Resistance Drill Tree Ring Analysis of American Sweetgum, Norway Spruce, and Pin Oak with the Application of a LOESS Regression

Tree ring analysis offers insight into the growth dynamics of trees throughout their lifetimes. Although conifers and deciduous trees differ in the composition of their annual rings, conifers being more homogenous than deciduous trees, the rings are produced under the same principles, with early cells being larger in diameter with thin cell walls and later cells being smaller in diameter with thicker walls (Fritts et al. 1991; Vaganov et al. 2005; Rathgeber et al. 2016). In temperate forest tree species, this boundary between the cessation of growth from the previous year and early growth of the current year can create a distinct boundary between each growing season.

Within annual rings the density of wood was found to vary, with early wood being less dense than late wood (Zobel and van Buijtenen 1989; Guilley et al. 1999). Models generated found density increases with a decrease in cell diameter and an increase in cell wall thickness, with these changes being the results of seasonal environmental variables (Buttò et al. 2021).

By examining and measuring the width of annual rings in either a cross-sectional disc or a core sample, we can determine changes in growth over time (Orozco-Aquilar et al. 2018; Li et al. 2024). A common nondestructive method of obtaining a sample for dendrochronology research is with the use of an increment borer. Coring, however, leaves a wound in the tree that can become susceptible to decay, or can breach a reaction zone, limiting progress of an existent decay from prior wounding (Fabiánová and Šil-hán 2021). The use of an increment borer was also found to cause staining of the wood around the wound (Lorenz 1944; Toole and Gammage 1959). The severity and spread of decay, discoloration, and wound closure is largely determined by the species and timing of the wound (Hepting et al. 1949; Eckstein and Dujesiefken 1999). However, a study found that over a 5-year period, the proper use of an increment borer for dendrochronology did not impact the growth rate or mortality of any tree in a variety of diameter classes of conifers (Palakit and Pumijumnong 2024). It was also found that over a 7-year period, coring did not cause an increase in mortality among hardwoods (Helcoski et al. 2019). In urban canopy management considerations, population density, infrastructure proximity, and surrounding property influence the acceptance of risk. There is both a desire to understand growth and vitality while also minimizing the wounding in tree management and growth monitoring (Hu et al. 2022; Joseph et al. 2023).

The process of collecting, processing, and storing core samples can be labor-intensive and timeconsuming (Gao et al. 2017), but an archive of physical evidence can be a priority, particularly for preserving the physical or chemical properties of the wood for future examination (Dunster 2014; Bassett et al. 2023). Where physical archiving is not required, alternative methods can reduce these drawbacks while maintaining accuracy in ring detection. Resistance drills provide one such alternative: offering a less invasive approach through leaving a smaller wound and a reduction in time spent collecting samples.

Resistance drills have been found to be capable of detecting annual tree rings by assessing wood density (Rinn et al. 1996; Yao et al. 2024). The resistance drill measures and records the resistance against both the rotation and inward movement of the needle as it bores through the tree (Downes et al. 2018; Sharapov et al. 2019). The needle is 3 mm at the head, 1.5 mm at the shaft, and 400 mm in length. A constant feed and rotation speed, set by the user, is maintained throughout the drilling process. The energy required to rotate and feed the electric motor, in increments of 0.1 mm, is recorded and translated into relative amplitude of 0% to 100% (Rinn et al. 1996). The differences in density of early wood and late wood allow for the observation and measurement of annual rings, with early wood being a lower density, and thus less energy required to drive the needle, than late wood (Rinn 1996, 2012a; Wang and Lin 2001; Yao et al. 2024).

Resistance drilling is done without the destructive sampling of disc collection and leaves behind a smaller hole (in the range of 3 mm) than an increment borer (often between 4.3 mm and 5.2 mm). While smaller in diameter, the wound still occurs and can thus provide opportunity for a decay event (Kersten and Schwartze 2005).

Sources of potential errors have been investigated, including drill bit sharpness, battery type and charge, air temperature, wood moisture content, and proximity to knots (Ukrainetz and O’Neill 2010). Ukrainetz and O’Neill (2010) found drill bit sharpness, battery type, and battery charge did not contribute significantly to differences in wood density as detected by the resistance drill, while they found air temperature, wood moisture content, and proximity to knots did contribute to changes in observed wood density. Solutions were offered. By drilling when air temperature is above freezing, only drilling live or only dead trees, and by maintaining at least 3-cm vertical distance from knots, the operator can avoid introducing more errors to the profile.

Other studies investigated drilling needle sharpness and its effects on drilling resistance values for predicting wood density (Sharapov et al. 2018; Gendvilas et al. 2025). When assessing various species and drill speed settings, it was found that these variables impacted the rate of needle wear and changes in drill resistance values (Gendvilas et al. 2025). It was also suggested, due to an uneven blunting of the cutting edges, that the resulting differences could lead to a deviation of the needle’s normally linear path (Sharapov et al. 2018). Deviation from a linear path or altering the angle of insertion will produce erroneous tree ring detection and measurements (Rinn 2012b), although resistance values remained reliable for certain species of softwoods for 600 m of drilling distance and less than 100-m drilling distance for certain hardwood species (Gendvilas et al. 2025).

Another source of potential error is the potential difficulty for resistance drills to precisely detect particularly narrow rings (Orozco-Aguilar et al. 2018; Oh et al. 2019). In a limited study across several species in Korea for dendrochronology purposes, a precision of 1 mm was suggested when using a specific software analytics package from one manufacturer while using a drill from a different manufacturer (Oh et al. 2019). Others have found resistance drills have difficulty detecting differences in early wood and late wood if the ring is less than 2 mm (Orozco-Aguilar et al. 2018). This can cause the resistance drill to underestimate a tree’s age and overestimate its growth rate. Despite these limitations, resistance drills can provide rapid data with reduced time and effort.

It has been suggested that there are 3 primary types of wood profiles that can be observed by resistance drills: temperate conifers, temperate ring-porous, and diffuse-porous/tropical woods (Rinn 2012b). Understanding these profiles is crucial for improving the accuracy of tree ring analysis across various species. The purpose of this study is to assess any differences and similarities in the resistance profiles and corresponding core samples of 3 species: Liquidambar styraciflua, Picea abies, and Quercus palustris.

The results varied among the 3 species. Table 1 contains the results of each tree and the rings counted by both the resistance drill and the core samples. Figures 1, 2, and 3 show a profile of the resistance drill with LOESS regression and its associated core sample for L. styraciflua, P. abies, and Q. palustris respectively. The third tree sampled for Q. palustris was found to be an outlier. Overall, the resistance drill tended to underestimate the number of rings in each sample when compared to the core sample, as was seen in Figure 4. The fourth Q. palustris was also found to have many consecutive annual rings less than 1 mm, and the resistance drill substantially underestimated the number of rings.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 1.

Liquidambar styraciflua tree 4. This is the resistance drill profile (solid line) with the applied LOESS regression (dashed line). The core sample associated with this resistance drill profile is included at the top. The core shows the diffuse porous profile of this species.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 2.

Picea abies tree 4. This is the resistance drill profile (solid line) with the applied LOESS regression (dashed line). The core sample associated with this resistance drill profile is included at the top. The core shows the distinct conifer ring boundaries of this species.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 3.

Quercus palustris tree 4. This is the resistance drill profile (solid line) with the applied LOESS regression (dashed line). The core sample associated with this resistance drill profile is included at the top. The core shows the ring porous profile of this species.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 4.

Scatterplot of rings counted in the core sample (x-axis) and rings detected by the resistance drill with LOESS regression (y-axis).

The average difference in the number of rings was lowest in L. styraciflua and greatest in Q. palustris. Picea abies had the lowest variance, indicating there was more consistency with errors in ring counting, while Q. palustris had substantially larger variability when comparing rings detected with the resistance drill to rings counted from the core samples.

Although Q. palustris had the greatest average difference and largest variability, it also had the largest number of rings less than or equal to 1 mm. This was verified by visually representing the resistance drill data with the locations of ring boundaries from the core samples and focusing on the region with many consecutive rings < 1 mm, as seen in the first 50 mm in Figure 5. This figure demonstrated the resistance drill would sometimes not detect the presence of smaller rings and thus classify multiple small rings as one larger ring.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 5.

Quercus palustris tree 5. This is the resistance drill profile (solid line) with the applied LOESS regression (dashed line). The core sample associated with this resistance drill profile is included at the top. The core shows the diffuse porous profile of this species. Many consecutive rings less than 1 mm in the core and the difficulty of their detection by the resistance drill can be seen in the first 50 mm of the profile.

The resistance drill demonstrated proficiency in detecting regularly spaced rings greater than 1 mm. Rings as small as 0.2 mm were successfully detected using the LOESS regression method. However, when multiple rings smaller than 1 mm were encountered consecutively, the resistance drill tended to lump them into one large ring, as seen throughout Figure 2. This study found that ring size and the 1-mm benchmark significantly influence the drill’s accuracy, whereas the species appeared to have a greater impact on the difference in amplitude between early wood and late wood.

Other methods of signal smoothing and cleaning have been successfully used: a Savitzky–Golay filter was used to remove the high and low frequency noise from resistance drill profiles in 5 conifer species (Xu et al. 2024). Another study, focused on Pinus tabuliformis, used a linear threshold method to automatically extract annual ring location and width (Li et al. 2024).

Some studies have made use of IML software to manually assess the resistance drill graphs and ring measurements (Orozco-Aguilar et al. 2018; Szewczyk et al. 2018; Oh et al. 2019). These studies included both conifer and deciduous tree species. All studies found that while reliable for assessing tree age and tree ring growth rates at the population level, due to the difficulties detecting smaller rings and subsequently underestimating tree age, this machine should not be used as a replacement for dendrochronology of individual trees.

In this study, the drill generally underestimated the number of rings present, as shown in Table 1 and Figure 2, where the resistance drill underestimated the number of rings in 9 out of the 15 samples. Underestimation could have occurred due to the needle deflecting and following a nonlinear path or drilling at an angle not reaching pith and missing rings along the drilling path (Rinn 2012b, 2013). There were also instances of rings not being detected due to the resistance drill profile not passing through the LOESS regression line, despite the resistance drill showing a clear dip in the region of a ring, as can be seen at 32-mm point in Figure 1. This could mean the LOESS regression might not be suitable for use in all tree species.

Quercus palustris tree 3 was identified as an outlier. The core sample for this tree showed numerous consecutive visual rings based on vessel distribution patterns and a large total number of visual rings compared to other trees sampled in the same area. Although the total number of visual rings on both sides of pith was approximately equal to the estimated total rings expected based on the age of the tree, the growth rate was substantially smaller than all other trees sampled.

The drill settings selected were based on operator knowledge of the machine’s limitations. This was done without a formal analysis of the settings’ potential impact on the sensitivity of the machine’s ability to detect annual rings. It has been recommended to use settings which achieve a stable range of 40% to 60% feed amplitude for tree ring analysis (Oh et al. 2019). However, it may be best to assess feed and rotation settings on a species level, as tree ring detection and analysis is best with high variability of amplitude between early wood and late wood with respect to protection of the drill’s motor (Sharapov et al. 2019; Gendvilas et al. 2024).

Various anomalies within the trees themselves could also cause an erroneous detection or missing of an annual ring. Unlike hand coring, a decay pocket may not be noticed by the user until the data is being processed or may be missed entirely if rings are being automatically detected using a regression. Furthermore, additional analysis is limited due to only 5 samples from 3 species having been tested, and individual species cannot represent the entirety of a wood type.

This study highlights the potential of resistance drills as an alternative method for tree ring analysis. The resistance drill demonstrated proficiency in detecting annual rings greater than 1 mm, with LOESS regression methods successfully identifying rings as small as 0.2 mm. However, the tendency to group multiple smaller rings into one larger ring remains a limitation, particularly when consecutive rings are less than 1 mm. While resistance drills showed consistent results across species, the variability observed in Q. palustris underscores the influence of individual tree characteristics and the challenges posed by smaller rings.

These findings agree with the suggestion that resistance drills are best suited for population-level assessments, where efficiency and reduced invasiveness could be prioritized over individual precision (Orozco-Aguilar et al. 2018). This study also supports the need for expanding optimal drill settings to other species to maximize amplitude differences in ring boundaries for clearer visual reference while minimizing potential damage to the drill from overloading the motor.

When a user considers the optimal method for assessing the data, it’s important to note that applying regressions to automatically extract annual rings is faster than manually extracting the annual rings from resistance drill data. However, automatic extraction removes user subjectivity, which could lead to inadvertently including or excluding annual rings. It may be an acceptable tradeoff to use automatic methods where data processing time is a high priority (Downes et al. 2018). When using manual methods, it may also be beneficial to assess various smoothing techniques to remove background noise and assist the user by making peaks from annual rings more easily recognized. Future work should also explore a broader range of species and evaluate alternative analytical approaches to address the limitations identified.

There are limited studies which assess the application of different methods of data extraction and their effectiveness based on the species or wood type. Future avenues for advancing the accuracy of resistance drills would be to determine if different regression methods are more or less accurate for specific species as well as a comparison of automatic vs. manual data assessment.

In summary, resistance drills hold promise as a complementary tool in dendrochronology, balancing the trade-off between accuracy, efficiency, and tree health. By building on these findings, researchers can further integrate resistance drills into environmental studies, expanding their applicability and reliability.

The authors reported no conflicts of interest.