Estimation of Short-Term Vegetation Recovery in Post-Fire Siberian Dwarf Pine (Pinus pumila) Shrublands Based on Sentinel-2 Data

Abstract

The frequency of wildfires ignited by lightning is increasing due to global climate change. Since the forest ecological recovery is influenced by numerous factors, the process of post-fire vegetation recovery in Siberian dwarf pine shrublands remains unclear and demands in-depth study. This paper explored the short-term recovery process of vegetation after two lightning-ignited fires in the Great Xing’an Mountains that occurred in 2017 and 2020, respectively. The study was aimed at presenting a monitoring approach for estimating the post-fire vegetation state and assessing the influence of various driving factors on vegetation recovery. Spectral indices were computed to evaluate forest vegetation recovery dynamics. The differences in vegetation recovery under various fire severity and topography conditions were also examined. Correlation analysis was employed to assess the influence of moisture content on the recovery of fire sites. The results show that fire severity, topographic features, and moisture content significantly impacted the rate of vegetation recovery. Specifically, regeneration takes place more rapidly on warm, high-altitude, and gentle slopes within highly and moderately burned areas. Additionally, areas marked by high moisture content demonstrate rapid recovery. Our study enriches the research cases of global wildfires and vegetation recovery and provides a scientific basis for forest management and the restoration of post-fire ecosystems.

1. Introduction

Wildfire constitutes a primary disturbance process in forest ecosystems [1,2]. Through altering community composition, age structure, energy flow, and nutrient cycling, wildfires exert a significant influence on regional carbon cycles, biodiversity, and global climate change [3,4,5,6,7,8,9]. The Sixth Assessment Report of the International Panel on Climate Change (IPCC) clearly stated that in 2020 global warming had reached 1.1 °C, above pre-industrial level, driven by more than a century of burning fossil fuels as well as unequal and unsustainable energy and land use. This has resulted in more frequent and more intense extreme weather events that have caused increasingly dangerous impacts on nature and people in every region of the world [10]. Wildfires have occurred frequently in recent years, such as the California forest fires in the United States in August 2020 [11], the Australian forest fires in September 2015 [12], and the forest fires that occurred in the northeastern forest areas of China from 2000 to 2018 [13], which have seriously affected the social, economic, and ecological benefits of forest areas [14].

With the provision of ecosystem services, the Great Xing’an Mountains represent an important component of the ecological security barrier in northern China [15]. In recent years, the occurrence of frequent lightning-ignited wildfires has adversely impacted the ecological conservation and restoration in this area [16,17,18]. The Siberian dwarf pine (Pinus pumila), an evergreen shrub species, is widely distributed in subalpine regions throughout northeastern Asia [19]. This species is highly flammable as it contains abundant volatile organic compounds in its needles, twigs, and seeds. The Siberian dwarf pine shrubland constitutes an important vegetation type in the Great Xing’an Mountains, where lightning-ignited fires account for 49.39% of the total number of wildfires [20]. With a substantial increase in burned areas, the natural recovery rate needs to be estimated to formulate regional strategies for post-fire forest management.

Traditional wildfire monitoring and assessment methods typically rely on field investigations. Nevertheless, characterizing post-fire vegetation recovery under diverse environmental conditions proves challenging through field investigation alone, especially in remote regions that are hard to access, such as for fires initiated by lightning strikes. The advancement of satellite remote sensing technology offers data sources and technical support for the monitoring of large-scale areas and continuous forest disturbances [21,22].

In contrast to conventional ground monitoring techniques, remote sensing images present several advantages, such as shorter imaging periods, broader monitoring ranges, and enhanced sensitivity towards forest disturbances [23,24]. The utilization of Sentinel-2 data holds considerable potential in effectively monitoring burned areas and vegetation dynamics [25]. On the one hand, the spatial resolution of Sentinel-2 data exceeds that of NOAA or MODIS data, rendering it suitable for studying regional-level forest disturbances [26]. Conversely, when contrasted with newer high-resolution satellite data and noctilucent remote sensing satellite data, the time span covered by the Sentinel-2 series is longer. This prolonged duration enables long-term and high-quality operations that facilitate continuous monitoring of the Earth’s surface with reliable data support.

Depending on the purpose of observing and studying post-fire vegetation recovery, diverse methods are adopted for assessment. These methods encompass image classification and the utilization of vegetation indices (VIs) [27,28]. Remote sensing imagery offers an opportunity to gather information regarding land use and land cover through image interpretation and classification. Spectral responses are employed in image classification to identify healthy vegetation within individual pixels [29]. The type or condition of surface features and their alterations can be evaluated by using multi-temporal imaging [30]. VIs are extensively utilized to assess the status of vegetation, including its recovery after natural or human-induced disturbances [31,32].

The Normalized Difference Vegetation Index (NDVI) is extensively utilized in post-fire recovery studies on account of its spectral combination-based approach. It effectively captures vegetation activity, density, and the absorption of photosynthetic radiation by vegetation [33,34]. Nevertheless, it can be influenced by the soil background and lacks sensitivity towards dense forest cover [35,36]. The Enhanced Vegetation Index (EVI) effectively addresses this concern by incorporating blue-band reflectance in its calculation formula [37]. The remotely sensed Moisture Stress Index (MSI) is employed for canopy stress analysis and is suitable for monitoring coniferous forests and assessing specific damages that cannot be detected using NIR/R vegetation indices [38]. The Normalized Burn Ratio (NBR) is another analytical method based on the spectral combination of forest fire disturbances. Using the dNBR—which is the difference value of the NBR before and after a fire—the burning area and severity can be extracted effectively [39].

Post-fire vegetation recovery is a complex phenomenon that cannot be evaluated through the application of a single, standardized spectral index. A multitude of factors, including climate conditions, initial plant mortality rates, soil properties, regional topography, and the composition of vegetation species collectively exert an influence on the rate of recovery [13,40]. Therefore, the impact of fire and subsequent recovery processes on vegetation varies across different biogeographical regions and is influenced by the type of vegetation as well as the pre-fire state [41].

In this study, we hypothesize that fire severity, topography conditions, and moisture content largely determine the post-fire vegetation regeneration. Therefore, the mean vegetation recovery rates on burned surfaces will vary between areas presenting different severity levels and topography conditions. Furthermore, moisture content will have different impacts on the recovery of fire sites. To test these hypotheses, we conducted an analysis of the vegetation recovery of Siberian dwarf pine shrublands following a wildfire in the Great Xing’an Mountains. Through short-term post-fire monitoring, this study aims to reveal the effects of fire on vegetation succession.

The main research objectives of this paper are as follows:

• To quantify the vegetation recovery process of Siberian dwarf pine shrublands through short-term observations;
• To explore the disparities in vegetation recovery under different fire severity levels and topography conditions;
• To explore the effect of moisture content on post-fire recovery using correlation analysis.

2. Materials and Methods

2.1. Study Site

The Huzhong Forestry Bureau (51°14′51″–52°25′28″ N, 122°36′58″–124°15′46″ E) is situated in the central portion of the Great Xing’an Mountains. The region lies within a cold temperate, continental monsoon climate zone, featuring long, cold winters and short, hot summers. The average annual precipitation amounts to approximately 460 mm. The frost-free period spans 80 to 90 days. The average annual temperature is −4.7 °C. The topography is mountainous, with elevations ranging from 360 m to 1511 m above sea level [42]. The forests in this area are predominantly dominated by a deciduous coniferous tree species, Dahurian larch (Larix gmelinii), and are mixed with some evergreen coniferous tree/shrub species—such as Korean spruce (Picea koraiensis), Scotch pine (Pinus sylvestris), and Siberian dwarf pine (Pinus pumila)—along with a few deciduous broadleaf species, like birch (Betula platyphylla) and aspen (Populus davidiana and Populus suaveolens). The Siberian dwarf pine is the most dominant species, accounting for over 80% of all vegetation coverage. The distribution of this species is influenced by the topographic and soil conditions [43]. It is typically mixed with Dahurian larch in open forests at altitudes of 800–1200 m or grows densely on rocky ridges at altitudes higher than 1200 m. The understory species consist of evergreen shrubs (e.g., Ledum and Vaccinium vitis-idaea), deciduous shrubs (e.g., Betula fruticose and Rhododendron dauricum), and some herbaceous plants (e.g., Chamaenerion angustifolium and Carex appendiculata) [44].

2.2. Fire History

In this region, wildfires are typically characterized by frequent surface fires and infrequent stand-replacing crown fires, with fire return intervals ranging from 30 to 120 years [45]. The causes of wildfires are weather conditions and deliberate factors. Due to climate change, high-temperature anomalies continue to occur, which leads to frequent lightning-ignited fires. According to the fire record statistics of the ministry of emergency management of the People’s Republic of China, from 2014 to 2024, as many as 77 fires were recorded in this area, among which 73 fires were caused by lightning (Figure 1) [46]. In this study, we focused on the fire events that occurred from 2014 to 2024 within the two largest lightning-ignited fires (Figure 2).

Figure 1. Number of wildfires from 2014 to 2024 in Huzhong Forestry Bureau. The blue and orange colors represent lightning-ignited fire and human-ignited fire, respectively.

Figure 2. Locations and vegetation types in the study areas. (a) indicated Daxigou, and (b) indicated Yalihe.

Case a:

From 23 to 24 June 2017, a lightning-ignited fire consumed an area of 212.4 ha of Siberian dwarf pine shrublands under severe drought conditions in Daxigou (Table 1, Figure 3a). The nearest remote automated weather station automatically recorded a maximum temperature of 34.5 °C, a mean daytime relative humidity of 62%, and a mean daytime wind speed of 1.1 m s⁻¹—predominantly from the northwest—with gusts reaching up to approximately 13.8 m s⁻¹. The dominant vegetation types in this study area are Siberian dwarf pine and Dahurian larch, with a minor presence of Scotch pine (Figure 2a).

Table 1. Location, time, and burning area of fire points in the Great Xing’an Mountains.

Location	Coordinates	Time	Burning Area (hm2)	Main Vegetation Types
Daxigou	52°0′51″ N–123°23′25″ E	23–24 June 2017	212.4	Siberian dwarf pine 80%, Dahurian larch 15%, and Scotch pine 5%
Yalihe	51°17′48″ N–123°33′07″ E	15 July 2020	12.96	Siberian dwarf pine 80% and Dahurian larch 20%

Figure 3. Photos of the study sites burning in Daxigou and Yalihe. (a) indicated that the fire in Daxigou was burning. The photo was taken on 23 June 2017. (b) indicated that the fire in Yalihe had just been extinguished. The photo was taken on 15 July 2020.

Case b:

On 15 July 2020, another lightning-ignited fire broke out in Yalihe and consumed 12.96 ha of Siberian dwarf pine shrublands (Table 1, Figure 3b). The nearest remote automated weather station automatically recorded a maximum temperature of 31 °C, a mean daytime relative humidity of 87.5%, and a mean daytime wind speed of 0.8 m s⁻¹—predominantly from the north—with gusts reaching up to approximately 16.2 m s⁻¹. The dominant vegetation types in this study area are Siberian dwarf pine and Dahurian larch (Figure 2b).

For over 30 years, the two study sites had neither been managed nor undergone a wildfire.

2.3. Data

Satellite data obtained from the Sentinel-2A and Sentinel-2B multispectral sensors of the European Space Agency Program for Earth Observation “Copernicus” [47] were utilized to assess post-fire vegetation recovery. The temporal resolution of each Sentinel-2 satellite is ten days, and their combined resolution is five days.

In Daxigou, the Sentinel-2 image obtained on 19 July 2016 was representative of the period before the fire event (23–24 June 2017), and the images acquired on 19 July 2017, 15 August 2018, 27 August 2020 and 2022, and 7 July 2024 were employed for the assessment of the forest vegetation state after the fire. In Yalihe, the Sentinel-2 image obtained on 30 August 2019 was representative of the period before the fire event (15 July 2020), and the images acquired on 12 August 2020, 2 August 2021, 29 August 2022, 10 July 2023, and 7 July 2024 were utilized for the assessment of the forest vegetation state after the fire.

A digital elevation model (DEM) with a spatial resolution of 12.5 m was employed to obtain altitudes, slopes, and aspects of slopes. This dataset, generated from the Japanese Earth observation satellite ALOS’s L-band synthetic aperture radar (PALSAR), is freely accessible via the Alaska Satellite Facility’s data portal (https://search.asf.alaska.edu/ (accessed on 20 October 2024)).

2.4. Methods

Based on remote sensing data, this paper quantified the short-term recovery process of vegetation on fire sites by monitoring post-fire vegetation recovery from wildfires over a period of 4–7 years. The differences in vegetation recovery under various topography conditions were also explored. Correlation analysis was employed to evaluate the influence of the moisture driving factor on the recovery of fire sites.

2.4.1. Vegetation Index

We utilized Sentinel-2 data with a spatial resolution of 10 m to calculate the indices listed in Table 2. Among these, the Normalized Difference Vegetation Index (NDVI) is a widely adopted vegetation index that demonstrates high sensitivity towards chlorophyll content, vegetation percentage, density, and dynamics [40,48]. However, it may become oversaturated in regions characterized by substantial above-ground biomass such as tropical rainforests [49,50,51]. In areas where vegetative cover is sparse, the soil’s brightness can cause alterations in the temporal patterns of NDVI and consequently introduce some uncertainty during parameter extraction [52]. To effectively address this concern, we have incorporated the Enhanced Vegetation Index (EVI), which takes into account atmospheric aerosol scattering and background radiation from soil by incorporating blue-band reflectance and employing mathematical transformations within its calculation formula [37].

Table 2. Spectral indices calculated in this study.

Index	Abbreviation	Formula
Normalized Difference Vegetation Index	NDVI	$NDVI={\displaystyle \frac{{\rho }_{\mathrm{nir}}-{\rho }_{red}}{{\rho }_{\mathrm{nir}}+{\rho }_{red}}}$
Enhanced Vegetation Index	EVI	$EVI=2.5\times {\displaystyle \frac{{\rho }_{\mathrm{nir}}-{\rho }_{red}}{{\rho }_{\mathrm{nir}}+6.0\times {\rho }_{red}-7.5\times {\rho }_{red}+1}}$
Moisture Stress Index	MSI	$MSI={\displaystyle \frac{{\rho }_{swir1}}{{\rho }_{nir}}}$
Normalized Burn Ratio	NBR	$NBR={\displaystyle \frac{{\rho }_{\mathrm{nir}}-{\rho }_{swir}}{{\rho }_{\mathrm{nir}}+{\rho }_{swir}}}$
Differenced Normalized Burn Ratio	dNBR	$dNBR=NB{R}_{pre-fire}-NB{R}_{post-fire}$

The remotely sensed Moisture Stress Index (MSI) is employed to analyze canopy stress, predict productivity, and model biophysics. It specifically identifies plant water stress in plants that are capable of adapting to low leaf water content through cellular adjustments. In the current study, the MSI proves particularly efficacious in monitoring coniferous vegetation and evaluating specific damages that NIR/R vegetation indices fail to detect [38]. When considering coniferous vegetation, the differences observed in the MSI between damaged and undamaged stands may not necessarily be ascribed to variations in LAI. The MSI ranges from 0 to over 3, with higher values indicating increased moisture stress.

The Normal Burn Ratio Index (NBR) [53] has garnered considerable attention in recent years due to its application in mapping burned areas [54]. Healthy vegetation exhibits extremely high reflection in the NIR and low reflection in the SWIR part of the spectrum. Conversely, burned areas exhibit low NIR reflectivity and high SWIR reflectivity, resulting in a marked difference between their spectral responses. A higher NBR value indicates healthy vegetation, while a lower value indicates bare land or burned areas [55]. The NIR band is an efficacious spectral band for monitoring vegetation, while the SWIR spectral band effectively represents moisture content in both soil and vegetation. Additionally, abrupt changes can be observed in the NBR of burned areas as wildfires significantly alter canopy structure and moisture levels. In contrast, unburned areas show minimal variation in NBR values [45,56].

Most studies use pre- and post-fire NBR images to estimate the burn severity from the NBR difference (dNBR). Higher dNBR values indicate severe damage, while areas with negative values of dNBR may indicate a lack of vegetation in the fire zone [52]. The US Geological Survey proposed a classification table to interpret the burn severity [57]. Based on this and according to the fire in the study areas, four modes were determined to classify the dNBR index and fire severity: dNBR with unburned severity (−0.1 to +0.1), low severity (+0.1 to +0.27), moderate severity (+0.27 to +0.66), and high severity (>+0.66).

Spectral indices were computed to assess the short-term recovery dynamics of forest vegetation in Daxigou from 2016 to 2024 and in Yalihe from 2019 to 2024.

All calculations of spectral indices were conducted in ENVI/IDL 8.8.

2.4.2. The Impact of Topography on Vegetation Recovery

The average VI values of the three years before the fire disturbance were employed as the average level of the vegetation index in the Daxigou and Yalihe study areas. The recovery level of the VI each year after the fire disturbance was calculated by the following formula [58]:

$$R_{vi}={\displaystyle \frac{V{I}_i}{V{I}_{mean}}}\times 100\%$$

(1)

where $R_{vi}$ represents the level of vegetation recovery, $V{I}_i$ represents the NDVI or EVI value for year i, and $V{I}_{mean}$ represents the average value of NDVI or EVI for the three years preceding the fire.

The mean recovery rates were calculated for various altitudes, slopes, and aspects across different fire severity levels during the studied years to analyze the impact of topography and fire severity on vegetation recovery.

2.4.3. Correlation Analysis

Correlation analysis is a statistical method used to determine the degree to which two or more variables are linearly related. It helps to understand the strength and direction of the relationship between variables. Pearson Correlation Coefficient is used to measure the linear correlation between two continuous variables. The formula for Pearson’s correlation coefficient (r) is:

$$\mathrm{r}={\displaystyle \frac{{\sum }_{i=1}^n\left(X_i-\overline{X}\right)\left(Y_i-\overline{Y}\right)}{\sqrt{{\sum }_{i=1}^n{\left(X_i-\overline{X}\right)}^2}\sqrt{{\sum }_{i=1}^n{\left(Y_i-\overline{Y}\right)}^2}}}$$

(2)

where r is the Pearson correlation coefficient, which ranges from −1 to 1. 1 indicates a perfect positive correlation, −1 indicates a perfect negative correlation, and 0 indicates no correlation. $X_i$ and $Y_i$ are the individual sample points indexed with i. $\overline{X}$ and $\overline{Y}$ are the mean values of X and Y, respectively. n is the number of sample points [59].

The correlation analysis was conducted using IBM SPSS Statistics software (version 24.0. Armonk, NY, USA: IBM Corp.), with MSI used as a variable representing the key post-fire characteristics of the affected areas. MSI indicates stress in ecosystems resulting from moisture deficiency.

3. Results

3.1. Forest Vegetation Recovery for the Study Area

According to Figure 4 and Table A1, in Daxigou, the areas with NDVI and EVI values ranging from 0.4 to 0.6 were predominant in the pre-fire year (2016). In the same year, the MSI category with values between 0.6 and 0.8 occupied the largest portion of the area. A considerable portion of the areas had NBR values between 0.4 and 0.6. In the year immediately following the fire (2017), the maximum distribution had territories with NDVI and EVI values between 0 and 0.1, indicating the destruction of the vegetation. This year was characterized by the most marked stress due to moisture deficiency. The MSI category with values between 1.4 and 1.6 had the largest share of the area. Meanwhile, the areas with negative NBR values predominated due to the destruction of the vegetation. The NDVI and EVI values witnessed the highest growth in the first post-fire year (2018). At this time, the areas in the range from 0.3 to 0.4 were dominant, indicating a positive recovery for all species. Due to the alleviation of moisture stress, 40.6% fell into the category with MSI between 1 and 1.2. The positive NBR values exhibited the highest growth between 2017 and 2018. In the following six years, from 2019 to 2024, the maximum distribution had areas with NDVI and EVI values between 0.3 and 0.5. During this period, the MSI value gradually declined, while the NBR value gradually increased. In 2024, the areas were concentrated in the NDVI and EVI categories between 0.4 and 0.5. The MSI category with values between 0.8 and 1 occupied the largest share of the area. A considerable portion of the areas had NBR values between 0.2 and 0.4.

Figure 4. Dynamics of the spectral indices calculated for the period of 2016–2024.

Similar to Daxigou, the forest vegetation in Yalihe exhibited the same recovery pattern after the occurrence of fire. The description was not presented in this paper.

3.2. Vegetation Recovery in Different Topographic Factors

3.2.1. Distribution of Fire Severity in the Study Areas

In Figure 5, the green, yellow, orange, and red colors represent unburned, and areas of low, moderate, and high burned severity, respectively. Moderate fire severity in both study areas covered the largest area. Examination of dNBR values showed that 104.36 hectares (49.13%) in Daxigou and 9.96 hectares (76.85%) were in the moderate burn severity class.

Figure 5. Distribution of burn severity in the study areas. (a) Indicates the Daxigou study area, and (b) indicates the Yalihe study area.

3.2.2. Vegetation Recovery in the Individual Slope Aspects

The dynamics of the recovery rates in the individual slope aspects for different severity levels were presented in Table 3. During the first and second recovery periods, high- and moderate-severity burned areas in Daxigou and Yalihe experienced rapid vegetation recovery. The NDVI and EVI recovery rates showed a substantial increase, ranging from 3% to 35%. In contrast, low-severity burned areas in Daxigou and Yalihe exhibited either negative or slow vegetation recovery, with NDVI and EVI recovery rates either decreasing or remaining stable within the range of −11% to 9%. During the third and fourth recovery periods, all severity levels in Daxigou and Yalihe demonstrated comparable recovery rates.

Table 3. Mean values of the recovery rates in the different slope aspects for different severity levels in the studied years and the dynamic changes during their periods. For Daxigou, the four periods represented: 2017–2018, 2018–2020, 2020–2022, and 2022–2024. For Yalihe, the four periods represented: 2020–2021, 2021–2022, 2022–2023, and 2023–2024.

Fire Severity	Aspect	2017	2018	2020	2022	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Daxigou (%)					Change (%)
High Severity	N	6	12	32	42	47	6	20	10	5
	NE	4	11	27	35	38	7	16	8	3
	E	5	16	35	46	50	11	19	11	4
	SE	5	34	34	40	41	29	0	6	1
	S	5	40	40	47	49	35	0	7	2
	SW	3	10	15	20	21	7	5	5	1
	W	2	5	12	16	17	3	7	4	1
	NW	3	6	17	21	23	3	11	4	2
Moderate Severity	N	28	36	49	56	63	8	13	7	7
	NE	27	32	47	56	63	5	15	9	7
	E	25	36	41	49	51	11	5	8	2
	SE	22	54	61	71	73	32	7	10	2
	S	21	53	59	69	71	32	6	10	2
	SW	24	45	54	64	68	21	9	10	4
	W	20	39	51	63	68	19	12	12	5
	NW	28	38	64	73	80	10	26	9	7
Low Severity	N	46	42	44	48	55	−4	2	4	7
	NE	38	35	35	39	44	−3	0	4	5
	E	34	34	35	39	43	0	1	4	4
	SE	26	26	28	31	34	0	2	3	3
	S	33	33	35	39	42	0	2	4	3
	SW	37	35	37	42	46	−2	2	5	4
	W	45	38	38	43	48	−7	0	5	5
	NW	44	34	38	45	49	−10	4	7	4
		Recovery of EVI in Daxigou (%)					Change (%)
High Severity	N	4	9	27	39	47	5	18	12	8
	NE	3	8	23	32	39	5	15	9	7
	E	4	12	31	44	51	8	19	13	7
	SE	3	31	32	40	44	28	1	8	4
	S	4	37	37	46	53	33	0	9	7
	SW	2	9	14	19	23	7	5	5	4
	W	1	4	11	15	18	3	7	4	3
	NW	2	5	15	21	24	3	10	6	3
Moderate Severity	N	22	30	45	52	65	8	15	7	13
	NE	21	26	42	52	65	5	16	10	13
	E	20	32	38	48	53	12	6	10	5
	SE	17	48	57	70	78	31	9	13	8
	S	16	48	55	68	77	32	7	13	9
	SW	19	40	50	63	72	21	10	13	9
	W	16	34	47	61	72	18	13	14	11
	NW	21	31	59	68	83	10	28	9	15
Low Severity	N	43	38	39	44	57	−5	1	5	13
	NE	35	32	31	36	45	−3	−1	5	9
	E	31	31	32	36	45	0	1	4	9
	SE	24	24	26	30	36	0	2	4	6
	S	30	31	32	37	44	1	1	5	7
	SW	34	31	33	39	48	−3	2	6	9
	W	43	34	34	40	50	−9	0	6	10
	NW	42	31	35	42	52	−11	4	7	10
Fire Severity	Aspect	2020	2021	2022	2023	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Yalihe (%)					Change (%)
High Severity	N	7	20	30	35	36	13	10	5	1
Moderate Severity	N	20	47	58	71	72	27	11	13	1
	NE	20	46	57	68	68	26	11	11	0
	E	23	46	53	64	63	23	7	11	−1
	W	6	14	18	20	20	8	4	2	0
	NW	16	47	62	76	77	31	15	14	1
Low Severity	N	28	27	34	40	41	−1	7	6	1
	NE	22	26	28	34	34	4	2	6	0
	E	54	57	60	67	67	3	3	7	0
	W	34	40	44	51	51	6	4	7	0
	NW	45	47	51	59	59	2	4	8	0
		Recovery of EVI in Yalihe (%)					Change (%)
High Severity	N	4	16	28	35	34	12	12	7	−1
Moderate Severity	N	14	40	54	71	70	26	14	17	−1
	NE	14	39	54	69	66	25	15	15	−3
	E	17	41	50	65	62	24	9	15	−3
	W	4	11	17	20	19	7	6	3	−1
	NW	11	39	57	76	75	28	18	19	−1
Low Severity	N	24	26	33	42	41	2	7	9	−1
	NE	19	25	28	35	35	6	3	7	0
	E	50	55	62	72	67	5	7	10	−5
	W	28	37	42	53	50	9	5	11	−3
	NW	40	46	50	62	59	6	4	12	−3

In burned areas with the same fire severity, rapid vegetation recovery was observed on the southeast- and south-facing slopes in Daxigou and the north-, northeast-, and northwest-facing slopes in Yalihe during the first recovery period. The NDVI and EVI recovery rates in moderate-severity burned areas increased significantly by 25–32%. During the second recovery period, the north-, northeast-, and northwest-facing slopes in both Daxigou and Yalihe showed the highest increments in NDVI and EVI recovery rates. In the third and fourth recovery periods in Daxigou and the third recovery period in Yalihe, all slope aspects exhibited similar recovery rates. However, during the fourth recovery period from 2023 to 2024 in Yalihe, the NDVI and EVI recovery rates for all slope aspects were negative.

In the seven years following the Daxigou fire event (2024), the recovery rates of NDVI and EVI on the southeast- and south-facing slopes in moderate-severity burned areas exceeded 70%. Similarly, within four years after the Yalihe fire event (2024), the recovery rates of NDVI and EVI on the north- and northwest-facing slopes in moderate-severity burned areas also exceeded 70%. These findings suggest that vegetation on the southeast- and south-facing slopes in Daxigou and the north- and northwest-facing slopes in Yalihe exhibited superior recovery compared to slopes with different aspects under the same fire severity.

3.2.3. Vegetation Recovery in Different Slopes

The dynamics of the recovery rates in different slopes for different severity levels were presented in Table 4. During the first recovery period, high- and moderate-severity burned areas in Daxigou and Yalihe experienced rapid vegetation recovery, as evidenced by substantial increases in NDVI and EVI recovery rates, ranging from 3% to 33%. In contrast, low-severity burned areas in these regions exhibited either negative or slow recovery, with NDVI and EVI recovery rates decreasing or remaining stable within the range of −4% to 7%. During the third and fourth recovery periods, all severity levels in Daxigou demonstrated significantly accelerated recovery rates.

Table 4. Mean values of the recovery rates in the different slopes for different severity levels in the studied years and the dynamic changes during their periods. For Daxigou, the four periods represented are 2017–2018, 2018–2020, 2020–2022, and 2022–2024. For Yalihe, the four periods represented are 2020–2021, 2021–2022, 2022–2023, and 2023–2024.

Fire Severity	Aspect	2017	2018	2020	2022	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Daxigou (%)					Change (%)
High Severity	2–6°	3	10	18	24	25	7	8	6	1
	6–15°	5	38	38	45	46	33	0	7	1
	15–25°	5	37	36	43	44	32	−1	7	1
	≥25°	4	30	31	36	37	26	1	5	1
Moderate Severity	≤2°	22	25	42	54	59	3	17	12	5
	2–6°	25	39	47	57	59	14	8	10	2
	6–15°	23	50	59	69	71	27	9	10	2
	15–25°	23	50	58	67	71	27	8	9	4
	≥25°	25	44	53	61	66	19	9	8	5
Low Severity	≤2°	50	47	46	51	61	−3	−1	5	10
	2–6°	32	31	31	35	39	−1	0	4	4
	6–15°	36	34	36	41	46	−2	2	5	5
	15–25°	39	36	37	41	46	−3	1	4	5
	≥25°	52	48	49	54	61	−4	1	5	7
		Recovery of EVI in Daxigou (%)					Change (%)
High Severity	2–6°	2	8	17	23	27	6	9	6	4
	6–15°	3	35	36	44	49	32	1	8	5
	15–25°	3	34	34	42	48	31	0	8	6
	≥25°	3	28	29	35	41	25	1	6	6
Moderate Severity	≤2°	17	20	37	50	61	3	17	13	11
	2–6°	20	34	43	55	61	14	9	12	6
	6–15°	18	44	55	67	76	26	11	12	9
	15–25°	18	45	54	65	77	27	9	11	12
	≥25°	19	39	50	59	71	20	11	9	12
Low Severity	≤2°	46	42	41	45	61	−4	−1	4	16
	2–6°	29	28	27	32	40	−1	−1	5	8
	6–15°	33	31	33	38	48	−2	2	5	10
	15–25°	36	33	34	38	48	−3	1	4	10
	≥25°	48	43	44	49	63	−5	1	5	14
Fire Severity	Aspect	2020	2021	2022	2023	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Yalihe (%)					Change (%)
High Severity	6–15°	6	18	27	31	32	12	9	4	1
Moderate Severity	≤2°	26	53	66	79	79	27	13	13	0
	2–6°	18	32	49	61	63	14	17	12	2
	6–15°	20	46	57	70	71	26	11	13	1
	15–25°	20	48	59	69	69	28	11	10	0
Low Severity	2–6°	35	38	41	47	46	3	3	6	−1
	6–15°	22	25	27	32	33	3	2	5	1
	15–25°	30	34	37	43	44	4	3	6	1
	≥25°	15	15	16	17	17	0	1	1	0
		Recovery of EVI in Yalihe (%)					Change (%)
High Severity	6–15°	4	14	25	31	30	10	11	6	−1
Moderate Severity	≤2°	19	45	61	78	75	26	16	17	−3
	2–6°	13	32	45	61	61	19	13	16	0
	6–15°	14	39	53	70	69	25	14	17	−1
	15–25°	14	42	55	69	67	28	13	14	−2
Low Severity	2–6°	31	37	40	50	47	6	3	10	−3
	6–15°	19	24	26	34	34	5	2	8	0
	15–25°	26	33	36	45	45	7	3	9	0
	≥25°	15	15	17	18	17	0	2	1	−1

In the high- and moderate-severity burned areas in Daxigou, as well as the moderate-severity burned areas in Yalihe, slopes ranging from 6 to 25° experienced rapid vegetation recovery during the first recovery period. The NDVI and EVI recovery rates showed a significant increase, exceeding 25%. In the following two periods, all slopes recovered at a comparable rate. However, during the fourth period from 2023 to 2024 in Yalihe, the recovery rates of EVI for all slopes presented negative values.

In 2024, the recovery rates of NDVI and EVI on slopes ranging from 6 to 25° with moderate burn severity in both study areas exceeded 65%.

3.2.4. Vegetation Recovery in Different Altitudes

The dynamics of the recovery rates in different altitudes for different severity levels were presented in Table 5. During the first recovery period, the high- and moderate-severity burned areas in Daxigou and Yalihe showed rapid vegetation restoration, as indicated by significant rises in NDVI and EVI recovery rates, which increased by 10% to 33%. Conversely, the low-severity burned areas in these regions experienced either negative or sluggish recovery, with NDVI and EVI recovery rates either declining or staying relatively stable between −4% and 6%. Throughout the subsequent second to fourth recovery periods, all burn severity levels in Daxigou and Yalihe exhibited similar recovery trends.

Table 5. Mean values of the recovery rates in the different altitudes for different severity levels in the studied years and the dynamic changes during their periods. For Daxigou, the four periods represented: 2017–2018, 2018–2020, 2020–2022, and 2022–2024. For Yalihe, the four periods represented: 2020–2021, 2021–2022, 2022–2023, and 2023–2024.

Fire Severity	Aspect	2017	2018	2020	2022	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Daxigou (%)					Change (%)
High Severity	750–800 m	3	19	23	27	25	16	4	4	−2
	800–850 m	4	27	29	33	33	23	2	4	0
	850–900 m	5	37	37	44	45	32	0	7	1
	900–950 m	5	38	37	43	45	33	−1	6	2
	950–1000 m	3	20	19	22	22	17	−1	3	0
	1000–1050 m	5	13	31	41	44	8	18	10	3
Moderate Severity	750–800 m	27	42	54	64	62	15	12	10	−2
	800–850 m	23	51	59	68	69	28	8	9	1
	850–900 m	22	53	57	67	69	31	4	10	2
	900–950 m	22	48	54	63	67	26	6	9	4
	950–1000 m	23	44	54	63	68	21	10	9	5
	1000–1050 m	17	29	45	56	61	12	16	11	5
Low Severity	750–800 m	28	28	28	30	35	0	0	2	5
	800–850 m	37	35	36	39	45	−2	1	3	6
	850–900 m	35	33	34	38	42	−2	1	4	4
	900–950 m	40	37	39	45	49	−3	2	6	4
	950–1000 m	32	30	32	36	40	−2	2	4	4
	1000–1050 m	17	17	19	21	22	0	2	2	1
		Recovery of EVI in Daxigou (%)					Change (%)
High Severity	750–800 m	2	17	22	27	26	15	5	5	−1
	800–850 m	2	24	27	33	35	22	3	6	2
	850–900 m	3	34	35	43	49	31	1	8	6
	900–950 m	3	35	34	42	49	32	−1	8	7
	950–1000 m	2	19	17	21	24	17	−2	4	3
	1000–1050 m	3	10	28	39	46	7	18	11	7
Moderate Severity	750–800 m	21	36	51	63	64	15	15	12	1
	800–850 m	17	45	55	67	74	28	10	12	7
	850–900 m	16	47	53	65	75	31	6	12	10
	900–950 m	17	42	50	62	73	25	8	12	11
	950–1000 m	18	39	50	61	72	21	11	11	11
	1000–1050 m	13	25	41	53	65	12	16	12	12
Low Severity	750–800 m	26	25	26	28	36	−1	1	2	8
	800–850 m	34	32	32	34	47	−2	0	2	13
	850–900 m	32	30	31	35	44	−2	1	4	9
	900–950 m	37	33	36	42	51	−4	3	6	9
	950–1000 m	30	28	29	34	41	−2	1	5	7
	1000–1050 m	17	17	18	20	23	0	1	2	3
Fire Severity	Aspect	2020	2021	2022	2023	2024	Period1	Period2	Period3	Period4
		Recovery of NDVI in Yalihe (%)					Change (%)
High Severity	1000–1050 m	6	18	27	31	32	12	9	4	1
Moderate Severity	950–1000 m	8	10	11	13	13	2	1	2	0
	1000–1050 m	20	47	60	71	72	27	13	11	1
	1050–1100 m	20	47	58	71	72	27	11	13	1
	1100–1150 m	17	40	44	56	56	23	4	12	0
Low Severity	950–1000 m	26	28	29	33	34	2	1	4	1
	1000–1050 m	26	29	32	37	38	3	3	5	1
	1050–1100 m	31	35	38	44	45	4	3	6	1
	1100–1150 m	18	22	22	28	28	4	0	6	0
		Recovery of EVI in Yalihe (%)					Change (%)
High Severity	1000–1050 m	4	14	25	31	30	10	11	6	−1
Moderate Severity	950–1000 m	6	9	11	14	13	3	2	3	−1
	1000–1050 m	14	40	56	70	69	26	16	14	−1
	1050–1100 m	14	40	54	71	70	26	14	17	−1
	1100–1150 m	12	35	41	58	55	23	6	17	−3
Low Severity	950–1000 m	23	28	28	35	35	5	0	7	0
	1000–1050 m	22	27	31	38	38	5	4	7	0
	1050–1100 m	27	33	37	47	45	6	4	10	−2
	1100–1150 m	16	21	22	30	29	5	1	8	−1

During the first period, both the altitudes ranging from 850 to 950 m in the high and moderate burn severity areas of Daxigou and the altitudes ranging from 1000 to 1100 m in the moderate burn severity areas of Yalihe experienced rapid vegetation recovery. The recovery rates of NDVI and EVI demonstrated a significant increase within the range of 25–33%. In the following two periods, all altitudes recovered at a comparable rate in different fire severity. However, during the fourth period from 2023 to 2024 in Yalihe, the recovery rates of EVI for all altitudes displayed negative values.

In 2024, the recovery rates of NDVI and EVI in areas with moderate burn severity exceeded 65% at altitudes ranging from 850 to 950 m in Daxigou and from 1000 to 1100 m in Yalihe.

3.3. Correlation Analysis

The correlation between MSI and the recovery of VI rates in turn manifested a decreasing trend with the advancement of the ecosystem restoration process (Figure 6). In the year immediately following the fire (2017 in Daxigou and 2020 in Yalihe), the correlation between vegetation recovery and moisture stress was the highest throughout the entire period of observation. This indicated that for the majority of the forest ecosystems, higher MSI values were associated with lower recovery rates of NDVI and EVI. In other words, the moisture deficiency caused by the fire was related to a lower vegetation recovery. The advancement of the ecosystem restoration processes mitigated this dependency. From the third to the seventh post-fire years (2020–2024), there was a significantly low correlation between the recovery rates of EVI in Daxigou and the MSI values (b3–b5 in Figure 6).

Figure 6. Correlation analysis between MSI and recovery of NDVI in Daxigou (a1–a5); MSI and recovery of EVI in Daxigou (b1–b5); MSI and recovery of NDVI in Yalihe (c1–c5); as well as MSI and recovery of EVI in Yalihe (d1–d5) during the studied years. The straight lines represent the regression lines.

4. Discussion

The dynamics of the spectral indices revealed the overall changes in the vegetation recovery. Generally, the areas with relatively high values of NDVI, EVI, and NBR, which corresponded to lower MSI values, had the largest territorial coverage in the pre-fire year (Figure 4, Table A1). In the first post-fire year, the areas were categorized into numerous classes due to diverse levels of vegetation damage in the fire-affected regions. With an increasing temporal distance from the fire event, an increasing proportion of the territories was concentrated in fewer and fewer classes, especially in the first 1–3 years after the fire. This indicated that this period had the fastest rate of vegetation recovery, which was consistent with previous studies [60,61,62]. In the different post-fire years, although there was a minor increase or decrease in the individual classes, the general trend of vegetation recovery was towards classes representing a superior state of vegetation (Figure 4, Table A1).

Our results indicated that fire severity significantly influenced post-fire vegetation recovery. During the first and second periods, vegetation in high- and moderate-severity burned areas of Daxigou and Yalihe exhibited rapid recovery. In contrast, low-severity burned areas in these regions experienced either negative or delayed recovery (Table 2, Table 3 and Table 4). This indicated that more severe fire damage was associated with a faster rate of vegetation recovery during the first three post-fire years. This result concurs with that of Hao et al. [58], who found that the recovery rate of high-severity combustion areas was the fastest within the first two years. The negative or delayed growth in low-severity burned areas can be attributed to delayed mortality. A similar trend has been observed in forested ecoregions in the western United States [63]. Previous studies have concluded that recovery in high-severity burned areas took significantly longer to reach pre-fire levels [52]. However, in this study, we found that moderate-severity burned areas reached pre-fire levels more rapidly than low burned areas. This indicated that delayed mortality can slow the recovery rate in low burned areas.

Topographic factors were found to have substantial impacts on post-fire vegetation recovery. The study sites, located in the Northern Hemisphere at mid to high latitudes, experienced notably weaker light conditions on northern slopes compared to southern slopes [13]. During the recovery period, vegetation on southeast- and south-facing slopes in Daxigou recovered rapidly (Table 3). This indicated that vegetation recovery was faster on warmer slopes in the first post-fire years. This result is in line with that of Wilson et al. [64], who discovered that vegetation had a higher recovery rate when the temperature was higher. However, in Yalihe, where sunny slopes were absent, this trend was not observed. Regarding slope gradients, vegetation on slopes ranging from 6 to 25° in both Daxigou and Yalihe showed rapid recovery during the recovery period (Table 4). These slopes benefit from efficient drainage systems and increased sunlight exposure, which promote seed germination [62]. Additionally, elevation was found to have a significant impact on post-fire vegetation recovery. Vegetation in high-altitude regions of Daxigou and Yalihe recovered rapidly (Table 5). This is because an increase in elevation leads to a higher amount of precipitation [65].

Taking into account the results and features of NDVI and EVI mentioned hitherto, these findings affirm that after 4–7 years of wildfires, the average recovery rate of EVI is higher than that of NDVI (Table 3, Table 4 and Table 5). EVI has been regarded as a modified vegetation index with high biomass sensitivity and vegetation monitoring capability through canopy background signal connection and mitigation of atmospheric effects [66,67,68,69]. Previous research studies have consistently demonstrated that the EVI surpasses the NDVI when analyzing post-fire recovery [70]. Consequently, researchers have employed EVI on a large scale to assess ecosystem disturbance in order to avoid saturation issues in areas with high biomass levels [71].

It cannot be disregarded that the recovery processes of forest vegetation were disrupted during the fourth recovery period in Yalihe, as attested by a decrease in the recovery rate of NDVI and EVI on all terrains (Table 3, Table 4 and Table 5). This disruption might have been caused by small local fires, although they were not recorded.

The correlation analysis demonstrated that the dependence between the recovery of vegetation and moisture content declined with the increase in time from the fire event (Figure 6). It was discovered that moisture content had a significant influence on vegetation recovery within the first three years after wildfires. Areas with high moisture content recovered rapidly.

In the final year of the research period (2024), vegetation indices in all burned areas recovered to over 75% of their pre-fire levels. To determine the extent of vegetation recovery in the Daxigou and Yalihe study sites, a field investigation was carried out on 20 May 2024 to validate the recuperation of burned areas. Our findings suggested that the dominant vegetation species regenerating at the periphery of the fire sites encompassed Betula platyphylla, Larix gmelinii, Pinus pumila, Rhododendron dauricum, and Vaccinium vitis-idaea (Figure 7a,b). In the center of the fire sites, there were numerous fallen logs and dead trees, with very little regenerated vegetation present (Figure 7c,d). In reality, the rate of vegetation recovery has not reached 75% of the pre-fire level.

Figure 7. Photos of the recovery status at the periphery and central of the fire sites in Daxigou and Yalihe. (a) indicated that the periphery of the Daxigou fire site, (b) indicated that the periphery of the Yalihe fire site, (c) indicated that the central of the Daxigou fire site, and (d) indicated that the central of the Yalihe fire site. The photo taken on 20 May 2024.

Even though more than 4–7 years have elapsed since the fire occurred, this study did not offer a timeframe for the complete recovery of the vegetation index to pre-fire levels. Hao et al. [58] discovered that it took at least 11 years for the vegetation index to recover to the pre-fire level after a wildfire in shrublands. Hence, we will continue to monitor the recovery progress of the study areas. Compared to the Copernicus 30 m Digital Elevation Model (DEM), the ALOS PALSAR DEM at 12.5 m offers higher spatial resolution, enabling more precise extraction of topographic information in the study area [72]. Future research directions can monitor post-fire forest recovery and vegetation regeneration over longer periods by combining data with higher spatial resolution (e.g., 1 m or 2 m) to explore the disparities between the recovery of vegetation indices and actual field restoration. Furthermore, in future studies, the impact of climate and soil change on vegetation recovery can be examined.

5. Conclusions

This study traced post-fire vegetation recovery dynamics by employing four groups of spectral vegetation indices and considering the fire severity and the topography features in two areas burned by lightning-ignited fires in the Great Xing’an Mountains. The NDVI, EVI, MSI, and NBR indicated general tendencies in post-fire vegetation dynamics. Our results suggest that the vegetation recovery situation in Siberian dwarf pine stands presented a positive trend year by year if no new fires occurred. The rate of vegetation recovery was controlled strongly by fire severity, topographic features, and moisture content. Vegetation demonstrated rapid recovery in areas with high and moderate burn severity, while low-severity burned areas experienced either delayed or negative recovery during the first three post-fire years. Large-scale delayed mortality was observed in low-severity burned areas. Vegetation in moderate-severity burned areas reached pre-fire levels more quickly compared to low-severity burned areas. Topographic features played a crucial role in post-fire regeneration, with vegetation on southeast- and south-facing slopes showing rapid recovery. Slopes ranging from 6 to 25° also exhibited rapid recovery, as did vegetation in high-altitude regions. Moisture content emerged as a critical factor influencing the recovery rate within the first three years after wildfires, with areas of higher moisture content recovering more rapidly. Our results could contribute to comprehending the heterogeneity of regional forest recovery and the trend of ecological recovery in fire-prone shrublands. This can provide a scientific basis for forest management and the restoration of post-fire ecosystems.