Spatiotemporal Dynamic Characteristics of Land Use in the Typical Watershed of Wenchuan Earthquake-Affected Areas—A Case Study in the Longxi River Basin

Abstract

Major earthquakes cause serious damage to ecosystem structure and function through their huge destructive force and subsequent geohazards. Understanding the characteristics of post-earthquake land use change is of great interest to assess the effects of ecological restoration in earthquake-affected areas. However, little is known about the consequences of land use change on a small scale due to incomplete, non-comprehensive, and sparse research data. Here, we used remote-sensing images to study the land use change characteristics of the Longxi River before and after the Wenchuan earthquake by calculating the land use dynamics degree, transition matrix, and gravity center of the Longxi River Basin from 2005 to 2015. The Wenchuan earthquake disaster did not affect the main dominance of forests. Grassland, the second dominant land type, was replaced by geohazard-induced unutilized land after the earthquake. Compared with pre-earthquake in 2005, the areas of cultivated land, forest, grassland, and water area decreased, while those of construction land and unutilized land increased in 2015. The single land use dynamic degree and spatial dynamic change degree show the highest transition intensity of other land use types to unutilized land between 2005 and 2009. However, the degrees of all land use types decreased between 2009 and 2015. Both banks of the lower reach of the Longxi River had high and developing comprehensive indexes of land use degrees during the study period. The gravity centers of grassland, construction land, water, and cultivated land changed from north to south, while forest and unutilized land had the opposite pattern. Our results provide useful information for ecological restoration, ecological security, and soil-erosion control in earthquake-affected areas.

1. Introduction

The land serving as an integral part of the ecosystem is the basis of human survival and is the material source for sustainable social development (Liu et al. [1]). Land use and land cover (LULC) change is one of the most important factors causing changes in ecological environment quality (Foley [2]), reflecting the interactions between human activities and natural environment (Li [3]; Peng et al. [4]; Sathiya et al. [5]). It also imperceptibly influences the change in geographical environment and ecosystem as the result of natural disasters, climate change, and human activity disturbances (Botezan et al. [6]; Rao et al. [7]; Piquer-Rodríguez et al. [8]; Liu et al. [9]). The effects of spatiotemporal LULC are important for understanding the dynamics and driving forces of landscape transitions in the environment (Nath et al. [10]; Pan et al. [11]; Serra et al. [12]). Undoubtedly, there are opposing results in the effects of LULC change. For example, some researchers found that LULC change caused by climate change and human activities can improve the ecological environment and accelerate the restoration of ecosystems (Ma et al. [13]). However, Faruque et al. [14] pointed out that LULC change in human-induced mangrove forests areas destroyed the ecosystem integrity and caused the reduction in the biodiversity in Bangladesh. Hence, it is necessary to critically evaluate the effects and control the operation of land use systems with regard to global environmental change and sustainable development (Lambin et al. [15]; Mooney et al. [16]).

In recent years, the integrated techniques of 3S technology can provide accurate and long-term information to detect the change characteristics of LULC change (Nath et al. [10]). Therefore, these techniques, combined with geography theory and landscape ecology theory, are widely used to study the spatiotemporal dynamic characteristics of LULC affected by geological disasters such as earthquakes, landslides, debris flows, and floods (Zhang et al. [17]; He et al. [18]; Aydöner et al. [19]; Sheykhmousa et al. [20]; Khan et al. [21]; Qiu et al. [22]), especially in mountainous regions where accessibility is difficult. LULC can change in response to earthquakes and help us predict geomorphological evolution, landscape patterns, and ecosystem functioning and processes (Singh et al. [23]; Dey et al. [24]; Singh et al. [25]; Nath et al. [10]). For example, Singh et al. [23] found that an east–northwesterly elongated zone of approximately 2100 km² suffered total destruction after the Gujarat earthquake of 26 January 2001. In addition, major earthquakes such as the 2008 Wenchuan earthquake can induce fault-existing areas to be seismically vulnerable for future landscape and land use planning (Zhuang et al. [26]; Sowter [27]; Nath et al. [10]). However, most previous studies mainly focused on large-scale LULC in typical earthquake-affected areas but did not consider how LULC changes respond to earthquakes on a small scale.

The Longxi River basin in Dujiangyan city of Sichuan province is disaster-prone because the rock strata were heavily fragmented by the induced long-time and large-scale geological movement of the Hongkou-Yingxiu Fault (Zou et al. [28]). It was also seriously affected by the Wenchuan earthquake with the highest intensity (XI degrees), resulting in frequent secondary disasters and subsequent deterioration in the ecological environment (Kang et al. [29]). Secondary disasters such as debris flows, landslides, collapses, and floods led to changes in geomorphology and river systems and serious soil erosion, which reduced the vegetation cover rate and reshaped the land use pattern (Liu et al. [30]). Previous studies showed that over 70% of damaged vegetation in typical areas affected by the sudden dislocation of the Wenchuan earthquake was distributed along the rivers within 5 km (Cui et al. [31]), indicating that great LULC changes occurred in earthquake-affected areas. However, to our knowledge, LULC changes in a typical basin after the earthquake remain poorly addressed. In addition, the Longxi River basin belongs to one of 35 giant panda reserves in China and plays an important role in the ecological barrier of the upper Yangtze River and water conservation for the Chengdu plain (Kang et al. [29]). Therefore, it is necessary to study the relationship between land use and cover change and natural recovery after the earthquake.

In this study, we sought to determine how changes in LULC were affected by the Wenchuan earthquake and subsequent natural recovery. We collected four different remote-sensing satellite images with high resolution as data resources and interpreted land use types in the Longxi River basin. Our main objectives were to (1) analyze the influencing extent of the Wenchuan earthquake on the land use pattern of the Longxi River basin, (2) determine the direction of land use and cover change under ecological recovery after the earthquake, and (3) demonstrate which factors combining with secondary disasters might control the process of land use and cover change. This study could improve our understanding of land use efficiency, ecological restoration processes, and land use optimization in earthquake-affected areas and help land managers choose better ways to carry out scientific land use planning in the future.

2. Materials and Methods

2.1. Study Area

Our study area is the Longxi River basin located in Longchi town, northwest of Dujiangyan City, Sichuan Province, China (31°1′28.06″–31°10′49.53″ N and 103°30′4.98″–103°36′2.08″ E) (Figure 1), at an altitude between 822 m to the south and 3291 m to the north above sea level, with a maximum horizontal distance of 11.1 km and an area of about 79.24 km². It is the first tributary of the upper reaches of the Minjiang River and is one of the three major debris-flow-monitoring demonstration areas in Sichuan Province due to frequent secondary disasters after the Wenchuan earthquake (Chen et al. [32]). Moderate and high mountains dominate its relief with an intensive surface incision from north to south, resulting in the “V”-shape canyon (Zou et al. [28]). The average slope gradient of a hillside is >20° and the gully gradient is between 28.7% and 62.6%, which provide great potential energy for the initiation of secondary disasters. The climate is subtropical humid monsoon. The Longxi River has a mean yearly rainfall of approximately 1100 mm, 80% of which occurs between May and October (Ding et al. [33]). According to the classifications of the International Union of Soil Sciences Working Group WRB (2015) [34], it has three soil types, including ali-udic argosols with the elevation of 822 to 1600 m, haplic acrisols with the elevation of 1600 to 2200 m, and haplic luvisols with the elevation of 2200 to 3291 m.

2.2. Data

In this study, to investigate the spatiotemporal variation of LULC in the Longxi River basin, information on LULC was interpreted from remote-sensing images, including SPOT-5 in 2009, Quick Bird in 2005 and 2011, and Worldview-2 in 2015 (

Table 1. Satellite imagery data information in the study area.

Satellite Imagery	Date	Intervals	Resolution/m	Coverage	Data Quality
Quick Bird	26 June 2005	Before the earthquake	0.61	Entire study area	Few clouds
SPOT-5	10 February 2009	1 year after the earthquake	2.5	Entire study area	Cloudless
Quick Bird	26 April 2011	3 years after the earthquake	0.61	Entire study area	Few clouds
Worldview-2	15 April 2015	7 years after the earthquake	0.46	Entire study area	Cloudless

), supplied by the Institute of Mountain Hazards and Environment, Chinese Academy of Sciences. We used band combination in Erdas 9.2 to receive the natural color image of SPOT-5 with the simulation of a blue band in a trained fitting model (Yang et al. [35]). Due to spatial incompatibility across 4 images, we used ENVI 5.3 version software to carry out image registration and fusion and projection coordinate transformation on the basis of world geodetic system (WGS) 1984 coordinate system. Then, based on primary-scale Chinese national topographic maps (1:50,000), the second-order polynomial model was used to ensure the accuracy of all image geometric corrections with 24 ground control points. After the geometric correction, we bounded the spatial inconsistency errors of all images within only 1 pixel. Combined with field survey data and image data, the method of human–computer interactive interpretation was used to interpret land use types in the Longxi River basin. In addition, we used shuttle radar topographic mission (SRTM) with 30 m × 30 m resolution for the digital elevation model (DEM) and extracted topographic features, including river systems, elevation, slope, and aspect.

2.3. Methodology

2.3.1. Establishment of Land Use Classification System

The land use classification system is a prerequisite for land use research (Jia et al. [36]). The Longxi River basin LULC consists of six categories, including cultivated land, forest, grassland, water, construction land, and unutilized land (mountains damaged by natural disasters, bare rock mass, and bare land) according to the FAO LULC categories (Faruque et al. [14]). We used the spectral library toolbox of ENVI 5.3 version software executed in Fujian Gewuzhitu Information Technology Co., Ltd. (Xiamen, China) to verify the spectral profiles of six categories (Nath et al. [10]). Over images for each category identification, 18 signature points were drawn, so a total of 108 signature points per each image were collected to present 6 categories used for image classification (Figure 2). We used a Garmin GPS machine to carry out a cross-check by field observation for 4 accessible signature points in each category in 2016. Then, we used maximum-likelihood classification (MLC) algorithm (Anderson et al. [37]) to carry out pixel-based supervised image classification to create 4 temporal LULC maps and perform polygon-to-point conversion using ArcGIS 10.9 software. For 4 remote-sensing satellite images, a total of 108 random points per image and corresponding 2.5 × 2.5 m polygons centered by them were selected and overlayed in Google Earth for visualization and validation. All selected points were identified as the exact land use type based on LULC categories. A similar approach was applied in the whole images for classification accuracy assessment, which was calculated as a contingency-errors matrix table.

2.3.2. Analysis Methods of Land Use Characteristics

The single land use dynamic degree (R₁) can reflect the change rate of different land use types in a certain period (Dong et al. [38]). We calculated it using the following equation (Equation (1)):

$$R_1=\frac{U_b-U_a}{U_a}\times \frac{1}{T}\times 100\%=\frac{\Delta U_{in}-\Delta U_{out}}{U_a}\times \frac{1}{T}\times 100\%$$

(1)

where U_a and U_b represent the area of certain land use types in the periods of a and b, respectively. ΔU_in is the sum of areas that other land use types change to a certain type; ΔU_out is the sum of areas that the certain type transits to other types; and T is the research period.

However, R₁ only reflects the annual area changes in LULC types rather than the spatial changes, so we proposed the spatial dynamic change degree (R₂) of a certain land use type using the following equation (Equation (2)):

$$R_2=\frac{\Delta U_{in}+\Delta U_{out}}{U_a}\times \frac{1}{T}\times 100\%$$

(2)

where ΔU_in, ΔU_out, and U_a are consistent with Equation (1).

In addition, to determine the change rate of the total area that was converted into other land use types, we calculated the comprehensive land use dynamic degree (S) using the following equation (Equation (3)) according to the methods described in Huang et al. [39]:

$$S=\left[{\displaystyle \sum _{i=1}^n\left(\Delta S_{i-j}/S_i\right)}\right]\times \frac{1}{T}\times 100\%$$

(3)

where i represents land use types in this study (n = 6); ΔS_i-j is the total area of the ith land use type converted to other land use types; and S_i is the total area of the ith land use type in the beginning of this study.

All of the above three indexes show land use change in total but not in detail. Previous research found that the degree of land use can be characterized by a comprehensive index reflecting the main land types affected by human and natural disturbances (Gong et al. [40]). To characterize the comprehensive level and change in land use in the Longxi River basin, we classified 6 land use types into 4 levels, including unutilized land level (scored as 1); forest, grass, and water land level (scored as 2); agricultural land level (scored as 3); and urban settlement land level (scored as 4), according to the method described in Gong et al. [41]. Higher scores indicate the greater intensity of human activities. Here, we calculated the comprehensive index of land use degree and the comprehensive change index of land use degree that then was visualized with a single grid (100 × 100 m) as a unit based on the grid analysis method in 2005, 2009, 2011, and 2015. The land use comprehensive index was calculated as the following equation (Equation (4)):

$$I=100\times {\displaystyle \sum _{j=1}^k{A}_j\times C_j}$$

(4)

The comprehensive change index of land use degree was calculated as the following equation (Equation (5)):

$$\Delta I_{b-a}=I_b-I_a=\left\{\left({\displaystyle \sum _{j=1}^k{A}_j\times C_{jb}}\right)-\left({\displaystyle \sum _{j=1}^k{A}_j\times C_{ja}}\right)\right\}\times 100$$

(5)

Among Equations (4) and (5), j is the land use level; k represents 4 land use levels in this study; I is the comprehensive index of land use; A_j is the j_th land use level; C_j is the area percentage of the jth land use level; ΔI_b-a is the comprehensive change index of land use; I_a and I_b are the indexes of land use degree at times a and b; C_ja and C_jb are the area percentages of jth land use level at times a and time b; and 100 indicate that the indexes are increased by 100 times to show the differences among them.

To reveal the direction and area of land use change under the intervention of natural factors and human activities, we used dynamic mapping and corresponding transition matrix tables to analyze the structural characteristics of land use change (Nath et al. [10]; Yang et al. [35]) in the Longxi River basin. Firstly, we transformed LULC themes interpreted from four different remote-sensing satellite images into raster data. Subsequently, we used the thematic change workflow tool in ENVI 5.3 to receive the dynamics of LULC change based on the transformed raster data during the period from 2007 to 2015. Finally, we set up transition matrix tables containing the transition dynamics of the LULC type to another type at a certain extent (Nath et al. [10]). The transition matrix is the following equation (Equation (6)):

$$S_{ij}=\left[\begin{array}{cccc}{S}_{11}& S_{12}& \cdots & S_{1n}\\ S_{21}& S_{22}& \cdots & S_{2n}\\ \cdots & \cdots & \cdots & \cdots \\ S_{n1}& S_{n2}& \cdots & S_{nn}\end{array}\right]$$

(6)

where S_ij is the land use status at the beginning and end of the study period and n is the number of land use types.

The spatial variation of land use types can be obtained by comparing the gravity center of various land use types in the early and late study periods (Huang et al. [42]). The calculation formula is listed as the following equation (Equation (7)):

$$X_t=\frac{{\displaystyle \sum _{i=1}^n}\left(A_{ti}\times X_i\right)}{{\displaystyle \sum _{i=1}^n}{A}_{ti}}{Y}_t=\frac{{\displaystyle \sum _{i=1}^n}\left(A_{ti}\times Y_i\right)}{{\displaystyle \sum _{i=1}^n}{A}_{ti}}$$

(7)

where X_t and Y_t represent the longitude and latitude of the gravity center of a certain land type in year t, respectively; A_ti represents the area of a certain land type in the ith (100 × 100 m) grid cell of the Longxi River basin; X_i and Y_i represent the mean longitude and latitude of geometric centers in the ith (100 × 100 m) grid cell of the Longxi River basin, respectively; and n is the sum of the grid cell numbers in the Longxi River basin.

3. Results

3.1. Land Use Change after the Wenchuan Earthquake

The land use types in the Longxi River basin consist of six categories, including cultivated land, forest, grassland, water, construction land, and unutilized land. The change in land type can be intuitively seen from Figure 3. As seen in

Table 2. Land use composition and change in the Longxi River basin from 2005 to 2015.

Year	Unit	Cultivated Land	Forest	Grassland	Water Area	Construction Land	Unutilized Land
2005	area/hm2	136.43	7253.47	336.98	14.31	63.63	119.65
	proportion/%	1.72	91.53	4.25	0.18	0.80	1.51
2009	area/hm2	85.10	6214.12	269.76	14.09	74.05	1267.49
	proportion/%	1.07	78.42	3.40	0.18	0.93	15.99
2011	area/hm2	66.15	6864.59	320.36	5.39	71.01	596.94
	proportion/%	0.83	86.63	4.04	0.07	0.90	7.53
2015	are/hm2	49.67	7186.50	220.57	6.01	79.21	382.49
	proportion/%	0.63	90.69	2.78	0.08	1.00	4.83
2005–2009	variation/hm2	−51.33	−1039.34	−67.22	−0.22	10.42	1147.85
	rate/%	−0.65	−13.12	−0.85	−0.003	0.13	14.48
2009–2011	variation/hm2	−18.95	650.47	50.60	−8.70	−3.04	−670.55
	rate/%	−0.24	8.21	0.64	−0.11	−0.04	−8.46
2011–2015	variation/hm2	−16.48	321.91	−99.79	0.62	8.20	−214.45
	rate/%	−0.21	4.06	−1.26	0.01	0.10	−2.71
2005–2015	variation/hm2	−86.76	−66.96	−116.40	−8.30	15.58	262.85
	rate/%	−1.09	−0.84	−1.47	−0.10	0.20	3.32

, forest dominated all land use types, accounting for 91.53% of the total area of the Longxi River basin before the earthquake in 2005. Though forest still dominated after the earthquake, unutilized land induced by secondary disasters replaced grassland as the second dominant land type, accounting for 15.99% in 2009, 7.53% in 2011, and 4.83% in 2015, respectively. Compared to 2005 before earthquake, the areas of all land types, except unutilized land and construction land, decreased in 2009; however, those of forest and grassland increased from 2009 to 2011, and those of forest, water area, and construction land increased from 2011 to 2015.

3.2. The Change Characteristics of Land Use Structure in the Longxi River Basin

3.2.1. Single Land Use Dynamic Degree

As seen in Figure 4, the change rates of land use types in the Longxi River basin showed differences during the period from 2005 to 2015. In detail, the single land use dynamic degrees of cultivated land were all negative during the periods from 2005 to 2009, from 2009 to 2011, from 2011 to 2015, and from 2005 to 2015. Grassland had a positive dynamic degree during the period from 2009 to 2011, but negative dynamic degrees during the periods from 2005 to 2009, from 2011 to 2015, and from 2005 to 2015, while construction land had the opposite pattern. Forest had positive dynamic degrees during the period from 2009 to 2011 and from 2011 to 2015, but negative dynamic degrees during the periods from 2005 to 2009 and from 2005 to 2015, while unutilized land had the opposite pattern. Water area only had a positive dynamic degree during the period from 2011 to 2015, but negative dynamic degrees during the periods from 2005 to 2009, from 2009 to 2011, and from 2005 to 2015.

3.2.2. The Spatial Dynamic Change Degrees of Land Use Types

As seen in

Table 3. Spatial dynamic change degrees of land use types in the Longxi River basin (%).

Land Type	Grassland	Cultivated Land	Construction Land	Forest	Water Area	Unutilized Land
2005–2009	40.22	32.12	44.96	5.85	2.10	263.08
2009–2011	87.21	50.53	52.32	8.80	32.51	37.84
2011–2015	26.66	19.51	27.28	2.62	16.42	20.92
2005–2015	14.85	11.74	18.93	1.23	6.33	34.34

, the spatial dynamic change degrees showed differences of unutilized land (263.08%) > construction land (44.96%) > grassland (40.22%) > cultivated land (32.12%) > forest (5.85%) > water area (2.10%) during the period from 2005 to 2009, indicating huge damage induced by secondary disasters to the original land use structure. However, the spatial dynamic change degrees were high in grassland, cultivated land, construction land, forest, and water area during the period from 2009 to 2011, showing that there were high transitions in such land types, while all land use types significantly decreased during the period from 2011 to 2015. Overall, the land use change in the Longxi River basin was intense and dominated by the transition intensity of other land use types to unutilized land during the period from 2005 to 2015.

3.2.3. The Comprehensive Land Use Dynamic Degree

The change rate of land use types increased and then decreased during the period from 2005 to 2015 in the Longxi River basin (

Table 4. The comprehensive land use dynamic degree in Longxi River basin (%).

Study Period	2005–2009	2009–2011	2011–2015	2005–2015
Comprehensive land use dynamic degree	81.37	162.55	64.73	39.35

). In detail, the comprehensive land use dynamic degrees were 81.37%, 162.55%, 64.73%, and 39.35% during the period from 2005 to 2009, from 2009 to 2011, from 2011 to 2015, and from 2005 to 2015, respectively.

3.2.4. The Comprehensive Change Index of Land Use Degree

The average comprehensive change index of land use degree in the basin was negative during the period from 2005 to 2009 but positive during the periods from 2009 to 2011 and from 2011 to 2015 (

Table 5. The comprehensive change index of land use degree in Longxi River basin.

Study Period	2005–2009	2009–2011	2011–2015	2005–2015
Comprehensive change index	−14.87	8.15	2.71	−4.02

). During the period from 2005 to 2015, the average comprehensive change index of land use degree decreased by 4.02. After the Wenchuan earthquake, the comprehensive index of land use degree in the Longxi River basin tended to be stable during the period from 2009 to 2015. The regions with a high comprehensive change index of land use degree were distributed in the northeast and south of the Longxi River, especially along both sides of the lower reaches of the Longxi River, showing a trend of concentration in the future (Figure 5).

3.3. The Gravity Center Changes in 6 Land Use Types

The gravity center change in each land use type in the basin is shown in Figure 6. The gravity center of grassland shifted from north to south during the period from 2005 to 2009 but shifted from south to north during the period from 2009 to 2011 and then from north to south again during the period from 2011 to 2015. Cultivated land, construction land, and water area shifted their gravity centers gradually from north to south during the study period from 2005 to 2015. However, unlike the other two types, the gravity center of cultivated land shifted from west to east during the period from 2005 to 2009. Compared to the other land use types, the gravity centers of forest and unutilized land had the opposite trend in that both of them shifted from south to north during the study period—though forest and unutilized land fluctuated during the periods from 2009 to 2011 and from 2011 to 2015, respectively.

3.4. The Characteristics of Land Use Transition Matrices

During the period from 2005 to 2009, the highest transition was from forest to unutilized land, accounting for 1087.88 hm² (

Table 6. Land use transition matrix from 2005 to 2015 in the Longxi River basin (hm²).

2005–2009	Grassland	Cultivated Land	Construction Land	Forest	Water Area	Unutilized Land	Total
Grassland	32.31	3.31	14.02	209.82	-	77.51	336.98
Cultivated land	17.85	23.12	9.06	64.61	-	21.79	136.43
Construction land	11.67	2.50	11.63	21.77	<0.01	16.05	63.63
Forest	192.92	54.78	31.89	5885.70	0.46	1087.88	7253.63
Water area	0.18	-	0.08	0.19	13.60	0.25	14.31
Unutilized land	14.84	1.39	7.36	32.03	0.03	64.01	119.65
Total	269.76	85.10	74.05	6214.13	14.09	1267.49	7924.62
2009–2011	Grassland	Cultivated land	Construction land	Forest	Water area	Unutilized land	Total
Grassland	59.83	11.08	8.92	173.26	0.15	16.53	269.76
Cultivated land	13.68	32.62	3.21	33.84	0.00	1.75	85.10
Construction land	12.58	1.63	33.79	20.10	0.02	5.93	74.05
Forest	77.68	18.65	13.77	5992.34	0.06	111.62	6214.12
Water area	0.15	-	0.01	0.26	5.16	8.51	14.09
Unutilized land	156.50	2.18	11.30	644.91	-	452.61	1267.49
Total	320.41	66.15	71.01	6864.71	5.39	596.94	7924.62
2011–2015	Grassland	Cultivated land	Construction land	Forest	Water area	Unutilized land	Total
Grassland	99.68	6.25	11.38	175.39	0.02	27.66	320.36
Cultivated land	10.10	32.10	1.34	22.13	0.13	0.35	66.15
Construction land	9.44	1.26	36.36	19.55	0.03	4.36	71.01
Forest	56.24	9.80	23.25	6666.11	0.07	109.12	6864.59
Water area	0.03	-	0.31	0.29	3.93	1.00	5.56
Unutilized land	45.09	0.26	6.74	303.03	1.83	240.00	596.94
Total	220.57	49.67	79.38	7186.50	6.01	382.49	7924.62
2005–2015	Grassland	Cultivated land	Construction land	Forest	Water area	Unutilized land	Total
Grassland	28.49	1.36	14.05	250.58	-	42.49	336.98
Cultivated land	17.56	12.97	9.61	81.40	0.04	14.85	136.43
Construction land	11.06	1.41	11.20	32.44	0.01	7.51	63.63
Forest	147.50	32.30	36.48	6772.54	0.32	264.33	7253.47
Water area	0.29	-	0.31	0.60	5.63	7.65	14.47
Unutilized land	15.68	1.63	7.73	48.95	-	45.66	119.65
Total	220.58	49.67	79.37	7186.50	6.01	382.49	7924.62

). The transition from grassland to forest (209.82 hm²) and the reverse one (192.92 hm²) were also high. As seen in Table 6, forest, cultivated land, and grassland, accounting for 7253.63 hm², 136.43 hm², and 336.98 hm² in 2005, decreased by 14.33% to 6214.13 hm², 37.62% to 85.10 hm², and 19.95% to 269.76 hm² in 2009, respectively. Unutilized land and construction land, accounting for 119.65 hm² and 63.63 hm² in 2005, increased by 959.33% to 1267.49 hm² and 16.38% to 74.05 hm² in 2009, respectively.

During the period from 2009 to 2011, the highest transition was from unutilized land to forest, accounting for 644.91 hm² (Table 6). The transitions from grassland to forest (173.26 hm²), from grassland to unutilized land (156.50 hm²), and from forest to unutilized land (111.62 hm²) were also high. As seen in Table 6, forest and grassland increased by 10.47% to 6864.71 hm² and 18.78% to 320.41 hm² in 2011, respectively. Unutilized land, cultivated land, water area, and construction land decreased by 52.90% to 596.94 hm², 22.27% to 66.15 hm², 61.76% to 5.39 hm², and 4.11% to 71.01 hm² in 2011, respectively.

During the period from 2011 to 2015, the highest transition was from unutilized land to forest, accounting for 303.03 hm² (Table 6). The transitions from grassland to forest (175.39 hm²) and from forest to unutilized land (109.12 hm²) were also high. As seen in Table 6, forest, construction land, and water area increased by 4.69% to 7186.50 hm², 11.78% to 79.38 hm², and 8.09% to 6.01 hm², respectively. Unutilized land, grassland, and cultivated land decreased by 35.92% to 382.49 hm², 31.15% to 220.57 hm², and 24.91% to 49.67 hm² in 2015, respectively.

Across the study period (2005–2015), the highest transition was from forest to unutilized land, accounting for 264.33 hm² (Table 6). The transition from grassland to forest (250.58 hm²) and the reverse one (147.50 hm²) were also high. As seen in Table 6, forest, cultivated land, grassland, and water area decreased by 0.92% to 7186.50 hm², 63.59% to 49.67 hm², 34.54% to 220.58 hm², and 58.47% to 6.01 hm² in 2015, respectively. Unutilized land and construction land increased by 219.76% to 382.49 hm² and 24.74% to 79.37 hm² in 2015, respectively.

4. Discussion

It is well known that earthquakes and secondary disasters cause major disturbances to the ecosystems in affected areas, resulting in soil physicochemical property changes (Vittoz et al. [43]; Lin et al. [44]), vegetation destruction (Cui et al. [31]; Wang et al. [45]), landscape fragmentation (Fichera et al. [46]), and LULC changes (Dewan and Yamaguchi [47]). Ecological restoration involving biological and engineering countermeasures is carried out to prevent further deterioration and to improve the recovery process in earthquake-affected areas (Lin et al. [44]; Millington et al. [48]; Kang et al. [29]). However, to our knowledge, there are few studies in the literature about LULC changes in a typical basin after an earthquake that reflect in more detail the response of ecosystems to major disturbances. Our study may provide a helpful source of evidence about spatiotemporal changes in LULC over multiple years and a more detailed content to document the process of ecological restoration in earthquake-affected areas.

4.1. Land Use Structure Change in the Longxi River Basin

After the Wenchuan earthquake, the areas of forest and unutilized land decreased and increased, respectively, in Dujiangyan city, to which the Longxi River basin belongs (Nath et al. [10]). Our study proved that the transition from forest to unutilized land was highest, accounting for 1087.88 hm² in the Longxi River basin. However, compared with Dujianyan city, the Longxi River basin had a higher recovery potential in forest because forest area increased gradually between 2011 and 2015, though it also transitioned to other land use types. This may be due to the great resistance and resilience of the large area of natural forests distributed in the basin. This result is consistent with the findings of Qiu et al. [49] and Knoke et al. [50], who observed that natural forests with more complex structures and richer biodiversity than plantations showed lesser degrees of earthquake impacts and higher recovery capacities. Overall, forest had a negative single land use dynamic degree after the earthquake until 2009 due to its destruction caused by the earthquake, while it had positive degrees during the periods from 2009 to 2011 and from 2011 to 2015 due to natural recovery. In addition, forest had the second lowest changes in spatial dynamic change degrees during the study period because it dominated all land use types in the Longxi River basin.

In general, environmental factors determine the distribution of cultivated land, while socioeconomic factors often control the direction, quantity, and speed of cultivated land transition (Nath et al. [10]). In this study, the area of cultivated land decreased continually during the study period due to the bury of secondary disasters such as landslides and debris flow induced by the earthquake and heavy rainfall in rainy seasons. This result is consistent with other studies demonstrating negative impacts of earthquakes and disasters on cultivated land transition (Li et al. [51]; Zou et al. [28]). In addition, the local government implemented policies for converting cultivated land into forest or grassland and for reconstruction for the relocation of victims, which also caused the reduction in cultivated land. Therefore, cultivated land had negative single land use dynamic degrees across the study period, with comparatively high spatial dynamic change degrees, showing high transitions to other land use types.

Unutilized land mainly consists of landslides, collapses, and bare land. It accounted for 119.65 hm² in 2005 and increased to 1267.49 hm² in 2009 due to a large amount of bare land caused by the Wenchuan earthquake, which had a seismic intensity of XI. This result is consistent with a previous investigation, which highlighted a large amount of bare land as the response of LULC to a major earthquake (Li et al. [52]). However, the area of unutilized land decreased gradually after 2009 due to the high resilience of natural forests as the dominant land use type and strong management by the Longxi Hongkou National Nature Reserve administration. Overall, unutilized land had the highest single land use dynamic degree after the earthquake until 2009 but had negative degrees during the periods from 2009 to 2011 and from 2011 to 2015 due to ecological restoration. Therefore, unutilized land had the highest spatial dynamic change degrees during the period from 2005 to 2009 but had a comparatively low degree after 2009.

The area of grassland reduced after the earthquake but increased due to natural recovery during the period from 2009 to 2011, while it reduced again during the period from 2011 to 2015 due to the transition to forest. During the period from 2009 to 2011, grassland was unstable and easy to transit to other land types under the influence of environment and ecosystem self-recovery because a part of it was transited from bare land induced by the earthquake and secondary disasters. However, the grassland transited from bare land became stable and evolved as forest after 7 years of natural recovery in 2015. This result is consistent with the findings of Luo et al. [53], who observed that grassland was mainly converted to forest after natural recovery in the Longmen fault zone. Therefore, grassland had a positive single land use dynamic degree with the highest spatial dynamic change degree during the period from 2009 to 2011, while it had negative single land use dynamic degrees with still comparatively high spatial dynamic change degrees during the periods from 2005 to 2009 and from 2011 to 2015.

Compared with 2005, the area of construction land increased in 2009 because the local government carried out the building reconstruction for victim settlement. However, due to the damage caused by a large scale of debris flows in 2010, the area of construction land reduced in 2011 (Xu et al. [54]). After the debris flows, the building reconstruction once again increased the area of construction land. This result is consistent with a previous study demonstrating the positive effect of reconstruction on construction land (Shi et al. [55]). Therefore, compared with grassland, construction land had the opposite pattern in the changes of single land use dynamic degrees across the study period but had a similar pattern in spatial dynamic change degrees.

Water area decreased slightly during the period from 2005 to 2009 but intensively during the period from 2009 to 2011. This is also due to the damage caused by debris flows in 2010. Debris flows created a large number of unconsolidated materials deposited onto the riverbed and debris flow accumulation fans on both banks of the river, resulting in the uplift of the Longxi River riverbed, reduction in water area, and difficulty restoring the damaged water area during the period from 2011 to 2015 (Yu et al. [56]). Therefore, water area had the highest single land use dynamic degree with a comparatively high spatial dynamic change degree during the period from 2009 to 2011.

Across the study period, the speed of land use change fluctuated in the Longxi River basin, reflecting the impacts of natural disasters, natural succession, and human activities in different periods. During the period from 2005 to 2009, land use change was mainly affected by secondary disasters, resulting in a value of 81.37% in comprehensive change index of land use degree. Forest was the main land use type transiting to other types. During the period from 2009 to 2011, land use change was mainly controlled by natural succession, reconstruction, and secondary disasters, resulting in the highest value of 162.55% in comprehensive change index of land use degree. Unutilized land was the main land use type transiting to other types. During the period from 2011 to 2015, land use change was mainly affected by natural succession and reconstruction, resulting in the lowest value of 64.73% in comprehensive change index of land use degree. Unutilized land was also the main land use type transiting to other types. During the period from 2005 to 2015, the comprehensive land use dynamic degree was 39.35%, which was higher than the findings of Shu et al. [57] and Liu et al. [58], who observed that the comprehensive land use degrees of the Salaxi demonstration area and the China–Vietnam border zone ranged from 1.33% to 2.72% and 0.43% to 4.50%, respectively. Therefore, land use types in the Longxi River basin changed relatively rapidly during the period from 2005 to 2015.

The comprehensive change index of land use degree also showed great change during the period from 2005 to 2015. It had an average value of −14.87 during the period from 2005 to 2009, suggesting that a high intensity of land use change occurred, especially a great transition from forest to unutilized land after the earthquake. After the post-earthquake 3 year recovery, the index reached 8.15, indicating that restoration in unutilized land increased the areas of grassland and forest. However, the index changed slowly during the period from 2011 to 2015, suggesting that the speed of natural recovery decreased. Across the study period, the index had an average value of −4.02, showing that negative effects of the earthquake on land use change still existed after a 7 year recovery. The regions with a high comprehensive change index of land use degree were distributed in the northeast and south of the Longxi River, especially along both sides of the lower reaches of the Longxi River, showing that flat land in the downstream alluvial fan is suitable for reconstruction in the basin due to its convenience for transportation, construction, and development. This result proved the conclusion of Zou et al. [59] that flat land in mountainous area had the potential for good economic foundation, convenient transportation, and a dense population.

4.2. The Changes of Land Use Gravity Centers

The gravity centers of land use types would undoubtedly migrate due to human activities and the changes in environmental factors (Chu et al. [60]). Our study found that the gravity centers of all land use types changed across the study period. The gravity center of grassland continued to move southward due to the reduction in its damaged area at high altitude and the transition of unutilized land at the early stage of natural succession.

The gravity center of cultivated land shifted from north to south but then suddenly eastward in 2009. This was due to the original distribution and victim settlement. Cultivated land was scattered in the middle and lower reaches of the basin before the earthquake. However, the destruction of cultivated land induced by the earthquake was severe in the middle reaches, resulting in its southward shift. In addition, the settlement of the population around Longchi town caused the reduction in cultivated land in the west of the basin, resulting in its sudden eastward shift.

The construction land scattered along the riverside for agritainment in the basin was seriously damaged by debris flows and was difficult to rebuild, resulting in its southward shift toward the lower reach of the Longxi River.

Forest, as the dominant land use type, was affected by many factors, including secondary disasters and natural recovery. Its gravity center was similar than unutilized land, and both of them shifted from south to north during the study period. However, forest shifted its gravity center from north to south during the period from 2009 to 2011 due to the reduction in forest area caused by debris flows in 2010.

Water area also transferred its gravity center from north to south. This may be due to the decrease in water storage in the Longchi lake in the north of the basin and the burying of the riverbed caused by unconsolidated materials of debris flows (Zou et al., 2019).

4.3. Land Use Transit Characteristics across the Study Period

During the period from 2005 to 2009, forest and grassland were destroyed after the earthquake, resulting in the increase in unutilized land; the areas of grassland, forest and water area decreased, while that of unutilized land increased. The transition between forest and grassland was high due to the simultaneous degradation of forest and grassland succession to forest. Construction land increased with the transition from forest, grassland, and cultivated land due to victim settlement.

After a 3 year recovery, the transitions among forest, grassland, and unutilized land still dominated the land use transit matrix. However, the areas of forest and grassland increased, while that of unutilized land decreased, though the debris flows that occurred in 2010 caused heavy losses (Yu et al. [56]). In particular, water area intensively decreased from 14.09 hm² in 2009 to 5.16 hm² in 2011, suggesting that the riverbed was buried by debris flows (Zou et al. [28]).

After 7 year recovery, the transitions among forest, grassland, and unutilized land still were the main transit types. However, Unlike in 2011, the area of grassland decreased in 2015, indicating that the Longxi River basin maintained positive succession due to the high resilience of natural forests and strong management by the Longxi Hongkou National Nature Reserve administration.

However, though the area of forest almost reached its original status in 2015, all land use types except unutilized land still had lower areas than in 2005 (Table 6), indicating that recovery was not complete 7 years after the earthquake (Zhao et al. [61]; Li et al. [62]). Hence, the management of land use, afforestation, and forest protection should be carried out to improve the restoration in the Longxi River basin.

In this study, we used four different remote-sensing satellite images to analyze the spatiotemporal dynamic characteristics of land use change in the Longxi River basin. However, we were not able to consider other environmental indicators (e.g., regional meteorological data, soil properties, population, and socioeconomic development data) as they were not concerned during the study period. Therefore, elucidating the effects of these and other indicators on land use change would require further research.

5. Conclusions

In this study, the GIS spatial analysis method was used to study the characteristics of land use change in the Longxi River basin before and after the 2008 Wenchuan earthquake. We calculated the dynamic degree of land use, the comprehensive change index of land use degree, and the transition matrix and the spatial variation of six land use types. Under the influence of the Wenchuan earthquake and secondary disasters, such as debris flows and landslides, the land use pattern of the Longxi River basin changed greatly. The post-earthquake reconstruction, the acceleration of urbanization, the formulation and release of national policies, the transformation of rural industries, and the forest natural recovery ability together affected the land use change. The following results were obtained:

(1)

Forest was the most dominant land use type in the basin across the study period, though unutilized land induced by secondary disasters became the second dominant land use type after the earthquake. Compared with pre-earthquake in 2005, the areas of cultivated land, forest, grassland, and water area decreased by 1.09%, 0.84%, 1.47%, and 0.10%, respectively, while those of construction land and unutilized land increased by 0.2% and 3.32%, respectively, in 2015.

(2)

The transition intensity of other land use types to unutilized land was highest during the period from 2005 to 2009, whereas spatial dynamic change degrees of all types decreased between 2009 and 2015.

(3)

High and developing comprehensive indexes of land use degree were distributed along both banks of the lower reach of the Longxi River during the study period.

(4)

The gravity centers of forest and unutilized land changed from south to north, while the other types had the opposite pattern.

Though some limitations exist, our study provides a reference to improve land use structure, soil-erosion control, and ecological rehabilitation in earthquake-affected areas. Future research should extend this work by monitoring LULC, using more remote-sensing satellite images to determine the change process of land use structure during ecological restoration.