UAVs Technology as a Complementary Tool in Post-Fire Vegetation Recovery Surveys in Mediterranean Fire-Prone Forests

Abstract

Forest fire is a recurring and serious environmental hazard, which is often due to the interaction between anthropogenic activities and climate change, despite having always characterized the vegetation landscape in the Mediterranean area. Alongside the required prevention and control works, there is an increasing need for post-fire monitoring. This is particularly relevant when it comes to natural or semi-natural forests, so that inappropriate reforestation is not undertaken without having well understood the dynamics of self-regeneration and the resilience of pre-existing phytocoenoses to fire. These monitoring operations often take a long time, but a quick picture of the severity of the fire and the response of the vegetation is also required. In this context, the data relating to post-fire surveys on a maritime pine forest of Monte Pisano (northern Tuscany, Italy), obtained from ground surveys and drone shots, are reported. This investigation showed: (1) UAV technology has proved to be very useful and fast, and it allows a good identification of tree/shrub species and forest structural parameters. (2) In these forests, repeated fires cause the disappearance of pine woods ad substitution with “pyroclimax” cork oak communities in equilibrium with a regime of repeated fires. (3) These first results, part of an ongoing investigation, contribute to a better understanding of the sylvogenetic processes following the passage of fire and can support the management of burned areas.

1. Introduction

The vegetation fire is one of the most relevant environmental factors that bring important transformations to vegetation landscapes, together with other natural disasters [1]. Fire has certainly been a strong factor in the transformation of the environment for millions of years [2], and it must be considered a natural element in shaping the vegetation landscape of the Mediterranean area and beyond since ancient times [3,4,5]. This results in a selection of species and phytocoenoses adapted to the periodic repetition of fires [6]. Nevertheless, wildfires in the Mediterranean and European areas are almost exclusively of anthropogenic origin [7,8], and it is an element of serious devastation in areas of great naturalistic value [9]. Furthermore, forest fires have affected regions in central and northern Europe that are not prone to fires, even though the burnt areas of the Mediterranean region have decreased slightly from 1980 to now [10,11].

This enlargement of the geographical area of fire can be linked to climatic change, which may represent a problematic new increase of the phenomenon [12] and of the risk of large fires in the Mediterranean area [13]. In this context, there are numerous studies on the flammability and susceptibility of different species and types of vegetation to fire, their ecological response, and the dynamics of post-fire regeneration [14,15,16,17,18]. Regarding this last aspect, all operations and methodologies relating to post-fire monitoring assume a particular significance. In this type of operation, both remote sensing and ground sensing, on a local scale and/or broad scale, are carried out, concerning the width of the burned surfaces as well.

Remote sensing is often based on satellite images or aerial photos, mainly with the use of false color and multispectral images to classify the severity of fires, the levels of regeneration of vegetative biomass, and the coverage rates of vegetation on the ground, at the level of small–medium scale [19,20,21]. The use of land surveying is necessary, however, to identify, at a large-scale level, the recovery process in terms of floristic composition, vegetation physiognomy, phytocoenosis structure, and species demography, which are diagnostic elements that are difficult to identify remotely. However, at the same time, surveys on the ground often take a long time due to the difficulties associated with the unreachability/roughness of places (for example, in areas with strong steepness).

In recent times, the use of high-definition images from UAVs has proven to be an important support for ground operations in agroforestry, but also in naturalistic fields. In fact, UAV technology enables the identification and mapping of numerous vegetational elements and forest species based on orthophotos and 3D models, thanks to dedicated software. In addition to this, UAVs have many advantages that make them suitable for forest use: low operating and material costs, high-intensity data collection, and the ability to house different types of sensors. Despite the advantages considered above, most UAVs still have some limitations: limited flight time (and consequently, small detectable areas), limited load capacity of technologies not yet adequately miniaturized, and finally, the user’s need for high-capacity data-processing systems [22,23,24,25,26]. However, these features still largely depend on the flight altitude used, the photographic resolution obtained, the image spectrum used, the morphology of the ground, and the vegetation structure [27].

Monitoring post-fire vegetation regrowth is important for understanding forest regeneration and detecting change in post-fire plant communities [16]. In this case study, results related to post-fire monitoring carried out with both ground-based floristic-vegetational surveys and with UAV technology, on repeatedly burned forest areas of Monte Pisano, north-western Tuscany (Italy) are reported.

This investigation aims to monitor the post-fire vegetation recovery process and to verify the actual complementarity of UAV technology to the ground vegetation surveys, comparing plots detected remotely and from the ground.

This not only in order to understand the dynamics of post-fire regeneration, but also the resilience of the different forest communities. The preliminary results obtained, comparing the remote and ground surveys, highlighted a notable capturing of the structural and vegetation data detectable on the ground with the use of UAV technology.

2. Material and Methods

2.1. Study Area

The study concerned a slope with a west/south-west exposure (coordinates of the centroid: 43°42′25″ N, 10°32′20″ E), falling within a larger area covered by fire in August 2009 and September 2018, in the southern sector of Monte Pisano, northern Tuscany, Italy (Figure 1). From the geological point of view, the entire slope is characterized by a matrix of quartz and conglomerate rocks (quartzites of M. Serra), which give rise to soils of sub-acidic nature, with almost no depth, in the cacuminal portions of the mountain, and a shallow depth in the watersheds (Tuscany Region, GEOscopio, https://www.regione.toscana.it/-/geoscopio, accessed on 15 March 2021).

As for the climate, the entire Monte Pisano area has a temperate-humid climate with summer aridity, with an average temperature of the coldest month between −3 °C and 18 °C, and that of the warmest month higher than 22 °C, and variable rainfall, depending on the slope and altitude, from 950 to 1250 mm per year [28].

Based on the bioclimatic classification of Pesaresi et al. [29], the study area’s bioclimate belongs to the temperate (submediterranean-oceanic) macro-bioclimate, lower mesotemperate thermotype, and lower humid ombrotype.

Before the 2009 fire, the plant landscape included pine forests of Pinus pinaster and large patches of low-lying post-fire forest [30]. The former consisted of the dominant layer of Pinus pinaster and a dominated layer of Erica arborea, Juniperus communis, Calluna vulgaris, and Arbutus unedo. The latter were characterized by a sparse regeneration of the aforementioned species and sparse covers of Quercus suber, Quercus ilex, Myrtus communis, and Phillyrea angustifolia.

The vegetation of pine forests was of Genisto pilosae-Pinion pinastri alliance (Biondi and Vagge 2015 [31]), while Q.suber phytocoenoses were of Fraxino orni-Quercion ilicis alliance (Biondi, Casavecchia, and Gigante 2003 [30]).

After the fire of 2009, in the spring of 2018, some phytosociological relevés were carried out on the same slope. These surveys revealed in part a fair regeneration of pine and in part a different evolution with a dominant cork oak [32].

Two years after the second fire, two neighboring areas (but with different resprouting) were selected and investigated. The choice of the two areas was determined for the following reasons:

• areas crossed twice by the fire;
• homogeneous vegetation existing before the two fires (maritime pine forests);
• homogeneous sites for morphology, exposure, and substrate;
• different vegetation regrowth after the first fire.

2.2. Remote Sensing and Ground Surveys

Two rectangular areas of about 1 ha have been identified inside forest areas crossed by fire in 2008 and 2018: Area 4: 9000 sqm, height 151–180 m above sea level (43°42′35.87″ N, 10°31′08.79″ E); Area 6: 9800 sqm, height 190–220 m above sea level (43°42′20.37″ N, 10°31′21.75″ E). These areas included or partially included two relevés, each of which were carried out in spring 2018 on the same mountain side (Figure 2).

The incomplete overlap of flights with the previous ground surveys was determined by the presence of high-voltage power lines. Both areas have an average slope of 38° and are approximately 500 m from each other. On 12 and 26 January 2021, two flights were carried out for Area 4 and Area 6, respectively, using a DJI quadcopter, model Phantom 4 Pro. The flights were planned on the lidar data of the Tuscany region, with a 1 m × 1 m mesh, Mission-Hub application, and based on 10 ground control points (GCPs) geo-referenced with Drone GPS. The UAV was equipped with its DJI-RGB camera (visible spectral range), FOV 84° 8.8 mm/24 mm (format 35 mm equivalent), sensor CMOS 1″ for 20 Mb per frame, image size: 3:2, and aspect ratio 5472 × 3648 pix. For flights, a set of 185 and 236 nadir images, respectively, for Area 4 and Area 6 were collected from a height of 25 m above ground level, at a flight speed of 2.5 m/s, with a forward-longitudinal-overlap of 75.5% and a side-overlap of 80.4%. The GSD (ground sampling distance) was 0.7 cm/pix. For the RGB camera settings, the shutter speed was 1/800 s, f was 4.33, and ISO was 100.

The assembly of the geo-referenced orthophotographic mosaic was performed using the Agisoft Metashape 17.5 software. Ten sample plots of 10 × 10 m were selected on the overall orthomosaic. They were identified using a random table based on the numeration of the various frames. These plots are the same ones used subsequently for the ground surveys. The drone images were processed with Agisoft Metashape to obtain the dense cloud points (DCP), the digital elevation model (DEM), and the 3D model (3DM), in addition to the orthophotomosaic (OPM). On these spatial data, for each plot, the physiognomic-structural characteristics were measured semi-automatically. The measurements made by DCP, DEM, and 3DM concerned: (I) the number of cork oak trees (N); (II) height of cork oak trees (H); (III) bare ground coverage (BG); (IV) herbaceous/shrubs layer (H-S) (<2 m) coverage; (V) crown of cork oak trees coverage (C); (VI) total vegetation coverage (TC); (VII) diameter of the cork trunks (DTT) at about 1.30 m from the base; (VIII) the number of burnt pine logs (on the ground) (NBPL); and (IX) diameter of the burnt pine logs (on the ground) (DBPL). OPM was used for expeditious identification of tree/shrub species, which was subsequently verified in the field.

Floristic-vegetational surveys were carried out in May–June 2021 for each plot. The method followed the Braun–Blanquet phytosociological approach [33,34]. In each survey plot, the fidelity of the tree-shrub species identification remotely and on the ground was verified. In addition, for comparison with remote measurements, diameters of the trunks of the cork oak and burnt pine logs were measured (tree caliper, 1.3 m from base). Plant species nomenclature follows Bartolucci et al. [35] and the subsequent updates summarized in the Portal to the Flora of Italy (http://dryades.units.it/floritaly/, accessed on 15 March 2020).

2.3. Data Analysis

One-way ANOVA and multivariate analysis by clustering (Ward’s method) were carried out on the structural data, to check for any differences or similarities between the two areas. The data set has been controlled for normality with Shapiro–Wilk test. Instead, for vegetational data, a multivariate analysis procedure was performed according to the “UPGMA cluster algorithm analysis” and applying the Bray/Curtis similarity coefficient. To detect a possible relationship between the current vegetation cover and the pre-fire pine cover, we applied a linear regression analysis to the total vegetation cover vs. the number of burnt pine logs on the ground. PAST 3.14 software was used for multivariate clustering analysis, and JMP Pro 16 software for one-way ANOVA and linear regression analysis.

3. Results

The reconstruction of the 3D model of the orthophoto mosaic, as well as the division into layers of the cloud points, allowed the measurement of the main physiognomic and structural characteristics of the vegetation (

Table 1. Mean (±S.E.) of structural data from UAV measurements (descriptive statistics): NT = number of trees; HT = height of trees; BG = coverage of bare ground; H-S = coverage of herbs and shrubs; C = coverage of trees crown; TC = total vegetation coverage; DTT = diameter of tree trunk; NBPL = number of burned pine logs (on the ground); DBPL = diameter of burned pine logs (on the ground). All trees are cork oaks; (*) sum of H-S minus overlap; lowercase letters of the first column are plots identifiers.

		NT	HT (m)	BG (sqm)	H-S (sqm)	C (sqm)	TC (sqm)	DTT (cm)	NBPL	DBPL (cm)
Area 4	a	2	2.9	70.6	22.7	3.3	29.4	3.95	13	7.6
	c	0	0	80.9	19.1	0	19.1	0	16	7.2
	d	1	3	81.1	17.2	1.7	18.9	12	16	10.3
	e	1	3.2	80.3	18.2	1.5	19.7	4.5	23	4.2
	f	2	4.5	77.4	20.9	1.05	22.5	6	18	6.4
	g	1	2.5	77.5	21	1.5	22.5	8	12	11.7
	h	1	4	64.6	33.6	1.8	35.4	7	17	6.7
	i	0	0	87	13	0	13	0	14	9.5
	m	2	4.5	76.2	19.4	2.2	23.8	5.2	20	10
	n	0	0	70.2	29.8	0	29.8	0	28	8
		1.4 ± 0.2	3.5 ± 0.3	76.6 ± 2	21.5 ± 1.9	1.8 ± 0.3	22.8 ± 2	6.6 ± 1	17.7 ± 1.5	8.1 ± 0.7
Area 6	a	4	7.4	13.9	36.8	49.3	86.1	19.9	1	31
	b	4	6.2	26.5	31.8	41.7	73.5	24.3	5	25
	c	5	6.5	6.9	39.3	53.8	93.1	24.6	2	19.5
	f	1	8	4.2	75.8	22	95.8	34.9	4	20.3
	h	1	5	29.4	68.1	10	70.6	16.6	7	22.7
	l	0	0	29.9	70.1	0	70.1	0	6	12.5
	n	5	5	12.6	44.8	42.6	87.4	13.6	7	16.1
	p	1	6	30.1	66.1	5	69.9	18	4	21
	r	3	5	5.3	66.9	27.8	94.7	16.6	3	18.5
	u	2	6	4.1	59.7	46.2	95.9	20	5	18
		2.8 ± 0.5	6.1 ± 0.3	16.3 ± 3.6	55.9 ± 5	33.1 ± 5.8	83.7 ± 3.6 (*)	20.9 ± 2	4.4 ± 0.6	20.4 ± 1.5

; Figure 3). Critical values (due to pixel resolution), such as cork logs and burnt pine logs diameters, were consistent with those measured in the field (

Table 2. Mean (±S.E.) of UAV measurements (_UAV) in comparison with ground survey measurements (_GS) (descriptive statistics): DTS = diameter of tree trunk (cork oaks); DBPL = diameter of burnt pine logs on the ground; lowercase letters of the first column are plots identifiers.

		DTS_UAV (cm)	DTS_GS (cm)	DBPL_UAV (cm)	DPBL_GS (cm)
Area 4	a	3.95	4.2	7.6	8.2
	c	0	0	7.2	7.8
	d	12	13.6	10.3	11.5
	e	4.5	5.2	4.2	4.7
	f	6	7.2	6.4	6.6
	g	8	7.5	11.7	12.2
	h	7	6.2	6.7	7.3
	i	0	0	9.5	9.3
	m	5.2	6.5	10	10.5
	n	0	0	8	8.2
		6.6 ± 1	7.2. ± 1	8.1 ± 0.7	8.6 ± 0.7
Area 6	a	19.9	21.9	31	31.5
	b	24.3	25.5	25	25.6
	c	24.6	23.1	19.5	18.7
	f	34.9	36.5	20.3	21
	h	16.6	17.7	22.7	23.5
	l	0	0	12.5	13.1
	n	13.6	12.8	16.1	17.3
	p	18	17.5	21	22.1
	r	16.6	17.8	18.5	19.2
	u	20	21.8	18	18.7
		20.9 ± 2	21.6 ± 2.2	20.4 ± 1.5	21.1 ± 1.6

). The high level of photographic resolution of the orthophotos (0.7 cm pixel) allowed good remote tree and shrub species detection (Figure 4).

Nevertheless, the smaller shrubs, juvenile trees, and herbaceous species, as well as the individuals under the foliage of the dominant layer, have been identified only with direct ground surveys (Table S1).

The one-way ANOVA applied to physiognomic/structural data collected remotely (but in the same way as those verified in the field) highlighted significant differences between the physiognomic and structural characteristics of the phytocoenoses of the two areas (Table 1; Figure 5). The number of individuals of Q. suber, the only tree exceeding 2 m in height, is much more numerous in Area 6 than in Area 4 (26 individuals vs. 10), and the height of the trees in Area 6 is on average double that of those in Area 4. This measure is consequently reflected in the average size of the individuals (crown area and trunk diameter), where there are significant differences in the size of the foliage and the diameter of the trunks. A considerable difference is observed between the two areas in the spatial dimensions of the foliage (33 sqm, Area 6 vs. 1.8 sqm, Area 4) and the diameters of the trunks (20.9 cm, Area 6 vs. 6.6 cm, Area 4) (Table 1; Figure 5).

The data seem to indicate a different age of the cork oak population. In the same way, evident differences are detectable in the covering of the herbaceous-shrubby mantle, which is reflected in the marked difference in the extension of the bare soil. Area 4 is mainly bare, with a very modest herbaceous/shrub layer and a practically absent tree layer. Area 6 has overall coverage levels, ranging from 70% to 95.9%; a substantial herbaceous/shrub layer; and an average coverage of the tree canopies of 30%, except for one plot.

Particularly significant are the data relating to the differences between the two areas in the number of burnt pine logs on the ground and their relative diameters. In Area 4, the number of burnt pine logs on the ground is four times greater than in Area 6, with an average diameter of the logs of about half that of those in Area 6 (Table 1 and Table 2; Figure 5).

In Area 4, the extremely low current coverage of the vegetation corresponds to a high number of burnt pine logs on the ground and a diameter not exceeding 11.7 cm. Vice versa, in Area 6, the high vegetation cover corresponds to a low number of burnt pine logs on the ground, with an average diameter of 20 cm (Table 1 and Table 2). Finally, Ward’s clustering applied to the measurement data reveals, with good significance (cophenetic correlation coefficient = 0.92), two different main clusters corresponding to the two areas (Figure 6a).

Ground measurements of the diameter of cork tree trunks and burnt pine logs on the ground turned out to be quite similar to those from the UAV. In fact, in any case, there was an average measurement discrepancy never exceeding one cm (Table 2).

The coverage data of each layer measured on the remote sensing data (bare ground, herbs and shrubs, tree crowns, total cover) reflects what we already observed from the phytosociological ground surveys (Table S1).

The coverage values of the herbaceous and shrubby layer obtained from the analysis of the images taken in late winter (through the processing of the dense cloud points) did not differ from the estimate made with the relevés in the field, despite the latter being carried out about three months after the flights. This was probably due to a combination of several factors: scarce spring rainfall, the scarce presence of soil, and probably, a decrease in the number of herbaceous seeds.

The phytosociological relevés showed differences between the two areas in terms of floristic composition and coverage (Table S1). The cluster analysis applied to the floristic-vegetational relevés shows, with good significance (cophenetic correlation coefficient = 0.79), two different main clusters corresponding to the two areas (Figure 6b), highlighting the presence of two different vegetation communities.

The number of species present in the two areas does not differ significantly (p = 0.22). Regarding the 24 species found, their distribution appears diversified between the two areas. Sixteen species, albeit with very different frequencies, are in common. Three species (Dittrichia viscosa, Pistacia lentiscus, Ulex europaeus) were found only in Area 4, and five (Fraxinus ornus, Coronilla emerus, Cytisus villosus, Tamus communis, Helichrysum italicum) were found only in Area 6.

In both areas, the tree layer (>2 m) is almost exclusively represented by Q. suber, apart from rare individuals of F. ornus and A. unedo that slightly exceed 2 m. In both areas, there are no arboreal individuals of P. pinaster, present as juveniles in Area 4, while it is practically absent in Area 6. In Area 4, Q. suber is present with extremely limited coverage, while in Area 6, the species appears with much greater coverage. The shrubby mantle is characterized in both cases by the prevalent species A. unedo, P. angustifolia, and S. aspera, but with a much more relevant coverage in Area 6 than in Area 4. E. arborea, although present in both areas, shows much higher coverage in Area 4. C. villosus is exclusively found in Area 6, with a high coverage value. Dittrichia viscosa is present exclusively in Area 4, although it has always a limited coverage index.

4. Discussion

The mixed methodology of ground survey and UAV technologies made it possible to trace with good precision a picture of the effects of the two fires in the two sample areas investigated. The techniques and methods of detection by UAVs are rapidly evolving, and RGB images are increasingly implemented by hyper/multispectral and lidar images [36]. In any case, the UAV-photogrammetric point cloud system in RGB proved in our case to be economical, quick, and functional for the purposes.

From the analysis of the data collected with ground surveys and remote measurements, it is possible to highlight two different plant communities in their floristic-vegetational and physiognomic aspects. Nevertheless, the current diversity can be interpreted as a different post-fire phase of the same secondary succession. The different adaptive strategy of the two target species in this post-fire context, the pine and the cork oak, becomes fundamental in this context.

Pinus pinaster and Quercus suber are indigenous species in this geographical area. However, the considerable extensions of the maritime-pine-forested areas of Monte Pisano are considered to be the result of ancient forest management [37] and recent plantings [38]. Quercus suber, on the other hand, although widespread on the western side of the mountain, only rarely gives rise to phytocoenoses in which it is dominant [30]. In this context, in the last 50 years, almost all the pine forests of Monte Pisano have been crossed by fire [39]. In many cases, the fires have occurred in the same area several times. In these areas, we observe an important diffusion of elements of Quercion, such as Q. suber, Q. ilex, Q. pubescens, and Fraxinus ornus [32].

The “active” response of pine by seed to the passage of fire is known in the literature for various contexts in the Mediterranean area [40,41]. In a low-intensity fire, high dissemination and re-growth occur [42]. In a high-intensity fire or crown fire, a negative effect on the success of post-burn occurs [43], although this does not prevent a normal re-establishment of P. pinaster forest [44]. This self-succession depends on the recurrence time of a fire for a pine forest already affected by the fire. P. pinaster, although it shows wide variability linked to different soil and climatic conditions, flowers at the age of 7–8 years [45].

If there is too short of a period between fires, there will still not be a sufficient number of sexually mature individuals, and the pine forest will decline or disappear [46]. Similarly, the “passive” response of cork oak is known, including vegetative regrowth from the aerial branches and its consequent ability to persist and spread over areas repeatedly affected by fires [47,48].

In this case study, in both areas, before the two fires, the entire slope was covered with a single pine forest, although with a differentiated contingent of evergreen sclerophylls [30]. After the fire of 2009, four surveys carried out nine years later had already highlighted a different type of recovery fundamentally linked to pine and cork oak: in one sector (Area 4), an important regeneration of pine and the presence of some cork oaks; in another (Area 6), a pronounced rarefaction of pine and a consistent presence of the cork [32]. This difference can be explained partly by the different intensity of the first fire along the slope and different occurrences in both areas of evergreen sclerophylls.

The 2021 data relating the number and diameter of burnt pines logs on the surface seem to confirm a different asset of arboreal vegetation in 2018.

In Area 4, the number and diameter of burnt pines logs on the surface suggest the presence of a young pine forest with large coverage at the time of the second fire. In Area 6, in contrast, the same data on burnt pine logs suggest a sparse cover of pine on expanding sclerophyll vegetation (Table 1).

Right now, the data indicate a further step towards the same vegetational aspects. In Area 4, the last fire has eliminated the young pine forest, and only a partial self-succession of pines is allowed. Although the sparse shrubs layer is still mainly represented by heather, other shrub species, such as those present with greater coverage in area 6, are detectable (Table S1). These species, together with the survived corks, can represent a new phase of post-fire regrowth.

In Area 6, the last fire has eliminated the sparse pine, allowing further development, in terms of coverage, of the pre-existing contingent of evergreen species (Q. suber, Q. ilex, A. unedo, P. angustifolia) that are supplanting the pine forest (Table S1). In short, what has been observed in Area 4 seems to represent a previous post-fire vegetational stage that was already in place for some time in Area 6. Nevertheless, in Area 4, the scarcity of tree coverage post-fire and the presence of an important heather shrub mantle (Table S1) could be an obstacle to a quick colonization by the cork, as observed in the Iberian peninsula [49,50]. In fact, in the Mediterranean area, the post-fire recovery of oak has been faster where it was already present to a dominant extent before the fire [51]. This may have influences and effects in the future, which may be discovered as the different coverage and specific compositions of the two areas are investigated.

Diachronic vegetation monitoring consists of surveys repeated over time with the same methodologies and on the same points [52]. The results obtained from this study are consequently to be considered preliminary, and only subsequent and periodic surveys will be able to confirm what was found. In any case, the data obtained seem to be in line with what has been observed in other regions of the Mediterranean, where fire in a pine forest, especially if repeated, gives rise to successions in which the pine forest is replaced by cork oak woods, passing through stages in which shrub communities may predominate [53,54,55].

5. Conclusions

In the burned area investigated, the evolutionary trend of the pine forests subject to fire seems to lead towards cork oak woods, albeit with different regenerative dynamics. These forests can represent a secondary succession in equilibrium with a regime of repeated fires. These dynamics are, in any case, determining an expansion of the neighboring phytocoenoses of Q. suber and Q. ilex.

If the ground survey remains essential for the certain identification of botanical entities, the drone photos, even if only on RGB, have proven to be of great use. The creation of a high-definition orthophotographic base, the digital technology of three-dimensional reading of the images, and their decomposition into different layers allowed the identification of the structural parameters in a rapid way. This takes on an important significance, especially when it is necessary to operate, as in this case, in rather inaccessible areas. Future remote monitoring planned in the same area will allow a better understanding of the post-fire dynamics of these phytocoenoses.