1. Introduction
Rainfall interception can be observed when precipitation falls above ground that is covered with vegetation. When rainfall reaches the tree canopy, some raindrops fall directly through the gaps in the canopy, while the rest are retained on leaves and branches. Drops retained in the canopy can later reach the ground by dripping, or can evaporate back into the atmosphere. The latter is known as rainfall interception or interception loss. The precipitation can reach the ground by flowing down the branches and stems (i.e., stemflow), or by falling through or from the canopy (i.e., throughfall).
Throughfall itself consists of three components: free throughfall, drip, and splash [
1,
2,
3,
4]. Free throughfall falls directly through the gaps in the tree canopy, has no contact with leaves or branches, and maintains the drop size distribution of an open rainfall [
1]. Dripping describes rainfall drops captured in the canopy that fall to the floor mainly due to saturation of the canopy, while splash describes drops falling from the canopy due to external influences such as wind or rainfall [
5]. Throughfall splash droplets are usually smaller than drops induced by dripping. Most often, a throughfall drop diameter of 1.5 mm has been measured as the threshold value determining a specific throughfall component, as smaller drops are splash-induced and larger drops are generated by dripping [
3,
6].
The amount of throughfall is influenced by various variables. In general, they can be divided into vegetation or meteorological variables. Some of the vegetation variables are phenophase, canopy storage capacity, bark structure, leaf area index (LAI), stem diameter at breast height (DBH), and projected canopy area, [
7,
8,
9,
10]. Meteorological variables are, among others, the amount of rainfall, the duration of a rainfall event, rainfall intensity, wind velocity and direction, air temperature, humidity, and the properties of raindrops [
11,
12,
13,
14,
15].
Raindrop properties, deffined by drop diameter and velocity, are used to describe the discrete nature of precipitation, which is often neglected in research connected with rainfall [
16]. The most important processes influenced by drop size distribution (DSD) are rainfall interception by vegetation canopies [
3,
4,
5,
6,
17,
18,
19] and soil erosion through raindrop impact [
20,
21,
22].
The processes of rainfall interception, throughfall, and stemflow depend on the various properties of raindrops [
17,
18]. The focus of past studies has been mainly on the investigation of throughfall DSD under trees [
5]. Throughfall has different DSDs under canopies of various tree species [
3], because the drips from different types of leaves or needles produce different throughfall DSDs [
23]. Splash and storage from leaves depend also on drop diameter of rainfall in the open and on leaf inclination [
4]. Additionally, throughfall drops are smaller under severe canopy vibrations, due to the high wind speed and high rainfall intensity [
3]. Throughfall drop characteristics—maximum diameter and median volume diameter—are also smaller when the canopy is foliated, as the phenophase is one of the most significant variables influencing throughfall DSD [
6].
However, the DSD of the rainfall in the open was rarely taken into account in earlier studies of rainfall partitioning (interception, throughfall, and stemflow). During the development of a stochastic interception model, Calder [
17] showed that the wetting of the canopies depends on the volume of the individual raindrops, as small raindrops increase the amount of water retained in the canopy. The influence of raindrop size on the predicted interception amount was also observed by Hall [
18]. Open rainfall DSDs were measured by Zabret et al. [
24], who showed that during a single event, an increase in drop diameter immediately increased throughfall under pine trees, which can, under specific conditions, exceed the amount of rainfall in the open.
The microstructure of rainfall varies in time and in space, and differs from event to event. It significantly influences parts of the hydrological cycle, yet it is often overlooked [
16]. Its influence on rainfall partitioning has been mainly analysed from the soil erosion perspective, in studies generally focused on throughfall DSD. Therefore, the main objectives of this study are as follows: (i) to analyse the influence of drop diameter and velocity of open rainfall on the amount of throughfall, (ii) to evaluate this influence for two tree species (i.e., birch and pine), and (iii) to evaluate this influence on rainfall events with different properties (e.g., rainfall amount, intensity).
4. Discussion
The results of this study demonstrate that during 236 rainfall events, throughfall accounted for 73% and 53% of rainfall under the observed birch and pine trees, respectively (
Figure 2). Similar results were reported by other researchers. Throughfall under urban deciduous trees was measured at 71% of rainfall under a blue gum tree in Melbourne, Australia [
35], and 74% of rainfall under a ginkgo tree in California, U.S. [
38]. Additionally, throughfall in a residential urban forest in North Carolina, U.S., ranged between 78% and 89% of rainfall [
39]. These values correspond well to the results of throughfall under the birch trees in the present study. Furthermore, the results of throughfall under the pine trees in this study agree with the annual proportions of throughfall ranging from 51% to 78% of rainfall in a pine forest in northern Britain [
40], and with the average throughfall of 66% in red pine plantations in southern Ontario, Canada [
41]. Similar values were also measured for coniferous Douglas fir trees and western red cedar trees in an urban area of British Columbia, Canada, for which throughfall resulted in 50% and 46% of rainfall, respectively [
42].
For the three analysed rainfall events (
Table 3), throughfall under the birch trees was lower than the average for this study (
Figure 2). However, throughfall under pine trees was higher than the average value during events B and C, and lower during event A. The duration of events B and C is almost equal, whereas event A was the shortest among the three considered events, and had a lower rainfall intensity (
Table 3). The increase of throughfall with a longer rainfall event duration and higher rainfall intensity was also observed in other studies [
11,
14,
15,
43,
44]. Additionally, the majority of rainfall was delivered at the beginning of events B and C, whereas the rainfall during event A was more evenly distributed (
Figure 3,
Figure 4 and
Figure 5). The high amount of rainfall at the beginning of the event resulted in a faster saturation of the tree canopy. Although drainage starts prior the complete canopy saturation, as reported by Klassen et al. [
45], it increases afterwards, resulting in higher throughfall during the rest of the event. The same was demonstrated also by Andre et al. [
13].
A detailed analysis of drop size distribution (DSD) demonstrates that the occurrence of larger raindrop diameters and higher drop velocities during all three events induced changes in throughfall under the pine trees (
Figure 3,
Figure 4 and
Figure 5). This corresponds with the findings of Calder [
17], Calder et al. [
46], and Hall et al. [
47], which found that throughfall is higher for larger raindrop diameters. A five-min time interval with larger drop diameters and velocities was observed more than 10 min after the beginning of events A and B. That interval increased throughfall under the pine trees by 47% and 25% during events A and B, respectively. Larger drop size distribution during event C was observed sooner, from minutes 4 to 7, as well as increased throughfall under the pine trees by 26%. A similar increase of 28% and 20% of throughfall under pine trees, after the onset of larger and faster raindrops during June and July 2014 events, respectively, was reported also by Zabret et al. [
24]. On the other hand, no influence from changes in the drop size spectrum on throughfall under the birch trees was observed in this study.
Analysis of event A demonstrates that an increase in drop diameter and velocity might have started the throughfall under both species of trees. During event B, throughfall under both the birch and pine trees started after 5 min, i.e., when 2 mm of rainfall had fallen. Throughfall under the pine trees was induced by 3 mm of rainfall, and under the birch trees by 7.8 mm, which were measured 4 and 7 min after the beginning of event C, respectively. However, during event A, throughfall under the birch trees started 12 min after the beginning of the rainfall event, while throughfall under the pine trees started after 14 min. At that time, the first increase in drop size spectra was observed (
Figure 3). The rainfall amount needed to induce throughfall during this event was only 1.2 mm and 1.8 mm for the birch and pine trees, respectively. A similar phenomenon of inducing throughfall by a cloud of larger drops was also shown in other studies [
3,
24]. For example, during the July 2014 event, throughfall under the pine trees started after 40 min from the beginning of the event, simultaneously with a drop diameter increase, when 1.8 mm of rainfall fell as well [
24].
The duration and rainfall distributions of events B and C were similar, as they were 69 and 71 min long, respectively, and the majority of rainfall was concentrated in the first 20 min of the events. However, event B delivered 15.2 mm of rainfall, and event C 33.4 mm of rainfall. Throughfall under the pine trees after the occurrence of larger and faster drops immediately increased in both cases, but it exceeded the amount of rainfall in the open only in the case of event B (
Figure 4). A similar phenomenon was also reported for the June 2014 event, which was 75 min long and delivered 10.2 mm of rainfall [
24]. This may be the consequence of the time delay in the increase in the drop size spectrum and the amount of rainfall delivered until then (
Figure 4 and
Figure 5). For event C, during which throughfall under the pine trees remained lower than the rainfall in the open, an increase in the drop diameter and velocities was observed 4 min after the beginning of the rainfall event, when 2 mm of rainfall had fallen and the tree canopy was not yet fully saturated. To compare to other studies, water storage capacity estimated for coniferous trees ranges up to 2.7 mm for red pine [
48] and up to 2.8 mm for sitka spruce canopy [
49]. However, an increase in drop size distribution during event B was observed after 13 min, when 6 mm of rainfall had already fallen. Additionally, drop diameter and velocity increase during the June 2014 event occurred after 15 min, when 4.6 mm of rainfall had fallen [
24]. By that time, the canopy storage capacity was full. Nanko et al. [
3] reported that the occurrence of larger and faster raindrops induces vibrations of the canopy, which is similar to wind stimulating dripping of stored water, and significantly increases throughfall, a fact which has also been reported by other authors [
3,
13,
38].
The increase in drop size and velocity spectrum influenced throughfall under the pine trees for each analysed event; however, no such response was observed for the birch trees. During all three considered events, the amount of throughfall under the birch trees maintained the same quantity from the start, and also when the rainfall began to decline. After that, the throughfall under the birch trees exceeded the rainfall in the open (
Figure 3,
Figure 4 and
Figure 5). Therefore, we can assume that the throughfall under the birch trees was the result of a saturated canopy, rather than the changes in drop diameter and velocity. The different response of pine and birch trees may be a result of different tree properties (
Table 1). Nanko et al. [
3] indicated that broad-leafed trees, such as sawtooth oak, have a larger water storage area per leaf than needle-leafed trees. Also, storage capacity of leaves was measured to be higher for larger drops than for smaller ones, as reported by other researchers [
4,
50]. Therefore, the birch tree (as a deciduous tree) retains larger drops on leaves, while pines react to drops with larger diameters with increased throughfall. Andre et al. [
13] reported that larger canopy storage capacity is also related to larger leaf area index (LAI), which was higher for birch than for pine trees during the analysed events in our study (
Table 1). Additionally, higher branch inclination increases throughfall, which is consistent with the results of Bassette and Bussiere [
4], as birch tree branches are oriented towards the sky, while pine tree branches lean towards the ground (
Table 1). Due to downwards branch inclination, intercepted rainfall may laterally translocate towards the canopy edge [
51], which may lead to the occurrence of drip points where throughfall exceeds 100% [
14,
52,
53]. However, the lateral translocation of the captured rain seems not to have had a significant influence on throughfall under the pine trees in our study. Furthermore, the properties of branches and stem are important, as Llorens and Gallart [
48] showed that the specific water retention capacity of branches and stem is six times higher than that of needles.
5. Conclusions
Continuous measurements of rainfall, throughfall, and drop size distribution (DSD) over three and a half years provided enough data to analyse the influence of open rainfall microstructure on throughfall. Similar to previous studies on interception, we showed that throughfall increases with the rainfall amount and duration. Additionally, rainfall distribution during the event influences the amount of throughfall. We demonstrated that an increase in drop diameter and velocity on the event scale is connected with an increase of throughfall under pine trees, whereas birch trees did not show any response. During event A, a 4% increase of average drop diameter and a 6% increase of average drop velocity increased throughfall under the pine trees by 32%. Birch trees have different properties to pine trees, which probably determined their response to changes in DSD. Birch trees’ larger leaf area index (LAI), upwards branch inclination, smooth bark, and finally, possession of leaves instead of needles, eliminated the effect of larger and faster raindrops. However, when the beginning of an event involved low intensity rainfall, the increase in drop velocity and diameter induced throughfall under both tree species. In comparison to events B and C, the late start of throughfall in event A, triggered only by 1.2 mm of rainfall, was observed simultaneously as an increase in drop size distribution.
The results of this study indicate that rainfall microstructure is an important variable influencing throughfall dynamics. The differences in the response of these two distinct tree species suggest that future research on the influence of open rainfall DSD on the throughfall amount and dynamics is needed. This may help to understand how rainfall DSD in various climate zones influences throughfall dynamics, and how it is affected by different tree canopy characteristics.