A GIS-Based Data Mining of Landuse Changes in the Yellow River

Basin, China

Kaidi Xiao, Fengxia Gu* and He Zhu

Department of Architecture and Urban Rural Planning, Shandong University of Science and Technology, Qingdao, China

Keywords: Yellow River Basin, Landuse Change, Landscape Pattern, Landscape Ecology, Spatiotemporal Evolution.

Abstract: This study provides an empirical example of data mining on landuse changes in the Yellow River Basin using

Geographic Information System (GIS). Time series of Landsat remote sensing images of 1980, 1990, 2000,

2010 and 2020 as well as ArcGIS software are used to calculate transfer matrixes of landuse changes and

estimate the degree of land development to mine information on landuse changes over time and across space.

Two points of key findings are outlined here. (1) The dominant class of landuse in the basin were grassland,

followed by cultivated land. The cultivated land area showed an overall decreasing trend with a wave of first

increasing and then decreasing over the study period. The areas of forest land, construction land, and

grassland had increased, while the areas of water and unused land had decreased. (2) The overall level of land

development in the basin was relatively high, but heterogeneity was also shown at different times and across

space.

1 INTRODUCTION

In recent years, Geographic Information Systems

(GIS) have been developed rapidly and its application

scope has been continuously expanding. For

example, it has been intensively applied to the studies

on landuse patterns in urban planning and land

resource management. Recently, many studies on the

landscape patterns of different watersheds have been

conducted, and the results show that the stability of

watershed landscapes is declining, which restricts the

sustainable development of regional economy and

society (Yang, etc,2020). However, previous

research focuses on local areas in space and a

relatively short period in time, which fails to reveal

the spatiotemporal changes in landscape patterns

under large-scale and long-term time series. There is

less research on the spatiotemporal dynamic

evolution at the entire watershed scale(Yao, 2013).

This paper takes the entire Yellow River Basin as

the study area and uses GIS technologies to analyse

five remote sensing images from 1980 to 2020 to

examine the spatiotemporal dynamic evolution

characteristics of the watershed landscape patterns.

The results provide insights on the relationship

between the landscape pattern and the environment,

human production and life. Knowledge advanced in

this study forms a scientific and reasonable basis for

the overall planning and management of the

watershed.

2 STUDY AREA

The spatial range of the Yellow River Basin involves

in 9 provinces, which set from 32°N-42°N and

96°E-119°E, and the area is about 7.95×105

km2，almost accounts for 8.3% of China(Chris,

2004).

The watershed area is a typical continental

climate, and the northwest is mainly the source water

area of the river. The glacier landform is developed

and the grassland area is wide. The central part is

mainly less landform, with broken terrain and fragile

ecological environment. Previous research shows that

the ecological function of the Yellow River Basin has

seriously degraded, and its ecological environment

protection and development are facing serious

challenges(Lu,etc. 2019).

Xiao, K., Gu, F. and Zhu, H.

A GIS-Based Data Mining of Landuse Changes in the Yellow River Basin, China.

DOI: 10.5220/0012877100004536

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Data Mining, E-Learning, and Information Systems (DMEIS 2024), pages 53-57

ISBN: 978-989-758-715-3

3 DATA SOURCES AND

RESEARCH METHODS

3.1 Data Sources

The basic data source for this study is Landsat TM

interpretation data, which are downloaded from the

Earth System Science Data Sharing Platform of the

National Earth System Science Data Center, with a

spatial resolution of 1km (www.geodata.cn/data).

Referring to the "Classification of Land Use Status"

standard, the 25 landuse classes in the Yellow River

Basin were reclassified into 6 types: cultivated land,

forest land, grassland, water area, construction land,

and unused land(Hu,2021). Other basic geographic

information element data were obtained from a

1:1000000 national basic geographic database.

3.2 Research Methods

3.2.1 The Landuse Transfer Matrix

The land use transfer matrix represents the transfer

structure, source and destination, and spatial

distribution characteristics of various landuse types in

the study area. It can quantitatively describe the

quantity, direction, and transfer rate of mutual

transformation between landuse types, reflect the

transformation process of the landscape in the region

from the beginning to the end of the study period, and

reveal the specific process of dynamic development

and change of the landscape pattern during over the

time(

Nassauer and Corry,2004).

The calculation formula is as follows (

Cheng, etc.,

2002):













…

nnn

111



(1)

Where, S

is the area (km

) converted from the

i-th type of landuse to the j-th type of landuse, n

represents the classification number, and vector S

...,

is the area of various land use types.

3.2.2 The Comprehensive Degree Index

The comprehensive degree index of landuse is an

evaluation and grading system that assigns a certain

grading index to reflect the depth of regional landuse

and the intensity of human activities on land. The

calculation formula is as follows(xia, etc., 2022):

)(100



××=

CAL

(2)

Where, 𝐴

represents the classification index of the

i-th landuse type in the study area (A

takes values of

1, 2, 3, and 4 if landuse types are unused land, forest

land or grass land or water, cultivated land, and

construction land, respectively), and 𝐶

represents

the percentage of the classified area of the i-th

landuse type to the total utilization area. The larger

the value of the comprehensive degree index of land

use, the higher the degree of the landuse

development(Chen,etc,2022 and Zuo,etc., 2022).

4 RESULTS AND ANALYSIS

4.1 Dynamic Change of Landuse

The results of landuse transfer matrix are shown in

Table 1. The change of landuse types between 1980

and 1990 is inconspicuous. The addition of

cultivation land is 877.43km

, transferred from grass

land and water, while the reduction of water area is

1253.69km

, which is either transferred to

cultivation land or grass land.

The change of landuse types between 1990 and

2000 is obvious. The addition of cultivation land is

2104.58km

, transferred from grass land. The

reduction of grass area is 1901.62km

, which is

transferred into cultivation land and others. The

addition of construction is 1164.70km

, transferred

from cultivation land.

From 2000 to 2010, however, cultivated land

decreased 2704.89km

, mainly transferring to

grassland. Grassland increased 4544.50km

, mainly

from cultivated land and unused land. Unused land

decreased by 1541.64km

, mainly transferring to

grassland. The change of other landuse types were

not significant. The water area decreased 642.37km2,

which was the remarkable value in the four periods.

The spatial change of landuse between 2010 and

2020 is significant. The cultivated land decreased by

9840.40km

, mainly transferring to grassland,

construction land and forest land, which attributed to

the implementation of ecological protection

measures and the rapid development of urbanization.

Forest land, water area and construction land

increased 2754.68km

, 1641.54km

and

12060.59km

, respectively. Among them, forest land

was mainly transferred from grassland and

cultivated land, which attributed to the

DMEIS 2024 - The International Conference on Data Mining, E-Learning, and Information Systems

Table 1: Land use type transfer matrix in four periods from 1980 to 2020 (Unit:km

Years

Landscape Type Cultivated Land Forest Land Grass Land Water Area

Construction

Lan

Unused

Lan

1980-

1990

Cultivated Land 209909.84 59.83 228.37 250.86 338.40 68.72

Forest Land 48.04 102680.22 73.56 7.16 4.00 2.09

Grass Land 683.47 104.04 386793.90 185.00 123.94 360.43

Water Area 800.32 40.18 709.27 12180.37 111.95 246.15

Construction Land 68.30 0.19 12.53 1.56 13019.95 0.42

Unused Land 317.83 30.83 261.28 226.31 20.56 66482.74

Area Increment 877.43 115.54 -48.40 -1253.69 526.14 -154.56

1990-

2000

Cultivated Land 207839.84 278.28 1915.78 279.21 1134.73 381.82

Forest Land 309.11 101449.28 1052.95 21.18 27.55 53.44

Grass Land 4198.90 821.25 380620.47 303.84 123.22 2059.32

Water Area 940.89 26.50 159.06 11588.34 14.37 122.46

Construction Land 186.06 0.45 26.30 0.48 13402.25 3.40

Unused Land 648.42 128.00 2430.17 86.75 19.62 63848.22

Area Increment 2104.58 -140.52 -1901.62 -544.88 1164.70 -683.84

2000-

2010

Cultivated Land 206365.65 1475.90 3774.96 739.07 1437.27 333.16

Forest Land 174.31 101818.91 526.35 39.49 97.40 51.22

Grass Land 2443.03 1683.34 378592.78 315.39 335.14 2829.54

Water Area 416.54 28.99 206.57 11322.65 34.73 271.13

Construction Land 192.18 7.92 68.08 21.85 14424.77 6.75

Unused Land 315.73 130.49 1566.11 186.61 66.38 64201.99

Area Increment -2704.89 953.87 4544.50 -642.37 -772.78 -1541.64

2010-

2020

Cultivated Land 126633.75 11326.23 54091.97 2767.64 12730.59 2005.07

Forest Land 10585.05 65313.81 26478.08 534.00 993.55 921.27

Grass Land 50588.63 27052.94 280664.98 3037.09 5019.51 17610.73

Water Area 2552.07 414.26 2832.68 5515.46 530.66 706.96

Construction Land 7595.85 498.08 2188.64 316.15 5514.08 239.43

Unused Land 2273.15 1313.94 24662.09 947.51 872.80 37360.99

Area Increment -9840.40 2754.68 387.91 1641.54 12060.59 -6764.02

implementation of the policy of returning farmland

to forest. Construction land was mainly transferred

from cultivated land. The area of unused land

decreased 6764.02km

, indicating the rapid

development of a large amount of unused land.

The most obvious changing landscape types over

the past 40 years in the study basin were firstly

cultivated land, then grass land, construction land,

and followed by unused land. The area of cultivated

land over the period showed a trend of firstly

increasing and then decreasing, mainly transferring

from and into grassland. The area of grassland

showed an increasing trend, mainly transferring from

cultivated land, then forest land, and followed by

unused land. The area of construction land also

showed an increasing trend, mainly transferred from

cultivated land and followed by forest land,

grassland and unused land. However, unused land

showed a decreasing trend, mainly transferring to

grassland.

A GIS-Based Data Mining of Landuse Changes in the Yellow River Basin, China

Figure 1: Evolution array of landscape types in the Yellow River Basin from 1980 to 2020.

Table 2: Comprehensive land use degree index of the Yellow River Basin from 1980 to 2020.

Years

Land Use Index

1980 1990 2000 2010 2020

8.80% 8.79% 8.79% 8.77% 8.70%

62.94% 62.78% 62.40% 62.90% 62.95%

26.32% 26.42% 26.66% 26.28% 24.83%

1.94% 2.01% 2.15% 2.53% 3.53%

221.41 221.64 222.18 223.51 223.19

4.2 The Evolution Process of

Landscape Types

The common evolution of the above six landscape

types resulted in the spatial changes of land use

pattern in the study area. To analyze the spatial

evolution process of the landscape patterns in the

Yellow River Basin from 1980 to 2020, four

transition maps of landuse types were generated

using ArcGIS software, as shown in Figure 1, where

the arrays in the map legends indicated the flow

directions of landuse type transition. The results

demonstrated that grassland was the most dynamic

landuse type in terms of transfer-in and transfer-out

types over the 40-year study period.

The evolution of landscape types can be divided

into two main periods: from 1980 to 2010 and from

2010 to 2020. Firstly, the main changes from 1980 to

2010 were grassland flowing into unused land,

grassland flowing into cultivated land and unused

land flowing into grassland. Secondly, the landscape

pattern changed dramatically, and the degree of

landscape transfer increased. Within the ecological

control line, there were mutual transfers among

cultivated land, forest land and grassland, and

mutual transfers between grassland and unused land.

The expansion of construction land mainly came

from cultivated land.

4.3 Analysis of Landuse Degree

Using the formula of landuse comprehensive degree

index and the land use classification standard of the

Yellow River Basin mentioned above, the landuse

comprehensive degree indexes of five years from

1980 to 2020 were calculated, which were presented

in Table 2.

DMEIS 2024 - The International Conference on Data Mining, E-Learning, and Information Systems

In the past 40 years, the landuse degree in the

study area of the Yellow River Basin has been

relatively high, with the index values above 220, the

minimum value of 221.41 in 1980, and the

maximum value of 223.51 in 2010.

From different time points, the degree of land use

in the Yellow River Basin was first in a period of

rapid development, and then slightly attenuated.

From 1980 to 2010, the land use index of the Yellow

River Basin kept rising steadily. The reason is that

the ecological protection measures mentioned above

were effective, and the natural ecological landscape

had been restored to a certain extent. From 2010 to

2020, the land use in the Yellow River Basin

declined, and the land use index decreased by 0.32,

indicating that there were unreasonable human

activities, which destroyed the original high

utilization degree.

It is suggested that the environmental

departments strengthen the land management policy

in the Yellow River Basin, especially for the

improvement of unused land such as sandy land, and

improve its classification level to improve the

overall land use degree index of the Yellow River

Basin. Hopefully, a sustainable land development

direction is forward on the way.

5 CONCLUSION

The landscape types that are dramatically changed in

the spatial evolution of the Yellow River Basin over

the past 40 years are farmland, grassland,

construction land, and unused land. The loss of

cultivated land area is mainly due to transferring into

grasslands. The gain of grassland area mainly

coming from cultivated land, followed by forest land

and unused land. The gain of unused land area

mainly converting from grassland, and the overall

unused land area has slightly decreased. The

construction land area continues to increase, mainly

from cultivated land.

The main types of transfer in and out were

grasslands, and the main transitional landscape type

was unused land. The two main lines of landscape

type evolution are: (1) from 1980 to 2010, grassland

flowed into unused land and cultivated land, and

unused land flowed into grassland; (2) the landscape

pattern has undergone significant changes from 2010

to 2020, with the transfer of cultivated land, forest

land, and grassland, as well as the transfer of

grassland and unused land. The overall level of land

development during the research period was

relatively high. The land use index continued to

steadily increase from 1980 to 2010, indicating that

ecological protection measures have been effective

and natural ecological landscapes have been restored

to a certain extent.

ACKNOWLEDGEMENT

Fund Project: Provincial College Student Innovation

and Entrepreneurship Training Program Project

(S202210424025).

REFERENCES

Yang, M., Gong, J. G., Zhao,Y., 2020. Analysis of

dynamic changes and trends in the landscape pattern of

the Baiyangdian Region, Acta Ecologica Sinica. vol.40,

no.20, pp.7165-7174.

Lu, D. D., Sun, D. Q., 2019. Development and

management tasks of the Yellow River Basin: a

preliminary understanding and suggestion, Acta

Geographica Sinica. vol.74, no.12, pp.2431-2436.

Hu, Y. X., Chen, D. C., etc., 2021. Landscape pattern

evolution and its impact on ecosystem services in the

Taihu Lake Region, Urban Problems.no.4, pp.95-103.

Nassauer, J. I., Corry, R. C., 2004. Using normative

scenarios in landscape ecology, Landscape Ecology.

vol. 19, no. 4, pp.343-356.

Cheng, Y., Li, Y. L., Z. B. Chang, etc., 2022. Landscape

ecological risk assessment based on land use change- a

case study of coastal zone of Guangdong Province,

Environmental Ecology. vol.4, no.11, pp.23-33.

Xia, Z. L. ,Lan, A. J., Fan, Z. M. , etc., 2022. Spatial and

temporal distribution characteristics of landscape

ecological risks in Karst Mountainous Watersheds

from perspective of production-living-ecological,

Bulletin of Soil and Water Conservation. pp.1-12.

Chen, Y., Zou, W., Karimian, H., etc., 2022. Landscape

ecological risk assessment and driving factor analysis in

Dongjiang River watershed, Chemosphere. vol.

307(P3).

Zuo, Q. T.,Wu, Q. S.,Jiang, L., etc., 2022. Multi-scale

regional definition and its application selection in the

Yellow River Basin, Hydro-Science and

Engineering.no. 5, pp.12-20.

Yao, W. Y., Ran D C, Chen, J. N., 2013. Recent changes

in runoff and sediment regimes and future projections

in the Yellow River basin. Advances in Water Science.

vol.24, no.5, pp. 607-616.

Chris, J. J., Mark, S. B., Robert, M., etc., 2004.

Quantifying patch distribution at multiple spatial

scales: applications to wildlife-habitat models,

Landscape Ecology, vol.19, no.8, pp.869-882.

A GIS-Based Data Mining of Landuse Changes in the Yellow River Basin, China