The tree is an essential natural resource for ecosystem balance. It regulates nature, climate and urban greening. Trees play a key role in local biodiversity as they release oxygen and reduce global warming. Trees regulate climate by mitigating urban heat islands. Trees behave as barriers for noise pollution and wind. They provide moisture to the atmosphere, which favours precipitation. They facilitate water infiltration into the soil contributing to the formation and maintenance of groundwater aquifers. They are also essential for soil building as their roots fix the soil preventing erosion. All trees are of importance, whether by their age, type, size and shape. Some of them are classified as remarkable or even monumental trees, and the law protects them [1
]. Therefore, it is fundamental to act for the preservation and sustainability of these living beings. This is a current social challenge [2
]. It is decisive to move from the anthropocentric conception of trees to ecocentric conception, that is, all living beings including humankind are interrelated and interdependent to keep the ecosphere equilibrium.
Contrary to what people thought for many years, trees react to physical and biological damages that lead to deterioration. The compartmentalization of decay in trees (CODIT) is a good example of this [6
]. Therefore, trees must be cared for in a way that enhances their own defence systems [6
]. In order to verify the health status of trees especially when more detailed information is required, invasive and even destructive techniques [7
] are used. Unfortunately, they interfere with the structural integrity of this living being. Therefore, it is urgent to disseminate and implement non-aggressive inspection techniques that preserve biological integrity and functionality [3
]. The basic rule must always be to start with less invasive techniques and only if required, to utilise the most aggressive ones, so that the damage produced in the tree [4
] is minimized.
Infrared thermography (IRT) is a non-invasive non-contact technique that relies on the detection of body heat emission [9
]. It measures continuously surface temperature in real-time [9
]. Aerial surveillance of the canopies to detect the distribution and spread of forest damage was the first use of IRT in trees [4
]. Later, this technique was applied to tree bark (branches and trunks). Now, it allows the evaluation of some types of damage in the lower trunk and to estimate the presumable cause that affects roots (the root system) [4
]. The latest thermal cameras capture thermal images of high resolution and sensitivity. It has been demonstrated that IRT in conjunction with visual inspection when applied to assessment, and monitoring of tree health provides reliable data [2
]. The IRT capabilities for tree inspection have not been exploited sufficiently yet, since the elaborate sap conductive tissues are practically on the surface of the living structure. The reading of tree surface temperature reveals specific variations when there is deterioration and voids inside the tree. The IRT observes the tree as a functional whole body and thermograms provide information to differentiate deteriorated tissues from healthy tissues. Other methods do not have this assessment capacity, much less the naked eye inspection. Other non-invasive methods survey the body by points, and then extrapolation applies to have an idea of the functional whole body. Through the naked eye, only advanced deterioration is detected, then a corrective measure is hardly effective, and the solution is tree felling.
Thermography allows the early detection of damage, while it is still not visually noticeable. Even more, it monitors the progress of pathology. The IRT is expeditious and non-invasive [2
]. It is, therefore, a powerful, fast and efficient tool to detect changes in the integrity of trees and branches, identifying if one of them should be removed. The differences in the thermal patterns of the tree surface indicate the deteriorated areas of the tree. The greater the differences in the thermal patterns of the trunk and branches, the worse the tree health condition [2
The methods different from IRT that acquire data for diagnosis are time-consuming, and require more hand labour, especially if the part of the tree to be examined cannot be reached from the ground [2
]. Some techniques, such as the resistograph, require perforation, and these holes may become pathways for pathogens [2
]. Other methods use X-rays or γ. A short summary of the other main methods for detecting tree deterioration is presented in Table 1
. The ionising radiation used in some methods conveys the perception that they are not safe for the health of living beings [3
]. Moreover, when IRT is compared to more sophisticated techniques, such as X-ray and γ, as well as tomographic acoustic techniques such as Picus and ArborSonic 3D, or even nuclear magnetic resonance [7
], IRT is the only one able to evaluate the health condition and functionality of tree tissues. That is, the assessment of the structural integrity to detect voids and deterioration inside the tree is possible because the tree is analysed as a functional whole body [4
]. IRT analyses the tree as a whole, in a holistic way, while other techniques provide information only on specified points, and the whole is obtained by extrapolation after a series of investigations [2
]. The authors raise concerns about avoiding errors when selecting the emissivity value for temperature reading calibration. However, more parameters are required for IRT quantitative readings. It is necessary to assess the value of reflected (or reflective) apparent temperature, which varies according to the angle between the camera and the tree surface, and the direction of radiation from the environment and sunrays. In some studies, the tree is cut into logs, and then the logs are perforated to simulate deterioration after the logs are sealed at both sides to maintain the water content. When observing the logs in thermograms, it is difficult to detect the holes and the larvae introduced inside them. Even with the same water content, the temperature of the logs does not behave like the tree because there is no flow of the sap, as this happens when the tree structure is alive [16
]. Some studies analyse trees of the same species affected by several pathologies. The main aim was to look for similar temperature patterns on the surface all along the trees applying IRT [18
]. The statistical analysis showed that there was no correlation [18
Considering the above, IRT is a well-established technique in a range of fields, such as industrial and building maintenance. Despite its merits, it is still a relatively recent technique in assessing tree health [4
] and remains residually implemented in agriculture. Due to the relatively scarce studies in this field [5
], more research to ensure its potential and applicability on a large scale is required. Thus, it is crucial more studies focus on contributing to show and analyse the complexity of the technique applied to trees, the new and atypical aspects of the problem and respond to some knowledge gaps. Accordingly, the present paper intends to detail some relevant features related to the applicability of the technology, thus contributing to turning the IRT technique as a decision-making tool to assess the health status of trees. To accomplish this, two sample trees, one of species Quercus pyrenaica
Willd and another of species Olea europaea
L., are qualitatively analysed from their thermal images.
The tree inspection was carried out applying qualitative IRT at the passive mode, that is, the heat source was the environment through solar radiation. The heat flows from the hotter to the colder zones of the tree. The irregularities of the thermal pattern on the tree surface can indicate the existence of defects, voids and deteriorated tissue. When it is found, deterioration as voids and defects occur in both the thermal properties of the constituents and the heat transfer process [2
Thermograms were recorded at different times along the day, from when direct solar radiation was hitting the trees until after sunset. Besides the thermograms, photographs were taken to support the visual inspection and thermogram interpretation. The thermograms were processed running FLIR software [23
]. The authors marked the spots at different locations over the tree trunk. The thermal patterns were analysed in order to correlate the temperature distribution with tree health. The atmospheric temperature and the relative humidity were measured at the time of the recording pf each thermogram, as well as the observation distance (between the thermal camera and the tree).
From a qualitative approach, although important, the value of emissivity is not preponderant. In fact, what is at issue is not the exact value of the measured temperature, but the temperature differences among the spots. The temperature values are comparable because the camera measured them with the same emissivity and reflected temperature.
Nevertheless, the emissivity value introduced was the one appropriate to the surface of the element under study. As the wood and bark are little reflective materials, the emissivity is high. Then, a high emissivity value was set in the camera. Thus, for both specimens, the emissivity value was 0.95. For each thermogram, the atmospheric temperature at the time of the capture was set in the camera. As it was a qualitative analysis, this study did not consider the reflected temperature. Further, it did not rain more than one week before surveying, then the thermograms did not register noises due to the humidity factor.
From the qualitative analysis, a more quantitative approach was taken, in particular, as regards to the comparison of recorded temperature values. It is important to understand thermal pattern differences among the trunks of similar diameter. The thermal patterns resulted from differences in the temperature along the surface. Naturally, this process introduces systematic errors that affect all points equally in each thermogram, whereby do not affect the result.
4. Results and Discussion
shows the thermogram of the Quercus pyrenaica
Willd tree recorded from the same point of view and at the same time as the photo shown in Figure 4
(on 21 January 2018, 2 p.m., winter in the northern hemisphere). The southern part of the tree (left side of the thermogram) is exposed to sunlight. Figure 7
shows another thermogram of the same tree at the same day but it was taken at 9:45 p.m. (4 h after sunset) from the same point of view. Table 3
shows the observation conditions, and the parameters set in the camera to capture the thermograms. Table 4
contains the temperatures of the spots marked in the thermograms.
shows areas under direct sun exposure (left zone of the tree) and areas in the shadow (right of the tree). On the left of the trunk is Sp1 under direct sun exposure at 32.5 °C, while Sp2 is in the shadow at 18 °C. The Sp3, on the bottom right, with 12 °C, point out an old cut trunk. Sp4 is in the shadow at 15 °C on a branch of larger diameter than the branch where Sp5 is located. Sp5 is under direct sun exposure at 23.5 °C.
shows the thermogram of the same tree captured 4 h after sunset. It can be seen that Sp1 is on the left trunk at 8.5 °C and Sp2 is on the central trunk at 8 °C. Sp3 is at 3 °C on the stump. Sp4 at 6 °C is located on a branch of larger diameter than the branch of Sp5. Sp5 is at 4.5 °C. The observation conditions and the parameters assumed to capture the thermogram are in Table 3
. Table 4
shows the spot temperatures in the thermogram.
shows a photograph (A) and a thermogram (B) of the tree Quercus pyrenaica
Willd taken at the same time. The capture angle (west) is different from previous images. A red arrow (B) shows a stump that remained from an old cut trunk identified as Sp3 in the thermograms of Figure 6
and Figure 7
shows the thermogram of the tree of species Olea europaea
L. described in Figure 5
and Table 2
. Table 5
shows the environmental conditions during the observation, and the parameters assumed. The spot temperatures in Figure 9
are detailed in Table 6
. The thermogram was recorded after sunset on a winter day (on 21 January 2018, at 5:50 p.m.; sunset at 5:35 p.m.). The atmospheric temperature was 18.0 °C and the relative humidity was 50%. The atmospheric temperature was 18.0 °C and the relative humidity was 50%. The thermogram was taken a distance of 5 m, using an emissivity of 0.95. A rainbow colour palette and a temperature range of 5.0 °C to 20.0 °C were used.
The Olea europaea
L. tree shows a large crack identified as Sp1 in the thermogram of Figure 9
. Sp1 temperature is slightly higher than the atmospheric temperature. This spot represents a zone of the tree that has been exposed to the sun for a longer time. Sp2 temperature is much lower than Sp1 temperature—the difference between them is 10 °C. The differences in the temperature at the same diameter trunk is an indicator of possible deterioration. Then, it is possible that the Sp2 zone has deteriorated. However, this zone was in the shadow for a longer time. In addition, Sp1 is in a crack area. Thus, the comparison of these two spots is not conclusive, but it reveals a clue of possible deterioration to take into account.
In the same Figure 9
, Sp2 is at a much lower temperature than Sp3. These spots are comparable because they are located on trunks of similar calibre. On the other hand, Sp2 was in the shadow for some time. Thus, the comparison is not conclusive, but it reveals a new clue to take into account for possible deterioration. Then, this study analysed the temperature of Sp2 in relation to the atmospheric temperature. The Sp2 temperature of 8.5 °C is much lower than the atmospheric temperature of 18 °C. This condition strengthens the possibility of deterioration at Sp2. Note that Sp3 is in the middle of the main trunk receiving sunlight. Although it already lost temperature, it was at 14 °C.
Sp4 in Figure 9
is at 10 °C, even exposed to direct sunlight, and it is on the same trunk as Sp3, which is at 14 °C. The two of them are located in trunks of the same diameter. The difference of 4 °C between the two points is a strong indicator of deterioration in Sp4. When comparing the temperature of Sp4 with the atmospheric temperature, the difference is 8 °C, which is relevant since Sp4 was exposed to direct solar radiation until just before the thermogram was recorded. This is another reason that indicates deterioration. Sp4 is approximately at the opening of an orifice. Then, it was expected the temperature to be higher than the registered temperature, and still higher than the values of the boundary zones, as is the case at the slot of Sp1. A probable cause for unhealthy conditions in that area is the hole at Sp4 that stores water from precipitation.
Sp5, Sp6 and Sp7 (Figure 9
) are on branches of approximately the same diameter. They are at lower temperatures than the main trunk, because they are of smaller calibre. The trunks of smaller calibre heat and cool faster than larger ones. Smaller trunks have a larger surface per volume unit, so they heat and cool faster. The thermal inertia of the trunks of larger diameter is greater because they have more mass. Thus, from the point of view of health, Sp5, Sp6 and Sp7 should be at approximately the same temperature if subjected to the same solar exposure. Sp7 should have a higher temperature since it received directly more sunlight. This did not happen. Sp6 temperature is 1 °C higher than Sp7, which is not significant. The most disturbing is Sp5, because it is at 9 ° C. This value is too low for a healthy area even if exposed to less time in the sun. Sp5 is at a much lower temperature than the atmospheric temperature, which is 18 °C. This is a strong indicator of possible deterioration.
Sp2, Sp4 and Sp5 conditions (Figure 9
) strongly indicate possible deterioration. This strong evidence was confirmed with observations of an enormous interior cavity linked to the crack (Figure 10
For doubtful cases, more thermography analysis would be recommended, e.g., at night, and if doubt remained, these spots become flags for the application of other diagnostic techniques and methods. Note that, these other diagnostic techniques do not need to be applied to the whole tree, but only in the identified points.
Epiphytic vegetation, such as lichens, is distributed more or less homogeneously throughout the tree and should not be responsible for the differences in the thermal pattern presented.
In IRT, as in all existing techniques, there are several conditions to take into account in order to achieve reliable results on tree analysis. That is: Exposure to sunlight and shadow; thermal contrast between the environment and the object targeted; the absence of water like rainfall; vegetation covers such as mosses and lichens; typical bark patterns of each species; the thermal comparison between the trunks of different calibre within the same tree [2
Thermographic observations to trees are often made against the sun, or to tree surfaces when they are under the incidence of solar radiation. Thermographic recordings of trees when the sun is in front of the camera and sunlight directly hitting the tree surface introduces recording noise. In these cases, the temperature differences are more influenced by reading errors (lens, light exposure and reflections) than by material properties and defects. The shaded or less-illuminated areas (lower direct solar exposure) exhibit lower temperature values than expected, suggesting that the tree has deteriorated [2
]. The IRT application requires a strong thermal contrast between the object to observe the environment and the objects that surround it. There must be a significant difference between the radiative power of the environment and the object analysed [9
The thermograms and photographs presented are illustrative of relevant aspects to be taken into account in the observation and thermographic analysis of tree salubrity. It can be seen that on the thermogram of Figure 6
, Sp1 and Sp5 are exposed to direct sunlight, and they are at a higher temperature than the atmospheric temperature, whereas the spots in the shadow are at a lower temperature than the atmospheric temperature. On the thermogram of Figure 7
, all temperatures are lower than the atmospheric temperature. The colour patterns of thermograms of Figure 6
and Figure 7
appear different. The various pattern colours correspond to different temperatures in the thermogram as shown in Table 4
. The temperature differences by themselves do not indicate deterioration, as they can result from sunlight exposure and shadows. Figure 7
shows a general homogeneous temperature distribution, in which the different shades of Sp3 in Figure 7
is on a tree stump. Even its diameter is larger than the other trunks and it is at the lowest temperature and considered lifeless, and therefore not functional. It is probably because it does not circulate sap and its water content is much lower than its surroundings. In the upper part of the tree, the branches are of smaller diameter, as in Sp5. The branches heat and cool faster. Therefore, at 2 p.m. (under the action of sun exposure) Sp3 is at a higher temperature than Sp4, and at 9:45 p.m. (night), the opposite is registered.
Under closer inspection, there are subtle differences of the temperature in the thermograms, which result from the bark pattern, pruning wounds, and epiphytic vegetation such as mosses and lichens. They appear as spots of slightly lower temperature than the adjacent ones. In any of these cases, these small thermal contrasts are not deterioration signals. Another case that leads to some doubts is the self-defence process (CODIT) that trees develop.
It is important to point out that thermographic inspection requires photographs because they are visual inspection tools (VTA) that facilitate the interpretation of thermograms. By improving the interpretation of the results, it is often possible to avoid invasive methods [8
Finally, it is relevant to highlight the main merits and weaknesses of the IRT technique. IRT has an enormous capacity to analyse the trees as a whole and differentiate functional tissue from dysfunctional tissue. This is crucial for the inspection of the vitality and health status, representing a fast, economical, nondestructive and environmentally friendly monitoring tool. However, as in other non-invasive methods, the main limitation is that IRT does not identify specifically the damage detected, i.e., does not identify the pathology, nor its causative agent. Nor can it also give precise indications of the magnitude of the damage. More studies are needed to optimise the technology and training, in order to make the system even more efficient and reliable.