Next Article in Journal
Estimating Carbon Stocks and Biomass Expansion Factors of Urban Greening Trees Using Terrestrial Laser Scanning
Next Article in Special Issue
Forest Burned Area Detection Using a Novel Spectral Index Based on Multi-Objective Optimization
Previous Article in Journal
Modelling Soil Organic Carbon as a Function of Topography and Stand Variables
Previous Article in Special Issue
Emissions Released by Forest Fuel in the Daxing’an Mountains, China
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 13
Issue 9
10.3390/f13091392
forests-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

How Environmental Factors Affect Forest Fire Occurrence in Yunnan Forest Region

by
Zheng Zhu
1,
Xiaofan Deng
1,
Fan Zhao
1,*,
Shiyou Li
2 and
Leiguang Wang
1
1
College of Big Data and Intelligent Engineering, Southwest Forestry University, Kunming 650224, China
2
College of Civil Engineering, Southwest Forestry University, Kunming 650224, China
*
Author to whom correspondence should be addressed.
Forests 2022, 13(9), 1392; https://doi.org/10.3390/f13091392
Submission received: 20 June 2022 / Revised: 16 August 2022 / Accepted: 26 August 2022 / Published: 31 August 2022
(This article belongs to the Special Issue Fire Ecology and Management in Forest)
Browse Figures
Review Reports Versions Notes

Abstract

:
Forest fire is an ecosystem regulating factor and affects the stability, renewal, and succession of forest ecosystems. However, uncontrolled forest fires can be harmful to the forest ecosystem and to the public at large. Although Yunnan, China is regarded as a global hotspot for forest fires, a general lack of understanding prevails there regarding the mechanisms and interactions that cause forest fires. A logistic regression model based on fire points in Yunnan detected by satellite in 2005–2019 was used to estimate how environmental factors in local areas affect forest fire events. The results show that meteorology is the dominant cause of the frequent forest fires in the area. Other factors of secondary importance are the daily minimum relative humidity and the daily maximum temperature. When using the logistic regression model based on the data of fire points in Yunnan over the period 2005–2019, the key threshold for the daily minimum relative humidity is 28.07% ± 11.85% and the daily maximum temperature is 21.23 ± 11.15 °C for a forest fire probability of 50%. In annual and monthly dynamic trends, the daily minimum relative humidity also plays a dominant role in which combustible substance load remains relatively stable from January to March, and the impact on forest fire becomes greater in April, May, and June, which plays a secondary role compared with the interannual climate. The maximum daily temperature ranks third in importance for forest fires. At the county level, minimum relative humidity and maximum temperature are the top two factors influencing forest fires, respectively. Meanwhile, the differences in forest fire points between counties correspond to the pathways of the two monsoons. This study applies quantitative expressions to reveal the important environmental factors and mechanisms that cause forest fires. The results provide a reference for monitoring and predicting forest fires.

1. Introduction

Forest fires are a key regulator of the ecological environment and an important disturbing factor in local forest landscapes [1,2,3]. Owing to climate change, the ecological benefits of forest fires and the main controlling factors are now of significant interest. Over the past decades, the effects of climate conditions, combustible substances, and human activities that cause forest fires have been investigated worldwide [4,5,6,7], with the focus being on the global scale of forest fire hotspots, including the northern forests of North America and northern Eurasia [8,9,10,11,12], tropical forest and savanna [13,14,15,16,17], and Mediterranean landscapes in Europe and Australia [18,19,20]. However, relatively few studies have considered forest fire in Southwestern China, despite this area having a high rate of forest fires. As a result, the mechanisms and interactions leading to forest fires in this region are poorly understood.
Remote-sensing multi-source data [21], combined remote sensing and ecosystem modeling [22], provincial statistical data [23], and county-level statistical data [24] show that Southwest China is a region with a high prevalence of forest fires. Yunnan is a province in Southwest China with frequent forest fires, where the main flammable species include the widely distributed Pinus yunnanensis forest and mixed forests of green broad-leaved trees [25]. In Yunnan, the monsoon is a key factor influencing forest fires. The humid summer and autumn climate is a major feature of monsoons from the western Pacific Ocean and the Indian Ocean [26]. In winter and spring, the monsoon recedes and is replaced by warm, dry southwesterly winds [27]. Both ground monitoring and remote sensing indicate that this is the season of frequent forest fires in Yunnan [28]. Human activity is another major cause of frequent forest fires in Yunnan. In fact, slash-and-burn agriculture has been practiced in Yunnan for a long time [29,30], and many Chinese cultural traditions (e.g., burning joss paper, burning incense, and setting off firecrackers) are potential sources of fire.
The environmental variables that affect forest fire occurrence in Yunnan have rarely been investigated, so limited information is available about the causes and effects of forest fires in Yunnan. By using a remote-sensing method, Li et al. [31] determined that the distance of residences and roads from the forest zone is a key factor in the occurrence of forest fires. Xiao et al. [32] investigated lake sediments in Yunnan and found evidence of forest fire events over the last 20,000 years. They pointed out that climate conditions, combustible substances, and human activities played a crucial role in the occurrence of forest fires. Based on three years of county-level forest fire records from various provinces of China, Ying et al. [24] concluded that the presence of combustible substances was the main factor influencing the occurrence of forest fires in Yunnan. However, these studies did not consider daily meteorological data and combustible vegetation of local areas, nor did they consider multiple natural and social factors, and their related studies did not explore the mechanisms by which spatial and temporal variation in the occurrence of forest fires is associated with certain critical environmental changes.
The increasingly significant characteristics of global climate change have brought great challenges to the accurate prediction of forest fires. Therefore, it is necessary to analyze the main influencing factors in forest fire occurrence based on the background characteristics of the study area and then revise or reconstruct the original forest fire prediction model on this basis to overcome this challenge. This study considers fire points in Yunnan monitored by satellite from 2005 to 2019 and the environmental conditions at the outset of forest fires, including daily weather data, combustible substance type, terrain, and social factors, to thoroughly investigate the occurrence of forest fires in Yunnan. We also derive quantitative expressions to understand important environmental factors that affect the occurrence of forest fires and the mechanisms that cause forest fires.
Based on the above reasons, this study will consider two points to fully explore forest fires in Yunnan: First, we comprehensively consider various environmental conditions when the forest fire occurs and use quantitative expressions to reveal the important environmental factors of forest fire occurrence. The main consideration is the threshold division of environmental factors; the second is to find out the main environmental factor variables by exploring the temporal and spatial trend of forest fires so as to determine the main environmental variables of forest fires in Yunnan. The study is divided into the following sections: (1) key environmental factors and their thresholds, (2) interannual trends of forest fires, (3) intra-annual trends of forest fires, and (4) trends of spatially heterogeneous forest fires. The results of the study provide suggestions for fire prevention.

2. Materials and Methods

2.1. Study Area

Yunnan is a province in Southwest China and the provincial capital is Kunming. Its administrative area includes eight prefecture-level cities and eight autonomous prefectures, with a total of 129 county-level divisions (Figure 1). Its geographical scope is 97°31′ E–106°11′ E, 21°8′ N–29°15′ N, and the overall terrain is generally high in the northwest and low in the southeast, with altitudes ranging from 76.4 m in the southeast to 6740 m in the northwest, dominated by plateau and mountainous areas, with typical vertical geographic features and rich species diversity, making it one of the top 25 biodiversity hotspots in the world [33,34].
Yunnan has a subtropical and tropical monsoon climate, influenced by monsoons from the Western Pacific Ocean and the Indian Ocean, with distinct wet and dry seasons. The dry season lasts from November to April, which is the high-frequency period of forest fires. Additionally, Yunnan has abundant forest resources; the National Bureau of Statistics of China reports that, as of 2017, Yunnan has 19.1419 million hm2 of forest area, a forest coverage rate of 50.03%, and a forest volume of 169.30919 million m3. The main tree species in the Yunnan Forest Zone are Pinus yunnanensis, Pinus armandii, Abies ernestii, and Quercus L.
According to the records of the Yunnan Provincial Fire Brigade, Yunnan had the most forest fire events in the country [35]. For this study, we downloaded the entire MODIS Collection 6.1 thermal anomaly dataset covering all of Yunnan from 2005 to 2019, which was obtained from the NASA Fire Information Resource Management System (NASA FIRMS) (https://earthdata.nasa.gov (accessed on 13 September2021)). The coordinates with 95% confidence level or higher were selected because the hotspots collected by satellites were not necessarily the actual fire points. A total of 8542 fire points were further confirmed by the actual ground fire points. Upon considering the long duration of large-scale wildfire, which was recorded by satellites, 4649 fire points (Figure 1) were finally confirmed by comparing their longitude and latitude range and time span and identifying points with a time span of no more than 24 h and a distance of no more than 0.01 in longitude or latitude coordinates as the same wildfire.

2.2. Environmental Factors

The terrain data were obtained from SRTM 90 m digital elevation model (DEM) digital altitude database (https://srtm.csi.cgiar.org/ (accessed on 12 September 2021)), which was mainly mapped by NASA and the National Mapping Agency of the US Department of Defense through the Shuttle radar terrain [36]. The DEM image of Yunnan was obtained by cropping the data in ArcGIS 10.7. The fire points were marked into DEM images according to longitude and latitude, and the corresponding altitude, slope, and aspect of each point were extracted. The slope aspect was divided into NS-aspect and EW-aspect. The NS-aspect and EW-aspect covered a range of 0° to 180°, and the direction was from north to south and from east to west. The meteorological data used were maximum temperature (Tmax), precipitation (Preci), minimum relative humidity (RHmin), maximum wind speed (WSmax), and sunshine duration (SSD), which were obtained from the website of the China National Meteorological Information Center (http://data.cma.cn/en (accessed on 8 June 2020)), which has 25 basic stations (Figure 1). Meanwhile, considering the combustion properties of local combustible substances, the annual cumulative gross primary productivity (GPP) of each fire point [37] was calculated and extracted, starting from the year corresponding to the date of the fire points. These data were obtained from NASA’s MO/YD17A2 product (https://search.earthdata.nasa.gov/ (accessed on 15 September 2021)).
Considering the difference in combustible substance types, 30 m global ground cover type data (https://www.webmap.cn/ (accessed on 15 September 2021)) was used to extract the combustible substance type corresponding to each fire point. The intensity of human activity was estimated by using the distances from fire points to residences and road networks, so the 1:250,000 National Basic Geographic Information Database (https://www.webmap.cn/ (accessed on 15 September 2021)) was imported into Arcgis 10.7 to extract the distances to residences and roads. Table 1 lists the 12 factors used in this study.

2.3. Research Method

Estimating the probability of wildfire from only existing data requires random non-fire points and fire points for the sample data. As per previous studies [37,38,39], the ratio of fire points to non-fire points was adjusted to 1:1.5. To do this, we added 6974 random non-fire points randomly generated in ArcGIS 10.7, and the dates of these non-fire points were randomly assigned using Excel to ensure that all points were unique in terms of time and location. The environmental factor was then extracted in the same way as the fire point samples. The logistic regression (LR) model reveals the relationship between wildfire ignition and environmental factor, and it has been widely applied to study wildfires [40,41]. Although researchers have used various methods to study forest fire probability ignition [42,43], the LR model has been used more often in empirical studies because of its simplicity and its reliable results. Most importantly, the LR model can be used to establish, based on the data, mathematical equations for the forest fire probabilities and other environmental variables and then to determine the threshold values of the corresponding variables for 50% forest fire probability and the partial derivatives when the forest fire probability is 50%. Therefore, an LR model with 12 environmental variables was established to clarify the relationship between each environmental variable and forest fire probability [44]:
ln ( P 1 − P ) = β 0 + ∑ j = 1 n β j x j
where P is the forest fire probability and 1 − P is the probability of no forest fire, β j is the regression coefficient of environment variable j, x j is environment variable j, and n is the total number of environment variables.
The receiver–operator characteristic curve is often used to evaluate the accuracy of the classification model and to calculate the optimized threshold [45]. It uses a horizontal coordinate of specificity (a positive class misclassified as a negative class) and a vertical coordinate of sensitivity (a positive class judged as a positive class), and the area under the curve (AUC) of the receiver–operator characteristic curve serves to evaluate the accuracy of the model. The range of AUC is 0–1, and the model has good predictive ability when AUC > 0.8 [46,47,48]. Given the possible collinearity problem between environmental variables, which decreases the accuracy of the model, we apply a multiple collinearity diagnosis to eliminate the relevant variables. The random forest algorithm serves to rank the importance of the environmental variables filtered by the model. The importance of each explained variable is expressed in terms of the decrease in the AUC when the variables are randomized 500 times [49,50].
The mathematical interpretation of the LR model leads to two important features of the derived mathematical equation, which should be further analyzed. The first important feature is the marginal threshold of each environmental variable when the forest fire probability is 50%, which can be calculated in the reverse direction by using mathematical equation. When the forest fire probability is 50%, the marginal value of each environmental variable can be calculated by using the practical values of all other variables in the model. This is the marginal threshold of the given environmental variable. The second important feature is to determine the partial derivative of the curve when the forest fire probability is 50%; this partial derivative reflects how the environmental variables affect the 50% forest fire probability. The larger the partial derivative, the greater the influence of environmental factors on the occurrence of forest fires. Finally, all statistical analyses were done using SPSS.26 (IBM, Chicago, IL, USA) and Rx 64 4.1.2 (Auckland University, Auckland, New Zealand).

3. Results

In this study, the study area and data samples were first determined, the LR model was further constructed to obtain the environmental factors with no significant correlation, and the corresponding marginal thresholds and corresponding partial derivatives were calculated. Finally, the relative importance of each factor was displayed by the random forest method. Proportionally, the following conclusions are explored from the total data samples of the whole year, the data samples of annual and monthly dynamics, and the data samples of the county scale.

3.1. Spatiotemporal Distribution and Causes of Fire Points

A total of 4649 forest fire events were detected in 129 counties of Yunnan Province during the period 2005–2019, with great spatial variability. Specifically, fire points were concentrated in Northwest Yunnan, Southwest Yunnan, and Southeast Yunnan (Figure 2a) On average, 310 fire events occurred per year, although 842 fires occurred in 2010, accounting for more than 18% of all forest fires (Figure 2c). Forest fires mainly occurred between December and June, with more than 99% of forest fires occurring in the first half of the year (Figure 2d). The fire season in Yunnan was from 1 December to 15 June, and the number of forest fires in March accounted for more than 26% of the annual number of forest fires.
Multicollinearity and significance tests (p < 0.05) were performed on the full samples, producing a total of 10 independently correlated variables (see Table 2). According to the results of the LR model and random forest ranking, it is shown that the AUC of the LR model constructed from the full sample was 90.82%, and the most important environmental factor was RHmin (Figure 3), which correlates negatively with forest fire occurrence and has a key threshold of 28.07% ± 11.85% (Table 2) at 50% forest fire probability. The correlation partial derivative indicates that a 1% decrease in RHmin increases the marginal forest fire probability by 3.04%. The second major factor is Tmax (Figure 3), followed by GPP, with the relative importance between the two being less than 0.1%. For these variables, the key threshold for 50% forest fire probability is 21.23 ± 11.15 °C and 2.58 × 103 ± 1.42 × 103 kg K, respectively. The associated rates of change of the partial derivative are 1.74% °C−1 and 0.03% (kg K)−1, which both correlate positively with forest fire probability. This section may be divided by subheadings. It should provide a concise and precise description of the experimental results, their interpretation, as well as the experimental conclusions that can be drawn.

3.2. The Annual and Monthly Dynamic Trends of Environmental Factors

In the period 2005–2019, the AUC of the LR model exceeds 90% in all years, and RHmin is the most important environmental factor affecting the forest fire probability in the model in all years (Figure 4a). GPP and Tmax are the second and the third most important environmental factors, respectively. In most years, the relative importance of RHmin exceeds 50% (Figure 3), and the relative importance of RHmin varies significantly between years, with 2010 having the largest relative importance, with 72% of the total, and 2018 having the lowest at 33.46% of the total. The Spearman correlation coefficient for RHmin and GPP indicates that no significant relationship exists between the two (p = 0.416), RHmin correlates negatively with Tmax (r = −0.67, p = 0.007), and the corresponding values of RHmin threshold and forest fire for 50% of the train were not significant, and their inter-annual variation is not significant (Figure 4b,c).
Between January and June, over 98% of forest fires occurred in the first half of the year (Figure 2c), whereas the results for the other months are not statistically significant because of the small amount of data. Therefore, we use herein the LR model from January to June to study the monthly dynamics of forest fires. RHmin is the most important environmental factor from January to May (Figure 5a), during which time the AUC of the related model exceeds 90% (except for April, which is 85%). GPP is the second-most important among the relevant models from April to May (Figure 5a), and it becomes the most important environmental factor in June, with an AUC of 93.5% for this model. No significant correlation exists between these two factors (p = 0.208), and the variation of RHmin and the threshold of 50% forest fire probability is more significant, reaching its maximum threshold in June. The GPP threshold at 50% forest fire probability is approximately 2.7 × 103 (kg K) (Figure 5b) and does not change significantly from January to June. The partial derivative of RHmin and GPP at 50% forest fire probability also does not change significantly (Figure 5b,c, respectively).

3.3. Spatial Heterogeneity of Environmental Factors

This study establishes 94 county-level LR models; 23 other counties have too few fire points to be statistically significant. Thus, the spatial heterogeneity of the environmental factors was analyzed by using the 94 abovementioned county-level LR models. The AUCs of all these county-level LR models exceed 90%, and the top three environmental factors for the county-level models are RHmin, Tmax, and altitude (Figure 6a). RHmin plays a dominant role in most county-level models, with an overall bias toward Northwest Yunnan, Southwest Yunnan, and Southeast Yunnan. Tmax is the dominant factor in Central Yunnan and in a few regions in Northwest Yunnan and Northeast Yunnan. Altitude is the dominant factor in Central Yunnan and in a few counties in Northeast and Southeast Yunnan. A correlation analysis of 63 county-level models reveals a significant negative correlation between RHmin and the relative importance coefficient of Tmax (r = −0.61, p = 0.003), whereas no significant relationship appears between the other two pairs of correlation factors. The correlation coefficients of RHmin and altitude are not significant in 54 counties (p = 0.372) and Tmax and altitude are not significantly correlated in 40 counties (p = 0.235; Figure 6b–d).
At 50% forest fire probability, the threshold of RHmin trends downward from Northwest Yunnan, through Southwest Yunnan and Southeast Yunnan, and finally to Central Yunnan (Figure 7a), where the partial derivative RHmin < −8 mainly occurs in Northwest Yunnan, Southeast Yunnan, Southwest Yunnan, and in the areas around the Lancang River and the Nujiang River, which indicates that for every 1% decrease in RHmin, the marginal forest fire probability of 50% in this region increases by at least 8% (Figure 7b). The high Tmax threshold mainly occurs in Central Yunnan, Northwest Yunnan, and Southwest Yunnan, decreasing in these three regions upon approaching the surrounding area. The highly correlated partial derivative is concentrated in Central Yunnan and Southwest Yunnan, and its marginal variation is much less than that of RHmin (Figure 7c,d). Compared with RHmin and Tmax, the altitude threshold for 50% forest fire probability is relatively large, and the degree of variation of the threshold and the related partial derivative reveals no obvious regularity. Overall, the variation of threshold and the related partial derivatives differ between counties (Figure 7e,f).

4. Discussion

Based on previous studies, two main driving mechanisms are responsible for igniting general forest fires [18,51]: one is drought, which generally triggers forest fires in areas with a copious and spatially continuous accumulation of combustible substance, and the other is the fire-source mechanism, which involves ignition of a combustible substance by other means, thus triggering a forest fire [39,52,53]. The latter mainly occurs in ecosystems with relatively little accumulation of combustible substance and relatively low productivity [54,55]. The climatic characteristics of Yunnan are mainly tropical and subtropical, which are more consistent with the drought-driven mechanism.
The results of this study show that RHmin is the most important factor causing forest fire, followed by maximum temperature, whereas other factors, such as GPP and altitude, make limited contributions (Figure 3). On an interannual scale, RHmin is also an important factor that causes forest fires throughout the year; the second most important interannual factor is GPP (Figure 4). In addition, the importance of RHmin undergoes a significant interannual variation, which indicates that other factors, especially vegetation cover and the accumulation of combustible substance, were relatively more variable over the time interval studied [56].
According to previous studies, most of the forest fires in the study were anthropogenic [57], suggesting that most of the farmland and forest in Southern China coexist in the form of a mosaic, with this mixed landscape being particularly common in mountainous areas. Forest fires due to agricultural activities were very common [58], and agricultural activities in Yunnan were also the main cause of forest fires during this study interval, especially at the peak of the farming season (April to June). GPP becomes the second-most-important factor affecting the occurrence of forest fire in April and May and surpasses RHmin as the most important factor in June. April to June is the growth period of most vegetation, so the combustible load increases during this period. The significantly higher summertime maximum temperature then makes forest fires easy to trigger. Traditional festivals, such as the Qingming Festival in China, which involves burning incense at graves, can then lead to numerous forest fires. In months other than June (i.e., January to May), RHmin is the main factor affecting the occurrence of forest fire, which indicates that the combustible substance in the region was suffering from a prolonged drought, thus establishing environmental conditions ideal for forest fires.
At the county level, RHmin and Tmax may play a spatially complementary role in determining the exact location of forest fires (Figure 6). With the influx into Yunnan of monsoons from the Western Pacific Ocean and the Indian Ocean, RHmin increases significantly [59,60], and the relative importance of Tmax becomes more pronounced in Central and Southwestern Yunnan, where frequent droughts occur [61,62]. Meanwhile, the threshold of RHmin decreases upon going from Northwest Yunnan to Southeast Yunnan, corresponding to the path of monsoons. Given the change in the threshold gradient, the load of vegetation and other combustible substance in this area is relatively low, therefore, drier climatic conditions are required to initiate forest fires, and the initiation of forest fire depends more on the combustibility level of the combustible substance. According to previous studies, the combustibility of different tree species varies significantly [63,64]; oil pine forests are more likely to ignite, which strongly affects the forest fire probability in the region. Altitude dominates as a fire-causing factor in a few county-level models, which might be due to human activities in these tropical forest zones, bringing fire sources and thus triggering forest fires [65]. Additionally, in alpine areas, arson also occurs to maintain grazing grassland [66].

5. Conclusions

In Yunnan, the occurrence of forest fires has been controlled by local environmental factors and is driven by meteorological characteristics, among which RHmin is the dominant factor. Based on the LR model established by data of fire points in Yunnan from 2005 to 2019, the key threshold of RHmin was 28.07% ± 11.85% at the 50% level of forest fire probability. RHmin was the most important environmental factor affecting the occurrence of forest fires in Yunnan over the years studied. The importance of Tmax for forest fire events ranks second. The key threshold of Tmax over the period studied was 21.23 ± 11.15 °C, and the peak forest fire period in Yunnan was in spring and summer, when Yunnan has its dry season. Thus, the combustible substance was relatively dry, so Tmax easily became a natural source of forest fire. RHmin and Tmax are spatially complementary, and the path of monsoons from the Western Pacific Ocean and the Indian Ocean corresponds to areas with the highest Tmax threshold. The results of this study allow us to use the key thresholds to determine the local early warning signs of forest fire and provide suggestions for preventing forest fires.
Based on this analysis of the results of this study, we propose several fire protection measures for Yunnan. First, in the LR model of data from Yunnan during 2005–2019, the most important environmental factor is RHmin. Therefore, forest fire monitoring and risk assessment should focus more on changes in the RHmin threshold and daily relative humidity, especially in the Eastern and Western regions of Yunnan, where monsoons have the greatest influence. Fire-prevention services should also design the distance between regional roads and the forest zone to improve forest fire prevention measures. Controlling the sources of fire in the forest zone is also a vital step to curb fire outbreaks. According to recent statistical data, human-caused forest fires have become more frequent [21], even exceeding the number of forest fires caused by climate change [67], and the use of fires for farming and culture in spring and summer, especially in the agricultural and forest lands, should be a major concern for the authorities. To better understand the combustibility of combustible substances in different regions, combustibility experiments should be implemented on different plant species to understand how they adapt to different degrees of forest fire, and also to raise public awareness of forest fire prevention through related scientific activities. Additionally, the management strategy should be tailored to the characteristics of each region (e.g., the strong monsoons of the Western and Southeastern parts of Yunnan), and meteorological data reports should be fully exploited. For the central plateau of Yunnan and the area where the three rivers merge, forestry engineering should be adopted to reduce the combustible substance load by appropriately adjusting the forest stand structure. For a tropical forest zone and alpine grassland areas, more reasonable management measures of the forest zone and livestock should be implemented to reduce forest fires.

Author Contributions

Conceptualization, Z.Z. and F.Z.; funding acquisition, F.Z.; writing—reviewing and editing, Z.Z., X.D. and F.Z.; field investigation, experiment, and data analysis, Z.Z., S.L., L.W. and F.Z.; All authors have read and agreed to the published version of the manuscript.

Funding

This study was financially supported by the National Natural Science Foundation of China (Grant No. 32160374) and Yunnan Fundamental Research Projects, China (Grant No. 20210AT070045).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Krawchuk, M.A.; Moritz, M.A. Constraints on global fire activity vary across a resource gradient. Ecology 2011, 92, 121–132. [Google Scholar] [CrossRef] [PubMed]
  2. Dantas, V.; Batalha, M.A.; Pausas, J.G. Fire drives functional thresholds on the savanna–forest transition. Ecology 2013, 94, 2454–2463. [Google Scholar] [CrossRef]
  3. Baker, A.G.; Catterall, C. Managing fire-dependent vegetation in Byron Shire, Australia: Are we restoring the keystone ecological process of fire? Ecol. Manag. Restor. 2016, 17, 47–55. [Google Scholar] [CrossRef]
  4. Chuvieco, E.; Giglio, L.; Justice, C. Global characterization of fire activity: Toward defining fire regimes from Earth observation data. Glob. Chang. Biol. 2010, 14, 1488–1502. [Google Scholar] [CrossRef]
  5. Bowman, D.M.; Balch, J.K.; Artaxo, P.; Bond, W.J.; Carlson, J.M.; Cochrane, M.A.; Antonio, C.M.; Defries, R.S.; Doyle, J.C.; Harrison, S.P.; et al. Fire in the Earth system. Science 2009, 324, 481–484. [Google Scholar] [CrossRef] [PubMed]
  6. Aldersley, A.; Murray, S.J.; Cornell, S.E. Global and regional analysis of climate and human drivers of wildfire. Sci. Total Environ. 2011, 409, 3472–3481. [Google Scholar] [CrossRef] [PubMed]
  7. Hantson, S.; Pueyo, S.; Chuvieco, E. Global fire size distribution is driven by human impact and climate. Glob. Ecol. Biogeogr. 2015, 24, 77–86. [Google Scholar] [CrossRef]
  8. Rother, M.T.; Grissino-Mayer, H.D. Climatic influences on fire regimes in ponderosa pine forests of the Zuni Mountains, NM, USA. For. Ecol. Manag. 2014, 32, 69–77. [Google Scholar] [CrossRef]
  9. Chen, H.; Hu, Y.; Chang, Y.; Bu, R.; Li, Y.; Liu, M. Changes of forest fire regime and landscape pattern under different harvesting modes in a boreal forest of Northeast China. J. Arid Land 2015, 7, 841–851. [Google Scholar] [CrossRef]
  10. Whitman, E.; Batllori, E.; Parisien, M.A.; Miller, C.; Coop, J.D.; Krawchuk, M.A.; Chong, G.W.; Haire, S.L. The climate space of fire regimes in north-western North America. J. Biogeogr. 2015, 42, 1736–1749. [Google Scholar] [CrossRef]
  11. Liu, Q.; Shan, Y.; Shu, L.; Sun, P.; Du, S. Spatial and temporal distribution of forest fire frequency and forest area burnt in Jilin Province, Northeast China. J. For. Res. 2018, 29, 1233–1239. [Google Scholar] [CrossRef]
  12. Parks, S.A.; Holsinger, L.M.; Panunto, M.H.; Jolly, W.M.; Dobrowski, S.Z.; Dillon, G.K. High-severity fire: Evaluating its key drivers and mapping its probability across western US forests. Environ. Res. Lett. 2018, 13, 044037. [Google Scholar] [CrossRef]
  13. Alencar, A.; Asner, G.P.; Knapp, D.; Zarin, D. Temporal variability of forest fires in eastern Amazonia. Ecol. Appl. 2011, 21, 2397–2412. [Google Scholar] [CrossRef] [PubMed]
  14. Nolte, C.; Agrawal, A. Linking management effectiveness indicators to observed effects of protected areas on fire occurrence in the Amazon rainforest. Conserv. Biol. 2013, 27, 155–165. [Google Scholar] [CrossRef]
  15. Smit, I.P.; Smit, C.F.; Govender, N.; Linde MV, D.; MacFadyen, S. Rainfall, geology and landscape position generate large-scale spatiotemporal fire pattern heterogeneity in an African savanna. Ecography 2013, 36, 447–459. [Google Scholar] [CrossRef]
  16. Alencar, A.A.; Brando, P.M.; Asner, G.P.; Putz, F.E. Landscape fragmentation, severe drought, and the new Amazon forest fire regime. Ecol. Appl. 2015, 25, 1493–1505. [Google Scholar] [CrossRef] [PubMed]
  17. Attorre, F.; Govender, N.; Hausmann, A.; Farcomeni, A.; Guillet, A.; Scepi, E.; Smit, P.J.; Vitale, M. Assessing the effect of management Chang.s and environmental features on the spatio-temporal pattern of fire in an African Savanna: Fire spatio-temporal pattern. J. Nat. Conserv. 2015, 28, 1–10. [Google Scholar] [CrossRef]
  18. Pausas, J.G.; Fernández-Muñoz, S. Fire regime Changs in the Western Mediterranean Basin: From fuel-limited to drought-driven fire regime. Clim. Chang. 2012, 110, 215–226. [Google Scholar] [CrossRef]
  19. Pausas, J.G.; Keeley, J.E. Abrupt climate-independent fire regime Chang.s. Ecosystems 2014, 17, 1109–1120. [Google Scholar] [CrossRef]
  20. Zhang, Y.; Lim, S.; Sharples, J.J. Modelling spatial patterns of wildfire occurrence in South-Eastern Australia. Geomat. Nat. Hazards Risk 2016, 7, 1800–1815. [Google Scholar] [CrossRef] [Green Version]
  21. Tian, X.; Zhao, F.; Shu, L.; Wang, M. Distribution characteristics and the influence factors of forest fires in China. For. Ecol. Manag. 2013, 310, 460–467. [Google Scholar] [CrossRef]
  22. Chen, A.; Tang, R.; Mao, J.; Yue, C.; Li, X.; Gao, M.; Shi, X.Y.; Jin, M.Z.; Ricciuto, D.; Rabin, S.; et al. Spatiotemporal dynamics of ecosystem fires and biomass burning-induced carbon emissions in China over the past two decades. Geogr. Sustain. 2020, 1, 47–58. [Google Scholar] [CrossRef]
  23. Yi, K.; Bao, Y.; Zhang, J. Spatial distribution and temporal variability of open fire in China. Int. J. Wildland Fire 2017, 26, 122–135. [Google Scholar] [CrossRef]
  24. Ying, L.; Han, J.; Du, Y.; Shen, Z. Forest fire characteristics in China: Spatial patterns and determinants with thresholds. For. Ecol. Manag. 2018, 424, 345–354. [Google Scholar] [CrossRef]
  25. Tang, C.Q.; He, L.Y.; Su, W.H.; Zhang, G.F.; Wang, H.C.; Peng, M.C.; Wu, Z.L.; Wang, C.Y. Regeneration, recovery and succession of a Pinus yunnanensis community five years after a mega-fire in central Yunnan, China. For. Ecol. Manag. 2013, 294, 188–196. [Google Scholar] [CrossRef]
  26. Tsai, C.L.; Behera, S.K.; Waseda, T. Indo-China monsoon indices. Sci. Rep. 2015, 5, 8107. [Google Scholar] [CrossRef]
  27. Egger, J.; Zuyu, T. A note on the dynamics of the quasi-stationary Kunming front. Meteorol. Atmos. Phys. 1992, 48, 225–229. [Google Scholar] [CrossRef]
  28. Ying, L.; Shen, Z.; Yang, M. Wildfire detection probability of MODIS fire products under the constraint of environmental factors: A study based on confirmed ground wildfire records. Remote Sens. 2019, 11, 3031. [Google Scholar] [CrossRef]
  29. Xiao, X.; Haberle, S.G.; Li, Y.; Liu, E.; Shen, J.; Zhang, E.; Yin, J.; Wang, S. Evidence of Holocene climatic Chang. and human impact in northwestern Yunnan Province: High-resolution pollen and charcoal records from Chenghai Lake, southwestern China. Holocene 2018, 28, 127–139. [Google Scholar] [CrossRef]
  30. Wang, S.; Chen, M.; Cao, R.; Cao, Q.; Zuo, Q.; Wang, P.; Yang, B.; Zhao, S. Contribution of plant litter and soil variables to organic carbon pools following tropical forest development after slash-and-burn agriculture. Land Degrad. Dev. 2019, 31, 1071–1077. [Google Scholar] [CrossRef]
  31. Ye, J.; Wu, M.; Deng, Z.; Xu, S.; Zhou, R.; Clarke, K.C. Modeling the spatial patterns of human wildfire ignition in Yunnan province, China. Appl. Geogr. 2017, 89, 150–162. [Google Scholar] [CrossRef]
  32. Xiao, X.; Haberle, S.G.; Shen, J.; Xue, B.; Burrows, M.; Wang, S. Postglacial fire history and interactions with vegetation and climate in southwestern Yunnan Province of China. Clim. Past 2017, 13, 613–627. [Google Scholar] [CrossRef]
  33. Myers, N.; Mittermeier, R.A.; Mittermeier, C.G.; Da Fonseca, G.A.; Kent, J. Biodiversity hotspots for conservation priorities. Nature 2000, 403, 853–858. [Google Scholar] [CrossRef] [PubMed]
  34. Yang, Y.; Tian, K.; Hao, J.; Pei, S.; Yang, Y. Biodiversity and biodiversity conservation in Yunnan, China. Biodivers. Conserv. 2004, 13, 813–826. [Google Scholar] [CrossRef]
  35. Hong, X.U.; Xiao, Y.; Yong, Z.H.U. Study on the forest fire danger forecast in Yunnan Province based on GIS and RS. J. Fujian For. Sci. Technol. 2007, 34, 85–88. [Google Scholar]
  36. Van Zyl, J.J. The Shuttle Radar Topography Mission (SRTM): A breakthrough in remote sensing of topography. Acta Astronaut. 2001, 48, 559–565. [Google Scholar] [CrossRef]
  37. Catry, F.X.; Rego, F.C.; Bação, F.L.; Moreira, F. Modeling and mapping wildfire ignition risk in Portugal. Int. J. Wildland Fire 2009, 18, 921–931. [Google Scholar] [CrossRef]
  38. Gray, M. Landscape-Scale Models and Maps of Fire Risk and Connectivity in the Lower Sonoran Desert. Ph.D. Thesis, Northern Arizona University, Flagstaff, AZ, USA, 2013. [Google Scholar]
  39. Guo, F.; Wang, G.; Su, Z.; Liang, H.; Wang, W.; Lin, F.; Liu, A. What drives forest fire in Fujian, China? Evidence from logistic regression and Random Forests. Int. J. Wildland Fire 2016, 25, 505–519. [Google Scholar] [CrossRef]
  40. Zeng, A.C.; Cai, Q.J.; Su, Z.W.; Guo, X.B.; Jin, Q.F.; Guo, F.T. Seasonal variation and driving factors of forest fire in Zhejiang Province, China, based on MODIS satellite hot spots. Ying Yong Sheng Tai Xue Bao J. Appl. Ecol. 2020, 31, 399–406. [Google Scholar]
  41. Guo, F.; Su, F.; Wang, G.; Sun, L.; Tigabu, M.; Yang, X.; Hu, H. Understanding fire drivers and relative impacts in different Chinese forest ecosystems. Sci. Total Environ. 2017, 605, 411–425. [Google Scholar] [CrossRef]
  42. Mitri, G.; Saba, S.; Nader, M.; McWethy, D. Developing Lebanon’s fire danger forecast. Int. J. Disaster Risk Reduct. 2017, 24, 332–339. [Google Scholar] [CrossRef]
  43. Elia, M.; Giannico, V.; Lafortezza, R.; Sanesi, G. Modeling fire ignition patterns in Mediterranean urban interfaces. Stoch. Environ. Res. Risk Assess. 2019, 33, 169–181. [Google Scholar] [CrossRef]
  44. Gao, C.; Lin, H.L.; Hu, H.Q.; Song, H. A review of models of forest fire occurrence prediction in China. Ying Yong Sheng Tai Xue Bao J. Appl. Ecol. 2020, 31, 3227–3240. [Google Scholar]
  45. Tien Bui, D.; Le, K.T.T.; Nguyen, V.C.; Le, H.D.; Revhaug, I. Tropical forest fire susceptibility mapping at the Cat Ba National Park Area, Hai Phong City, Vietnam, using GIS-based kernel logistic regression. Remote Sens. 2016, 8, 347. [Google Scholar] [CrossRef]
  46. Guo, F.; Su, Z.; Tigabu, M.; Yang, X.; Lin, F.; Liang, H.; Wang, G. Spatial modelling of fire drivers in urban-forest ecosystems in China. Forests 2017, 8, 180. [Google Scholar] [CrossRef]
  47. Hanley, J.A.; McNeil, B.J. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982, 143, 29–36. [Google Scholar] [CrossRef]
  48. Vilar del Hoyo, L.; Martín Isabel, M.P.; Martínez Vega, F.J. Logistic regression models for human-caused wildfire risk estimation: Analysing the effect of the spatial accuracy in fire occurrence data. Eur. J. For. Res. 2011, 130, 983–996. [Google Scholar] [CrossRef]
  49. Taylor, A.E. Statistical Enhancement of Support Vector Machines; Oregon State University: Corvallis, OR, USA, 2009. [Google Scholar]
  50. Janitza, S.; Strobl, C.; Boulesteix, A.L. An AUC-based permutation variable importance measure for random forests. BMC Bioinform. 2013, 14, 119. [Google Scholar] [CrossRef]
  51. Pausas, J.G.; Bradstock, R.A. Fire persistence traits of plants along a productivity and disturbance gradient in mediterranean shrublands of south-east Australia. Glob. Ecol. Biogeogr. 2007, 16, 330–340. [Google Scholar] [CrossRef]
  52. Abatzoglou, J.T.; Williams, A.P. Impact of anthropogenic climate Chang. on wildfire across western US forests. Proc. Natl. Acad. Sci. USA 2016, 113, 11770–11775. [Google Scholar] [CrossRef]
  53. Hanes, C.C.; Wang, X.; Jain, P.; Parisien, M.A.; Little, J.M.; Flannigan, M.D. Fire-regime Chang.s in Canada over the last half century. Can. J. For. Res. 2019, 49, 256–269. [Google Scholar] [CrossRef]
  54. Bajocco, S.; Koutsias, N.; Ricotta, C. Linking fire ignitions hotspots and fuel phenology: The importance of being seasonal. Ecol. Indic. 2017, 82, 433–440. [Google Scholar] [CrossRef]
  55. Alcasena, F.J.; Salis, M.; Ager, A.A.; Arca, B.; Molina, D.; Spano, D. Assessing landscape scale wildfire exposure for highly valued resources in a Mediterranean area. Environ. Manag. 2015, 55, 1200–1216. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  56. Han, J.; Shen, Z.; Li, Y.; Luo, C.; Xu, Q.; Yang, K.; Zhang, Z. Beta diversity patterns of post-fire forests in central Yunnan Plateau, southwest China: Disturbances intensify the priority effect in the community assembly. Front. Plant Sci. 2018, 9, 1000. [Google Scholar] [CrossRef]
  57. Huang, X.; Li, M.; Li, J.; Song, Y. A high-resolution emission inventory of crop burning in fields in China based on MODIS Thermal Anomalies/Fire products. Atmos. Environ. 2012, 60, 9–15. [Google Scholar] [CrossRef]
  58. van Wees, D.; van Der Werf, G.R.; Randerson, J.T.; Andela, N.; Chen, Y.; Morton, D.C. The role of fire in global forest loss dynamics. Glob. Chang. Biol. 2021, 27, 2377. [Google Scholar] [CrossRef]
  59. Lee, S.S.; Seo, Y.W.; Ha, K.J.; Jhun, J.G. Impact of the western North Pacific subtropical high on the East Asian monsoon precipitation and the Indian Ocean precipitation in the boreal summertime. Asia-Pac. J. Atmos. Sci. 2013, 49, 171–182. [Google Scholar] [CrossRef]
  60. Lee, S.S.; Wang, B. Regional boreal summer intraseasonal oscillation over Indian Ocean and Western Pacific: Comparison and predictability study. Clim. Dyn. 2016, 46, 2213–2229. [Google Scholar] [CrossRef]
  61. Zhao, Y.; Hu, Y.M.; Zhang, X.C.; Wang, L.; Chen, F.; Zhao, Z. Spatial pattern of farming distance in rural area using ESDA. Sci Geogr. Sin. 2016, 36, 760–765. [Google Scholar]
  62. Yan, Z.; Zhang, Y.; Zhou, Z.; Han, N. The spatio-temporal variability of droughts using the standardized precipitation index in Yunnan, China. Nat. Hazards 2017, 88, 1023–1042. [Google Scholar] [CrossRef]
  63. Li, S.Y.; Ma, C.L.; Luo, W.B.; Li, S.H.; Wang, X.F.; Yuan, J.J.; Yang, L. Combustibility ordering and sorting of 35 woody plant species in Kunming area. Chin. J. Ecol. 2008, 27, 867–873. [Google Scholar]
  64. LI, S.; Chang, N.; Guan, X.; Wang, Q.H.; Shu, L.F. Combustion characteristics of live branches of 15 common woody plant species in Kunming, Yunnan Province. Chin. J. Ecol. 2012, 31, 276–281. [Google Scholar]
  65. Cattau, M.E.; Harrison, M.E.; Shinyo, I.; Tungau, S.; Uriarte, M.; DeFries, R. Sources of anthropogenic fire ignitions on the peat-swamp landscape in Kalimantan, Indonesia. Glob. Environ. Chang. 2016, 39, 205–219. [Google Scholar] [CrossRef]
  66. Buntaine, M.T.; Mullen, R.B.; Lassoie, J.P. Human use and conservation planning in Alpine areas of Northwestern Yunnan, China. Environ. Dev. Sustain. 2007, 9, 305–324. [Google Scholar] [CrossRef]
  67. Liu, Z.; Yang, J.; Chang, Y.; Weisberg, P.J.; He, H.S. Spatial patterns and drivers of fire occurrence and its future trend under climate Chang. in a boreal forest of Northeast China. Glob. Chang. Biol. 2012, 18, 2041–2056. [Google Scholar] [CrossRef]
Forests 13 01392 g001 550
Figure 1. Geographical location of Yunnan and locations of major meteorological stations in Yunnan.
Figure 1. Geographical location of Yunnan and locations of major meteorological stations in Yunnan.
Forests 13 01392 g001
Forests 13 01392 g002a 550Forests 13 01392 g002b 550
Figure 2. (a) Distribution of fire points in forest as detected by satellite in Yunnan, China, 2005–2019. (b) Random non-fire locations generated by ArcGIS. (c) Number of fire points for each year, 2005–2019. (d) Number of fire points for each month, January–June, averaged over the years 2005–2019.
Figure 2. (a) Distribution of fire points in forest as detected by satellite in Yunnan, China, 2005–2019. (b) Random non-fire locations generated by ArcGIS. (c) Number of fire points for each year, 2005–2019. (d) Number of fire points for each month, January–June, averaged over the years 2005–2019.
Forests 13 01392 g002aForests 13 01392 g002b
Forests 13 01392 g003 550
Figure 3. Ranking of relative importance of environmental variables.
Figure 3. Ranking of relative importance of environmental variables.
Forests 13 01392 g003
Forests 13 01392 g004a 550Forests 13 01392 g004b 550
Figure 4. (a) Ranking of relative importance of environmental variables for each year. (b) Interannual thresholds for the most important environmental variables for each year. (c) Interannual partial derivatives of the most important environmental variables for each year.
Figure 4. (a) Ranking of relative importance of environmental variables for each year. (b) Interannual thresholds for the most important environmental variables for each year. (c) Interannual partial derivatives of the most important environmental variables for each year.
Forests 13 01392 g004aForests 13 01392 g004b
Forests 13 01392 g005a 550Forests 13 01392 g005b 550
Figure 5. (a) Ranking of relative importance of environmental variables by month in the first half of the year. (b) Thresholds of RHmin and GPP for each month in the first half of the year. (c) Partial derivatives of RHmin and GPP in each month of the first half of the year.
Figure 5. (a) Ranking of relative importance of environmental variables by month in the first half of the year. (b) Thresholds of RHmin and GPP for each month in the first half of the year. (c) Partial derivatives of RHmin and GPP in each month of the first half of the year.
Forests 13 01392 g005aForests 13 01392 g005b
Forests 13 01392 g006 550
Figure 6. (a) Most important environmental variables in different counties. (b) Ranking of relative importance of RHmin in each county. (c) Ranking of relative importance of Tmax in each county. (d) Ranking of relative importance of altitude in each county.
Figure 6. (a) Most important environmental variables in different counties. (b) Ranking of relative importance of RHmin in each county. (c) Ranking of relative importance of Tmax in each county. (d) Ranking of relative importance of altitude in each county.
Forests 13 01392 g006
Forests 13 01392 g007a 550Forests 13 01392 g007b 550
Figure 7. (a) Ranking of RHmin thresholds by county. (b) Ranking of partial derivatives of RHmin by county. (c) Ranking of Tmax thresholds by county. (d) Ranking of partial derivatives of Tmax by county. (e) Ranking of altitude thresholds by county. (f) Ranking of partial derivatives of altitude by county.
Figure 7. (a) Ranking of RHmin thresholds by county. (b) Ranking of partial derivatives of RHmin by county. (c) Ranking of Tmax thresholds by county. (d) Ranking of partial derivatives of Tmax by county. (e) Ranking of altitude thresholds by county. (f) Ranking of partial derivatives of altitude by county.
Forests 13 01392 g007aForests 13 01392 g007b
Table
Table 1. Data source.
Table 1. Data source.
Variable TypeVariable FactorData SourcesUnit
Precipitation mm
Maximum
temperature		 °C
MeteorologicalMinimum
relative humidity		China National Meteorological Information CenterRelative humidity (RH%)
Maximum wind
speed		 m/s
Sunshine duration h
Altitude m
TerrainSlopeDigital elevation database°
NS-Aspect °
EW-Aspect °
Combustible substanceGPPNASAkg C
Human activityDistance from residence1:250,000 National Basic Geographic Information Databasem
Distance from road m
Table
Table 2. Fitting eigenvalues for full data logistic regression model.
Table 2. Fitting eigenvalues for full data logistic regression model.
VariablesRegression CoefficientsThreshold (±Standard Deviation)
of 50% IgnitionProbability		Partial
Derivative
Precipitation−1.52 × 10−18.16 (±6.46)−3.79
Minimum relative humidity−1.22 × 10−128.07 (±11.85)−3.05
Sunshine duration4.3 × 10−27.32 (±3.43)1.07
Maximum temperature6.8 × 10−221.23 (±11.15)1.70
Maximum wind speed1.34 × 10−15.32 (±2.19)3.35
Distance from residence2.26 × 10−41.33 × 103 (±1.07 × 103)5.00 × 10−3
Distance from road9.57 × 10−52.39 × 103 (±1.79 × 103)2.50 × 10−3
Slope9.63 × 10−346.32 (±24.52)2.50 × 10−1
Altitude−6.49 × 10−52.45 × 103 (1.12 × 103)1.50 × 10−3
GPP1.28 × 10−32.59 × 103 (1.22 × 103)3.25 × 10−2
Constant−2.23--
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Zhu, Z.; Deng, X.; Zhao, F.; Li, S.; Wang, L. How Environmental Factors Affect Forest Fire Occurrence in Yunnan Forest Region. Forests 2022, 13, 1392. https://doi.org/10.3390/f13091392

AMA Style

Zhu Z, Deng X, Zhao F, Li S, Wang L. How Environmental Factors Affect Forest Fire Occurrence in Yunnan Forest Region. Forests. 2022; 13(9):1392. https://doi.org/10.3390/f13091392

Chicago/Turabian Style

Zhu, Zheng, Xiaofan Deng, Fan Zhao, Shiyou Li, and Leiguang Wang. 2022. "How Environmental Factors Affect Forest Fire Occurrence in Yunnan Forest Region" Forests 13, no. 9: 1392. https://doi.org/10.3390/f13091392

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop