Available via license: CC BY
Content may be subject to copyright.
Forest Fire Susceptibility Modeling Using a Convolutional Neural
Network for Yunnan Province of China
Guoli Zhang
1
·Ming Wang
1,2
·Kai Liu
1,2
©The Author(s) 2019
Abstract Forest fires have caused considerable losses to
ecologies, societies, and economies worldwide. To mini-
mize these losses and reduce forest fires, modeling and
predicting the occurrence of forest fires are meaningful
because they can support forest fire prevention and man-
agement. In recent years, the convolutional neural network
(CNN) has become an important state-of-the-art deep
learning algorithm, and its implementation has enriched
many fields. Therefore, we proposed a spatial prediction
model for forest fire susceptibility using a CNN. Past forest
fire locations in Yunnan Province, China, from 2002 to
2010, and a set of 14 forest fire influencing factors were
mapped using a geographic information system. Over-
sampling was applied to eliminate the class imbalance, and
proportional stratified sampling was used to construct the
training/validation sample libraries. A CNN architecture
that is suitable for the prediction of forest fire susceptibility
was designed and hyperparameters were optimized to
improve the prediction accuracy. Then, the test dataset was
fed into the trained model to construct the spatial predic-
tion map of forest fire susceptibility in Yunnan Province.
Finally, the prediction performance of the proposed model
was assessed using several statistical measures—Wilcoxon
signed-rank test, receiver operating characteristic curve,
and area under the curve (AUC). The results confirmed the
higher accuracy of the proposed CNN model (AUC 0.86)
than those of the random forests, support vector machine,
multilayer perceptron neural network, and kernel logistic
regression benchmark classifiers. The CNN has stronger
fitting and classification abilities and can make full use of
neighborhood information, which is a promising alternative
for the spatial prediction of forest fire susceptibility. This
research extends the application of CNN to the prediction
of forest fire susceptibility.
Keywords China · Convolutional neural
network · Forest fire susceptibility · Geographic
information system · Machine learning
1 Introduction
Forest fires have caused considerable losses in global forest
resources and people’s lives and property, seriously
impacting the global ecological balance, and have received
considerable attention from countries worldwide. In recent
years, with global warming, industrialization, and human
interventions, the frequency and severity of forest fires
have been increasing significantly in many parts of the
world (Crimmins 2006; Running 2006; Hantson et al.
2015). Regional forest fire susceptibility is often affected
by many factors and has typical nonlinear and complex
characteristics; therefore, it is still a difficult task to
develop forest fire prediction models with satisfactory
accuracy (Ngoc Thach et al. 2018). Various approaches
have been developed for modeling forest fire susceptibility,
ranging from physics-based methods to statistical and
machine learning (ML) methods (Dimuccio et al. 2011;
Tien Bui et al. 2017; Leuenberger et al. 2018; Hong et al.
2019; Jaafari et al. 2019). Compared to traditional
&Ming Wang
wangming@bnu.edu.cn
1
State Key Laboratory of Earth Surface Processes and
Resource Ecology / Academy of Disaster Reduction and
Emergency Management, Faculty of Geographical Science,
Beijing Normal University, Beijing 100875, China
2
Key Laboratory of Environmental Change and Natural
Disasters, Ministry of Education, Beijing Normal University,
Beijing 100875, China
123
Int J Disaster Risk Sci www.ijdrs.com
https://doi.org/10.1007/s13753-019-00233-1 www.springer.com/13753
qualitative and statistical analysis methods, ML approaches
have shown the ability to provide better results for the
spatial prediction of wildfires (Bar Massada et al. 2013). In
the last decade, various ML algorithms—such as artificial
neural networks (Dimuccio et al. 2011; Bisquert et al.
2012; Satir et al. 2016), random forests (RF) (Oliveira et al.
2012; Arpaci et al. 2014; Pourtaghi et al. 2016), support
vector machine (SVM) (Hong et al. 2018), multilayer
perceptron neural network (MLP) (Vasconcelos et al. 2001;
Satir et al. 2016), kernel logistic regression (KLR) (Hong
et al. 2015; Bui, Le, et al. 2016), naive Bayes (Elmas and
So
¨nmez 2011; Jaafari et al. 2018), gradient boosted deci-
sion trees (Sachdeva et al. 2018), and particle swarm
optimized neural fuzzy (Tien Bui et al. 2017)—have been
successfully developed and widely applied for producing
wildfire susceptibility maps. Comparative studies of mul-
tiple ML algorithms have also been employed (Pourtaghi
et al. 2016; Cao et al. 2017; Ngoc Thach et al. 2018; Kim
et al. 2019). Therefore, advanced ML approaches are very
promising for forest fire spatial prediction. However, the
ML approaches mentioned above are pixel-based classifiers
with shallow architectures, which do not make use of the
spatial patterns that are implicit in images (Zhang et al.
2018). In addition, these classifiers directly classify the
input data without feature extraction, and representative
features cannot be mined from the input data to improve
classification accuracy.
Deep learning (DL) methods (Hinton and Salakhutdinov
2006; Lecun et al. 2015) have recently received more
attention and achieved remarkable success. Deep learning
algorithms attempt to discover multiple representation
levels (Schmidhuber 2015) and have been broadly applied
in areas such as object recognition and detection, speech
recognition, and natural language processing (Lecun et al.
2015). The convolutional neural network (CNN) (LeCun
et al. 1998), which has been recognized as one of the most
successful and widely used DL algorithms, has produced
significant improvements in the latest studies in areas such
as disaster damage detection (Muhammad et al. 2018;
Vetrivel et al. 2018), remotely sensed image classification
(Liu and Abd-Elrahman 2018; Zhang et al. 2018), and
landslide susceptibility mapping (Wang et al. 2019).
However, none of these studies evaluated the effectiveness
of CNN in the prediction of forest fire susceptibility. The
first law of geography (Tobler 1970) emphasizes that near
things are more closely related than distant things. Whether
a pixel is an ignition point should not only consider the
situation of the pixel itself, but also consider other pixels
within a certain range around the pixel. While pixel-based
classifiers may overlook certain information in spatial
patterns, the contextual-based CNN explores the complex
spatial patterns that are implicit in images (Zhang et al.
2018). The CNN can make full use of contextual
information (that is, neighborhood information) and can
discover multiple levels of representations from input data,
which is more suitable for the evolution of fire event spatial
characteristics. The DL process reveals the deep features
and can distinguish the differences between different geo-
graphical units. Therefore, there is a certain practical sig-
nificance in studying the application of the CNN algorithm
in forest fire susceptibility analysis.
Forest fire susceptibility, in this article, is defined as the
probability estimation of fire occurrence in a region. The
main objective of this study is to utilize contextual-based
CNN with deep architectures for the spatial prediction of
regional forest fire susceptibility in Yunnan Province,
China. The forest fire susceptibility model was established
based on a CNN and the hyperparameters of the model
were optimized to improve the prediction accuracy. The
performance of the proposed model was compared with
benchmark methods using several statistical measures—
Wilcoxon signed-rank test (WSRT), receiver operating
characteristic (ROC) curve, and area under the curve
(AUC).
2 Study Area and Data Collection
Yunnan Province is located in southwestern China (Fig. 1).
It is a mountainous region with over 90% mountain and
plateau landscape interspersed with less than 10% small,
scattered valley basins. The terrain slopes downward from
northwest to southeast, and elevation ranges from 0 to
6135 m. The region has a plateau-type tropical monsoon
climate, and average temperatures in the summer and
winter are 19–22 °C and 6–8 °C, respectively. The distri-
bution of precipitation throughout the seasons and regions
is extremely uneven. The winter and spring seasons from
November to April of the following year are dry seasons
with precipitation accounting for only 20% or less of the
1100 mm annual precipitation. Yunnan has abundant and
diverse forest resources, including tropical rainforests,
seasonal rainforests, subtropical evergreen broad-leaved
forests, and temperate coniferous forests. Compared with
other regions in China, the frequency of forest fires in
Yunnan is relatively high (Ying et al. 2018). The occur-
rence of forest fires in Yunnan has a strong seasonal pat-
tern, mainly concentrated in the spring from mid-February
to mid-May.
Compiling a forest fire inventory is a mandatory task for
forest fire susceptibility modeling (Tien Bui et al. 2017). In
this study, a forest fire event map was prepared using
multiple resources including historical fire reports and the
interpretation of satellite images (see Cao et al. (2017) for
details). From 2002 to 2010, a total of 7675 fires occurred,
and the number of fires in the spring was 4428, accounting
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
for 58% of the total number of fires. The occurrence of
forest fires is affected by many factors, and selecting the
appropriate influencing factors is important (Pew and
Larsen 2001; Guo et al. 2016). Fourteen forest fire influ-
encing factors were selected, and all datasets were prepared
in the form of raster/vector data, as described and listed in
Table 1.
3 Methodology
This section first provides a preview of the CNN algorithm,
and then elaborates the procedure of the susceptibility
model development through a series of processes including
data preprocessing and sample library generation, model
architectures design and parameter adjustment, and per-
formance evaluation. Finally, the detailed methodological
flowchart is described.
3.1 Preview of Convolutional Neural Network
Convolutional neural network (CNN) is one of the most
notable DL approaches and has exhibited robust perfor-
mance in feature learning for image classification and
recognition. It is a feed-forward neural network whose
parameters are trained by using the classic stochastic gra-
dient descent based on the backpropagation algorithm (Hu
et al. 2015).
Generally, the CNN consists of several building blocks
—convolutional, pooling, and fully connected layers (Ya-
mashita et al. 2018). The different types of processing
layers play different roles. The convolutional layers, which
perform linear convolution operations between the input
tensor and a set of filters, output the feature maps. Typi-
cally, each feature map is then followed by a nonlinear
activation function. The rectified linear unit (ReLU), which
performs the nonlinear transformation of the feature map
generated by the convolution layer and introduces nonlin-
earity into the system, is the most commonly used activa-
tion function.
The purpose of the convolution operation is to extract
different input layer features and achieve weight sharing.
The input and output of each stage are sets of arrays called
feature maps. For example, if the input is a 2-dimensional
image x, the input is first decomposed into a sequential array
x={x1,x2,…,xN}. The convolutional layer is defined as:
yj¼fb
jþX
i
kij xi
!
where yjdenotes the jth output for the convolutional layer
and xidenotes each input feature map. kij denotes the
convolutional kernel with the ith input map xi. * Denotes
the discrete convolution operator, bjdenotes a trainable
bias, and fis the nonlinear activation.
The pooling layers perform a subsampling operation to
reduce the dimensions of the feature maps. According to
Fig. 1 Location of Yunnan Province, the study area, in China
123
Int J Disaster Risk Sci
the maximum and average functions, the pooling layer can
be divided into the max-pooling and average-pooling lay-
ers. The fully connected layers, which are the flat feed-
forward neural network layers, provide high-level
abstraction features. They are often used at the end of the
network architecture and create the final nonlinear com-
binations of features for making the predictions by the
network. The activation function for the last fully con-
nected layer needs to be selected reasonably based on given
tasks. The softmax or sigmoid function can be used to
compute the posterior probability for each grid cell.
3.2 Data Preprocessing and Sample Library
Generation
First, appropriate forest fire influencing factors were
selected, and variables raster datasets (VRDs) and ignition
raster datasets (IRDs) were constructed through prepro-
cessing. Then, the sample libraries were collected from the
established IRDs and VRDs using an appropriate sampling
method.
3.2.1 Forest Fire Influencing Factors
Four categories of forest fire influencing factors were
considered, including topography-related, climate-related,
vegetation-related, and human activities-related variables.
ArcGIS 10.5 was employed for handling geographic data.
The effect of the topography has been considered as a
significant feature in forest fire assessment (Renard et al.
2012; Adab et al. 2013). Three topography-related influ-
encing factors—elevation, slope, and aspect—were
retrieved from the digital elevation model (DEM). The
DEM was produced by the Advanced Spaceborne Thermal
Emission and Reflection Radiometer (ASTER) GDEM
version 2 (with a 30 m pixel size). Surface roughness was
obtained from the Climate Forecast System Reanalysis
(CFSR). The climatic characteristics of an area affect the
occurrence and intensity of forest fires (Moritz et al. 2012).
Six meteorologically related influencing factors—average
temperature, average precipitation, average wind speed,
maximum temperature, specific humidity, and precipitation
rate—were obtained from the CFSR. The spatial resolution
is approximately 0.3 degree (0.312°90.312°) with a 6 h
temporal resolution. These influencing factors were cal-
culated as the spring means from March through May. Six
meteorologically related factors were finally mapped with
the inverse distance weighted (IDW) interpolation method.
The Moderate Resolution Imaging Spectroradiometer
(MODIS) normalized difference vegetation index (NDVI)
has also been identified as an important variable in forest
fire modeling (Bajocco et al. 2015). The NDVI values
reflect the vegetation’s health and essentially the fuel load
distribution (Yi et al. 2013). The MODIS NDVI monthly
Table 1 Data description of forest fire influencing factors
No. Data Scale/resolution original Unit Original data format Source
1 Elevation 30 m m Raster ASTER GDEM
a
2 Slope 30 m degree Raster
3 Aspect 30 m degree Raster
4 Average temperature 0.312°(lat/long) °C NetCDF CFSR
b
5 Average precipitation 0.312°(lat/long) kg/m
2
NetCDF
6 Surface roughness 0.312°(lat/long) Ratio NetCDF
7 Average wind speed 0.312°(lat/long) m/s NetCDF
8 Maximum temperature 0.312°(lat/long) °C NetCDF
9 Specific humidity 0.312°(lat/long) Ratio NetCDF
10 Precipitation rate 0.312°(lat/long) mm/h NetCDF
11 Forest coverage ratio 1:1,000,000 Ratio Vector Vegetation map (Zhang 2007)
12 NDVI 500 m Ratio Raster Geospatial Data Cloud
c
13 Distance to roads 1:40,000,000 km Vector Basic geographic data
d
14 Distance to rivers 1:40,000,000 km Vector Basic geographic data
e
ASTER GDEM advanced spaceborne thermal emission and reflection radiometer global digital elevation map; CFSR climate forecast system
reanalysis; NetCDF network common data form; NDVI normalized difference vegetation index
a
https://search.earthdata.nasa.gov
b
https://rda.ucar.edu
c
http://www.gscloud.cn
d
http://www.ngcc.cn
e
http://www.ngcc.cn
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
synthetic products with a 500 m resolution were obtained
from the Geospatial Data Cloud website. The clip and
raster calculator tools in ArcGIS10.5 were applied to cal-
culate the annual spring NDVI of Yunnan for the period
2002–2010. The NDVI map for each year was calculated
by taking the average NDVI of the spring months (March,
April, and May). The forest coverage data for Yunnan that
were used in this research were derived from the vegetation
map of the People’s Republic of China (1:1,000,000). The
forest coverage ratio map was calculated by the ratio of the
forest area of each pixel to the total area of the pixel. The
distance from the river network and the distance from the
road network (highways, main roads, and local roads) were
obtained by applying the buffer tool in the proximity
toolbox, which calculates the Euclidean distance from the
road and river networks. Then, maps of the proximity to
roads and rivers were produced. Figure 2shows 11 maps of
the 14 influencing factors in 2010. Not shown in Fig. 2are
maximum temperature, specific humidity, and precipitation
rate, which are discussed in more detail in Sect. 4.1.
3.2.2 Influencing Factor Evaluators
Variable selection is particularly important in the predic-
tion of forest fire susceptibility. The high dimensionality of
the training dataset may complicate the prediction process
and decrease the prediction accuracy. In this study, multi-
collinearity analysis and an information gain ratio (IGR)
were selected to evaluate the forest fire influencing factors.
Multicollinearity analysis (O’Brien 2007) was applied to
estimate the correlation between the forest fire influencing
factors. Two measures of variance—inflation factors (VIF)
and tolerances (TOL)—were used to identify the degree of
multicollinearities among the forest fire influencing factors.
A factor is considered to be multicollinear if its tolerance is
less than 0.1 or the VIF value is greater than 10 (Colkesen
et al. 2016). The IGR is an effective approach for selecting
an optimal subset of variables that can represent the whole
dataset to improve the prediction performance in forest fire
susceptibility mapping (Dash and Liu 1997; Jaafari et al.
2018). The average merit (AM) reveals the importance of
forest fire influencing factors in predicting forest fire
occurrence (Jaafari et al. 2018).
3.2.3 Building Variables Raster Datasets and Ignition
Raster Datasets
The min–max normalization process was conducted for the
influencing factor maps to avoid the potential bias caused
by the unbalanced magnitudes of factors. Rasterization was
performed by the ArcGIS model builder tool to convert all
maps to a raster format with the same pixel size (5000 m9
5000 m), the same data type (8 bit-unsigned integer), and
the same coordinate system (WGS 1984 Web Mercator),
resulting in 18,718 cells in total. Finally, the composite
bands tool was used to combine all the factor maps in a
year into one raster, which was named the VRD. All the
data of influencing factors from 2002 to 2010 were pro-
cessed in the same way and a total of nine VRDs were
finally established.
Class imbalance refers to the number of events in the
nonfire class being much greater than that in the fire class
(Lo
´pez et al. 2013). Class imbalance has a detrimental
impact on the classification performance of the CNN and
oversampling has been proved to be a robust solution for
solving this problem in DL (Buda et al. 2018). In addition,
according to the spatial characteristics of forest fire events,
a certain area near the pixel of a fire vector point may be a
fire-prone area. Therefore, to solve the class imbalance,
buffer analysis was used with the existing fire vector points
(Cao et al. 2017) to increase the number of events in the
fire class. The pixels in the 5 km buffer zone were
resampled to 1 (fire), and the pixels outside the buffer zone
were resampled to 0 (nonfire). Then, the generated raster
data—the IRD—were obtained. All fire datasets from 2002
to 2010 were processed, and nine IRDs for the corre-
sponding years were obtained. All IRDs use the same pixel
size, data type, and coordinate system as the VRDs.
3.2.4 Sample Library Generation
Since the CNN conducts its effective training in a fully
supervised manner, the training and validation sample
libraries must be constructed first. A binary classification
method was adopted for the susceptibility analysis of forest
fires, and samples were classified to either the forest fire
class or the nonfire class, with 1 representing the fire class
and 0 representing the nonfire class.
The CNN has many learnable parameters to estimate;
thus, this predictor requires more data to achieve sufficient
training. Table 2shows that there is a large annual varia-
tion in the actual number of fire points in 2002–2009 (data
from 2010 were used as the test dataset and therefore not
included in the training and validation process). A simple
random sampling method may lead to sample quantity
imbalances, while small samples would lead to overfitting
the model. Therefore, to generate more samples and keep
the sample quantity balanced, proportional stratified sam-
pling was adopted. The total number of forest fires was
multiplied by the sampling rate (0.8) to determine the
number of fire samples in every year. The same number of
nonfire samples was randomly selected, which constitutes a
total of 49,706 training samples.
After the number of samples needed for each year is
determined (Table 2), a list of XY coordinates for fire or
nonfire class was randomly generated and the locations of
123
Int J Disaster Risk Sci
Fig. 2 Forest fire influencing factor maps in 2010 for Yunnan Province, China. NDVI normalized difference vegetation index
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
the sample points were recorded—the list contains the
same number of coordinates as the sample quantity of each
year (Table 2) (XY was defined using the row and column
number of the IRD). Then the same number of windows
were defined, each with the size of 25925 pixels and the
center of the window was the XY coordinates in the XY
coordinate list. Then, the VRD in the window with 25925
pixels was extracted as a 3-dimensional array, n9n9c,
where ndenotes the row and column of each input patch
and crepresents the number of forest fire influencing fac-
tors. Each sample consisted of two parts: a 3-dimensional
array containing influencing factors; and a corresponding
ground truth label from the IRD in the central XY coor-
dinates. The dataset was divided randomly into two parts:
(1) 80% as the training samples (49,706 pixels); and (2) the
remaining 20% as the validation samples (12,426 pixels).
3.3 The Proposed Convolutional Neural Network
Architectures and Related Parameters
The architecture of the proposed convolutional neural
network (CNN) model was completed by referring to the
AlexNet model (Krizhevsky et al. 2013) and the architec-
ture and hyperparameters were tuned based on our datasets.
As mentioned above, each input patch was a 3-dimensional
data representation of size n9n9c, taking 25 925 911 as
an example. Figure 3shows the main architecture of the
prediction model of the CNN for forest fire susceptibility.
There were a total of three convolution layers, two pooling
layers, and three fully connected layers. The first three
consecutive convolution layers had 64, 128, and 256 ker-
nels, with uniform kernel sizes of 393. Each convolution
layer was followed by an activation function (ReLU) and a
pooling layer. Zero padding was employed to retain in-
plane dimensions. All of the pooling layers perform max-
pooling and summarize a 292 neighborhood with a stride
Fig. 3 The architectural design of the proposed convolutional neural network (CNN) model (C1–C3 are convolutional layers and FC1–FC3 are
fully connected layers. The values on the right and below C1–C3 indicate the number of filters and their sizes. The values below FC1–FC3
indicate their dimensions, that is, the number of neurons in the fully connected layer.). ReLU rectified linear unit
Table 2 Training sample statistics
Year Actual fire points Samples in the fire buffer Sampling rate Number of fire samples Number of nonfire samples
2002 158 1536 0.8 1229 1229
2003 267 2687 0.8 2150 2150
2004 522 4234 0.8 3387 3387
2005 620 4708 0.8 3766 3766
2006 625 4801 0.8 3841 3841
2007 835 6092 0.8 4874 4874
2008 339 3014 0.8 2411 2411
2009 491 3994 0.8 3195 3195
total 3857 31,066 24,853 24,853
123
Int J Disaster Risk Sci
of 2 pixels. At the end, the next three weight layers were
fully connected layers with 128, 64, and 32 neurons each.
Finally, the output of the last fully connected layer was fed
into a 2-way classifier with an activation function named
softmax, which computes the probabilities for the two
classes’ labels.
The parameters in CNN, which are automatically
learned during the training process, refer to kernels in the
convolution layers and weights in the fully connected
layers. Training the CNN network is a process to find
appropriate parameters to minimize the error between the
predicted results and the ground truth labels on a training
dataset. The CNN converted each input patch from the
original pixel values to the final probability classification
results, and the parameters were calculated by a loss
function through feed-forward propagation. The learnable
parameters were updated according to the loss value by
using the stochastic gradient descent based on the back-
propagation algorithm.
The hyperparameters (Table 3) are the variables that
need to be set before the training process begins. Dropout is
a recently introduced regularization technique and the fully
connected layers are followed by dropout rates of 0.5 to
mitigate overfitting. Bergstra and Bengio (2012) argued
that random searches are more efficient for hyperparameter
optimization than grid searches and manual searches. A
random search was used to optimize the hyperparameters
and improve the accuracy and speed of the model. Adam,
an efficient stochastic optimization algorithm based on the
gradient (Kingma and Ba 2014), was selected as the
optimizer.
3.4 Performance Evaluation
The evaluation criteria are a key factor in assessing the
classification performance and guiding the classifier mod-
eling (Sokolova and Lapalme 2009). In this article, a two-
class classification method is modeled to predict forest fire
susceptibility. Thus, five statistical measures including
overall accuracy, specificity, sensitivity, positive predictive
value (PPV), and negative predictive value (NPV) are
employed to appraise the classification capability (Tien Bui
et al. 2017). The five statistical measures are computed in
the following manner:
Overall accuracy ¼TP þTN
TP þTN þFP þFN ;
Specificity ¼TN
FP þTN
Sensitivity ¼TP
TP þFN ;PPV ¼TP
FP þTP ;
NPV ¼TN
FN þTN
where TP (true positive) and TN (true negative) are the
number of samples that are correctly classified as positive
(fire class) and negative (nonfire class) observations,
respectively. FP (false positive) and FN (false negative) are
the number of samples that are misclassified. Sensitivity is
the percentage of positive (fire class) observations that are
correctly classified whereas specificity is the percentage of
negative (nonfire class) observations that are correctly
identified.
The ROC curve has been increasingly utilized to eval-
uate and validate the global performance assessment of the
prediction models in ML and data mining research (Pour-
taghi et al. 2016; Satir et al. 2016). It depicts the trade-offs
between the TPs and FPs rather than arbitrarily selecting a
particular threshold (Freeman and Moisen 2008). A ROC
plot is constructed by plotting the TP rate (TPrate, sensi-
tivity) on the Y-axis against the FP rate (FPrate, 1.0—
specificity) for all possible thresholds from 0 to 1 on the X-
axis. The ROC plot for a good classifier tends to rise
sharply at the origin and then level off near the maximum
value of 1. A trivial classifier will result in a ROC graph
near the diagonal where the TP rate is equal to the FP rate
for all thresholds. The AUC is generally considered to be
an important index to quantitatively assess the overall
accuracy of the classifier’s performance. An AUC value
near 0.5 means that the predictive ability of the model is
completely random and a value of 1.0 represents a perfect
Table 3 A list of parameters and hyperparameters utilized in the convolutional neural network (CNN) model
Parameters Hyperparameters
Convolution layer Kernels Kernel size: 393; number of kernels: 64, 128, 256; stride =1; padding; activation function=ReLU
Pooling layer None Pooling methods: max-pooling; filter size: 292; stride = 2
Fully connected
layer
Weights Number of weights; activation function: Softmax
Others Model architecture; initialize weights; optimizer; loss function; window size; epochs; learning rate; dropout
ReLU rectified linear unit
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
prediction without misclassification. The closer the AUC
value is to 1, the better the performance of the forest fire
prediction model. The AUC measure is computed by
obtaining only the area of the graphic:
AUC ¼1þTPrate FPrate
2
3.5 Technical Process for Predicting Forest Fire
Susceptibility
The technical process can be summarized into the follow-
ing five steps. The detailed procedure of this study is
depicted in Fig. 4.
Step 1 Construct the fire and nonfire inventory maps and
create maps of the influencing factors that can poten-
tially influence the ignition susceptibility.
Fig. 4 Methodological flowchart employed in this study. CNN convolutional neural network; ROC receiver operation characteristic
123
Int J Disaster Risk Sci
Step 2 Preprocess all datasets and generate the train-
ing/validation samples.
Step 3. Design the architecture of the CNN model,
optimize the CNN hyperparameters, and train the CNN
classifier.
Step 4 Predict forest fire susceptibility using the VRD of
2010.
Step 5 Evaluate the performance of the proposed model.
The CNN predication model was constructed under a
graphics processing unit (GPU) acceleration environment
using the Keras DL framework that uses TensorFlow as a
backend, which is a Python-based DL library. The system
configuration used in the lab environment is as follows:
Intel Core i7 CPU, 16 GB RAM, Windows10 OS, and an
NVIDIA GeForce GTX 1070 with 12 GB of onboard
memory.
4 Results
This section first shows the results of multicollinearity
analysis and an information gain ratio (IGR) for the
selection of forest fire influencing factors. The loss and the
accuracy in the training/validation phases were tracked.
Then, the test dataset was fed into the trained model and
the prediction map of ignition probabilities was constructed
by the CNN model. Finally, the performance of the pro-
posed model was compared with benchmark methods.
4.1 Relative Importance Analysis of Influencing
Factors
According to the results of a multicollinearity analysis of
the 14 forest fire influencing factors in Table 4, three fac-
tors—precipitation rate, specific humidity, and maximum
temperature—did not satisfy the critical values, suggesting
the existence of multicollinearity and should be excluded
from further analyses.
For the IGR method, the factors with a higher value of
average merit (AM) indicate a stronger prediction ability of
the model. However, factors with AM values equal to or
less than 0 indicate a “null” contribution to the forest fire
susceptibility model and should be excluded from further
analysis (Bui, Tuan, et al. 2016). The results listed in Fig. 5
show that the AM values of all remaining 11 influencing
factors are greater than 0, indicating that all these influ-
encing factors contribute to the model and should be
retained in the following prediction process. Temperature
Table 4 Multicollinearity analysis of forest fire influencing factors
No. Forest fire influencing factor TOL VIF
1 Elevation 0.288 3.47
2 Slope 0.819 1.221
3 Aspect 0.999 1.001
4 Average temperature 0.157 6.372
5 Average precipitation 0.4 2.5
6 Surface roughness 0.199 5.016
7 Average wind speed 0.106 9.443
8 Forest coverage ratio 0.881 1.134
9 NDVI 0.608 1.645
10 Distance to roads 0.948 1.055
11 Distance to rivers 0.932 1.073
12 Maximum temperature 0.014 69.653
13 Specific humidity 0.017 58.696
14 Precipitation rate 0.062 16.078
NDVI normalized difference vetation index; TOL tolerances; VIF
variance inflation factor
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Distance to rivers
Distance to roads
Forest coverage ratio
Aspect
Slope
NDVI
Elevation
Precipitation
Surface roughness
Wind speed
Temperature
AM value
Forest fire influencing factors
Fig. 5 Average merit (AM) of
each forest fire influencing
factor using the information
gain ratio (IGR) method
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
is the most important factor for forest fire susceptibility,
with the highest AM value of (0.139). It is followed by
wind speed (0.131), surface roughness (0.112), precipita-
tion (0.107), and elevation (0.102).
4.2 Model Accuracy
The training process was divided into training and valida-
tion phases. The validation dataset was used to monitor the
model classification performance after the end of each
epoch, and the validation results were used as the basis for
whether the training process should be terminated earlier or
the hyperparameters should be fine-tuned. The loss and
accuracy were two important indicators for evaluating the
effect of model training. Callback functions were applied
to adjust the training state and statistics during model
training, including “EarlyStopping,” “Re-
duceLROnPlateau,” and “ModelCheckpoint.”
Figure 6shows the loss and the training/validation
accuracy using TensorBoard for visualization. After the
training phase, the training accuracy was close to 91%, the
training loss no longer decreased after the 100th epoch and
tended to fit, and the minimum loss value was 0.3. After the
80th epoch, the validation loss no longer decreased and
tended to converge. However, the validation loss rose
slightly after the 120th epoch, suggesting overfitting may
have occurred. Immediately, the EarlyStopping function
ended the training process to decrease the phenomenon of
overfitting. The validation accuracy of the model reached
82%, and the minimum validation loss was 0.45.
The slow decline curve of validation loss indicates that
the model was well fitted. Overfitting did not occur because
the training process was effectively terminated by the
“EarlyStopping” function. After the training process, the
model corresponding to the epoch with the lowest loss of
the validating dataset in Check_pointer was selected as the
final classification model.
4.3 Susceptibility Map
After the training process, a classification model was
achieved. A test dataset—the VRD of 2010—was then
used at the end of the project to evaluate the performance
of the final model. The VRD of 2010 was fed into the
classification model to predict forest fire susceptibility. The
model resulted in probabilities for both the fire and nonfire
classes on each pixel in the generated prediction map.
Notably, although the CNN was designed to predict a
single label from a small input patch of size 25 925 911,
the CNN was trained to predict all pixels in the new VRD
since the sliding window was densely overlapped and
covered the entire VRD during the reasoning phase. The
sum of the probabilities of the fire and nonfire class was 1
on each pixel. The probability of the fire class was chosen
as the final predicted probability value, and then we pro-
duced a map of the probability values for all pixels by
converting the array data into an image using the libtiff
package in Python. The forest fire susceptibility map was
finally produced by dividing the image into five levels
using the natural breaks method in ArcGIS10.5. Five
Fig. 6 a,bRepresent the
training accuracy and the
convergence graph of the
training loss, respectively; c,
drepresent the validation
accuracy and the convergence
graph of the validation loss,
respectively
123
Int J Disaster Risk Sci
susceptible classes were identified—very low, low, mod-
erate, high, and very high—for constructing the forest fire
susceptibility map (Fig. 7).
From the predicted susceptibility results, we can see that
the highest forest fire ignition susceptibilities are mainly
distributed in the south and northwest of Yunnan Province.
The areas with the lowest forest fire ignition susceptibilities
are in the central and northeastern parts of Yunnan.
4.4 Model Comparison
The usability of the proposed model was compared with
benchmark methods random forests (RF), support vector
machine (SVM), multilayer perceptron neural network
(MLP), and kernel logistic regression (KLR). The four
benchmark methods were built and implemented using the
scikit-learn package, and the grid search method was used
to optimize the hyperparameters. The main hyperparame-
ters utilized in the benchmark methods are listed in
Table 5. Recursive feature elimination with cross-valida-
tion was employed to perform automatic tuning of the
number of features selected for SVM and KLR.
The performance of the five models was evaluated and
compared for both the training and validation datasets of
2009. Table 6shows that the CNN had the higher speci-
ficity (93.77%), sensitivity (97.84%), PPV (94.02%), NPV
(97.75%), and overall accuracy (95.81%) for the training
dataset than the four benchmark methods. The proposed
CNN model had the highest overall accuracy (87.92%) for
the validation dataset, followed by the RF (84.36%), KLR
(81.23%), SVM (80.04%), and MLP (78.47%).
Figure 8illustrates the ignition susceptibility map con-
structed by the benchmark models using the same test
datasets. The values denote the probabilities of each pixel
having an ignition occurrence. To enhance the compara-
bility of the prediction results for the CNN and benchmark
models, the probability map was divided into five classes
with the same natural breaks method as CNN. It can be
observed that in the prediction maps of SVM, MLP, and
KLR, forest fire susceptibility of most areas in the north-
west of Yunnan Province was categorized as very low and
low. In contrast, CNN and RF can well-identify the ignition
probabilities in the northwest. For the southern region of
Yunnan Province, compared with the predicted results of
the CNN model, the RF model divided most regions of
southern Yunnan into very high susceptible zones, while
the results of the SVM, MLP, and KLR models predicted
high susceptibility only in the southwest region of Yunnan
Province. The high fire occurrence region in the southeast
was not clearly identified in the predicted results.
Figure 9shows that the very high and very low sus-
ceptibility classes in the CNN model account for 77.51% of
the total area, which was the highest proportion among all
the models; whereas the remaining three classes of high,
moderate, and low have the lowest proportion of all
models, accounting for 7.15%, 6.33%, and 9.01% of the
total area, respectively. The results show that the proposed
CNN model can effectively divide the very high and very
low susceptible zones. However, the probability prediction
results of the benchmark models had many areas within
medium and high susceptibility zones, and there was no
clear determination regarding zones with high forest fire
susceptibility; thus, what threshold segmentation method
should be used to divide the fire warning areas needs to be
considered. Most traditional ML methods will face this
problem after obtaining the probability prediction results.
In contrast, the probability results obtained by the CNN are
more distinct in their spatial pattern, and the statistical
values reflect a high and low bipolar distribution, which is
more advantageous in the division of the areas with fire
warnings and those without fire warnings and can reduce
the influence of the threshold segmentation of the delin-
eation of fire warning areas. Thus, the result of the CNN
model is better for forest fire predictions.
To confirm the statistical significance of the prediction
performance between the proposed CNN model and the
benchmark models, the nonparametric WSRT (Wilcoxon
1945) was employed for paired comparisons. The null-
hypothesis (H
0
) was that there was no significant difference
at 95% confidence intervals of two prediction models. If
the pvalue was less than the significance level (0.05), H
0
was rejected, and a significant difference exists in the
Fig. 7 Forest fire susceptibility map derived from the convolutional
neural network (CNN) model for Yunnan Province, China, in 2010
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
models (Tien Bui et al. 2019). The analysis was performed
using SPSS (Statistical Package for the Social Sciences),
and the Wilcoxon signed-rank test (WSRT) results are
reported in Table 7. The pvalue of all pairwise compar-
isons was less than 0.05, confirming that the classification
performances of the proposed model and benchmark
models are significantly different.
The ROC curves for the five models are depicted in
Fig. 10. They were drawn using the 2010 prediction maps
of the ignition probabilities and the corresponding IRD.
The AUC provides a single measure of a classifier’s per-
formance to evaluate which model is better on average.
The AUC of the CNN model is 0.86 (Fig. 10), indicating
that the global fit of the model with the testing dataset is
86%, followed by RF (0.82), SVM (0.79), MLP (0.78), and
Fig. 8 Prediction maps of the ignition probabilities for the benchmark models for Yunnan Province, China. KLR kernel logistic regression; MLP
multilayer perceptron neural network; RF random forests; SVM support vector machine
123
Int J Disaster Risk Sci
KLR (0.78). The ROC curves clearly show that the CNN
model has the highest prediction performance.
5 Discussion
This section first describes the advantages of CNN com-
pared with benchmark methods, then discusses the selec-
tion of the model architecture and window size, and finally
discusses the differences between the CNN and traditional
ML algorithms.
5.1 Advantage of the Convolutional Neural Network
Method
Compared with the benchmark methods, the CNN has the
following advantages. First, because the CNN can consider
the correlation of adjacent spatial information, it has
advantages in the study of problems with spatial and geo-
graphical correlation characteristics. Second, the CNN
preserves the spatial relationships between pixels by
learning the internal feature representations from factor
vectors. The process of DL reveals the deep features and
can distinguish the differences between different geo-
graphical units. The CNN was used to conduct multiple
convolution and pooling operations to extract the charac-
teristics. As the convolutions and pooling increased, these
features became more advanced and more abstract. These
abstract features depicted the degree of forest fire suscep-
tibility, which was the decisive factor for determining
forest fire susceptibility. Third, the CNN reduces the
number of weights that need to be trained and the com-
putational complexity of the network through weight
sharing.
5.2 Model Sensitivity
The architecture of the CNN model should be selected in
accordance with the quantity of sample data and the
problem complexity. For those with small quantities of
sample data and simple classification problems, the com-
plex structure was prone to model overfitting. For those
with large quantities of sample data and complex classifi-
cation problems, the simple structure was prone to model
nonconvergence. Both of these problems should be avoided
as much as possible in the training of the CNN model.
The selection of the window size should be consistent
with the maximum geospatial area that affects the forest
fire susceptibility of the center window pixel. A larger
window size means that the pixels of a larger geographical
unit impact the center window pixel. In fact, the pixels that
affect the fire susceptibility of the center window pixel
have a certain geographical spatial range. Therefore, the
selection of the window size must be appropriate. After the
experiments, the 25925 window size was most reasonable.
First, this size corresponds to the maximum geographical
spatial range that affects the fire susceptibility of a pixel.
Second, the size of this window is smaller than that of the
commonly used windows (for example, 2249224) in
image processing because forest fire probability prediction
and image classification are completely different
Table 5 A list of the main hyperparameters utilized in the benchmark methods
Benchmark
methods
Hyperparameters
RF Number of trees: 160; max_features: “sqrt”; bootstrap: True; max_depth=20
SVM Penalty factor (C): 100; kernel function = “RBF”; gamma =1
MLP Number of hidden layers: 1; momentum: 0.2; learning rate: 0.001; iteration= 300; alpha = 0.01; solver =”Adam”; activation
=”ReLU”
KLR Kernel function=“RBF”; tuning parameter (δ): 0.02; regularize parameter (C): 0.025
RF random forests; SVM support vector machine; MLP multilayer perceptron neural network; KLR kernel logistic regression; RBF radial basis
function; ReLU rectified linear unit
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
very low low moderate high very high
Percent of susceptibility classes
Forest fire susceptibility classes
CNN
RF
SVM
MLP
KLR
Fig. 9 Percentages of different forest fire susceptibility classes. CNN
convolutional neural network; RF random forests; SVM support
vector machine; MLP multilayer perceptron neural network; KLR
kernel logistic regression
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
applications, and the CNN model that is used for image
processing cannot be completely duplicated.
It is necessary to discuss the differences between the
CNN and traditional ML algorithms. The first is the
dependency of the CNN model on the data and hardware.
The CNN requires a large number of training samples, and
its performance increases as the scale of data increases.
Because the CNN inherently performs a large number of
matrix multiplication operations, they rely heavily on high-
end machines compared to traditional ML algorithms that
can run on low-end machines. The second is that CNN can
automatically explore high-level features from raw data.
This is a very distinctive part of DL and a major step ahead
of traditional ML. Moreover, the CNN model shows a
strong generalization ability compared with the benchmark
methods. In terms of time efficiency, because the CNN has
a large number of parameters to learn, the training time of
the CNN is longer than those of traditional ML models.
Because of the reusability of the CNN neural network after
Table 6 Training and validation performance comparison
Phase Performance Prediction model
CNN RF SVM MLP KLR
Training TP 3126 3053 2860 2800 2940
TN 2996 2796 2322 2349 2372
FP 199 399 873 846 823
FN 69 142 335 395 345
Sensitivity (%) 97.84 95.56 89.51 87.64 89.5
Specificity (%) 93.77 87.51 72.68 73.52 74.24
PPV (%) 94.02 88.44 76.61 76.8 78.13
NPV (%) 97.75 95.17 87.39 85.6 87.3
Overall accuracy (%) 95.81 91.53 81.1 80.58 81.98
Validation TP 739 580 532 491 577
TN 666 768 747 763 721
FP 133 31 52 36 78
FN 60 219 267 308 222
Sensitivity (%) 92.49 72.59 66.58 61.45 72.22
Specificity (%) 83.35 96.12 93.49 95.49 90.24
PPV (%) 84.75 94.93 91.1 93.17 88.09
NPV (%) 91.74 77.81 73.67 71.24 76.46
Overall accuracy (%) 87.92 84.36 80.04 78.47 81.23
CNN convolutional neural network; FP false positive; FN false negative; KLR kernel logistic regression; MLP multilayer perceptron neural
network; NPV negative predictive value; PPV positive predictive value; RF random forests; SVM support vector machine; TN true negative; TP
true positive
Table 7 Wilcoxon signed-rank test (two-tailed)
No. Pairwise comparison pvalue Significance
1 CNN versus RF \0.0001 Yes
2 CNN versus SLM \0.0001 Yes
3 CNN versus MLP \0.0001 Yes
4 CNN versus KLR \0.0001 Yes
CNN convolutional neural network; RF random forests; SVM support
vector machine; MLP multilayer perceptron neural network; KLR
kernel logistic regression
Fig. 10 Receiver operating characteristic (ROC) curves and area
under the curve (AUCs) of the five models. CNN convolutional neural
network; RF random forests; SVM support vector machine; MLP
multilayer perceptron neural network; KLR kernel logistic regression
123
Int J Disaster Risk Sci
initial training, there is still considerable room for
improving the efficiency in the later training time. The
prediction time of the CNN is relatively short when using
GPU-accelerated computing technology. Finally, although
CNN has shown excellent performance, the interpretability
is its deficiency, which needs to be further studied.
6 Conclusion
In this article, we investigated a CNN with deep architec-
tures for the spatial prediction of forest fire susceptibility in
Yunnan Province, China. Past forest fire locations from
2002 to 2010 were extracted and a set of 14 forest fire
influencing factors were optimized using multicollinearity
analysis and the IGR technique. We explored the prepro-
cessing methods for forest fire influencing factors and the
methods for generating effective training/validation sample
libraries. The CNN architecture suitable for the prediction
of forest fire susceptibility in the study area was designed,
and hyperparameters were optimized to improve the pre-
diction accuracy. Several common methods, such as more
training samples, regularization (dropout), batch normal-
ization, and reduced architecture complexity, were used in
the CNN model to mitigate overfitting. Then, the test
dataset was fed into the trained model and the prediction
map of ignition probabilities was constructed by the CNN
model. Finally, the performance of the proposed model was
compared with traditional ML methods using several sta-
tistical measures, including WSRT, ROC, and AUC.
Through this research, we found that the CNN model
performs better than the benchmark methods. The CNN
model (AUC=0.86) has higher predictive power than the
benchmark methods according to the ROC–AUC. The
probability result obtained by the CNN can clearly distin-
guish the very high and very low susceptible zones, and the
susceptibility spatial pattern was very distinct. The CNN
model shows a strong generalization ability and the pre-
diction time of the CNN was relatively short when using
GPU-accelerated computing technology. In conclusion, the
CNN has the advantages of considering neighborhood
information, extracting deep features, sharing weights, and
pooling operations, which allow the CNN to obtain better
prediction results. The CNN model will have important
practical application value for forest fire prevention plan-
ning and forest management.
There are still some limitations in the research. For
example, the influence of different CNN architectures—
such as VGG-net (Visual Geometry Group Network), RES-
net Residential Energy Services Network), and GoogLeNet
—on forest fire prediction results have not been studied in
depth. In addition, more actual data are needed for the
experimental verification of the method. In recent years,
the application of CNNs has become increasingly exten-
sive. Many different variants of the architecture have been
derived and many ensemble classifiers have been proposed.
Comparing various classifiers and exploring the most
suitable models to improve forest fire prediction should be
investigated in the future.
Acknowledgements This research was supported by the National
Key Research and Development Plan (2017YFC1502902) and
National Natural Science Foundation of China (41621601). The
financial support is highly appreciated. We thank Yinxue Cao for her
help in getting the data from Climate Forecast System Reanalysis. We
are also grateful to the anonymous reviewers and the editors for their
constructive comments.
Open Access This article is distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://crea
tivecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
made.
References
Adab, H., K.D. Kanniah, and K. Solaimani. 2013. Modeling forest fire
risk in the northeast of Iran using remote sensing and GIS
techniques. Natural Hazards 65(3): 1723–1743.
Arpaci, A., B. Malowerschnig, O. Sass, and H. Vacik. 2014. Using
multi variate data mining techniques for estimating fire suscep-
tibility of Tyrolean forests. Applied Geography 53: 258–270.
Bajocco, S., E. Dragoz, I. Gitas, D. Smiraglia, L. Salvati, and C.
Ricotta. 2015. Mapping forest fuels through vegetation phenol-
ogy: The role of coarse-resolution satellite time-series. PLoS
ONE 10(3): 1–14.
Bar Massada, A., A.D. Syphard, S., I. Stewart, and V.C. Radeloff.
2013. Wildfire ignition-distribution modelling: a comparative
study in the Huron–Manistee National Forest, Michigan, USA.
International Journal of Wildland Fire 22(2): 174–183.
Bergstra, J., and Y. Bengio. 2012. Random search for hyper-
parameter optimization. Journal of Machine Learning Research
13(1): 281–305.
Bisquert, M., E. Caselles, J.M. Sa
´nchez, and V. Caselles. 2012.
Application of artificial neural networks and logistic regression
to the prediction of forest fire danger in Galicia using MODIS
data. International Journal of Wildland Fire 21(8): 1025–1029.
Buda, M., A. Maki, and M.A. Mazurowski. 2018. A systematic study
of the class imbalance problem in convolutional neural networks.
Neural Networks 106: 249–259.
Bui, D.T., K.T.T. Le, V.C. Nguyen, H.D. Le, and I. Revhaug. 2016.
Tropical forest fire susceptibility mapping at the Cat Ba National
Park area, Hai Phong City, Vietnam, using GIS-based Kernel
logistic regression. Remote Sensing 8(4): 1–15.
Bui, D.T., T.A. Tuan, H. Klempe, B. Pradhan, and I. Revhaug. 2016.
Spatial prediction models for shallow landslide hazards: A
comparative assessment of the efficacy of support vector
machines, artificial neural networks, kernel logistic regression,
and logistic model tree. Landslides 13(2): 361–378.
Cao, Y., M. Wang, and K. Liu. 2017. Wildfire susceptibility
assessment in Southern China: A comparison of multiple
methods. International Journal of Disaster Risk Science 8(2):
164–181.
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
Colkesen, I., E.K. Sahin, and T. Kavzoglu. 2016. Susceptibility
mapping of shallow landslides using kernel-based Gaussian
process, support vector machines and logistic regression. Journal
of African Earth Sciences 118: 53–64.
Crimmins, M.A. 2006. Synoptic climatology of extreme fire-weather
conditions across the southwest United States. International
Journal of Climatology 26(8): 1001–1016.
Dash, M., and H. Liu. 1997. Feature selection for classification.
Intelligent Data Analysis 1(1–4): 131–156.
Dimuccio, L.A., R. Ferreira, L. Cunha, and A. Campar de Almeida.
2011. Regional forest-fire susceptibility analysis in central
Portugal using a probabilistic ratings procedure and artificial
neural network weights assignment. International Journal of
Wildland Fire 20(6): 776–791.
Elmas, C¸ ., and Y. So
¨nmez. 2011. A data fusion framework with novel
hybrid algorithm for multi-agent Decision Support System for
Forest Fire. Expert Systems with Applications 38(8): 9225–9236.
Freeman, E.A., and G.G. Moisen. 2008. A comparison of the
performance of threshold criteria for binary classification in
terms of predicted prevalence and kappa. Ecological Modelling
217(1): 48–58.
Guo, F., Z. Su, G. Wang, L. Sun, F. Lin, and A. Liu. 2016. Wildfire
ignition in the forests of southeast China: Identifying drivers and
spatial distribution to predict wildfire likelihood. Applied Geog-
raphy 66: 12–21.
Hantson, S., S. Pueyo, and E. Chuvieco. 2015. Global fire size
distribution is driven by human impact and climate. Global
Ecology and Biogeography 24(1): 77–86.
Hinton, G.E., and R.R. Salakhutdinov. 2006. Reducing the dimen-
sionality of data with neural networks. Science 313(5786): 504–
507.
Hong, H., A. Jaafari, and E.K. Zenner. 2019. Predicting spatial
patterns of wildfire susceptibility in the Huichang County,
China: An integrated model to analysis of landscape indicators.
Ecological Indicators 101: 878–891.
Hong, H., B. Pradhan, C. Xu, and D. Tien Bui. 2015. Spatial
prediction of landslide hazard at the Yihuang area (China) using
two-class kernel logistic regression, alternating decision tree and
support vector machines. Catena 133: 266–281.
Hong, H., P. Tsangaratos, I. Ilia, J. Liu, A.X. Zhu, and C. Xu. 2018.
Applying genetic algorithms to set the optimal combination of
forest fire related variables and model forest fire susceptibility
based on data mining models. The case of Dayu County, China.
Science of the Total Environment 630: 1044–1056.
Hu, F., G.S. Xia, J. Hu, and L. Zhang. 2015. Transferring deep
convolutional neural networks for the scene classification of
high-resolution remote sensing imagery. Remote Sensing 7(11):
14680–14707.
Jaafari, A., E.K. Zenner, M. Panahi, and H. Shahabi. 2019. Hybrid
artificial intelligence models based on a neuro-fuzzy system and
metaheuristic optimization algorithms for spatial prediction of
wildfire probability. Agricultural and Forest Meteorology 266–
267: 198–207.
Jaafari, A., E.K. Zenner, and B.T. Pham. 2018. Wildfire spatial
pattern analysis in the Zagros Mountains, Iran: A comparative
study of decision tree based classifiers. Ecological Informatics
43: 200–211.
Kim, S.J., C.H. Lim, G.S. Kim, J. Lee, T. Geiger, O. Rahmati, Y. Son,
and W.K. Lee. 2019. Multi-temporal analysis of forest fire
probability using socio-economic and environmental variables.
Remote Sensing 11(1): Article 86.
Kingma, D.P., and J. Ba. 2014. Adam: A method for stochastic
optimization. Presented as a conference paper at the 3rd
International Conference for Learning Representations, San
Diego, 2015. arXiv preprint abs:1412.6980. Ithaca, NY: Cornell
University.
Krizhevsky, A., I. Sutskever, G.E. Hinton. 2013. ImageNet classifi-
cation with deep convolutional neural networks. In Proceedings
of the 26th Annual Conference on Neural Information Processing
Systems (NIPS), 3–6 December 2012, Lake Tahoe, Nevada,
USA, ed. F. Pereira, C.J.C. Burges, L. Bottou, and K.Q.
Weinberger, Vol. 2, 1097–1105.
Lecun, Y., Y. Bengio, and G. Hinton. 2015. Deep learning. Nature
521(7553): 436–444.
Lecun, Y., L. Bottou, Y. Bengio, and P. Haffner. 1998. Gradient-
based learning applied to document recognition. Proceedings of
IEEE 86(11): 2278–2324.
Leuenberger, M., J. Parente, M. Tonini, M.G. Pereira, and M.
Kanevski. 2018. Wildfire susceptibility mapping: Deterministic
vs. stochastic approaches. Environmental Modelling and Soft-
ware 101: 194–203.
Liu, T., and A. Abd-Elrahman. 2018. Deep convolutional neural
network training enrichment using multi-view object-based
analysis of Unmanned Aerial systems imagery for wetlands
classification. ISPRS Journal of Photogrammetry and Remote
Sensing 139: 154–170.
Lo
´pez, V., A. Ferna
´ndez, S. Garcı
´a, V. Palade, and F. Herrera. 2013.
An insight into classification with imbalanced data: Empirical
results and current trends on using data intrinsic characteristics.
Information Sciences 250: 113–141.
Moritz, M. A., M.-A. Parisien, E. Batllori, M. A. Krawchuk, J. Van
Dorn, D.J. Ganz, and K. Hayhoe. 2012. Climate change and
disruptions to global fire activity. Ecosphere 3(6): Article 49.
Muhammad, K., J. Ahmad, and S.W. Baik. 2018. Early fire detection
using convolutional neural networks during surveillance for
effective disaster management. Neurocomputing 288(C): 30–42.
Ngoc Thach, N., D. Bao-Toan Ngo, P. Xuan-Canh, N. Hong-Thi, B.
Hang Thi, H. Nhat-Duc, and T.B. Dieu. 2018. Spatial pattern
assessment of tropical forest fire danger at Thuan Chau area
(Vietnam) using GIS-based advanced machine learning algo-
rithms: A comparative study. Ecological Informatics 46: 74–85.
O’Brien, R.M. 2007. A caution regarding rules of thumb for variance
inflation factors. Quality and Quantity 41(5): 673–690.
Oliveira, S., F. Oehler, J. San-Miguel-Ayanz, A. Camia, and J.M.C.
Pereira. 2012. Modeling spatial patterns of fire occurrence in
Mediterranean Europe using Multiple Regression and Random
Forest. Forest Ecology and Management 275: 117–129.
Pew, K.L., and C.P.S. Larsen. 2001. GIS analysis of spatial and
temporal patterns of human-caused wildfires in the temperate
rain forest of Vancouver Island, Canada. Forest Ecology and
Management 140(1): 1–18.
Pourtaghi, Z.S., H.R. Pourghasemi, R. Aretano, and T. Semeraro.
2016. Investigation of general indicators influencing on forest
fire and its susceptibility modeling using different data mining
techniques. Ecological Indicators 64: 72–84.
Renard, Q., R. P
´lissier, B.R. Ramesh, and N. Kodandapani. 2012.
Environmental susceptibility model for predicting forest fire
occurrence in the Western Ghats of India. International Journal
of Wildland Fire 21(4): 368–379.
Running, S.W. 2006. Is global warming causing more, larger
wildfires? Science 313(5789): 927–928.
Sachdeva, S., T. Bhatia, and A.K. Verma. 2018. GIS-based evolu-
tionary optimized Gradient Boosted Decision Trees for forest fire
susceptibility mapping. Natural Hazards 92(3): 1399–1418.
Satir, O., S. Berberoglu, and C. Donmez. 2016. Mapping regional
forest fire probability using artificial neural network model in a
Mediterranean forest ecosystem. Geomatics, Natural Hazards
and Risk 7(5): 1645–1658.
Schmidhuber, J. 2015. Deep Learning in neural networks: An
overview. Neural Networks 61: 85–117.
123
Int J Disaster Risk Sci
Sokolova, M., and G. Lapalme. 2009. A systematic analysis of
performance measures for classification tasks. Information
Processing and Management 45(4): 427–437.
Tien Bui, D., Q.T. Bui, Q.P. Nguyen, B. Pradhan, H. Nampak, and P.
T. Trinh. 2017. A hybrid artificial intelligence approach using
GIS-based neural-fuzzy inference system and particle swarm
optimization for forest fire susceptibility modeling at a tropical
area. Agricultural and Forest Meteorology 233: 32–44.
Tien Bui, D., N.D. Hoang, and P. Samui. 2019. Spatial pattern
analysis and prediction of forest fire using new machine learning
approach of Multivariate Adaptive Regression Splines and
Differential Flower Pollination optimization: A case study at
Lao Cai province (Viet Nam). Journal of Environmental
Management 237: 476–487.
Tobler, W.R. 1970. A computer movie simulating urban growth in the
Detroit region. Economic Geography 46(sup1): 234–240.
Vasconcelos, M.J. P. de, S. Silva, M. Tome
´, M. Alvim, and J. M. C.
Perelra. 2001. Spatial Prediction of Fire Ignition Probabilities:
Comparing Logistic Regression and Neural Networks. Pho-
togrammetric Engineering & Remote Sensing 67(1): 73–81.
Vetrivel, A., M. Gerke, N. Kerle, F. Nex, and G. Vosselman. 2018.
Disaster damage detection through synergistic use of deep
learning and 3D point cloud features derived from very high
resolution oblique aerial images, and multiple-kernel-learning.
ISPRS Journal of Photogrammetry and Remote Sensing 140: 45–
59.
Wang, Y., Z. Fang, and H. Hong. 2019. Comparison of convolutional
neural networks for landslide susceptibility mapping in Yanshan
County, China. Science of the Total Environment 666: 975–993.
Wilcoxon, F. 1945. Individual comparisons by ranking methods.
Biometrics Bulletin 1(6):80–83.
Yamashita, R., M. Nishio, R K.G. Do, and K. Togashi. 2018.
Convolutional neural networks: An overview and application in
radiology. Insights into Imaging 9(4): 611–629.
Yi, K., H. Tani, J. Zhang, M. Guo, X. Wang, and G. Zhong. 2013.
Long-term satellite detection of post-fire vegetation trends in
boreal forests of China. Remote Sensing 5(12): 6938–6957.
Ying, L., J. Han, Y. Du, and Z. Shen. 2018. Forest fire characteristics
in China: Spatial patterns and determinants with thresholds.
Forest Ecology and Management 424: 345–354.
Zhang, X. 2007. Vegetation map of the People’s Republic of China
(1:1000000). Beijing: Geology Press (in Chinese).
Zhang, C., X. Pan, H. Li, A. Gardiner, I. Sargent, J. Hare, and P.M.
Atkinson. 2018. A hybrid MLP-CNN classifier for very fine
resolution remotely sensed image classification. ISPRS Journal
of Photogrammetry and Remote Sensing 140: 133–144.
123
Zhang et al. Forest Fire Susceptibility Modeling Using a Convolutional Neural Network
