Sustainability 14 03797

sustainability
Article
Water Information Extraction Based on Multi-Model RF
Algorithm and Sentinel-2 Image Data
Zhiqi Jiang 1 , Yijun Wen 2, *, Gui Zhang 3 and Xin Wu 4
1 School of Forestry, Central South University of Forestry and Technology, Changsha 410004, China;
jiangzhiqi2022@163.com
2 Faculty of Science, Central South University of Forestry and Technology, Changsha 410004, China
3 National Forest Fire Prevention Virtual Simulation Laboratory Teaching Center, Changsha 410004, China;
csfu3s@163.com
4 Key Laboratory of Digital Dongting Lake of Hunan Province, Changsha 410004, China; csuftfire@csuft.edu.cn
* Correspondence: wenyishalong@163.com; Tel.: +86-18207370571
Abstract: For the Sentinel-2 multispectral satellite image remote sensing data, due to the rich spatial
information, the traditional water body extraction methods cannot meet the needs of practical
applications. In this study, a random forest-based RF_16 optimal combination model algorithm is
proposed to extract water bodies. The research process uses Sentinel-2 multispectral satellite images
and DEM data as the basic data, collected 24 characteristic variable indicators (B2, B3, B4, B8, B11, B12,
NDVI, MSAVI, B5, B6, B7, B8A, NDI45, MCARI, REIP, S2REP, IRECI, PSSRa, NDWI, MNDWI, LSWI,
DEM, SLOPE, SLOPE ASPECT), and constructed four combined models with different input variables.
After analysis, it was determined that RF_16 was the optimal combination for extracting water body
information in the study area. Model. The results show that: (1) The characteristic variables that
have an important impact on the accuracy of the model are the improved normalized difference

water index (MNDWI), band B2 (Blue), normalized water index (NDWI), B4 (Red), B3 (Green), and
Citation: Jiang, Z.; Wen, Y.; Zhang, band B5 (Vegetation Red-Edge 1); (2) The water extraction accuracy of the optimal combined model
G.; Wu, X. Water Information RF_16 can reach 93.16%, and the Kappa coefficient is 0.8214. The overall accuracy is 0.12% better than
Extraction Based on Multi-Model RF the traditional Relief F algorithm. The RF_16 method based on the optimal combination model of
Algorithm and Sentinel-2 Image Data. random forest is an effective means to obtain high-precision water body information in the study
Sustainability 2022, 14, 3797. https:// area. It can effectively reduce the “salt and pepper effect” and the influence of mixed pixels such as
doi.org/10.3390/su14073797 water and shadows on the water extraction accuracy.
Academic Editors: Zheng Duan,
Babak Mohammadi and Keywords: random forest; Sentinel-2; red-edge remote sensing data; water extraction
Amir Mosavi
Received: 21 January 2022

Accepted: 15 March 2022
1. Introduction
Published: 23 March 2022
As an important part of the earth’s water cycle, land surface water, such as rivers, lakes,
Publisher’s Note: MDPI stays neutral
and reservoirs, are irreplaceable for the global ecosystem and climate system [1]. Surveying
with regard to jurisdictional claims in
land surface water bodies and delineating their spatial distribution are meaningful to
published maps and institutional affil-
the understanding of hydrology processes and the management of water resources [2].
iations.
Remote sensing monitoring has become an important method for monitoring surface water
resources in recent years. Compared with traditional field survey methods, it has great
advantages of macroscopicity, instantaneity, dynamism, and cost effectiveness.
Copyright: © 2022 by the authors.
With the continuous development of research technology, the application of machine
Licensee MDPI, Basel, Switzerland. learning to various disciplines is becoming more and more extensive, such as support vec-
This article is an open access article tor machine (SVM) [3,4], random forest (RF) [5–7], artificial neural networks (ANNs) [8,9],
distributed under the terms and classification and regression tree (CART) [10,11], and so on. There are also many machine
conditions of the Creative Commons learning algorithms based on different applications, such as that used by Shaban and other
Attribution (CC BY) license (https:// researchers based on multi-objective optimization algorithms to predict the compressive
creativecommons.org/licenses/by/ strength of pozzolanic materials RAC [12]. Among them, the random forest classification
4.0/). algorithm demonstrates its outstanding ability in dealing with large data sets, and it also
Sustainability 2022, 14, 3797. https://doi.org/10.3390/su14073797 https://www.mdpi.com/journal/sustainability

Sustainability 2022, 14, 3797 2 of 19
shares strong anti-noise and anti-overfitting abilities and shows sound effect in classifica-
tion [13]. Therefore, more and more scholars apply the random forest machine learning
method to remote sensing image classification and terrain characteristic extraction research,
for example, Zheng Xiaorou et al. used the random forest algorithm, CART decision tree
algorithm, and support vector machine to classify land use. Comparing the classification
results of several algorithms, it was found that the classification results of the random forest
algorithm were more accurate [14]; using Sentinel-1 and Sentinel-2 data, Tufail Rahat and
others used a machine learning random forest classification algorithm to classify land use
categories, and the results showed an overall accuracy of 97% and Kappa coefficient of
0.97 [15]; VF Rodriguez-Galiano et al. applied the random forest algorithm to land cover
classification in a complex area to classify 14 different land classes in southern Spain. The
results show that the algorithm obtains accurate land cover classification results with an
overall accuracy of 92% and a Kappa coefficient of 0.92. RF is robust to noise reduction
and noise in training data [16]. Raziye Hale TOPALOĞLU and other researchers compared
and analyzed classification accuracies of land cover/use maps created from Sentinel-2 and
Landsat-8 data. According to overall classification accuracies, it can be seen that, for both
MLC and SVM classification methods, the image classification accuracies of the Sentinel-2
dataset is better than the Landsat-8 dataset. Although the Sentinel-2 dataset resampled to
30 m using Landsat-8 as a base, the overall classification accuracies of Sentinel-2 are much
higher than Landsat-8 data [17]; in addition, a large number of studies have shown that the
introduction of terrain factors can significantly improve the classification accuracy of land
use or water body extraction [18,19].
Forkuor and other researchers compared and explored the synergistic use of Landsat-8
and Sentinel-2 data in mapping land use and land cover (LULC) in rural Burkina Faso.
Specifically, the contribution of the red-edge bands of Sentinel-2 in improving LULC
mapping was examined. It was found that the classification of the Sentinel-2 red-edge
bands alone produced better and comparable results to Landsat-8 and the other Sentinel-2
bands, respectively. The results of this study demonstrate the added value of the Sentinel-2
red-edge bands and encourage multi-sensorial approaches to LULC mapping in West
Africa [20]. Elhadi Adam and Hristos Tyralis used the RF algorithm to study the land
use classification of cultivation areas and mountain areas [21,22], and the accuracy of
the results exceeded 80%. Swapan Talukdar and Pankaj Singha examined six machine
learning algorithms, namely random forest (RF), support vector machine (SVM), artificial
neural network (ANN), fuzzy adaptive resonance theory-supervised predictive mapping
(Fuzzy ARTMAP), spectral angle mapper (SAM), and Mahalanobis distance (MD). Accuracy
assessment was performed using Kappa coefficient, receiver operational curve (RoC), index-
based validation, and root mean square error (RMSE). Results of the Kappa coefficient
show that all the classifiers have a similar accuracy level with minor variation, but the RF
algorithm has the highest accuracy at 0.89 and the MD algorithm (parametric classifier) has
the least accuracy at 0.82. Finally, this review concludes that the RF algorithm is the best
machine learning LULC classifier among the six examined algorithms [23]; in addition, a
large number of studies have shown that the introduction of terrain factors can significantly
improve the efficiency of land use and the water extraction classification accuracy [24,25].
In the study of water information extraction, compared with other remote sensing data,
Sentinel-2A image has a higher extraction accuracy. For example, the three multispectral
bands of the Spot5 satellite have a spatial resolution of 10 m and a temporal resolution of
26 days. Despite having high spatial resolution, the satellite has a small number of bands
and a limited application field. Although it can meet the needs of water extraction, the data
acquisition of it is paid with a high price of image, which cannot be widely used by most
researchers. Landsat-8 has a re-entry period of 16 days and a spatial resolution of 30 m in
the visible band. Although it has seven spectral bands, a wide range of applications, and
free access to a large amount of data, LandSat80 s spatial resolution has some limitations [26].
The temporal resolution of the Gaofen-1 satellite reaches 4 days, and its spatial resolution
of free data is 16 m. Sentinel-2A has a single-star re-entry period of 10 days, and 5 days
for double-star; its spatial resolution of multispectral four bands is 10 m. In contrast,

Sentinel-2A has considerable spatial and temporal resolution with free data acquisition.
It has great application prospects in water extraction. At the same time, compared with
traditional remote sensing data, Sentinel-2A remote sensing data adds four red-edge bands
(B5, B6, B7, B8A) that are closely related to chlorophyll content. These bands are used in
land use cover classification, wetland extraction, and crops. The research on identification
has achieved good results.
Although the Sentinel-2A image has its own data advantages, due to the rich spatial
information of the Sentinel-2 multispectral satellite image remote sensing data, the tradi-
tional water extraction methods cannot meet the needs of practical applications. In this
paper, a random forest-based RF_16 optimal combination model algorithm is proposed
to extract water bodies. In the research process, Sentinel-2 multispectral satellite images
and DEM data were used as the basic data, and four kinds of variables were collected,
including traditional remote sensing data, red-edge remote sensing data, water body index
factor, and terrain factor; a total of 24 characteristic variable indicators (B2, B3, B4, B8, B11,
B12, NDVI, MSAVI, B5, B6, B7, B8A, NDI45, MCARI, REIP, S2REP, IRECI, PSSRa, NDWI,
MNDWI, LSWI, DEM, SLOPE, SLOPE ASPECT) using these four combined models with
different input variables were constructed using characteristic indicators, and the influence
and contribution of each variable to the extracted water body were analyzed at the same
time. After analysis, it was determined that RF_16 is the optimal combination model for
extracting water body information in the study area.
2. Materials and Methods

2.1. Study Area
Xiangtan City is located in the central eastern part of Hunan Province, with a total area
of 5015 km2 , spanning 111◦ 580 –113◦ 050 E, 27◦ 200 5500 –28◦ 050 4000 N [27], with a subtropical
warm and humid climate, and vegetation dominated by plantation forests. The city’s main
roads are in the shape of a ring, the outer ring mainly consisting of Xianglian Avenue,
Furong Avenue, East Erhuan Road, East Jiuzhao Road, West Jiuzhao Road, East Jinpeng
Road, Baoshui Road, Shimatou Road, 107 Provincial Road, Shanghai-Kunming Expressway,
Xuchang-Guangzhou Expressway, West Dapeng Road, and Xiangzhang Road; the inner
ring is composed of Hutan Road, North Jianshe Road, South Shuangyong Road, Middle
Shuangyong Road, and Qingyuan Road. According to the administrative boundaries, the
area can be further divided into Yuhu District, Xiangtan County, and the rural-urban fringe
in Yuetang District (see Figure 1). The study area belongs to a typical subtropical warm
Sustainability 2022, 14, x FOR PEER REVIEW
and humid climate zone, with distinct seasons and rich rainfall. The winter 4 of 20 here is cold,
and summer is hot.
Figure 1.1.Locations
Figure Locationsof the study
of the areas.areas.
study
2.2. Data Sources and Pre-Processing
Xiangtan City has distinct seasons, and the higher quality images are mainly concen-
trated in summer and autumn from July to August. Therefore, the study selects nine
scenes of Sentinel-2 image data on 3 July 2020, 10 July 2020, 17 July 2020, 24 July 2020, 29
2.2. Data Sources and Pre-Processing

Xiangtan City has distinct seasons, and the higher quality images are mainly con-
centrated in summer and autumn from July to August. Therefore, the study selects nine
scenes of Sentinel-2 image data on 3 July 2020, 10 July 2020, 17 July 2020, 24 July 2020,
29 July 2020, 5 August 2020, 12 August 2020, and 19 August 2020. The Sentinel-2 data can be
downloaded for free from the ESA website (https://scihub.copernicus.eu/dhus/#/home).
Other supporting data in the study include: 30 m resolution DEM products, DEM data
downloaded from Geospatial Data Cloud (https://scihub.copernicus.eu/dhus/#/home),
and 2018 Xiangtan city land use status vector data (http://www.globallandcover.com/)
terrain factors (including elevation, slope, and slope aspect). Through DEM extraction,
the current land use status map and Google Earth data were used for the collection and
accuracy evaluation of the land class sample points.
Data Pre-Processing
Sentinel-2 data pre-processing mainly includes atmospheric correction, image regis-
tration, and image fusion. Since the downloaded Sentinel-2 L1C image is an atmospheric
apparent reflectance product that has been geometrically corrected, the SNAP software
plug-in Sen2Cor-2.4.0 is used to perform atmospheric correction to obtain the real re-
flectance data of the ground objects. SNAP software is available for free download through
ESA’s (European Space Agency) data distribution website (https://scihub.copernicus.eu/).
Then, the bilinear resampling was performed in SNAP, and the original 60 m resolution
band data was removed. The Sentinel-2-related band parameters used are shown in Table 1.
The data come from the following website: https://eros.usgs.gov/sentinel-2. Finally, the
author uses ENVI’s “Seamless Mosaic” tool for stitching and ENVI’s “Subset Data from
ROIs” tool for cropping to obtain the image data of Xiangtan City.
Table 1. Sentinel-2 Data Band Parameters.
Center Wavelength Spectral Width Spatial Resolution

Sentinel-2 Bands
(nm) (nm) (m)
Band 1-Coastal aerosol 443 20 60
Band 2-Blue 490 65 10
Band 3-Green 560 35 10
Band 4-Red 665 30 10
Band 5-Vegetation
705 15 20
Red-Edge 1
Band 6-Vegetation
740 15 20
Red-Edge 2
Band 7-Vegetation
783 20 20
Red-Edge 3
Band 8-NIR 842 115 10
Band 8a-Vegetation
865 20 20
Red-Edge 4
Band 11-SWIR1 1610 90 20
Band 12-SWIR2 2190 180 20
3. Study Methods and Steps

3.1. Study Methods
3.1.1. Random Forest Algorithm
The random forest algorithm (RF algorithm) is a machine learning algorithm based
on the decision tree combination composition proposed by Breiman. It is suitable for
processing high-dimensional data and is not easy to produce overfitting. The basic principle
of the algorithm is as follows:
(1) Through the Bootstrap Method with resampling, N groups of Bagging are randomly
selected from the original data set with replacement. The size of each Bagging is about 2/3
of the original data; and the size of the test data set is about 1/3 of the original data, which
is called out of bag (OOB) data.
(2) Then, according to the minimum principle of the Gini Coefficient, N groups of
Bagging are randomly selected to form N decision trees, and the subset of each node
variable after internal splitting is used to construct multiple CART decision trees and form
a random forest. The definition formula of the Gini Coefficient is shown in Equation (1),
where T is a given data set, Ci is a sample randomly selected and identified as a certain
category, and f (Ci , T )/| T | is the probability that the selected sample is Ci .
Gini = ∑ ∑ ( f (Ci , T )/|T |)( f (Cj , T )/|T |) (1)

i6= j
In the equation, T is a given data set, Ci is a sample randomly selected and identified as
a certain category, and f (Ci , T )/| T | represents the probability that the selected sample is Ci .
(3) The generated random forest classifier classifies the data. For the accuracy estima-
tion, when each sample belongs to the OBB sample, its votes are counted each time, and
the votes of the majority will decide the classification category. Since OBB samples are not
involved in the establishment of decision trees, they can be used to estimate prediction
errors, so as to use OBB errors to evaluate model performance and quantify the importance
of each characteristic variable [28]. The importance discrimination of characteristic vari-
ables can improve the accuracy of water extraction. The importance of variables is defined
in Equation (2):
1 t j
V k j = ∑ ei − ei (2)
N i =1
In the Equation (2), V k j is the importance of j characteristic variables, N is the

generated decision tree, ei is the OOB error of the i-th decision tree, and eij is the new OOB
error calculated after randomly changing the j-th characteristic variable.
3.1.2. Accuracy Verification Algorithm

Accuracy evaluation is an effective means of judging the quality of water extraction
methods, so as to determine the water extraction methods suitable for the study area.
The accuracy evaluation method usually adopts the confusion matrix. The mathematical
expression equation of confusion matrix is shown in (3):
 
x11 x12 ... x1n
 x21 x22 ... x23 
CM =  (3)
 
.. .. .. .. 
 . . . . 
xn1 xn2 ... xnn
where the xij (i, j = 1, 2, . . . n) element represents the probability that the training sample is
divided into class J, the i-th row of the matrix represents the probability that the i-th type
n
of training sample is identified as other types, obviously ∑ xij = 1, while the j-th column
j =1
represents the probability that each type of training sample is identified as class j by the
sensor, and the diagonal element represents the probability that the class i sample in the
training sample is accurately identified.
The Kappa coefficient is a statistical method to determine the classification accuracy,
and it is a commonly used parameter in the consistency test. The closer the Kappa value is
to 1, the better the representation, and vice versa, the worse the representation.
N. ∑ri xij − ∑( xi+ · x+i )

Kappa = (4)
N 2 − ∑ ( xi + · x +i )
where r is the number of rows of the error matrix; xij is the value on the main diagonal; xi+
and x+i is row i and column j, respectively, and N is the total number of samples.
C
Q= (5)
A
where Q is the overall classification accuracy, C is the exact number of pixels, and A
is the total.
3.2. The Extraction of Characteristic Variables

Based on Sentinel-2 image data and terrain data, this study extracted 24 characteristic
indicators. The spectral band data included B2-B8A, B11, and B12 bands, and the role
of the red-edge band in land use classification was taken into consideration [29]. The
authors divide it into 4 types: (1) traditional remote sensing data: B2, B3, B4, B8, B11, B12,
NDVI, and MSAVI; (2) red-edge remote sensing data: B5, B6, B7, B8A, NDI45, MCARI,
REIP, S2REP, IRECI, and PSSRa; (3) water indices factors: NDWI, MNDWI, and LSWI; and
(4) terrain factors: elevation, slope, and slope aspect. The relevant calculation equation
and indices descriptions are shown in Table 2. The spectral reflectance curve of ground
objects corresponding to each characteristic variable is shown in Figure 2. Considering the
important role of the red-edge band in land use classification, these spectral characteristic
indices are divided into four categories. At the same time, in order to test the classification
effect of various variables, four models are constructed in the study: (1) Model A: traditional
remote sensing data; (2) Model B: traditional remote sensing data + red-edge remote sensing
data; (3) Model C: traditional remote sensing data + red-edge remote sensing data + water
indices factor; and (4) Model D: traditional remote sensing data + red-edge remote sensing
data + water indices factor + terrain factor. By comparing and analyzing the importance
ranking of characteristic variables, the author selects the best classification scheme.
Table 2. Characterization of spectral indices.
Abbreviations Full Names Calculation Formulas Types Citations

Normalized Difference Vegetation Traditional Spectral
NDVI (B8 − B4)/(B8 + B4) [30]
Indices Indices Characteristics
Traditional Spectral
LSWI Land Surface Water Indices (B8 − B11)/(B8 + B11) [31]
Indices Characteristics
Normalized Difference Water Traditional Spectral
NDWI (B8 − B4)/(B8 + B4) [32]
Modified Soil Adjusted Vegetation 0.5 × (2 × (B8+1) − sqrt((2 × Traditional Spectral
MSAVI [33]
Indices B8 + 1)2 − 8 × (B8 − B4))) Indices Characteristics
Modified Normalized Difference Traditional Spectral
MNDWI (B3 − B11)/(B3 + B11) [34]
Water Indices Indices Characteristics
Red-Edge Spectral
NDI45 Normalized Difference Indices (B5 − B4)/(B5 + B4) [35]
Indices Characteristics
Modified Chlorophyll Absorption [(B5 − B4) − 0.2 × (B5 − B3)] Red-Edge Spectral
MCARI [36]
Ratio Indices × (B5 − B4) Indices Characteristics
700 + 40 × ((B4 + B7)/2 − Red-Edge Spectral
REIP Red-Edge Inflection Point Indices [37]
B5)/(B6 − B5) Indices Characteristics
The Sentinel-2 Red-Edge Position 705 + 35 × ((B4 + B7)/2 − Red-Edge Spectral
S2REP [38]
Indices B5)/(B6 − B5) Indices Characteristics
Inverted Red-Edge Chlorophyll Red-Edge Spectral
IRECI (B7 − B4)/(B5/B6) [39]
Pigment Specific Simple Red-Edge Spectral
PSSRa B7/B4 [40]
Ratio(chlorophyll) Indices. Indices Characteristics
REIP [37]
dices B5) Indices Characteristics
The Sentinel-2 Red-Edge Posi- 705 + 35 × ((B4 + B7)/2 − B5)/(B6 − Red-Edge Spectral
S2REP [38]
tion Indices B5) Indices Characteristics
Inverted Red-Edge Chloro- Red-Edge Spectral
IRECI (B7 − B4)/(B5/B6) [39]
phyll Indices Indices Characteristics
Pigment Specific Simple Ra- Red-Edge Spectral
PSSRa B7/B4 [40]
tio(chlorophyll) Indices. Indices Characteristics
Figure2.2.Spectral
Figure Spectralcurve
curveof
ofsurface
surfacefeatures.
features.
3.3.
3.3. Sample
Sample Selection
Selection
The
The study
studycombines
combines the thevector
vector data
dataand andSentinel-2
Sentinel-2 data
data ofofXiangtan
Xiangtan City’s
City’sland
landuse
use
status
statusin in2018
2018to to collect
collect land
land samples
samples within
within itsits boundaries.
boundaries. To To ensure
ensurethe
the scientificity
scientificityand
and
accuracy
accuracy of of the
thestudy
studyandandtotomake
makethethesample
samplefully fullysatisfy
satisfythethecharacteristics
characteristicsofofrandom
random
distribution,
distribution, the authors used the following approach for sample selection: (1) since
the authors used the following approach for sample selection: (1) since the
the
Sentinel-2
Sentinel-2 datadata has
has been
been resampled
resampled to to aa spatial
spatial resolution
resolution of of 10
10m,m,ArcGIS
ArcGIS10.4
10.4software
software
isis used
used toto generate
generate aa grid
grid of 30mm××
of30 3030m;m; (2)(2) perform
perform spatial
spatial overlay
overlay analysis
analysis on grid
on grid and
and 2
current land use vector data, and filter attribute data with an area equal to 100100
current land use vector data, and filter attribute data with an area equal to m2;m(3);
(3) vector
vector data
data conversion
conversion was
was performed
performed to to convert
convert the
the surfacedata
surface datainto
intopoint
pointdata,
data,and
anda
atotal
totalofofabout
about300,000
300,000 points were extracted. On this basis, in order to
points were extracted. On this basis, in order to improve the effi-improve the
efficiency of random forest, the authors selected 10,000 sample
ciency of random forest, the authors selected 10,000 sample points by ArcGIS randompoints by ArcGIS random
sampling
sampling tool. Theselected
tool. The selectedsample
samplepoints
pointsadequately
adequately cover
cover thethe study
study area
area andandmeetmeet
the
the requirement of uniform distribution. In order to further ensure the accuracy of the
data, the author imported the samples into Google Earth and eliminated outliers. There
are 7226 samples finally retained, including 1390 water areas, 1849 cultivated land areas,
2987 forest land areas, 219 grass land areas, 156 wetlands, 249 other land areas, and
376 construction land areas.
The samples are divided into two parts, 70% of them are used as the training sample set
for modeling, and the remaining 30% are used as the validation set and do not participate
in the model construction. Additionally, they are used to evaluate the classification effect of
each model. The number of sample points for each object classification is shown in Table 3.
The classification accuracy is evaluated by the overall accuracy (OA), Kappa coefficient,
producer’s accuracy (PA), and user’s accuracy (UA). Among them, Kappa coefficient is an
index to measure the classification accuracy, which can measure the consistency between
the classification results of the model and the true results at the same time. Its value ranges
from −1 to 1, and the closer it is to 1, the more consistent the two are; when it is close to 0,
it indicates that the two are consistent with each other as expected by chance, which means
the consistency is poor; when it is close to −1, it indicates that the consistency between the
two is very low, which means the consistency is extremely poor.
Table 3. Number of samples for each class.
Training Total Number of

Characteristic Class Validation Samples
Samples Samples
Water 927 463 1390
Cultivated land 1233 616 1849
Forest land 1996 991 2987
Grass land 146 73 219
Wetland 104 52 156
Construction land 251 125 376
Other land 166 83 249
4. Results
4.1. Results of Classification and Accuracy Evaluation
In order to improve the accuracy of classification, this study adopts a grid-search
based on OOB error value. This method employs ntree and mtry to search for parameter
optimization. By employing ntree and mtry, grid-search uses M and N values. By multi-
plying M and N, it disciplines different RF classifiers and estimates its learning accuracy
according to OOB error value. In this way, the exact combination with the highest learning
accuracy is regarded as the optimal parameter, whose advantage is to assure the optimal
solution in the delineated grid and avoid significant errors. The RF classification algorithm
and grid-search algorithm are both implemented in Matlab 2012B language platform. The
grid-search allows us to search for the optimal parameter according to the RF algorithm for
the dour models. The scope of mtry parameter optimization for model A is (2, 3, 4, 5); for
model B it is (3, 5, 7, 9); for model C it is (5, 10, 15); and for model D it is (6, 12, 18). The scope
of ntree parameter optimization for all models is unified as (25, 50, 75, 100). The mtry and
ntree optimal parameters for the four models are, respectively, (4, 100), (7, 100), (10, 100),
and (12, 100). The above parameters are used to execute RF classification algorithm, and the
results are shown in Figure 3. Figure 4 is a schematic diagram of water areas superimposed
by a false color image. The overall accuracy of the four models are, respectively, 81.24%,
86.39%, 89.71%, and 92.37%. With the increasing of parameters, the overall accuracy and
Kappa coefficient increase accordingly. The results show that the addition of terrain factors
and red-edge remote sensing data can improve the accuracy of classification recognition.
The conclusions are in line with those of Forkuor [29] and A. Htitiou [40]. Figure 3 is
accuracy comparison of four models (model A, model B, model C, and model D).
From Table 4, we can see that as the parameter increases, the classification accuracy
varies for different land. Therefore, the figure of local area classification results is selected.
Three regions of the figure will be selected randomly so as to make a specific contrast
analysis. Figure 4 is the standard false color image of Sentinel-2 data in Xiangtan City.
Figure 4a–c are images of three local areas in Xiangtan City. Figures 5–7 are classification
images corresponding with three local areas. According to the contrast of the three figures,
we can see that the classification accuracy of model A is lower than that of model B. The
minor error is mainly due to the wrong classification of wetland and cultivation land.
Based on the traditional remote sensing image data, model B employs the red-edge remote
sensing data B5, B6, B7, B8A, NDI45, MCARI, REIP, S2REP, IRECI, and PSSRa. Therefore,
its producer’s accuracy has increased 5.47%, and the user’s accuracy increases 6.89%.
Meanwhile, red-edge remote sensing data can effectively decrease the influences of soil
background noise, making the water extraction accuracy increase from 80.63% to 82.16%.
Despite this, the accuracy enhancement is not noticeable. The reason is that the red-edge
band can more accurately identify different kinds of vegetation. According to the contrast
between A1 and D1 of Figure 5, A2 and D2 of Figure 6, and A3 and D3 of Figure 7, we can
see clearly that the black and white dots in model A are fewer than those of model D. The
whole figure is clearer and prettier. However, the producer’s accuracy and user’s accuracy
of building land have decreased 1.62% and 2.2%, respectively. It might be the image by
red-edge band.
Sustainability 2022,14,
Sustainability2022, 14,3797
x FOR PEER REVIEW 99 of
of 19
20
Sustainability 2022, 14, x FOR PEER REVIEW 10 of 20
Grassland 82.36 76.52 84.51 79.84 87.12 81.65 90.12 83.54

Wetland 80.15 73.28 85.62 80.17 87.45 81.56 90.22 93.14
Building land 82.03 73.49 80.41 71.29 83.46 79.02 86.52 81.39
Others 85.87 81.37 88.41 81.61 91.67 84.62 94.67 88.62
The overall
81.24 86.39 89.71 92.37
accuracy (OA)%
Kappa coefficient 0.8016 (a) Model A0.809 2 (b) Model
0.811 4 B 0.811 6
Compared with model B, model C is visually clearer. The black and white dots are
fewer and characteristics are more evenly spaced. (See Figures 5C1, 6C2 and 7C3). Mean-
while, the water index factors of NDWI, MNDWI, and LSWI are added. Among them, the
improved water index factor MNDWI can strengthen open waters and dampen or even
remove the black and white dots caused by clusters of buildings, vegetation, and soil. In
addition, the clattered objects on the ground have been effectively merged. The extraction
of linear objects on the ground is also more effective. The water extraction accuracy has
increased from 82.16% to 91.39%. The overall accuracy has increased from 86.39 to 89.71%.
Compared with model C, because of the addition of terrain factors, model D’s accu-
racy has been(c)enhanced.
Model C The shadows of mountain areas and
(d) Model D the wrong classification of
waters are noticeably decreased (see Figure 7C3). The addition of topographic factors such
Figure3.
3. Accuracy
Accuracy comparison
comparisonof
offour
fourmodels
models(model
(modelA,
A, modelB,B,model
modelC,
C,and
andmodel
modelD).
D).
Figure
as elevation, slope, and aspect can effectively reducemodel
water misclassification.
From Table 4, we can see that as the parameter increases, the classification accuracy
varies for different land. Therefore, the figure of local area classification results is selected.
Three regions of the figure will be selected randomly so as to make a specific contrast
analysis. Figure 4 is the standard false color image of Sentinel-2 data in Xiangtan City.
Figure 4a–c are images of three local areas in Xiangtan City. Figures 5–7 are classification
images corresponding with three local areas. According to the contrast of the three figures,
we can see that the classification accuracy of model A is lower than that of model B. The
minor error is mainly due to the wrong classification of wetland and cultivation land.
Based on the traditional remote sensing image data, model B employs the red-edge remote
sensing data B5, B6, B7, B8A, NDI45, MCARI, REIP, S2REP, IRECI, and PSSRa. Therefore,
its producer’s accuracy has increased 5.47%, and the user’s accuracy increases 6.89%.
Meanwhile, red-edge remote sensing data can effectively decrease the influences of soil
background noise, making the water extraction accuracy increase from 80.63% to 82.16%.
Despite this, the accuracy enhancement is not noticeable. The reason is that the red-edge
band can more accurately identify different kinds of vegetation. According to the contrast
between A1 and D1 of Figure 5, A2 and D2 of Figure 6, and A3 and D3 of Figure 7, we can
see clearly that the black and white dots in model A are fewer than those of model D. The
whole figure is clearer and prettier. However, the producer’s accuracy and user’s accuracy
Figure
Figure 4.
4. Standard
of building land false
Standard have color
false image
decreased
color of1.62%
image Sentinel-2
and data
2.2%,
of Sentinel-2 inrespectively.
Xiangtan
data City It
in Xiangtan and three
might
City local
andbe the
threearea images
image
local by
area
in (a–c). band.
red-edge
images in (a–c).
Table 4. Classification accuracy statistics of models A, B, C, and D.
Model A Model B Model C Model D

Producer’s User’s Producer’s User’s Producer’s User’s Producer’s User’s
Types
Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
(PA) (UA) (PA) (UA) (PA) (UA) (PA) (UA)
Water 80.63 83.22 82.16 86.22 91.39 94.12 92.16 95.22
Cultivation land 85.43 79.26 88.52 81.53 89.21 82.46 91.25 84.53
Figure 5. The classification result map of the four models corresponding to the first area. There is a
Forest 90.45 92.76 92.87 93.98 93.07 94.19 93.98 95.06
lot of noise in (A1), the noise of (B1) is significantly reduced, the water body extracted by (C1) is
more complete, and the water body extracted by (D1) has the best effect.
racy has been enhanced. The shadows of mountain areas and the wrong classification of
waters are noticeably decreased (see Figure 7C3). The addition of topographic factors such
as elevation, slope, and aspect can effectively reduce water misclassification.
Table 4. Classification accuracy statistics of models A, B, C, and D.
Model A Model B Model C Model D

Producer’s User’s Producer’s User’s Producer’s User’s Producer’s User’s
Types
Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
(PA) (UA) (PA) (UA) (PA) (UA) (PA) (UA)
Water 80.63 83.22 82.16 86.22 91.39 94.12 92.16 95.22
Cultivation land 85.43 79.26 88.52 81.53 89.21 82.46 91.25 84.53
Forest 90.45 92.76 92.87 93.98 93.07 94.19 93.98 95.06
Grassland 82.36 76.52 84.51 79.84 87.12 81.65 90.12 83.54
Wetland 80.15 73.28 85.62 80.17 87.45 81.56 90.22 93.14
Building land 82.03 73.49 80.41 71.29 83.46 79.02 86.52 81.39
Others 85.87 81.37 88.41 81.61 91.67 84.62 94.67 88.62
The overall
81.24 86.39 89.71 92.37
accuracy (OA)%
Kappa coefficient 0.8016 0.8092
Figure 4. Standard false color image of Sentinel-2 data 0.8114 0.8116area images
in Xiangtan City and three local
in (a–c).
Figure 5.
Figure 5. The
The classification
classification result
result map
mapof
ofthe
thefour
fourmodels
modelscorresponding
correspondingtotothe
thefirst
firstarea.
area.There
Thereis is
a
alotlotofof
Sustainability 2022, 14, x FOR PEER REVIEW noise inin
noise (A1), the
(A1), noise
the ofof
noise (B1) is is
(B1) significantly reduced,
significantly thethe
reduced, water body
water extracted
body by by
extracted (C1) is 20
11(C1)
of is
ability 2022, 14, x FOR PEER REVIEW more complete, and the water body extracted by (D1) has the best effect. 11 of 20
more complete, and the water body extracted by (D1) has the best effect.
Figure
Figure 6.6. The
The classification result map
classification result map ofof the
the four
four models
models corresponding
correspondingto tothe
thesecond
secondarea.
area.There
Thereis
Figure 6. The classification
is a lot of result
noise in map of
(A2), the
the fourofmodels
noise (B2) is corresponding
significantly to the second
reduced, the area.body
water There
extracted by (C2)is
a lot of noise in (A2), the noise of (B2) is significantly reduced, the water body extracted by (C2)
is a lot of noise is
in more
(A2), complete,
the noise of (B2)
and theiswater
significantly reduced,by
body extracted the water
(D2) hasbody extracted
the best effect. by (C2) adding the
Because
more complete, and the water body extracted by (D2) has the best effect. Because adding the water
is more complete, andbody
water the water
indexbody
factorextracted
to modelby C (D2) has the best
can improve effect. body
the water Because addingaccuracy.
extraction the
water body index body index
factor factor to
to model model
C can C can the
improve improve
waterthe water
body body extraction
extraction accuracy.accuracy.
Figure 7. The classification result map of the four models corresponding to the third area. There is
Figure 7. The classification
aFigure
lot of 7.
noiseresult
The in map
(A3), theofnoise
classificationthe four models
of (B3)
result map corresponding
isofsignificantly
the tocorresponding
reduced,
four models thethe
third area.
water toThere
body is area.
theextracted
third by There
(C3) isis
a lot of noise in more
(A3), complete,
the noise of (B3)
and theiswater
significantly reduced,by
body extracted the water
(D3) hasbody extracted
the best effect. by (C3) isadding the water
Because
a lot of noise in (A3), the noise of (B3) is significantly reduced, the water body extracted by (C3) is
more complete,body and the water
index body
factor to extracted
model C canby (D3) has the
improve the best
watereffect.
bodyBecause adding
extraction the water
accuracy. Adding terrain
more complete, and the water body extracted by (D3) has the best effect. Because adding the water
body index factor to model
factors C canDimprove
to Model can reducethethe
water body extraction
misclassification of accuracy.
hillshade Adding
and waterterrain
bodies.
factors to Modelbody D canindex factor
reduce thetomisclassification
model C can improve the water
of hillshade andbody
waterextraction
bodies. accuracy. Adding terrain factors
to Model D can reduce the misclassification
4.2. The Evaluation of Characteristic Importance of hillshade and water bodies.
4.2. The Evaluation of
OneCharacteristic Importance
of the advantages of random forest classification method is its contribution and
One of the advantagesofofevaluating
importance random forest classification
characteristic method
variables is its
[41]. contribution
Figure 6 showsand
the importance
importance ofranking
evaluating characteristic variables [41]. Figure 6 shows the importance
of characteristic variables according to OOB error. It results in the best model
ranking of characteristic
being modelvariables according
D. According to thetofigure,
OOB the
error. It resultsand
importance in the best model
contribution of each variable
being model D. According
can to the(See
be evaluated figure, the 8).
Figure importance and band
The original contribution
data can ofmainly
each variable
be seen in the front
Compared with model B, model C is visually clearer. The black and white dots
are fewer and characteristics are more evenly spaced. (See Figures 5C1, 6C2 and 7C3).
Meanwhile, the water index factors of NDWI, MNDWI, and LSWI are added. Among them,
the improved water index factor MNDWI can strengthen open waters and dampen or even
remove the black and white dots caused by clusters of buildings, vegetation, and soil. In
addition, the clattered objects on the ground have been effectively merged. The extraction
of linear objects on the ground is also more effective. The water extraction accuracy has
increased from 82.16% to 91.39%. The overall accuracy has increased from 86.39 to 89.71%.
Compared with model C, because of the addition of terrain factors, model D’s accuracy
has been enhanced. The shadows of mountain areas and the wrong classification of waters
are noticeably decreased (see Figure 7C3). The addition of topographic factors such as
elevation, slope, and aspect can effectively reduce water misclassification.
4.2. The Evaluation of Characteristic Importance

One of the advantages of random forest classification method is its contribution and
importance of evaluating characteristic variables [41]. Figure 6 shows the importance
ranking of characteristic variables according to OOB error. It results in the best model
being model D. According to the figure, the importance and contribution of each variable
can be evaluated (See Figure 8). The original band data can mainly be seen in the front
position of the ranking, so the normalized difference water index (MNDWI) and band B2
(Blue), normalized water index (NDWI), B4 (Red), B3 (Green), and band B5 (Vegetation
Red-Edge 1) are improved. The importance of blue band (B2) is far higher than that of other
bands. Because the blue band is sensitive to soil background reflection, it can effectively
differentiate soil from vegetation. MNDWI has the noticeable effect of extracting water
information [42]. For the classification of land use and land coverage, the characteristics
of original band and the importance of MNDWI have been identified by many scholars.
However, due to the different study areas and study time, the categories of objects and the
growth state of plants are different. Therefore, the importance ranking differs. In general,
they are among the former position [43]. Among the importance ranking, the following are
traditional remote sensing dates (NDVI, B12, B11, NDI45, PSSRa, LSWI, and MSAVI). Most
red-edge remote sensing data are among the behind position of the ranking. Additionally,
the importance of terrain factors (elevation, slope, and aspect) is also lower.
VALUE
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Figure 8. Characteristic variables importance value of model D.

The
The accuracy
accuracy ofof the
the random
random forest
forest classification
classification results
results of
of 24
24 characteristic
characteristic combina-
combina-
tions is evaluated, and the relationship diagram between the number of
tions is evaluated, and the relationship diagram between the number of characteristics characteristics
and
and the
the classification
classification accuracy
accuracy was
was generated
generated (see
(see Figure 9). Conclusions
Figure 9). Conclusions cancan be
be reached
reached
that as characteristics increase, the variables with higher importance and contributions
that as characteristics increase, the variables with higher importance and contributions are
first used as the input data of random forest models. The correlation of characteristic var-
iables is low. The overall accuracy and Kappa coefficient demonstrate a sharp increase. In
the first period, as the visible band and red-edge index band are input into the random
forest classification model, the overall accuracy and Kappa coefficient quickly increased,
respectively, from 54.52% and 0.36 to 89.98% and 0.779. The tendency of growing classifi-
The accuracy of the random forest classification results of 24 characteristic combina-

Sustainability 2022,tions
14, 3797
is
evaluated, and the relationship diagram between the number of characteristics 12 of 19
and the classification accuracy was generated (see Figure 9). Conclusions can be reached
that as characteristics increase, the variables with higher importance and contributions are
first used as theare
input
firstdata
usedofasrandom
the inputforest
datamodels. The correlation
of random forest models.of characteristic
The correlation var-of characteristic
iables is low. The overall is
variables accuracy
low. Theand Kappa
overall coefficient
accuracy and demonstrate a sharp
Kappa coefficient increase. Ina sharp increase.
demonstrate
the first period,Inasthe
thefirst
visible band
period, as and red-edge
the visible bandindex
and band are input
red-edge index into
bandthearerandom
input into the random
forestmodel,
forest classification classification model,
the overall the overall
accuracy accuracy
and Kappa and Kappa
coefficient coefficient
quickly increased, quickly increased,
respectively,
respectively, from 54.52% and from
0.3654.52%
to 89.98%andand
0.360.779.
to 89.98% and 0.779.ofThe
The tendency tendency
growing of growing classifi-
classifi-
cation accuracycation
is rapid.accuracy
In the is rapid.period,
middle In the middle period,
the increase the increase
fluctuates. When fluctuates.
the number When the number
of reaches
of characteristics characteristics
16, thereaches 16, theand
classification classification and Kappa
Kappa coefficient reachcoefficient
the highest reach the highest
level, 92.57% andlevel, 92.57%
0.8116, and 0.8116,
respectively. respectively.
When Whennumbers
characteristic characteristic
increasenumbers
from 16 increase
to from 16 to
24, the
24, the data become data become
redundant, andredundant,
more time and more time
is needed due tois needed due to the interconnectivity
the interconnectivity of of
different The
different characteristics. characteristics.
performanceThe performance
of classifier of classifier
is worse is worse
than before. than the
Therefore, before. Therefore,
phenomenon ofthe phenomenon
over-fitting ofclassification
of the over-fitting of the classification
results is in the trend results is in the trend
of decreasing fluc- of decreasing
tuation of the classification accuracy. For this reason, in order to ensure the classificationthe classification
fluctuation of the classification accuracy. For this reason, in order to ensure
accuracy
accuracy and improve theand improve the
classification classification
efficiency, efficiency,
the research the research
carried carried
out further out further feature
feature
variable
variable selection selectionthe
and reduced and reduced the
dimension of 24dimension of 24 feature
feature variables. Two variables.
methods are Two methods are
mainly used formainly used for
comparative comparative analysis.
analysis.
Figure 9. The relationship

Figure 9. between the number
The relationship of features
between and classification
the number accuracy.
of features and classification accuracy.
(1) Choose based on variable importance estimates. According to the importance of

the feature variables of model D and the first 16 feature variables of model D (MNDWI, B2,
NDWI, B3, B5, NDVI, B12, B11, NDI45, PSSRa, LSWI, B7, S2REP, B8A, B8, and REIP), the
new model is recorded as RF_16.
(2) The 24 feature variables are screened by the Relief F algorithm. The Relief F
algorithm is a feature selection algorithm that considers the interdependence of multiple
variables and features. This algorithm can effectively deal with noisy multi-class and
regression problems. The Relief F algorithm evaluates the classification ability of the
feature through the “hypothesis interval”, considering the class distance and the intra-class
distance comprehensively. If the class distance is greater than the intra-class distance, the
weight is increased; the weights are continuously updated through the class spacing and
intra-class distance, and feature selection is performed according to the final calculated
weights. The weight update formula is shown in Equation (3), assuming that the interval θ
refers to when the sample category is kept unchanged. For the maximum distance that the
classification decision surface can move, see Formula (4).
1 ∑c6=class( xi ) di f f f [ xi , M( xi )] di f f f [ xi , H ( xi )]
( )
i i −1
wf = wf + − (6)
n (n − 1) × mclass(xi ) mclass( xi )
1
θ= [k x − M( x )k − k x − H ( x )k] (7)
2
In the formula, diff () is the distance between different samples, n is the number of
samples, f is the feature of the evaluation, i is the randomly selected sample, and M(x), H(x)
are the same and the closest to the sample x, respectively, and are adjacent sample points.
The Relief F algorithm is used to select the 24 feature variables of model D, and the
first 16 variables that are more related to the target class are reserved, and the model with
the same number as RF_16 is constructed, which is recorded as Relief-16. The variable com-
bination model Relief-16 based on the Relief F algorithm dimensionality reduction, and the
variable combination model RF_16 based on variable importance estimation dimensionality
reduction are classified by random forest algorithm, and the algorithm execution time, clas-
sification accuracy, and Kappa coefficient are compared, see Table 5. The RF_16 algorithm
is compared with Relief-16 and model D, and the overall classification accuracy is increased
by 0.12% and 0.79%, respectively, and the Kappa coefficient is 0.0048 and 0.0108 higher than
that of Relief-16 and model D, respectively. The lowest classification accuracy of cultivated
land is 90.12%. In terms of the cultivated land classification accuracy of the Relief-16 model,
the highest is 92.21%, which shows that the RF_16 algorithm is not the best cultivated land
extraction model. However, the ground type with the highest classification accuracy of the
RF_16 model is the water body, which is 94.16%. Compared with the Relief-16 algorithm
and the model D water body extraction accuracy, the extraction accuracy is increased by
1.77% and 2%, respectively. The extraction accuracy of other ground object types is higher
than 91%. The classification effect is generally better. Therefore, in comparison of the three
models, the RF_16 model has the best effect in classifying land use information, especially
in terms of water body extraction.
Table 5. Classification accuracy statistics of models RF_16, Relief-16, and model D.
RF_16 Relief-16 Model D

Producer’s User’s Producer’s User’s Producer’s User’s
Types
Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
(PA) (UA) (PA) (UA) (PA) (UA)
Water 94.16 96.22 92.39 95.42 92.16 95.22
Cultivation land 90.12 88.53 92.21 85.46 91.25 84.53
Forest 93.87 94.98 93.17 94.19 93.98 95.06
Grassland 91.51 88.34 91.12 84.65 90.12 83.54
Wetland 93.62 90.17 92.45 91.56 90.22 93.14
Building land 91.41 89.29 89.56 85.02 86.52 81.39
Others 94.81 92.61 91.67 89.62 94.67 88.62
The overall
93.16 93.04 92.37
accuracy (OA)%
Kappa coefficient 0.8224 0.8176 0.8116
The classification results and differences are shown in Figure 10. Figure 10a is a
partial map of the classification effect of the Relief-16 model, and Figure 10b is a partial
map of the same position of the classification effect of the RF_16 model. According to the
comparison of areas 1, 3, 5, 6, and 7 in the figure, it can be seen that the extraction of ground
objects from the classification results of the RF_16 model is more complete, reducing the
misclassification of wetlands and woodlands. In areas 2 and 4, the misclassification of water
bodies is reduced, and the water body extraction is more complete and continuous. At the
same time, in area 4, it can be seen that based on the RF_16 model, the misclassification
of building land and water bodies is reduced, and the overall classification effect is better,
which further proves that the RF_16 model has strong adaptability in water body extraction.
objects from the classification results of the RF_16 model is more complete, reducing the
misclassification of wetlands and woodlands. In areas 2 and 4, the misclassification of wa-
ter bodies is reduced, and the water body extraction is more complete and continuous. At
the same time, in area 4, it can be seen that based on the RF_16 model, the misclassification
of building land and water bodies is reduced, and the overall classification effect is better,
which further proves that the RF_16 model has strong adaptability in water body extrac-
tion.
(a) Relief-16 (b) RF_16

Figure 10. Comparison of classification results of models RF_16 and Relief-16.
Figure 10. Comparison of classification results of models RF_16 and Relief-16.
Figure 11 is theFigure
land use
11 isclassification
the land usemap of Xiangtan
classification map City obtained by
of Xiangtan Citya random
obtained by a random
forest algorithm based on the RF_16 model. It can be seen from this figure
forest algorithm based on the RF_16 model. It can be seen from this figure that the land that the land
use classification results of Xiangtan City based on the RF_16 model algorithm have good
use classification results of Xiangtan City based on the RF_16 model algorithm have good
spatial feasibility, especially
spatial theespecially
feasibility, extracted the
water body information,
extracted which is more
water body information, whichcom-is more complete
plete and moreand accurate. It shows Itthat
more accurate. adding
shows thatred-edge band information
adding red-edge in land use
band information in land use cover
cover can improve the classification
can improve accuracy,accuracy,
the classification while adding
whilea adding
vegetation index canindex
a vegetation effec- can effectively
tively extract vegetation information,
extract vegetation and adding
information, a watera body
and adding waterindex
body factor can greatly
index factor can greatly improve
improve the accuracy of water body extraction. The extracted water body information
the accuracy of water body extraction. The extracted water body information data is
data is superimposedsuperimposed
on the on truethe trueremote
color color remote
sensing sensing
imageimage
data ofdata of Xiangtan
Xiangtan City, City,
as as shown in
shown in Figure Figure 12, which
12, which can provide
can provide betterbetter information
information for future
for future waterwater
bodybody research, databases,
research,
databases, andand technical
technical background
background support.
support.
Figure 11.
Figure 11.Classification
Classificationresults of image.
results of image.
Figure 12.
Figure 12.Xiangtan
XiangtanCity Water
City Information
Water Results
Information Map.Map.
Results
4.3. Accuracy Evaluation

To further evaluate RF_16, the water extraction effect of the 16 model will be based
on RF_16. Compare the water extraction results of the model with the water extraction
results of NDWI, MNDWI, and NDVI, and evaluate the accuracy of the calculation results
4.3. Accuracy Evaluation

To further evaluate RF_16, the water extraction effect of the 16 model will be based on
RF_16. Compare the water extraction results of the model with the water extraction results
of NDWI, MNDWI, and NDVI, and evaluate the accuracy of the calculation results of these
methods, respectively. See Table 6 for the accuracy evaluation of each method.
Table 6. Evaluation of extraction accuracy of water extraction by different methods.
Kappa Missed False Extraction The Overall

Methods
Coefficient Extraction Rate/% Rate /% Accuracy (OA)%
NDVI 0.8046 2.86 13.72 89.22
NDWI 0.8112 1.94 11.24 90.12
MNDWI 0.8206 1.89 10.56 92.57
RF_16 0.8224 0.48 7.22 93.16
Comprehensive analysis based on the RF_16 model method has the lowest missed
detection rate and a low false extraction rate; the missed detection rate of NDVI is the
highest, and the false extraction rate is also high; the missed detection rates of NDWI
and MNDWI are the same, but the false extraction rate of mNDVI is lower than that of
NDVI, indicating that the water extraction effect is better than that of NDVI, but based
on the RF_16 model method, it has the lowest false extraction rate. Simultaneously, based
on RF_16, the water extraction effect of the 16 model method is the best, but it is based
on RF_. When extracting water bodies by the model method, the selection of samples
takes time and energy, and the quality of samples will directly affect the quality of the
RF-based water body. The extraction accuracy of the model, and so the selection of samples
in water extraction, is very key, but the operation of the water index method is simple
and saves time.
In order to reflect the effect of water body extraction by the RF_16 model concretely
and clearly and compare it with the water body effect extracted by NDVI, NDWI, and
MNDWI, we selected the water body extraction effect maps of three local areas in Xiangtan
City. Table 7 shows the different water bodies in Xiangtan City and a schematic diagram
of the results of the extraction method. It can be seen from the table that the water bodies
extracted by NDVI, NDWI, and MNDWI obviously have certain missed detections and
false detections. In particular, the NDVI normalized vegetation index extraction results
of water bodies have the highest missed detection rate and the highest false detection
rate. Especially, NDVI, NDWI, and MNDWI all mistakenly classify hill shadows as water
bodies, while the RF_16 model does not. Comparative analysis of the four methods of
water body extraction shows that the RF_16 model extracts the water body extraction
results with the least noise, and the lowest missed detection rate and false detection rate.
Therefore, comparing the four methods, the overall extraction accuracy of the RF_16 model
is the highest.
andfalse
and
and
and
and false
false
false
false and detections.
detections.
detections.
detections.
detections. In
false detections. Inparticular,
In
detections.
InIn particular,
particular,
In particular,
particular,
particular, theNDVI
the
the
particular,
thethe NDVI
NDVI
NDVI
NDVI normalized
normalized
normalized
thenormalized
NDVI normalized
normalized
normalized vegetation
vegetation
vegetation
normalized
vegetation
vegetation indexextraction
index
index
vegetation
index
index extraction
extraction re-
re-
re-
index extraction
extraction
extraction extraction
re-
re- re
and
andfalse
and
sults
sults false
false
of
of and
water
sults detections.
water false
detections.
detections.
of bodies
water
bodies In Inparticular,
In
have
bodies
have the
the In
particular,
particular,
highest
have the
highest thethe
the NDVI
NDVI
NDVI
missed
highest
missed the NDVI
normalized
normalized
detection
missed
detection rate
detection
ratevegetation
vegetation
vegetation
and
and the
rate
the index
vegetation
highest
and index
index
the
highest extraction
falseindex
extraction
extraction
highest
false detection
false re-
detection re-
re-
detectionre
sults
sults
sults
sults
sultsof
of of
of
of water
water
water
sults
water
water of bodies
bodies
bodies
water
bodies
bodies have
have
have
bodies
have
have the
thethe
the
the havehighest
highest
highest
highest
highestthe missed
missed
missed
highest
missed
missed detection
detection
detection
missed
detection
detection rate
rate
rate
detection
rate
rate and
and
and
and
and rate
thethe
the
the
the highest
highest
highest
and
highest
highestthe false
false
false
highest
false
false detection
detection
detection
false
detection
detection detection
detectio
sultsof
sults
sults ofwater
of water
sults
water bodies
bodies
of water
bodies havethe
have
bodies
have the
the havehighest
highest
highestthe misseddetection
missed
highest
missed detection
missed
detection rateand
rate
detection
rate and
and the
ratethehighest
the highest
and
highestthe falsedetection
false
highest
false detection
false
detection detection
rate.
rate.Especially,
rate.
rate.
rate. rate.
Especially,
Especially,
Especially,
Especially,
rate. NDVI,
Especially,NDVI,NDVI,
NDVI,
NDVI,
NDVI,
Especially, NDWI,
NDWI,
NDWI,
NDWI,
NDWI,
NDVI, and
NDWI,andMNDWI
and
and
and
NDWI, and
MNDWI
MNDWI
MNDWI
MNDWI and all
MNDWI allmistakenly
all
all
all
MNDWI all
mistakenly
mistakenly
mistakenly
mistakenly
all classify
mistakenly
classify
classify
classify
classify
mistakenly hill
hillshadows
classify
hill
hill
hill hill
shadows
shadows
shadows
shadows
classify hill asaswater
shadows
as
as
as water
water
water
water
shadows asaswater
wate
rate.
rate.
rate.
rate.
rate. Especially,
Especially,
Especially,
Especially,
rate.
Especially, NDVI,
NDVI,
NDVI,
NDVI,
Especially,
NDVI, NDWI,
NDWI,
NDWI,
NDWI,
NDVI,
NDWI, and
and
andand
NDWI,
and MNDWI
MNDWI
MNDWI
MNDWI and
MNDWI all
all
all
all
MNDWI
all mistakenly
mistakenly
mistakenly
mistakenly
all
mistakenly classify
classify
classify
classify
mistakenly
classify hill
hill
hill
hill
classify
hill shadows
shadows
shadows
shadowshill
shadows as
as
asaswater
water
water
water
shadows
as water as wate
bodies,
bodies,
bodies,
bodies,
bodies, while
whilethe
bodies,
while
while
while
bodies, while
the
the
the
the RF_16
while the
RF_16
RF_16
RF_16
RF_16 the model
RF_16
modelmodel
model
model
model
RF_16 does
doesnot.
does
does
does
model does
not.
not.
not.
not. Comparative
not. Comparative
Comparative
Comparative
Comparative
Comparative
does not. analysis
analysis
analysis
analysis
analysis
Comparative of
ofofthe
analysis
of
of thefour
the
the
the
analysis of
fourthe
four
four
four
of methods
four
methods
methods
methods
methods
the four of
methods
of
of
of
of
methods of
bodies,
bodies,
bodies,
bodies,
bodies,
water
water body
while
while
while
bodies,
while
while
water
body
the
body
the
thethe
the
extraction
extraction
RF_16
RF_16
RF_16
while RF_16
RF_16 the
shows
extraction
shows
model
model
model
RF_16
model
model that
shows
that
does
does
does
model
does
does
the
the
not.
that
not.
not.
does
not.
not.
RF_16the
RF_16
Comparative
Comparative
Comparative
Comparative
Comparative
model
RF_16
model extracts
model
extracts
analysis
analysis
analysis
not. Comparative analysis
analysis
the
extracts
the
of
water
water
of
ofof
the
the
ofthe
analysisthe
the
the
body
four
water
body
four
four
of
four
four methods
methods
methods
the four
methods
methods
extraction
body
extraction re-of re-oo
of
of
ofof
methods
extraction
re-
water
water
water
water
water body
body
body
bodywater
body extraction
extraction
extraction
body extraction
extraction
extraction shows
shows
shows
extraction
shows
shows that
that
that
shows
that
that thethe
the
the
the RF_16
RF_16
RF_16
that
RF_16
RF_16 model
model
model
themodel extracts
extracts
extracts
RF_16extracts
model
model model
extracts
extracts the
the
the
extracts
the
the water
water
water
water
water the body
body
bodywater
body
body extraction
extraction
extraction re-
re-
re-
body extraction
extraction
extraction extraction
re-
re- re
water
water
water
sults
sults body
with water
body
body
sults
with theextraction
with
the body
extraction
extraction
least
leastthe shows
noise,
least
noise, shows
shows and that
noise,
and shows
that
that
the
the the
the
the
lowest
and RF_16
that
the
lowest the
RF_16
RF_16
missed
lowestRF_16
model
model
missed model
extracts
extracts
detection
missed
detection the
thewater
extracts
the
rate
detection
rate water
water
and
and the
rate bodywater
body
body
false and
false extraction
body
extraction
extraction
detection
false
detection re-
re-
19re- rate.
rate.
detection
rate. re
Sustainability 2022, 14, 3797 sults
sults
sults
sults
sults with
with
with
with
with sults thethe
the
the
the with least
least
least
least
least thenoise,
noise,
noise,
least
noise,
noise, andand
and
andnoise,
and the
the
the
the
the and lowest
lowest
lowest
lowest
lowest the missed
missed
missed
lowest
missed
missed detection
detection
detection
missed
detection
detection rate
rate
rate
detection
rate
rate and
and
and
and
and rate false
false
false
false
false and detection
detection
detection
false
detection
detection
of rate.
16detection
rate.
rate.
rate.
rate. rate
sults
sultswith
sults
Therefore, sults
with
with thethe
the least
with
comparingleast
least noise,
the least
noise,
noise,
the andnoise,
and
and the
the
the lowest
and themissed
lowest
lowest lowest
missed
missed detection
missed
detection
detection rate
detection
rate
rate and
and
and false
rate and
false
false detection
false
detection
detection rate.
detection rate
rate.RF_16
rate.
Therefore,
Therefore,
Therefore,
Therefore,
Therefore,
Therefore,
Therefore,
comparing
comparing
comparing
comparing
Therefore,
comparing
comparing
comparingthe
comparing
thethefour
the
the
the
four
four
four
the
four
four
four
methods,
four
fourmethods,
methods,
methods,
methods,
methods,
the
methods,
methods,
the
theoverall
the
the
the
methods,
the
the
the
overall
overall
overall
overall
overall
overall
extraction
theoverall
extraction
extraction
extraction
extraction
overall
extraction
extraction
accuracy
extraction
accuracy
accuracy
accuracy
accuracy
extraction
accuracy
accuracy
of
accuracy
of
of
of ofthe
of
accuracy
of
the
the
the
the
the
RF_16
theof the
RF_16
RF_16
RF_16
RF_16
of
RF_16
RF_16the RF_16
RF_1
Therefore,
Therefore,
Therefore, Therefore, comparing
comparing
comparing the
comparing the four
four the methods,
methods,
four the
the
methods, overall
overall
the extraction
extraction
overall
the four methods, the overall extraction accuracy of the RF_16 accuracy
accuracy
extraction of of
accuracy the
the RF_16
RF_16
of the
model
modelis
model
model
model is isthe
model
isis thehighest.
the
the
the
model is isthe
highest.thehighest.
highest.
highest.
highest. highest.
model
model
modelis
model
model isis
is the
modelthe
isthe highest.
highest.
thehighest.
the highest.
is the
highest. highest.
Table 7. Comparison of different water extraction methods.
Table
Table7.
Table
Table
Table 7.7.Comparison
Table
7.7. Comparison
Comparison
Comparison
Comparison of
ofofdifferent
7.7.Comparison
of
of different
different
different water
waterextraction
ofofdifferent
different water
water
water water
extraction
extraction
extraction
extraction methods.
extraction methods.
methods.
methods.
methods.
methods.
Table
Table 7.7.
Table7.
Table 7. Table
Comparison
Table Comparison
Comparison
7.Comparison
Comparison
Table 7. ofof
of
of different
7. Comparison
Comparison
different
ofdifferent
different
of different
water
water water
extraction
of different
different extraction
extraction
waterextraction
water extraction
water methods.
water extraction
extraction
methods.
methods.
methods. methods.
methods.
methods.
Table Comparison different water extraction methods.
Methods
Methods Extraction
Extraction Effect Effect FirstArea
First Area SecondSecond
Second Area Area
Area Third
Third Area
Methods
Methods
Methods
Methods
Methods
MethodsMethods Extraction
Extraction
Extraction
Extraction
Extraction
Extraction
Methods Extraction
Effect
Effect
Extraction
Effect
Extraction
Effect
Effect Effect
Effect Effect
First
First
First
First
Area
FirstArea
First Area
Area
AreaFirstArea
Area
First Area
Area
Second
Second
Second
Area
SecondArea
Second Area Area
Area
Second
Area
Second Area ThirdArea
Third
Third
Third
Third
Third
Area
Area
Area
Area
Third
Area
Area
Area
Third Area
Area
Methods
MethodsMethods
Methods
Methods Extraction Effect
Effect
Extraction
Extraction
Extraction Effect Effect
Effect First
First
First Area
FirstAreaFirst
Area
Area Second
Second
Second Area
SecondArea
Second
Area Area
Area Third
Third
Third Area
ThirdArea
Third
Area
Area
NDVI
NDVI NDVI
NDVI
NDVI
NDVI
NDVI
NDVI
NDVI
NDVI NDVI
NDVI
NDVI NDVI
NDVI
NDWI
NDWI
NDWI
NDWI
NDWI NDWI
NDWI
NDWI
NDWI
NDWI
NDWI NDWI
NDWI
NDWI
NDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
MNDWI
RF_16
RF_16 RF_16
RF_16
RF_16
RF_16
RF_16
RF_16
RF_16
RF_16 RF_16
RF_16
RF_16
RF_16
RF_16
5.
5.
5. 5.Discussion
5. 5.5.Discussion
Discussion
Discussion
Discussion
Discussion Discussion
5.5.
5.5. Discussion
Discussion
5.Discussion
5. Discussion
Discussion
Discussion
With
With
5. Discussion
With
With
With the
With
the
the
the
the continuous
With thethecontinuous
continuous
continuous
continuous
continuous progress
progress
progress
progress
progress
continuous of
ofofremote
progress
of
of
progress remote
remote
remote
remote sensing
ofofremote
sensing
sensing
sensing
sensing
remote technology,
sensing
technology,
technology,
technology,
technology,
sensing the
technology, theability
the
the
the
technology, the
ability
ability
ability of ofremote
theability
ability of
of
of remoteofofremote
remote
remote
remote
ability remot
With
With
With
With
With the
the
the With
the
the continuous
continuous
continuous
the
continuous
continuous progress
continuousprogress
progress
progress
progress of
ofof
progress
of
of remote
remote
remote
remote
remoteof sensing
sensing
sensing
remote
sensing
sensing technology,
technology,
technology,
sensing
technology,
technology, the
thethe
technology,
the ability
ability
theability the
ability
ability of
ofof remote
remote
ofremote
ability
of remote
remote of remote
sensing
sensing
sensing
sensing
sensing data
sensing
data
data
data
Withsensing
the
data acquisition
data
acquisition
acquisition
acquisition
continuous
acquisition
data has
acquisitionhas
has
has become
progress
has
acquisition has
become
become
become
become has stronger
become
stronger
stronger
stronger
of become
remote
stronger and
stronger
and
and
and
sensing
and
stronger stronger.
and
stronger.
stronger.
stronger.
technology,
stronger.
and The
stronger.
The
The
The
The
stronger. Sentinel-2A
The
Sentinel-2A
Sentinel-2A
Sentinel-2A
theThe ability
Sentinel-2A satellite
Sentinel-2A
satellite
of satellite
satellite
remote
satellite
Sentinel-2A re-
satellite
re- re-
re-
re-
re-
satellite re
sensing
sensing
sensing
sensing
sensing data
data
data
sensing
data
data acquisition
acquisition
acquisition
data
acquisition
acquisition has
has
has
acquisition
has
has become
become
become
becomehas
become stronger
stronger
stronger
become
stronger
stronger and
and
and
stronger
and
and stronger.
stronger.
stronger.
and spatial
stronger.
stronger. The
The
The
stronger.
The Sentinel-2A
Sentinel-2A
TheSentinel-2A
The
Sentinel-2A
Sentinel-2A satellite
satellite
satellite
Sentinel-2A
satellite
satellite re-
re-
re-
satellite
re-
re- re
mote
mote
mote
sensing
mote
mote sensing
mote
sensing
sensing
data
sensing
sensing
mote data
sensing
data
data
acquisition
data
data
sensing has
hasdata
has
has
has rich
rich
rich
data has
rich
has
rich spectral
rich
spectral
spectral
become
spectral
spectral
has rich bands
spectral
bands
bands
stronger
bands
bands
spectral and
bands
and
and
and
andbandshigh
high
and and
high
high
high and spatial
high
spatial
spatial
stronger.
spatial
spatial
high resolution,
resolution,
resolution,
The
resolution,
resolution,
spatial providing
resolution,
providing
providing
Sentinel-2A
providing
providing
resolution, aaaaamulti-
providing multi-
satellite
providing multi-
multi-
multi- a amulti-
multi
mote
mote
mote
mote sensing
sensing
motesensing
mote feature
sensing
sensing data
data
data
datahas
sensing
data has
has
has
has datarich
rich
rich
richhas
rich spectral
spectral
spectralrich
spectral
spectral bands
bands
bands
spectral
bands
bands and
and
and
bands
and
and high
high
high
high
high and spatial
spatial
spatial
spatial
spatial resolution,
resolution,
resolution,
highsensingspatial
resolution,
resolution, providing
providing
providing
resolution,
providing
providing a
providing multi-
aaaamulti-
multi-
multi-
multi- a multi
dimensional
dimensional
dimensional
dimensional
remote sensing
dimensional
dimensional feature
feature
data
feature
feature
dimensional has space
feature
space
space for
space
for
rich spectral
space
space
feature for
for
for land
space land
land
land for
land cover
bands
for land
cover
cover
cover
cover and
land remote
cover
highremote
remote
remote
remote
remote
cover sensing
sensing
sensing
spatial
sensing
sensing
remote classification.
classification.
classification.
resolution,
classification.
classification.
sensing However,
classification.
providing a However,
However,
However,
However,
However,
classification. multi- thethe the
the
the
the
However, th
dimensional
dimensional
dimensional
dimensional
dimensional feature
feature
feature
feature
dimensional feature space
space
space
space
feature
space for
forfor
for
for land
land
land
space land
land cover
cover
cover
cover
forcover
land remote
remote
remote
remote
cover
remote sensing
sensing
sensing
sensing
remote
sensing classification.
classification.
classification.
classification.
sensing However,
However,
However,
However,
classification.
classification. However, the
the
the
However,the
the th
multi-dimensional
multi-dimensional
multi-dimensional
multi-dimensional
dimensional
multi-dimensional
multi-dimensional feature
feature
feature
space
feature
feature space
feature
space
forspace
space
space
featureland easily
space
easily
easily
easilycover
easily
space causes
easily
causes
causes
remote
causes
causes information
causes
information
information
sensing
information
information
easilyinformation
causes information redundancy
redundancy
redundancy redundancy
classification.
redundancy
redundancy
information during
during
during
However,
during
during
redundancy classifica-
during
classifica-
classifica-
the
classifica-
classifica-
during classifica-
classifica
multi-dimensional
multi-dimensional
multi-dimensional
multi-dimensional
multi-dimensional
feature
featurespace
feature
feature space
space
feature
space easily
spaceeasily easily
easily
space
easily causes
causes
causes
causes
easily
causes information
information
information
causes
information redundancy
redundancy
redundancy
redundancy
informationredundancy during
redundancyduring classifica-
classifica-
duringclassifica-
during
during classifica-
during
classifica- classifica
tion,
tion,which
tion, tion,
which
which
multi-dimensional
tion,
tion, which
which
tion, leads
which
leads
leads
leads
leads
which to
to the
leads
feature
toto thereduction
the
toleads
the
the to the
reduction
reduction
space
reduction
reduction
to the of
reduction
easily
ofofclassification
of
of
reduction of
classification
classification
causes
classification
classification
of accuracy
classification
accuracy
informationaccuracy
accuracy
accuracy
classification and
andalgorithm
accuracy
and
redundancy
and
and
accuracy and
algorithm
algorithm
algorithm
algorithm
and running
algorithm
duringrunning
running speed.
running
classifica-
running
running
algorithm speed.speed.
speed.
speed.
speed.
running speed
tion,
tion,
tion,
tion, which
which
tion,which
which
which leads
leads
tion,leads
which
leads
leads to
toto
to
to the
thethe
leads
the reduction
reduction
thereduction
tofor of
of
ofof
thedimensionality
reduction
reduction reduction classification
classification
ofclassification accuracy
accuracy
accuracy
of classification
classification
classification accuracy
accuracy and
and
and
accuracy
and
and algorithm
algorithm
algorithm
and
algorithm
algorithm running
running
running
algorithm
running
running speed.
speed.
speed.
running
speed.
speed. speed
Commonly
Commonly
tion, Commonly
which
Commonly
Commonly
Commonly used
used
leads used
used
used
Commonly tomethods
used
methods
the
methods
methods
methods
used methods
reduction for
for offor
fordimensionality
for
methods dimensionality
dimensionality
classification
dimensionality
dimensionality
for reduction
reduction
accuracy
reduction
reduction
reduction
dimensionality or
reduction
or
ororfeature
and
or
reduction featureor
algorithm
feature
feature
feature or optimization
feature optimization
optimization
running
optimization
optimization
optimization
feature include
include
speed.
optimization include
include
include include
includ
Commonly
Commonly
Commonly
Commonly
Commonly Commonly used
used
used
usedmethods
used methods
methods
used
methods
methods for
for
for
methods
for
for dimensionality
dimensionality
dimensionality
for dimensionality
dimensionality
dimensionality reduction
reduction
reduction
reduction
reduction or
oror
reduction
or feature
feature
orfeature
feature
feature optimization
optimization
optimization
orDifferent
feature
optimization
include
include
include include
thetheReliefF
the
Commonly
the
the the
ReliefF
ReliefF
ReliefF
ReliefF algorithm
theReliefF
algorithm
used methods
algorithm
algorithm
algorithm
ReliefF [44],
algorithm[44],
[44],
[44],
[44],
algorithm principal
for [44],
principal
dimensionality
principal
principal
principal component
[44],principalcomponent
component
component
component
principal component analysis
analysis
reductionanalysis
analysis
analysis
component [45],
[45],etc.
analysis
[45],
oranalysis
feature
[45],
[45], [45],
etc.
etc.
etc.
etc. etc.
Different
optimization
Different
Different
Different
[45], methods
etc.Different
methods
include
methods
methods
methods
Different are
methodsare are
are
are
are
methods ar
the
the
the
the
the ReliefF
ReliefF
ReliefF
the
ReliefF
ReliefF algorithm
algorithm
algorithm
ReliefF
algorithm
algorithm [44],
[44],
[44],
algorithm
[44],
[44], principal
principal
principal
[44],
principal
principal component
component
component
principal
component
component component analysis
analysis
analysis
analysis
analysis [45],
[45],
[45],
analysis
[45], etc.
etc.
etc.
etc. Different
Different
Different
[45], etc.
Different methods
methods
methods
Different
methods are
are
are
methods
are are
suitable
thesuitable
ReliefF
suitable
suitable
suitable for
suitable
for
for
for
for different
suitable for
different
algorithm
different
different
differentfor datasets
different
datasets
[44], principal
datasets
datasets
datasets
different and
datasets
and
and
and
and
datasets scenarios,
and
scenarios,
component
scenarios,
scenarios,
scenarios,
and and
scenarios,
and
and
and
and
scenarios, aaaaaunified
analysis and
unified
unified
unified
unified
and a[45],
[45],a
etc.
conclusion
unified
conclusion
etc.
conclusion
conclusion
conclusion
unified
Different
Different has
conclusion
has
has
has
has
conclusion
methods
not
nothas
methods
not
not
not yet
yet
yet
has yet
yet been
not
been
are are
been
been
been
not yet
yet been
been
suitable
suitable
suitable
suitable
suitable for
for different
different
fordifferent
for
suitable different datasets
datasets
datasets
datasets
for different and
and
and scenarios,
scenarios,
andscenarios,
datasets scenarios, and
andaaaaaunified
and
and
and scenarios, unified
unified
unified conclusion
and a conclusionconclusion
conclusion
unified has
has not
hasnot
has
conclusion not yet
nothas yet
yet been
yetbeenbeen
been
not bee
yet been
suitable forfor different
different datasets
datasets and and scenarios,
scenarios, and and unified
a unified conclusion
conclusion hashas notnot
yetyet beenbeen
formed. As an ensemble learning method, random forest has the characteristics of high
efficiency and high accuracy. It can not only ensure high accuracy but also fast speed in
medium- and high-resolution image classification and has the ability of feature selection.
While making full use of the rich spectral information and spatial information of
Sentinel-2A data, and by reducing the dimensionality of the high-dimensional feature
variable model and using the out-of-bag (OBB) error, the important feature combina-
tions are prioritized and finally selected. The optimal combined model RF_16 includes
16 characteristic variables of MNDWI, B2, NDWI, B3, B5, NDVI, B12, B11, NDI45, PSSRa,
LSWI, B7, S2REP, B8A, B8, and REIP. Among these spectral characteristic variables, the
visible light band plays an important role in land use classification, and the vegetation
red-edge band of Sentinel-2A also shows a higher value, which is consistent with the results
obtained by Immitzer et al. [46]. Especially in the extraction of water bodies, adding a
water body index factor and vegetation index factor can effectively improve the accuracy
of water body extraction. Therefore, the importance ranking of MNDWI and NDWI factors
is relatively high, and their importance is higher in wetland-type extraction.
Although the application research of Sentinel-2 data based on random forest in water
body extraction in this paper has achieved considerable results, due to the limitations of
various objective conditions, there are still many problems that need to be further explored,
for example, compared with other classification methods, trying to use Sentinel-2 data in
different phases further determines the accuracy of water body extraction based on feature-
optimized random forest method. At the same time, although the classification method
based on machine learning can achieve better classification accuracy, it still cannot solve the
problem of mixed pixels. In future research, the decomposition of mixed pixels can be used
for land use classification to extract water bodies. At the same time, comprehensive analysis
using RF_16, the water extraction effect of 16 algorithm, is the best, but the limitation is
that the selection of samples is time consuming and energy consuming. The operation of
the water index method is simple and relatively time saving. The extraction method of
water body can be selected according to the actual needs.
6. Conclusions
In order to improve the water body extraction accuracy of remote sensing images,
this study is based on Sentinel-2 image data, and four kinds of variables are extracted,
including traditional remote sensing data, red-edge remote sensing data, water body index
factor, and terrain factor; a total of 24 characteristic variable indicators build four combined
models with different input variables. The importance of different characteristic variables
in random forest and the reliability and practical value of the importance of characteristic
variables calculated by random forest are studied, and the importance of each variable is
analyzed. The water body is extracted based on the RF_16 optimal combination model
algorithm after feature optimization and the traditional Relief F feature screening algorithm.
After analysis, it is determined that RF_16 is the optimal combination model for extracting
water body information in the study area. The random forest classification results are
involved in the subsequent accuracy evaluation. To extract the water body information in
the study area, the results show that:
(1) When red-edge remote sensing data, water index factors, and terrain factors are
added to traditional sensing data models in sequence, the overall accuracies are, respec-
tively, 81.24%, 86.39%, 89.71%, and 92.37%; Kappa coefficients are, respectively, 0.8016,
0.8092, 0.8114, and 0.8116.
(2) The characteristic variables that have important influence on model accuracy, such
as normalized difference water index (MNDWI), band B2 (Blue), normalized water index
(NDWI), B4 (Red), B3 (Green), and band B5 (Vegetation Red-Edge 1), have been improved.
(3) The optimal combination model RF_16 has a water extraction accuracy of 93.16%
and a Kappa coefficient of 0.8214. The overall accuracy is 0.12% better than the traditional
Relief F algorithm. The RF_16 method based on the optimal combination model of random
forest is an effective means to obtain high-precision water body information in the study
area. It can effectively reduce the “salt and pepper effect” and the influence of mixed pixels
such as water and shadows on the water extraction accuracy.
In conclusion, different feature variables of random forest have different contributions
to the model and have different effects on the classification accuracy. The importance of
the calculated feature variables is reliable, and the classification accuracy after screening
important feature variables is improved, which gives it a good practical value. Feature-
based optimization of the RF_16 model can play an important role in optimizing the
extraction of water body information, and also provide a scientific basis and decision
support for the management of water body information in the future.
Author Contributions: Y.W. conceived and designed the study. Z.J. wrote the first draft, performed
the data analysis, and collected all the study data. G.Z. and X.W. provided critical insights in editing
the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported in part by the Science and Technology Innovation Platform and
Talent Plan Project of Hunan Province under Grant 2017TP1022 and in part by the Hunan Province
Emergency Management Technology Project (2020YJ007).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The author would like to thank the teachers for their careful guidance and the
students for their kind help.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Li, M.; Liang, H.; Guo, J.; Zhu, A. Automated Extraction of Lake Water Bodies in Complex Geographical Environments by Fusing
Sentinel-1/2 Data. Water Res. 2022, 14, 30. [CrossRef]
2. Du, Y.; Zhang, Y.; Ling, F.; Wang, Q.; Li, W.; Li, X. Water Bodies’ Mapping from Sentinel-2 Imagery with Modified Normalized
Difference Water Index at 10 m Spatial Resolution Produced by Sharpening the SWIR Band. Remote Sens. 2016, 8, 354. [CrossRef]
3. Loukika, K.N.; Keesara, V.R.; Sridhar, V. Analysis of Land Use and Land Cover Using Machine Learning Algorithms on Google
Earth Engine for Munneru River Basin, India. Sustainability 2021, 13, 13758. [CrossRef]
4. Zhu, Y.; Zhou, J.; Qiu, H.; Li, J.; Zhang, Q. Operation Rule Derivation of Hydropower Reservoirs by Support Vector Machine
Based on Grey Relational Analysis. Water Res. 2021, 13, 2518. [CrossRef]
5. Liu, H.; Jiang, Q.; Ma, Y.; Yang, Q.; Shi, P.; Zhang, S.; Tan, Y.; Xi, J.; Zhang, Y.; Liu, B. Object-Based Multigrained Cascade Forest
Method for Wetland Classification Using Sentinel-2 and Radarsat-2 Imagery. Water Res. 2022, 14, 82. [CrossRef]
6. Daho, M.; Settouti, N.; Bechar, M.; Boublenza, A.; Chikh, M.A. A new correlation-based approach for ensemble selection in
random forests. Int. J. Intell. Comput. Cybern. 2021, 14, 251–268. [CrossRef]
7. He, C.; Wei, J.; Song, Y.; Luo, J.-J. Seasonal Prediction of Summer Precipitation in the Middle and Lower Reaches of the Yangtze
River Valley: Comparison of Machine Learning and Climate Model Predictions. Water Res. 2021, 13, 3294. [CrossRef]
8. Han, H.; Choi, C.; Kim, J.; Morrison, R.R.; Jung, J.; Kim, H.S. Multiple-Depth Soil Moisture Estimates Using Artificial Neural
Network and Long Short-Term Memory Models. Water Res. 2021, 13, 2584. [CrossRef]
9. Chatziantoniou, A.; Psomiadis, E.; Petropoulos, G.P. Co-Orbital Sentinel 1 and 2 for LULC Mapping with Emphasis on Wetlands
in a Mediterranean Setting Based on Machine Learning. Remote Sens. 2017, 9, 1259. [CrossRef]
10. Elmahdy, S.; Ali, T.; Mohamed, M. Regional Mapping of Groundwater Potential in Ar Rub Al Khali, Arabian Peninsula Using the
Classification and Regression Trees Model. Remote Sens. 2021, 13, 2300. [CrossRef]
11. Weber, R. Liver-related deaths in persons infected with the human immunodeficiency virus: The D:A:D study. Arch. Internal Med.
2006, 166, 1632–1641. [CrossRef]
12. Wafaa, M.S.; Khalid, E.; Jian, Y.; Shen, S.L. A multi-objective optimization algorithm for forecasting the compressive strength of
RAC with pozzolanic materials. J. Clean. Prod. 2021, 321, 129355. [CrossRef]
13. Ham, J.; Chen, Y.; Crawford, M.; Ghosh, J. Investigation of the random forest framework for classification of hyperspectral data.
IEEE Trans Geosci. Remote Sens. 2005, 43, 492–501. [CrossRef]
14. Zheng, X.; Jia, J.; Guo, S.; Chen, J.; Sun, L.; Xiong, Y.; Xu, W. Full Parameter Time Complexity (FPTC): A Method to Evaluate
the Running Time of Machine Learning Classifiers for Land Use/Land Cover Classification. IEEE J. Stars 2021, 14, 2222–2235.
[CrossRef]
15. Tufail, R.; Ahmad, A.; Javed, M.; Sajid, R. A machine learning approach for accurate crop type mapping using combined SAR and
optical time series data. Adv. Space Res. 2022, 69, 331–346. [CrossRef]
16. Rodriguez-Galiano, V.F.; Ghimire, B.; Rogan, J.; Chica-Olmo, M.; Rigol-Sanchez, J.P. An assessment of the effectiveness of a
random forest classifier for land-cover classification. ISPRS J. Photogramm. Remote Sens. 2011, 67, 93–104. [CrossRef]
17. Topalolu, R.H.; Sertel, E.; Musaolu, N. Assessment of classification accuracies of sentinel-2 and landsat-8 data for land cover/use
mapping. Int. Arch. Photogramm. Remote Sens. Spat. Inf. 2016, 41, 1055–1059. [CrossRef]
18. Tan, Y.; Chen, H.; Xiao, W.; Meng, F.; He, T. Influence of farmland marginalization in mountainous and hilly areas on land use
changes at the county level. Sci. Total Environ. 2021, 794, 149576. [CrossRef]
19. Naushad, R.; Kaur, T.; Ghaderpour, E. Deep Transfer Learning for Land Use and Land Cover Classification: A Comparative Study.
Sensors 2021, 21, 8083. [CrossRef]
20. Forkuor, G.; Dimobe, K.; Serme, I.; Tondoh, J. Landsat-8 vs. Sentinel-2: Examining the added value of sentinel-20 s red-edge bands
to land-use and land-cover mapping in Burkina Faso. GISCI Remote Sens. 2018, 55, 331–354. [CrossRef]
21. Adam, E.; Mutanga, O.; Odindi, J.; Abdel-Rahman, E.M. Land-Use/Cover Classification in a Heterogeneous Coastal Landscape
Using RapidEye Imagery: Evaluating the Performance of Random Forest and Support Vector Machines Classifiers. Int. J. Remote
Sens. 2014, 5, 3440–3458. [CrossRef]
22. Tyralis, H.; Papacharalampous, G.; Langousis, A. A Brief Review of Random Forests for Water Scientists and Practitioners and
Their Recent History in Water Resources. Water Res. 2019, 11, 910. [CrossRef]
23. Talukdar, S.; Singha, P.; Mahato, S.; Pal, S.; Liou, Y.A.; Rahman, A. Land-use land-cover classification by machine learning
classifiers for satellite observations—A review. Remote Sens. 2020, 12, 1135. [CrossRef]
24. Clark, M.L. Comparison of simulated hyperspectral HyspIRI and multispectral Landsat 8 and Sentinel-2 imagery for multi-
seasonal, regional land-cover mapping. Remote Sens. Environ. 2017, 200, 311–325. [CrossRef]
25. Htitiou, A.; Boudhar, A.; Lebrini, Y. The Performance of Random Forest Classification Based on Phenological Metrics Derived
from Sentinel-2 and Landsat 8 to Map Crop Cover in an Irrigated Semi-arid Region. Remote Sens. Earth Syst. Sci. 2019, 2, 208–224.
[CrossRef]
26. Zhang, H.K.; Roy, D.P.; Yan, L.; Li, Z.; Huang, H.; Vermote, E.; Skakun, S.; Roger, J.C. Characterization of Sentinel-2A and
Landsat-8 top of atmosphere, surface, and nadir BRDF adjusted reflectance and NDVI differences. Remote Sens. Environ. 2018,
215, 482–494. [CrossRef]
27. Wang, J.Y.; Su, X.T.; Wang, Q.; Hao, X.Y. Land use change and its impact on ecological service value in the urban-rural interlaced
zone in Xiangtan City. Hunan Agric. Sci. 2021, 6, 93–97. [CrossRef]
28. Ghosh, A.; Sharma, R.; Joshi, P.K. Random forest classification of urban landscape using Landsat archive and ancillary data:
Combining seasonal maps with decision level fusion. Appl. Geogr. 2014, 48, 31–41. [CrossRef]
29. Ahmad, A. Performance Evaluation of Sentinel-2 and Landsat 8 OLI Data for Land Cover/Use Classification Using a Comparison
between Machine Learning Algorithms. Remote Sens. 2021, 13, 1349. [CrossRef]
30. DeFries, R.S.; Townshend, J.R.G. NDVI-derived land cover classifications at a global scale. Int. J. Remote Sens. 1994, 15, 3567–3586.
[CrossRef]
31. Zhang, Y.S.; Xie, Y.L. Vegetation Water Content Estimation Using NDVI and LSWI from MODIS Images. Sci. Geogr. Sin. 2008, 28,
72–76. [CrossRef]
32. Gao, B. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens.
Environ. 1995, 58, 257–266. [CrossRef]
33. Haboudane, D.; Miller, J.R.; Pattey, E.; Zarco-Tejada, P.J.; Strachan, I.B. Hyperspectral vegetation indices and Novel Algorithms
for Predicting Green LAI of crop canopies: Modeling and Validation in the Context of Precision Agriculture. Remote Sens. Environ.
2004, 90, 337–352. [CrossRef]
34. Gautam, V.K.; Gaurav, P.K.; Murugan, P. Assessment of Surface Water Dynamicsin Bangalore Using WRI, NDWI, MNDWI,
Supervised Classification and K-T Transformation. Aquat. Procedia 2015, 4, 739–746. [CrossRef]
35. Zhang, G.F.; Yue, C.R.; Zhang, W.Q. Inversion of aboveground biomass in Dianchi Lakeside Wetland based on Sentinel data.
J. Terahertz Sci. Electron. Inf. 2020, 18, 142–149.
36. Wu, C.; Niu, Z.; Tang, Q.; Huang, W. Estimating chlorophyll content from hyperspectral vegetation indices: Modeling and
validation. Agric. For. Meteorol. 2008, 148, 1230–1241. [CrossRef]
37. Baret, F.; Guyot, G.; Begue, A.; Maurel, P.; Podaireet, A. Complementarity of middle-infrared with visible and near-infrared
reflectance for monitoring wheat canopies. Remote Sens. Environ. 1988, 26, 213–225. [CrossRef]
38. Mercier, A.; Betbeder, J.; Baudry, J.; Roux, V.L.; Hubert-Moy, L. Evaluation of sentinel-1 & 2 time series for predicting wheat and
rapeseed phenological stages. ISPRS J. Photogramm. 2020, 163, 231–256. [CrossRef]
39. Frampton, W.J.; Dash, J.; Watmough, G.; Milton, E.J. Evaluating the capabilities of Sentinel-2 for quantitative estimation of
biophysical variables in vegetation. ISPRS J. Photogramm. 2013, 82, 83–92. [CrossRef]
40. Blackburn, G.A. Quantifying chlorophylls and caroteniods at leaf and canopy scales: An evaluation of some hyperspectral
approaches. Remote Sens. Environ. 1998, 66, 273–278. [CrossRef]
41. Strobl, C.; Boulesteix, A.L.; Zeileis, A.; Hothorn, T. Bias in random forest variable importance measures: Illustrations, sources and
a solution. BMC Bioinform. 2007, 8, 25. [CrossRef]
42. Sivanpillai, R.; Jacobs, K.M.; Mattilio, C.M.; Piskorski, E.V. Rapid flood inundation mapping by differencing water indices from
pre-and post-flood landsat images. Front. Earth Sci. 2021, 15, 11. [CrossRef]
43. Zhang, X.Y.; Li, R.; Zhen, Z.; Zhao, Y.H. Landsat-8 Remote Sensing Image Forest Vegetation Classification Based on Random
Forest Model. J. Northeast. For. Univ. 2016, 44, 53–57. [CrossRef]
44. Song, Y.; Si, W.; Dai, F.; Yang, G. Weighted ReliefF with threshold constraints of feature selection for imbalanced data classification.
Concurr. Comput. 2020, 32, 14. [CrossRef]
45. Long, Z.W.; Ji, Z.H.; Pei, Y.S. Characteristics and distribution of phosphorus in surface sediments of a shallow lake. J. Environ. Sci.
2022, 124, 50–60. [CrossRef]
46. Immitzer, M.; Vuolo, F.; Atzberger, C. First Experience with Sentinel-2 Data for Crop and Tree Species Classifications in Central
Europe. Remote Sens. 2016, 8, 166. [CrossRef]

Sustainability 14 03797

Uploaded by

Copyright:

Available Formats

You might also like

Sustainability 14 03797

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sustainability 14 03797

Uploaded by

Copyright:

Available Formats

sustainability

Received: 21 January 2022

Sustainability 2022, 14, 3797. https://doi.org/10.3390/su14073797 https://www.mdpi.com/journal/sustainability

for double-star; its spatial resolution of multispectral four bands is 10 m. In contrast,

2. Materials and Methods

and summer is hot.

2.2. Data Sources and Pre-Processing

Table 1. Sentinel-2 Data Band Parameters.

Center Wavelength Spectral Width Spatial Resolution

3. Study Methods and Steps

Gini = ∑ ∑ ( f (Ci , T )/|T |)( f (Cj , T )/|T |) (1)

In the Equation (2), V k j is the importance of j characteristic variables, N is the

3.1.2. Accuracy Verification Algorithm

N. ∑ri xij − ∑( xi+ · x+i )

3.2. The Extraction of Characteristic Variables

Table 2. Characterization of spectral indices.

Abbreviations Full Names Calculation Formulas Types Citations

Table 3. Number of samples for each class.

Training Total Number of

Sustainability 2022, 14, x FOR PEER REVIEW 10 of 20

Grassland 82.36 76.52 84.51 79.84 87.12 81.65 90.12 83.54

Table 4. Classification accuracy statistics of models A, B, C, and D.

Model A Model B Model C Model D

Sustainability 2022, 14, 3797 10 of 19

Table 4. Classification accuracy statistics of models A, B, C, and D.

Model A Model B Model C Model D

4.2. The Evaluation of Characteristic Importance

Figure 8. Characteristic variables importance value of model D.

The accuracy of the random forest classification results of 24 characteristic combina-

Figure 9. The relationship

(1) Choose based on variable importance estimates. According to the importance of

Table 5. Classification accuracy statistics of models RF_16, Relief-16, and model D.

RF_16 Relief-16 Model D

(a) Relief-16 (b) RF_16

4.3. Accuracy Evaluation

4.3. Accuracy Evaluation

Kappa Missed False Extraction The Overall

You might also like