Volume 3, No. 7 July
2024 (1499-1511)
p-ISSN
2980-4868 | e-ISSN 2980-4841
https://ajesh.ph/index.php/gp
Analysis of Prediction of Land Availability
in Residential Areas Using the Cellular Automata Method in Batangan District
Rizka Nurhaimi
Ayuningtias1*, Ayomi Dita
Rarasati2
1,2Universitas Indonesia, Depok, West Java,
Indonesia
Email: rizkaayunungtias@gmail.com1*, ayomi@eng.ui.ac.id2
ABSTRACT
The land cover change will be impacted by population
growth and its activities. This research aims to predict land cover change in
the Batangan sub-district from 2013 to 2033. The method used is quantitative,
based on satellite image data, followed by cellular automata modeling using the
MOLUSCE plugin on QGIS software. The land cover themes used are open areas,
built-up areas, forests, water bodies, agriculture and livestock, and
transportation. Spatial variables that are considered to influence the
prediction results are road networks, built-up areas, agricultural land, and
water bodies. The predicted land area that can be used for settlements in 2033
is 272.81 hectares, while the land required for settlements based on population
projections with additional labor due to the existence of industrial allotment
areas is 840.74 hectares. This results in a shortage of land for the
development of residential areas of 567.93 hectares, which should be a concern
for the local government in making policies for residential areas in the
Batangan sub-district. The implications of this land cover change also need to
be considered in the environmental and social contexts.
Keywords:
Land Cover, Prediction,
Settlements.
INTRODUCTION
Land functions to meet human needs and their activities. Human
activities are growing rapidly and diversely, which is not balanced by the
limited availability of land, resulting in the convention of undeveloped land becoming
built land Land use change refers to human activities carried out to meet
socio-economic needs such as settlements and infrastructure
The
increase in population encourages changes in land use and land cover leading to
the expansion of built-up areas and cultivation areas by reducing the area of
forest areas
Approaches to the method of modelling land
cover change include MCE (multi-criteria evaluation), MLP (multi-layer
perceptron), SLEUTH (slope, land use, exclusion, urban extent, transportation,
and hillshade), LR (logistic regression), ANN (artificial neural network), and
GWR (geographic weighted regression). The ANN method is one of the transition
potential models contained in MOLUSCE (Modules for
Land Use Change Simulations) developed for QGIS
Previous research was conducted by Hapsary; Subiyanto; and Firdaus
(2021) to
predict land change using the artificial
neural network (ANN) approach contained in the MOLUSCE plugins in QGIS
software. In this study, using the ANN method produced a kappa accuracy value
greater than the logistic regression method
Currently, the company PT Hwa Seung Indonesia has been built there
with a workforce need of 26,602 workers. The growth of industrial
companies and immigrant labour,
will affect the need for land for residential areas but is limited to other
land functions, especially agricultural and aquatic functions
RESEARCH METHODS
Study Location
Batangan
District is one of the sub-districts in Pati Regency located in the eastern
coastal area. The north is bordered by the Java Sea, the east is bordered by
Rembang Regency, the south is bordered by Jaken and Jakenan Districts, and the
west is bordered by Juwana District. This sub-district is one of the largest
salt production centres
in Indonesia spread across several villages such as Pecangaan Village,
Mangunlegi Village, Lengkong Village, Jembangan Village, Bumimulyo Village,
Ketitangwetan Village, and Raci Village
The
method in this study is quantitative with a spatial approach. This approach
uses MOLUSCE which is one of the plugins in QGIS software. The steps in land
cover prediction are (1) input of land cover data in 2013 and 2018 as well as
spatial variable data, (2) evaluating correlations using Pearson's correlation
method, (3) transition potential modelling, (4) cellular automata simulations,
and (5) validation
Pearson's
correlation can be calculated using Equation
1. Where rxy is Pearson's correlation coefficient, ∑xy is the sum of the
multiplication of the variables x and y, n is the number of samples, and S is
the standard deviation. The coefficient value has a range between -1 to 1. The
correlative value of -1 is negative between the two variables, the value of 0
has no correlation, and the value of 1 has a positive correlation
Kappa
accuracy test to measure the accuracy of prediction of land cover change
assisted by confusion matrix
Kappa
= ................................... 2
Kappa
Loc = ........................... 3
Kappa
Histo = ......................... 4
Table 1. Kappa Test
|
Coefficient value |
Interpretation |
|
< 20 |
Poor |
|
2,21 – 0,40 |
Fair |
|
0,41 – 0,60 |
Moderate |
|
0,61 – 0,80 |
Good |
|
> 0.81 |
Very Good |
Population
Projections
The analysis determines the population projection in the
next 10 years by using an exponential method that has a higher level of
accuracy for assuming continuous population growth. The following is the exponential
projection formula in equation 5:
Pn
= P0ern or Pt = P0ert .........................
5
Information:
Pn or Pt = total population in
year n or t
P0 = total
population in the first year
r = population growth rate
n or t = time period in years
e = 2.7182818 (the prime number of the natural
algorithm system)
Data
The
data used is a land cover map that has been processed from a SPOT6/7 image map
to a raster type map. There are 6 types of land cover at the study site, namely
open areas, fasum building areas (public facilities), forest areas, aquatic
areas, agricultural and livestock areas, and transportation areas. The
development of land cover area can be seen in table 1. from 2013, 208 and 2023.
Table 2. Land cover area of Batangan District
in 2013, 2018, and 2023
|
Land Cover |
2013 (ha) |
2018 (ha) |
2023 (ha) |
|
Open Area |
919,67 |
975,90 |
975,90 |
|
Fasum Building |
262,85 |
270,95 |
270,95 |
|
Forest |
31,71 |
31,72 |
31,72 |
|
Waters |
2067,94 |
2057,68 |
2057,68 |
|
Agriculture and Livestock |
2430,94 |
2376,10 |
2376,10 |
|
Transportation |
38,68 |
39,43 |
39,43 |
Source: SPOT image mosaic
data 6/7, 2024
The
land cover depicted from 2013, 2018, and 2023 can be seen in figure 2. As
illustrated that in the north of the study location is a water area that has an
area of almost 36%. Meanwhile, the area of agriculture and livestock is 41% in
the central to southern part. Land cover changes were not very visible
spatially in 2018 and 2023.
Figure 1. Land cover map of Batangan District
Source: SPOT image mosaic data 6/7, 2024
Discussion
The
spatial variables used need to be processed in ArcGIS software using the Euclidean distance method
with an output cell size of 5 because the map used is a 1:5,000 detail scale
map as shown in Figure 4. Euclidean distance is used to calculate the distance
from each cell at its nearest source. This tool is used on each spatial variable
in this study. The result is a raster map in Figure 5. As basic data used for
the land cover prediction process. Each of the spatial variable maps after
analysis produced 5 distance criteria. The results of the Euclidean distance analysis
can be seen in the chart Table
2.
Figure 2. Euclidean Distance in ArcGIS
Source: author, 2024
Table 3. Euclidean Distance Results
|
The results of the
building distance map show that in the northern left and south ends where
there are few buildings so there is a red color.
|
On the road distance map,
it can be seen that the northern part has a red color which means that there
are no roads compared to other areas at the study site.
|
|
The further north it goes,
the redder the color which indicates that the agricultural area only exists
from the central part to the south.
|
The salt area is
concentrated in the central to northern part so that the central part to the
south and the northern part on the left do not have these areas.
|
Source: author, 2024
The
results of Pearson's correlation (table 4.) in the modeling for the variables
of salt areas with roads and salt areas with agricultural areas have negative
values, which means that between the two variables there is a negative
correlation. The salt area and the road have a negative correlation because
there is no road network in the salt area so that the two variables do not
correlate with each other. Agricultural areas and salt areas also have a
negative correlation because they are located in two separate areas, the salt
area is in the north area while the agricultural area is in the southern area.
Table 4. Pearson Correlation Results
|
Spatial Variables |
Road |
Salt Areas |
Building |
Agricultural Area |
|
Road |
.. |
-0,35 |
0,29 |
0,75 |
|
Salt Areas |
|
.. |
0,22 |
-0,57 |
|
Building |
|
|
.. |
0,24 |
|
Agricultural Area |
|
|
|
.. |
Source: author, 2024
The
change in area (table 5.) that occurred in the land cover in 2013 and 2018 from
this modelling showed a change in the decrease in the land cover of the
built-up area of buildings and public facilities by -0.96%. The area of aquatic
land cover also decreased by -0.17%. The land cover that has increased in area
is the open area land cover of 0.97%.
Table 5. Area changes on the 2013 and 2018
maps
|
Land Cover Theme |
2013 (sq. meter) |
2018 (sq. meter) |
∆ (sq. meter) |
2013% |
2018% |
∆ % |
|
Open Area |
9159925,00 |
9716375,00 |
556450,00 |
15,93 |
16,89 |
0,97 |
|
Fasum Building |
24290300,00 |
23740350,00 |
-549950,00 |
42,23 |
41,27 |
-0,96 |
|
Forest |
317025,00 |
317025,00 |
0,00 |
0,55 |
0,55 |
0,00 |
|
Waters |
20637825,00 |
20538000,00 |
-99825,00 |
35,88 |
35,71 |
-0,17 |
|
Agriculture and Livestock |
425775,00 |
433925,00 |
8150,00 |
0,74 |
0,75 |
0,01 |
|
Transportation |
2688050,00 |
2773225,00 |
85175,00 |
4,67 |
4,82 |
0,15 |
Source: author, 2024
In
the transition potential modeling stage, the method used is an artificial
neural network (multi-layer perception). The neighbourhood used is 1 pixel,
learning rate 0.100, maximum iterations 1000, hidden layers 10, and momentum
0.050. The validation result of kappa is 0.91 so that this model can be used
for the automata cellular simulation model to predict land cover maps in 2033
Table 6. Transition Potential Modelling
|
Neighbourhood |
1 px |
|
Learning rate |
0,100 |
|
Maximum Iterations |
1000 |
|
Hidden layers |
10 |
|
Momentum |
0,050 |
|
∆ Overall accuracy |
-0,00693 |
|
Min. validation overall error |
0,01745 |
|
Current validation kappa |
0,91297 |
Source: author, 2024
Figure 3. Spatial Variables
Source: author, 2024
The
result of the Kappa correction test in prediction modeling is 1 with a percentage
of 100% (table 7.). The results interpret that the map modeling carried out has
a very good correction rate. So that it is well-validated and can be used
for modeling simulations.
Table 7. Area changes on the 2013 and 2018
maps
|
% of correctness |
100 |
|
Kappa (overal) |
1 |
|
Kappa (histo) |
1 |
|
Kappa (filter) |
1 |
Source: author, 2024
Changes
in the land cover area of the modeling process can be seen in the table above
in the land cover of buildings, public facilities, and aquatic land. The open
area has increased by 60.71 hectares. The types of open areas at the study site
are sports fields, cemeteries, grasslands, yards, hardened surfaces/fields,
shrubs, vacant land, and mixed plants. The land cover that experienced the
largest decrease in area was agriculture and livestock covering an area of
54.84 hectares as shown in Table
8. and shown in figure 6.
Table 8. Changes in land cover area on the
2013 and 2018 maps
|
Land Cover |
Area (hectares) |
Gap 2013-2023 |
|
|
2013 |
2033 |
||
|
Open Area |
919,67 |
980,38 |
60,71 |
|
Fasum Building |
262,85 |
272,81 |
9,96 |
|
Forest |
31,71 |
30,01 |
-1,71 |
|
Waters |
2067,94 |
2053,27 |
-14,67 |
|
Agriculture and Livestock |
2430,94 |
2376,10 |
-54,84 |
|
Transportation |
38,68 |
39,22 |
0,53 |
Source: author's analysis,
2024
Figure 4. Spatial Variables
Source: author, 2024
CONCLUSION
The
land area based on the prediction results of cellular automata that can be used
as a residential area is 272.81 hectares. Meanwhile,
the area of land needs in residential areas is according to the projected
population with an increase in labor of 840.74 hectares. The lack of land
availability for residential areas of 567.93 hectares needs attention from the
local government to provide a policy for residential areas in Batangan District
in the form of vertical residential areas by considering economic and social
conditions that need to be studied more deeply in the next study.
Atef, I., Ahmed, W., Abdel-Maguid,
R. H., Baraka, M., Darwish, W., & Senousi, A. M. (2023). Land Use and Land
Cover Simulation Based on Integration of Artificial Neural Networks with
Cellular Automata-Markov Chain Models Applied to El-Fayoum Governorate. ISPRS
Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences,
10, 771–777.
Blissag, B., Yebdri, D., &
Kessar, C. (2024). Spatiotemporal change analysis of LULC using remote sensing
and CA-ANN approach in the Hodna basin, NE of Algeria. Physics and
Chemistry of the Earth, 133.
https://doi.org/10.1016/j.pce.2023.103535
DPMPTSP Pati. (2023). Data
proyek 2021-2023.
Hapsary, M. S. A., Subiyanto, S.,
& Firdaus, H. S. (2021). Analisis Prediksi Perubahan Penggunaan Lahan
dengan Pendekatan Artificial Neural Network dan Regresi Logistik di Kota
Balikpapan. Jurnal Geodesi Undip.
Hossain, M. A., & Huggins, R.
(2021). The environmental and social impacts of unplanned and rapid
industrialization in suburban areas: the case of the Greater Dhaka Region,
Bangladesh. Environment and Urbanization ASIA, 12(1), 73–89.
Iskandar, B., Saidah, Kurnia, A.
A., Jauhari, A., & Zannah, F. (2024). Modeling Land Cover Change Using
MOLUSCE in Kahayan Tengah Forest Management Unit, Kalimantan Tengah. Jurnal
Sylva Lestari, 12(2), 242–257.
https://doi.org/10.23960/jsl.v12i2.865
Kumar, S., Singh, V., & Saroha,
J. (2023). Interpretation of land use/land cover dynamics with the application
of geospatial techniques in sarbari khad watershed of Himachal Pradesh, India.
GeoJournal, 88(3), 2623–2633.
https://doi.org/10.1007/s10708-022-10769-3
Leni, D., Dwiharzandis, A.,
Sumiati, R., & Afriyani, S. (2023). Feature Selection Based on Pearson
Correlation in Building Energy Efficiency Modeling. In Jurnal Teknika Sains
(Vol. 08).
Lukas, P., Melesse, A. M., &
Kenea, T. T. (2023). Prediction of Future Land Use/Land Cover Changes Using a
Coupled CA-ANN Model in the Upper Omo–Gibe River Basin, Ethiopia. Remote
Sensing, 15(4). https://doi.org/10.3390/rs15041148
Monsaputra, M. (2023). Analisis
perubahan penggunaan lahan pertanian menjadi perumahan di kota Padang Panjang.
Tunas Agraria, 6(1), 1–11. https://doi.org/10.31292/jta.v6i1.200
Muhammad, A. M., Rombang, J. A.,
Saroinsong, F. B., Studi, P., Kehutanan, I., Pertanian, J. B., Pertanian, F.,
& Ratulangi, S. (2016). Identifikasi Jenis Tutupan Lahan di Kawasan
KPHK Poigar dengen Metode Maximum Likelihood.
Muslim, S., Utomo, R. P., &
Permana, C. T. H. (2023). Perubahan penggunaan lahan dan dan pola spasial
tutupan lahan di sekitar Kawasan Industri Purwosuman, Sragen. Region :
Jurnal Pembangunan Wilayah Dan Perencanaan Partisipatif, 18(1), 38.
https://doi.org/10.20961/region.v18i1.53755
Naikoo, M. W., Rihan, M., &
Ishtiaque, M. (2020). Analyses of land use land cover (LULC) change and
built-up expansion in the suburb of a metropolitan city: Spatio-temporal
analysis of Delhi NCR using landsat datasets. Journal of Urban Management,
9(3), 347–359.
Nhu, V.-H., Shirzadi, A., Shahabi,
H., Singh, S. K., Al-Ansari, N., Clague, J. J., Jaafari, A., Chen, W., Miraki,
S., & Dou, J. (2020). Shallow landslide susceptibility mapping: A
comparison between logistic model tree, logistic regression, naïve bayes tree,
artificial neural network, and support vector machine algorithms. International
Journal of Environmental Research and Public Health, 17(8), 2749.
Obiahu, O. H., Yan, Z., &
Uchenna, U. B. (2021). Spatiotemporal analysis of land use land cover changes
and built-up expansion projection in predominantly dystric nitosol of Ebonyi
state, Southeastern, Nigeria. Environmental Challenges, 4,
100145.
Penfound, E., & Vaz, E. (2024).
Modelling future wetland loss with land use landcover change simulation in the
Greater Toronto and Hamilton Area: The importance of continued greenbelt
development restrictions. Habitat International, 143.
https://doi.org/10.1016/j.habitatint.2023.102974
Rahmi, L. (2017). Analisis Proyeksi
Pertumbuhan+Penduduk Terhadap Kondisi Ketenagakerjaan Di Kota Sawahlunto
Sumatera Barat. Georafflesia.
Sankarrao, L., Ghose, D. K., &
Rathinsamy, M. (2021). Predicting land-use change: Intercomparison of
different hybrid machine learning models. Environmental Modelling &
Software, 145, 105207.
Ulfa, M., Rochwulaningsih, Y.,
Soedarto, J., Semarang, S. H., & Tengah -Indonesia, J. (2021). Pola
Hubungan Produksi dalam Usaha Garam Rakyat di Kecamatan Batangan, Kabupaten
Pati (Vol. 2, Issue 2).
United Nations. (2022). World
Population Prospects 2022 World Population Prospects 2022 Summary of Results.
https://population.un.org/wpp/
Wang, S. W., Munkhnasan, L., &
Lee, W.-K. (2021). Land use and land cover change detection and prediction in
Bhutan’s high altitude city of Thimphu, using cellular automata and Markov
chain. Environmental Challenges, 2, 100017.
|
Copyright holder: Rizka Nurhaimi Ayuningtias, Ayomi Dita Rarasati (2024) |
|
First publication right: Asian Journal of Engineering, Social and Health
(AJESH) |
|
This article is licensed under:
|
