For cross-dating the samples, different programs are used in addition to the renewed visual check of the drill cores under the optical microscope. These programs are COFECHA (Holmes, 1983) and TSAPWin (Rinntech, Heidelberg).
To prepare a location chronology, it is first necessary to check whether errors have occurred during the dating process. For this purpose, COFECHA-Win (Holmes, 1983, 1999) and TSAP-Win (RINN, 2005; RINN, 2007) software is used.
COFECHA works with a “segment time series correlation technology”, and is used to find dating errors (Grissino-Mayer, 2001). The raw data of the individual trees provided by WinDENDRO are compared with each other; the software detects errors in the synchronisation between the individual trees. The “Segment length to examine” function makes it easier to limit any errors to a shorter time period. However, the segment length and overlap selected should not be too small, so as not to change the pattern representative of the tree and the statistical significance. In the program output, the series intercorrelation is stated for the individual annual segment blocks, together with the intercorrelation value when inserting or removing segments. In this way, missing or incorrectly set annual rings can be exactly determined using COFECHA. In addition, TSAP-Win (Time Series Analysis and Presentation) is used in the “segment-oriented cross-date check”. As with WinDENDRO, the synchronisation of the curves can be visually checked using TSAP. To not only make a visual assessment of the synchronisation, the statistical parameters calculated by TSAP such as synchronicity, cross-dating index (CDI), t value and r value are used.