Nothing Special   »   [go: up one dir, main page]
Next Article in Journal
Landscape–Ecological Problems Resulting from Spatial Conflicts of Interest in the Poľana Biosphere Reserve
Previous Article in Journal
The Importance of Urban Greening Spaces for Avian Communities in an Urbanized Landscape
Previous Article in Special Issue
Evaluation of the Hydrological Response to Land Use Change Scenarios in Urban and Non-Urban Mountain Basins in Ecuador
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Land
Volume 14
Issue 2
10.3390/land14020401
land-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Monitoring and Analysis of Relocation and Reclamation of Residential Areas Based on Multiple Remote Sensing Indices

by
Huiping Huang
1,2,
Yingqi Wang
1,2,
Chao Yuan
1,
Wenlu Zhu
1,2 and
Yichen Tian
1,2,*
1
Aerospace Information Research Institute, Chinese Academy of Sciences, No. 9 Dengzhuang South Road, Haidian District, Beijing 100094, China
2
University of Chinese Academy of Sciences, No.19(A) Yuquan Road, Shijingshan District, Beijing 100049, China
*
Author to whom correspondence should be addressed.
Land 2025, 14(2), 401; https://doi.org/10.3390/land14020401
Submission received: 19 January 2025 / Revised: 10 February 2025 / Accepted: 13 February 2025 / Published: 14 February 2025
(This article belongs to the Special Issue Land Use Change from Non-urban to Urban Areas: Problems, Challenges and Opportunities II)
Browse Figures
Figure 1
<p>Satellite and UAV images of different stages in the process of village relocation and reclamation.</p> "> Figure 2
<p>Schematic diagram of relocation areas in the Yellow River floodplain region ((<b>a</b>) China; (<b>b</b>) part of the Yellow River floodplain area; (<b>c</b>) the study area of this research).</p> "> Figure 3
<p>Schematic diagram of the 2015 human activity data and village boundary extraction results.</p> "> Figure 4
<p>Technical roadmap.</p> "> Figure 5
<p>Short-wave infrared image (Fengqiu County): (<b>a</b>) is from Landsat-8 OLI, and (<b>b</b>) is the fitted GaoFen-2 short-wave infrared image.</p> "> Figure 6
<p>Optimal thresholds for extracting construction land and vegetation using EASI and kNDVI.</p> "> Figure 7
<p>Results of relocation and reclamation status identification and overall progress statistics ((<b>a</b>) identification results of village relocation and reclamation status; (<b>b</b>) remote sensing imagery in 2015 (before relocation and reclamation); (<b>c</b>) remote sensing imagery in 2023 (after relocation and reclamation); (<b>d</b>) the proportion of villages undergoing relocation and reclamation).</p> "> Figure 8
<p>Comparison of drone imagery from field surveys showing (<b>a</b>) idle land, (<b>b</b>) resident-initiated reclamation, and (<b>c</b>) fully reclaimed land.</p> "> Figure 9
<p>Schematic diagram of the implementation effects of relocation and reclamation projects based on EASI and kNDVI.</p> "> Figure 10
<p>Spatial distribution of relocation and reclamation progress of residential areas in Henan floodplain area.</p> ">
Versions Notes

Abstract

:
The relocation of residents from high-risk areas is a critical measure to address safety and development issues in the floodplain regions of Henan Province in China. Whether the old villages can be reclaimed as farmland after demolition concerns Henan Province’s ability to maintain its farmland red line. This paper integrated multiple remote sensing indices and proposed a remote sensing identification method for monitoring the progress status of village relocation and reclamation that adapted to data characteristics and application scenarios. Firstly, it addressed the issue of missing target bands in GF-2 (GaoFen-2) by employing a band downscaling method; secondly, it combined building and vegetation indices to identify changes in land cover in the old villages within the floodplain, analyzing the implementation effects of the relocation and reclamation policies. Results showed that using a Random Forest regression model to generate a 4 m resolution shortwave infrared band not only retains the original target band information of Landsat-8 but also enhances the spatial detail of the images. Based on the optimal thresholds of multiple remote sensing indices, combined with human footprint data and POI (Points of Interest) identified village boundaries, the overall accuracy of identifying the progress status of resident relocation and reclamation reached 93.5%. In the floodplain region of Henan, the implementation effect of resident relocation was relatively good, with an old village demolition rate of 77%, yet the farmland reclamation rate was only 23%, indicating significant challenges in land conversion, lagging well behind the pilot program schedule requirements. Overall, this study made two primary contributions. First, to distinguish between rural construction and bare soil, thereby improving the accuracy of construction land extraction, an Enhanced Artifical Surface Index (EASI) was proposed. Second, the monitoring results of land use changes were transformed from pixel-level to village-level, and this framework can be extended to other specific land use change monitoring scenarios, demonstrating broad application potential.

1. Introduction

The Yellow River is China’s second-longest river and is revered as the “mother river” of the Chinese nation, being the most suitable settlement area for ancient Chinese inhabitants [1]. The floodplain region downstream of the Yellow River is part of the river channel, serving not only as a place for flood discharge, flood retention, and sedimentation but also as the fundamental space for the livelihood and production activities of the local population. For a long time, due to constraints from its unique geographical environment and threats of flooding during the flood season, the public infrastructure in the floodplain has been weak, social development has lagged, and economic and social progress has been relatively slow [2]. Since 2015, Henan Province has carried out two rounds of pilots, implementing flood control safety measures, such as relocation and on-site flood avoidance for 57,000 people. In 2017, the state formulated the “Henan Yellow River Floodplain Resident Relocation Plan” (hereinafter referred to as the Three-Year Plan), which planned to relocate 243,200 people living in low-lying areas with prominent flood risks within three years, completing the relocation task by 2020, with demolition and reclamation work to be finished within three months after relocation [3]. The Three-Year Plan clearly states that “the demolition of old villages and land reclamation are key to the success or failure of the project, insisting on synchronously advancing relocation with demolition and reclamation”. Through the reclamation and organization of land from demolished old villages, larger, more regular farmland plots can be created, improving agricultural production conditions, enhancing land use efficiency and agricultural productivity benefits, increasing the quantity and quality of farmland, and providing strong guarantees for ensuring food security.
The timely reclamation into farmland of the 246 old villages demolished under the two pilot programs and the Three-Year Plan is crucial for achieving the goals of relocating floodplain residents, ensuring their stability, enabling development, and promoting prosperity. It also concerns Henan Province’s ability to maintain its farmland red line. Therefore, there is an urgent need to use scientific and technological methods to understand the current progress of land reclamation in the relocated old villages within Henan floodplain region, analyze the proportion of regional land reclamation, and identify the primary factors affecting land reclamation, ensuring that the demolition of old villages and their conversion into farmland are completed smoothly and on schedule.
Remote sensing technology, with its unique advantages of long-term sequential observation, multi-angle monitoring, and fine spatial resolution, has become a crucial research tool for accurately extracting information on land cover distribution and changes [4,5]. High-resolution remote sensing images can capture the detailed characteristics of land cover information, reveal internal structures, and finely depict the state changes in regional land cover forms, further highlighting its potential in serving regional land consolidation and promoting high-quality development [6,7,8]. The process of demolishing old villages and reclaiming land in floodplain areas represents different stages of land cover type changes, which present differentiated spectral and textural information on high-resolution remote sensing images (Figure 1). Before demolition, the land cover type of old villages is rural construction land; during the period when houses have been demolished but reclamation has not yet occurred, the land cover type is bare soil or sparse vegetation; after reclamation, the land cover type becomes farmland.
In remote sensing image classification methods, pixel-based remote sensing index approaches are characterized by their simplicity, speed, and high versatility, making them widely applied in the extraction of land cover information [9,10,11,12]. Figure 1 shows that the land cover types at different stages of the demolition and reclamation of old villages are construction land, bare soil, and farmland, respectively. Scholars have proposed various remote sensing indices for extracting these three types of land cover [13,14]. For example, commonly used indices for identifying construction land include NDBI (Normalized Difference Built-Up Index) and MNDBI (Modified Normalized Difference Built-Up Index); for identifying bare soil, indices like BSI (Bare Soil Index) and NDII (Normalized Difference Infrared Index) are often utilized; and for recognizing farmland or vegetation, indices such as NDVI (Normalized Difference Vegetation Index) and EVI (Enhanced Vegetation Index) are typically employed. However, these indices may not be suitable for rural areas. The specific reasons and solutions are detailed in the following sections.
Confusion between rural bare soil information and built-up land information, as well as spatial resolution that does not meet the needs of precise identification, have long been challenges in remote sensing research both domestically and internationally [15,16]. To address the spectral similarity issue of high-reflectance buildings and bare soil in remote sensing images, Xu et al. utilized a dual-index approach with the Normalized Difference Soil Index (NDSI) and the Normalized Difference Impervious Surface Index (NDISI), effectively resolving the confusion between bare soil and built-up land in Landsat imagery [17]. This provided a reliable technical means for extracting bare soil from Landsat imagery. However, Landsat imagery struggles to meet the requirements for large-scale detailed bare soil mapping, while higher spatial resolution images often have lower spectral resolution and lack some key bands necessary for index construction. Particularly in rural areas, where characteristics differ significantly from urban built-up land, existing building indices cannot satisfy the needs of extracting rural built-up land. Therefore, high-resolution mapping of rural built-up land remains an urgent challenge to be addressed. Zhu et al. improved the confusion between red rooftops and bare soil by fitting a 4 m spatial resolution shortwave infrared band and combining it with the Artifical Surface Index (ASI) [18] and the Red Roof Index (RRI), achieving high-precision extraction of rural built-up land [19].
After the complete execution of relocation and reclamation work, the land use type would transform into farmland. One of the challenges in extracting farmland information from remote sensing images lies in the diversity of farmland types, the richness of cultivated crop species, and the differences in crop growth stages, which can easily lead to confusion between farmland and other land categories [20]. The kNDVI (kernel Normalized Difference Vegetation Index), proposed in 2021, employs machine learning kernel methodology to linearize NDVI, thereby avoiding issues related to the nonlinearity and saturation of vegetation indices [21]. Research indicates that kNDVI exhibits stronger resistance to saturation, bias, and complex phenological cycles, showing greater robustness against noise and stability across spatial and temporal scales [22].
After addressing the shortcomings of high-resolution identification of rural construction land and farmland, it becomes feasible to conduct status recognition of relocation and reclamation based on land use types as a core technology. However, identifying the state of relocation and reclamation differs from simple land use classification in several key aspects: (1) Determining the status of relocation and reclamation requires considering land use types both before and after policy implementation. (2) The transition from pixel-scale to village-scale land use type analysis has not been addressed in existing research. (3) There is no comprehensive index proposed for extracting construction land that is applicable to areas with high proportions of rural and bare soil. Therefore, the primary objectives of this study are to address the main issues related to land use type identification involved in rural relocation and reclamation projects and to monitor such activities at the village scale.
Based on the aforementioned background, this study aimed to utilize high-resolution band fitting technology, along with EASI and kNDVI, to propose a method for identifying the progress of relocation and reclamation in the Henan floodplain region that meets the demands of high-resolution applications. Firstly, to address the issue that most remote sensing index calculations depend on shortwave infrared (SWIR) bands, which are missing from China’s GF-2 (GaoFen-2) high-resolution imagery, we employed a band downscaling method to fit and generate high-resolution SWIR bands. This approach enables the broader application of remote sensing index methods to high-resolution remote sensing images. Secondly, the high-resolution SWIR band obtained through the downscaling method was applied to our proposed Enhanced Artificial Surface Index (EASI) for construction land and the kernel Normalized Difference Vegetation Index (kNDVI). Finally, using villages as the unit of analysis, we applied these remote sensing indices to identify the land cover conditions of planned demolition villages in the Henan floodplain, thereby assessing whether old villages have been reclaimed into farmland as scheduled after demolition. This allowed us to analyze the implementation effects of the two pilot programs and the Three-Year Plan for relocation and reclamation.

2. Study Area and Data Processing

2.1. Study Area

This study selected the Yellow River floodplain region of Henan Province, China, as the research area. The Henan floodplain covers an area of approximately 2116 square kilometers, with 2.28 million mu (approximately 152,000 hectares) of farmland and a resident population of 1.254 million people. The region encompasses parts of six provincial-level municipalities under the jurisdiction of Henan Province—Zhengzhou, Kaifeng, Luoyang, Jiaozuo, Xinxiang, and Puyang—and involves 17 counties (districts), 59 towns, and 1172 natural villages.
The Henan floodplain region is extensive, with a large population, and its safety infrastructure is severely lagging. In the event of a flood exceeding the 20-year recurrence interval, most of the high-lying areas and all low-lying areas would be inundated by floodwaters. According to the flood control standards and principles outlined in the State Council-approved “Comprehensive Plan for the Yellow River Basin (2012–2030)”, there are 1.037 million people not meeting the 20-year flood protection standard, among whom 833,000 are at high risk of flood inundation. To address the urgent need for relocating residents from high-risk areas for safety reasons, the scope of relocation and reclamation under the two pilot programs and the Three-Year Plan does not encompass all villages within the Henan floodplain. Instead, it focuses on fifty-nine towns across nine counties (districts) in four provincial-level municipalities: Zhengzhou, Kaifeng, Xinxiang, and Puyang. This area has been designated as the study area for this research (Figure 2). The first batch of pilots involved four towns in three counties, covering fourteen villages with 4676 households and 16,718 people. The second batch of pilots involved nine towns in six counties, covering twenty-five villages with 11,132 households and 40,132 people. The Three-Year Plan involves thirty-four towns in eight counties, covering two hundred seven villages and 243,200 people.

2.2. Data Processing

2.2.1. Remote Sensing Data

Remote sensing images from the Landsat-8 OLI satellite for the study area with cloud cover less than 10% were obtained from a remote sensing cloud platform [23] for May to August 2023, with a spatial resolution of 30 m. A de-cloud function was used for cloud masking, and pixel-level fusion reconstruction was applied to generate an image set with minimal cloud cover. Using the median synthesis method, the image set was synthesized into a single image. Meanwhile, GaoFen-2 satellite images with cloud cover less than 10% from May to August 2023 were acquired from the China Center for Resources Satellite Data and Application, with a spatial resolution of 4 m. Preprocessing operations such as orthorectification, atmospheric correction, and mosaicking were performed on the GaoFen-2 images. In this study, we selected the B2 (Blue), B3 (Green), B4 (Red), and B8 (Near Infrared) bands at 30-m resolution from the Landsat-8 OLI imagery as the core variables for regression analysis; meanwhile, the B11 (Shortwave Infrared 1) and B12 (Shortwave Infrared 2) bands were designated as target bands to construct a Random Forest Regression model. For the GaoFen-2 imagery, the 4-m resolution B1 (Blue), B2 (Green), B3 (Red), and B4 (Near Infrared) bands were utilized for shortwave infrared band fitting. All these high-resolution bands were then applied in remote sensing index calculations and the identification of relocation and reclamation status.
During the process of land cover information identification, water bodies are often treated as noise or background information and need to be removed [24]. The Henan floodplain region includes sections of the Yellow River, fish ponds, ditches, and other water bodies. In this study, a mask was created by calculating water indices and performing visual interpretation to remove the majority of water bodies.

2.2.2. Human Footprint Data

The 2015 World Settlement Footprint (WSF) dataset was jointly released in July 2020 by the German Aerospace Center (DLR) and the Earth Observation Center (EOC). The WSF dataset utilized remote sensing satellite data from Landsat-8 OLI, Sentinel-1, and Sentinel-2 for the years 2014–2015. Based on Random Forest classification methods, the resultant dataset has a spatial resolution of 10 m [25].
The two pilot batches and the Three-Year Plan for relocation and reclamation in the Henan floodplain area began in 2015. This study used human settlement footprint data from 2015 to generate the outlines of old village boundaries before resident relocation, comparing the status of these villages (undemolished construction land) before and after relocation (various types). The effectiveness of the relocation and reclamation policies was evaluated by analyzing changes in 246 villages. To conduct this analysis, the administrative boundaries of nine districts and counties involved in the relocation and reclamation efforts in Henan floodplain were used to clip the WSF dataset, obtaining human settlement footprint data within the study area for 2015 (Figure 3). The human activity footprint data are raster data. To ensure the completeness of village boundary identification in subsequent steps, a morphological-based preprocessing method was applied before vectorization. In ENVI 5.3 software, morphological closing operations were performed, which involve dilation followed by erosion. Closing operations expand noise regions and connect them with surrounding background areas before eroding them away, thereby removing noise and fine details from the image and achieving smoother edges. This preprocessing step ensured that the raster images were appropriately smoothed and noise-reduced, facilitating accurate vectorization of village boundaries.

2.2.3. Point of Interest (POI)

POI data are an important component of geographic big data and play a significant role in location-based services and regional scene understanding [26]. In this study, POI data were obtained through the Application Programming Interface (API) provided by AMap (https://www.amap.com, accessed on 1 March 2022), primarily using place names and addresses to acquire the locations and names of rural settlements in the Henan floodplain area. This study adopted a proximity analysis method to identify rural settlements within patches of human activity footprint. Prior to conducting the proximity analysis, the POI data for the Yellow River floodplain were cleaned and filtered to select 1901 rural settlement POIs containing village names, and the human activity footprint data were vectorized.
The vectorized polygon data from the 2015 human activity footprint did not include a property for identifying patch names. In this study, we utilized the location and name information of rural settlements from POI (Points of Interest) address data to obtain village names for the vector patches of the human activity footprint using a proximity method. This process determined the locations of 246 villages planned for relocation in the Yellow River floodplain area (Figure 3), which is essential for subsequent evaluation of the implementation effects of relocation and reclamation. The proximity analysis method assumes that the place name type of a village patch is determined by the nearest POI within or around the plot. The methodology is as follows: First, calculate the centroid of each village patch; second, determine the control point through proximity analysis, where the control point is the closest POI to the centroid of the village patch; finally, complete attribute attachment by associating the place name of the control point with its respective village patch. Using nearest neighbor analysis, initial classification results can be obtained, where larger village patches correspond to one POI place name. A small number of villages may consist of multiple patches, leading to potential classification inaccuracies, which require manual revision.

3. Methods

  • Data preparation. Firstly, acquire and preprocess the involved 30-m resolution Landsat-8 OLI images and 4-m resolution GaoFen-2 images. Secondly, apply morphological closing operations on World Settlement Footprint to smooth boundaries and fill internal holes. Cleanse and filter Points of Interest (POI) data to retain POIs related to rural settlements, and obtain the contour scope of old villages in reclamation areas based on proximity analysis methods.
  • Estimation of short-wave infrared band at 4-Meter Resolution. Establish a Random Forest regression model between the two short-wave infrared bands SWIR1 and SWIR2 of Landsat-8 OLI and its blue, green, red, and near-infrared bands. Use this model with the blue, green, red, and near-infrared bands of the GaoFen-2 image as input to estimate the values of the two short-wave infrared bands SWIR1 and SWIR2 for the GaoFen-2 image.
  • Identification of village relocation and reclamation status. Based on the estimated short-wave infrared bands from the Random Forest regression model and the original visible and near-infrared bands of GaoFen-2, calculate the values of EASI and kNDVI. Perform optimal threshold selection for both index images separately, then statistically analyze the proportion of pixels that conform to the optimal thresholds on a per-village basis. This process identifies the status of relocation and reclamation (whether old villages have not been relocated, have been relocated but not reclaimed or reclaimed spontaneously, or have been officially reclaimed) (Figure 4).
Land 14 00401 g004
Figure 4. Technical roadmap.
Figure 4. Technical roadmap.
Land 14 00401 g004

3.1. Short-Wave Infrared (SWIR) Band Fitting

Due to the absence of short-wave infrared (SWIR) bands in GaoFen-2 imagery, this study adopted a band downscaling method to fit and generate SWIR bands at 4-m resolution. The downscaling method is based on theories and technologies of image processing, data fusion, and statistical modeling, which can be categorized into physical models and statistical models [27]. Among these, the Random Forest model within statistical models was widely applied due to its high accuracy and robustness [28,29]. The regression kernel correlates lower spatial resolution SWIR bands with higher spatial resolution visible/near-infrared bands, achieving the purpose of downscaling lower-resolution bands [30,31].
This study selected the Random Forest machine learning model to establish the relationship between the regression kernel and target bands. The regression kernel consisted of multispectral surface reflectance data, including blue, green, red, and near-infrared (NIR) bands, while the target bands were the two short-wave infrared bands. Using Landsat-8 OLI as the low-resolution remote sensing imagery and GaoFen-2 as the high-resolution imagery, the specific method for fitting the high-resolution target bands is as follows. The Landsat-8 OLI regression kernel was used as independent variables and the Landsat-8 OLI target bands as dependent variables to establish a Random Forest machine learning model. Then, the GaoFen-2 regression kernel was input as independent variables into this model to predict and generate the 4-m resolution SWIR bands for GaoFen-2.
This study used two image quality metrics, Contrast and Information Entropy, to evaluate the differences in image texture detail information, thereby examining the effectiveness of the band downscaling [32,33]. The formulas for calculating Contrast and Information Entropy are as follows:
C O N = ∑ i = 0 255 i 2 p Δ ( i )
H = − ∑ i = 0 255 p Δ ( i ) log 2 ( p Δ ( i ) )
where CON represents Contrast, H represents Information Entropy, i represents the gray level, and p Δ ( i ) represents the frequency of pixels with gray level i appearing in the image. Before calculating these two metrics, the image values were stretched to the grayscale range of [0, 255].

3.2. Remote Sensing Index Calculation

3.2.1. Enhanced Artifical Surface Index

This study developed an Enhanced Artificial Surface Index (EASI) for the extraction of rural construction land. EASI integrates the existing Artificial Surface Index (ASI) with a Simple Soil Index (SSI) proposed in this study. Compared to ASI, EASI better suppresses bare soil confusion within construction land, addressing the issue of distinguishing between idle or temporarily unseeded land and construction land in rural areas. The formula for calculating EASI is shown below. In the formula, f is a normalization function that standardizes the values of ASI and SSI to the range [0, 1]; α is a weighting coefficient, which was set to 0.5 in this study.
E A S I = f A S I − α × f S S I
The ASI is composed of four remote sensing indices: AF, VSF, SSF, and MF (Equation (4)), used for extracting information on construction land [18]. In the formula, f is a normalization function that standardizes the values of AF, VSF, SSF, and MF to the range [0, 1].
A S I = f A F × f V S F × f S S F × f M F
AF is the Artificial Surface Factor (Equation (5)), where ρ N I R and ρ B l u e represent the surface reflectance of the near-infrared and blue bands, respectively, from multispectral remote sensing imagery. The normalized ratio calculation can reduce observational errors and uncertainties in estimation.
A F = ρ N I R − ρ B l u e ρ N I R + ρ B l u e
VSF is the Vegetation Suppressing Factor (Equation (6)), calculated based on the NDVI (Equation (7)) and the MSAVI (Equation (8)). In these formulas, ρ N I R and ρ Re d represent the surface reflectance of the near-infrared and red bands, respectively, from multispectral remote sensing imagery.
V S F = 1 − N D V I × M S A V I
N D V I = ρ N I R − ρ Re d ρ N I R + ρ Re d
M S A V I = 2 × ρ N I R + 1 − 2 × ρ N I R + 1 2 − 8 × ρ N I R − ρ Re d 2
SSF is the Soil Suppressing Factor (Equation (12)), calculated based on the EMBI (Equation (10)). The EMBI is constructed on the foundation of the MBI and the MNDWI. In these formulas, ρ N I R , ρ G r e e n , ρ S W I R 1 , and ρ S W I R 2 represent the surface reflectance of the near-infrared, green, short-wave infrared 1, and short-wave infrared 2 bands, respectively, from multispectral remote sensing imagery.
S S F = 1 − E M B I
M B I = ρ S W I R 1 − ρ S W I R 2 − ρ N I R ρ S W I R 1 + ρ S W I R 2 + ρ N I R + 0.5
M N D W I = ρ G r e e n − ρ S W I R 1 ρ G r e e n + ρ S W I R 1
E M B I = M B I − M N D W I − 0.5 M B I + M N D W I + 1.5
MF is the Modulation Factor (Equation (13)). In this formula, ρ B l u e , ρ G r e e n , ρ N I R and ρ S W I R 1 represent the surface reflectance of the blue, green, near-infrared, and short-wave infrared 1 bands, respectively, from multispectral remote sensing imagery.
M F = ρ B l u e − ρ G r e e n − ρ N I R − ρ S W I R 1 ρ B l u e − ρ G r e e n + ρ N I R − ρ S W I R 1
This study developed a Simple Soil Index (SSI) (Equation (14)) primarily to identify the land leveling status of areas that were previously old village sites but have not yet been reclaimed, where the land cover type is bare soil. By utilizing visible light bands, this index enhances the representation of large patches of bare soil, facilitating the extraction of idle or temporarily unseeded bare land in rural areas. In the formula, ρ Re d , ρ G r e e n , and ρ B l u e represent the surface reflectance of the red, green, and blue bands, respectively, from multispectral remote sensing imagery.
S S I = ρ Re d − ρ G r e e n + ρ B l u e

3.2.2. Kernel Normalized Difference Vegetation Index

This study selected the kNDVI to identify whether old village sites have been converted to farmland after demolition. The kNDVI employs a specific kernel function that accounts for all higher-order relationships between the NIR and red reflectance bands, not just first-order relationships. Therefore, it can effectively alleviate saturation issues and exhibits higher performance robustness compared to NDVI. A relatively simple and practical expression for kNDVI is shown in Equation (15).
k N D V I = tanh ρ N I R − ρ Re d 2 σ 2
ρ N I R and ρ Re d represent the surface reflectance of the near-infrared and red bands, respectively. σ is a length-scale parameter specified for each particular application, indicating the sensitivity of the index to sparse or dense vegetation areas. In this study, σ was set to 0.5 × ρ N I R + ρ Re d .

3.3. Identification of Relocation and Reclamation Status

To identify construction land in villages that have not been demolished and farmland after reclamation has been completed, this study applied Formulas (3) and (15) to enhance the information from high-resolution remote sensing images. Subsequently, village-level statistics were conducted based on the pixel recognition results. Through repeated experiments, we found that setting the threshold at 60% most effectively determined the status of villages. Therefore, for both construction land and vegetation, if more than 60% of the pixels within a village meet the classification criteria, the village is classified according to its corresponding land use type. If over 60% of the pixels within a village are identified as construction land, the village is considered undemolished. If over 60% of the pixels within a village are identified as vegetation, the village is considered reclaimed. If a village does not meet either of the above conditions, it is categorized as unreclaimed or partially reclaimed. Ultimately, this method allows for the identification of the relocation and reclamation status for all villages listed.

4. Results and Analysis

4.1. Evaluation of Short-Wave Infrared Downscaling Fitting Results

The Random Forest machine learning model established between the regression kernel and target bands based on 30-m resolution Landsat-8 OLI remote sensing imagery successfully constructed a statistically invariant relationship across scales. The R-squared values for the two SWIR bands were greater than 0.9 on the training set and exceeded 0.85 on the testing set, indicating strong performance of our regression models. Compared to Landsat-8 OLI, the downscaling fitting results for the 4-m short-wave infrared bands (Figure 5) maintained stable information entropy while significantly enhancing contrast (Table 1). This indicates that the downscaling fitting process effectively preserves original information while significantly improving spatial detail in the image.

4.2. Remote Sensing Identification and Accuracy Analysis of Relocation and Reclamation Status

4.2.1. Optimal Threshold Selection and Remote Sensing Identification Results for Relocation and Reclamation

Based on the results of multiple remote sensing indices calculated for the Henan floodplain area, 1500 sample points were selected for repeated trials. The optimal threshold for identifying construction land in villages using the EASI was determined to be 0.47, while the optimal threshold for recognizing farmland after reclamation using the kNDVI was set at 0.22 (Figure 6).
Using the EASI and kNDVI indices, we achieved the identification of relocation and reclamation status for 246 villages across two batches and the Three-Year Plan in the Henan floodplain area (Table 2). The results are shown in Figure 7. All the involved villages are distributed along both sides of the Yellow River, with villages becoming more scattered and numerous towards the northeast direction. As of August 2023, there were 56 undemolished villages and 190 relocated villages, of which 134 had not completed reclamation, and only 56 had completed reclamation.

4.2.2. Accuracy Analysis of Remote Sensing Identification for Relocation and Reclamation Status of Villages

The accuracy verification adopted a method combining field surveys with high-resolution imagery validation.
  • Ultra-high resolution imagery was used to validate the identification accuracy of optimal thresholds for multiple remote sensing indices. Taking the identification of construction land using EASI as an example, 400 sample points were randomly selected in the study area (300 points identified as construction land and 100 points as non-construction land). The true values of these sample points were obtained using ultra-high resolution imagery data, and a confusion matrix was constructed for accuracy validation. The overall classification accuracy reached 93%, and the Kappa coefficient reached 0.82 (Table 3), indicating that the use of multiple remote sensing indices for identifying rural construction land in the floodplain area is effective.
  • Field surveys were conducted to verify the accuracy of identifying relocation and reclamation progress in villages. A total of 82 villages were sampled from two batches and the Three-Year Plan for planned demolition (Table 4). In the first batch, 13 villages were inspected, achieving an accuracy of 96.15% for identifying demolition and reclamation status. In the second batch, 17 villages were inspected, achieving an accuracy of 94.12%. Under the Three-Year Plan, 52 villages were inspected, achieving an accuracy of 93.26%. By weighting the accuracies according to the proportion of villages inspected in each phase, the overall monitoring accuracy was determined to be 93.5%.
Compared to the reclamation of farmland, the remote sensing identification accuracy for village demolition is higher. The main reasons are twofold: (1) In remote sensing images, village demolition is characterized by the reduction or disappearance of original residential houses and changes in land cover types to bare soil after land leveling, abandoned land overgrown with weeds but not yet reclaimed, and regular farmland after reclamation. These changes result in significant spectral and textural differences, making them relatively easy to identify. (2) Reclamation of original villages into farmland appears as vegetation/crop cover in remote sensing images. This can lead to misclassification of regularly shaped plots that have not been reclaimed but are overgrown with weeds, or areas where villagers have planted scattered crops on their own initiative, as already reclaimed land. Therefore, the accuracy of reclamation monitoring is lower. Verification revealed that a very few villages misidentified as reclaimed were not part of government-organized farmland reclamation projects but had naturally grown weeds (Figure 8a) or crops planted sporadically by villagers (Figure 8b). Especially in winter remote sensing images, where vegetation is less prominent, these areas can more easily be confused with surrounding large-scale reclaimed farmland (Figure 8c).

4.3. Implementation Effects of Relocation and Reclamation

As of the time of remote sensing image acquisition in August 2023, the implementation of resident relocation and reclamation in the floodplain area for two pilot batches and the Three-Year Plan did not meet the expected outcomes. The demolition rate of old villages was 77%, while the farmland reclamation rate was only 23%, indicating that the actual progress lagged behind the requirements of the pilot plans. We have also created schematic diagrams of typical villages to more clearly illustrate the research results at different stages of relocation and reclamation (Figure 9).
The first batch of pilot projects in the Henan floodplain area achieved significant results, with a relocation rate of 100% and a reclamation rate of 86%. The first batch of pilot plans for resident relocation in the Henan floodplain area was scheduled to complete relocation by 2017, involving four towns across three counties: Lankao County, Fengqiu County, and Fan County. In 2017, apart from Puligudui Village in Zhangzhuang Township of Fan County and Xueguozhuang Village in Li Zhuang Town of Fengqiu County, all other 12 villages completed their relocation work, achieving a 100% relocation rate. By 2019, farmland reclamation had been completed with an 86% reclamation rate. In Fan County’s Zhangzhuang Township and Chenzhuang Town, after the demolition of five villages, the reclaimed land was mainly converted into vegetable bases, followed by farmland. Puligudui Village in Zhangzhuang Township of Fan County was the only village that did not complete demolition by 2017 among the first batch of pilots. It started demolition in 2021, completed land leveling in 2022, but had not finished reclamation as of August 2023. Xueguozhuang Village in Li Zhuang Town of Fengqiu County completed relocation before 2017 but had also not finished reclamation by August 2023.
Resident relocation in the second batch of pilot projects in Henan floodplain area has been largely completed, but the reclamation rate is low at only 36%. Henan Province implemented the second batch of pilot projects for resident relocation in the floodplain area from 2017 to 2019, involving six counties (Lankao County, Fengqiu County, Changyuan County, Zhongmu County, Taiqian County, and Puyang County) across nine towns and twenty-five villages. As of August 2023, out of the twenty-five villages in the second batch of pilots, twenty-three villages (92%) have completed house demolition, with only two villages remaining undemolished: Zhuanghu Village and Nanxi Village. Among the relocated villages, only nine villages (36%) have completed reclamation. Specifically, nine villages in Guying Town of Lankao County, Wuqiu Township of Changyuan County, Langchenggang Town of Zhongmu County, and Liyuan Township of Puyang County have finished both relocation and reclamation. Meanwhile, fourteen villages in Chenqiao Town of Fengqiu County, Wuba Town and Sunkou Town of Taiqian County, and XichengTown and Xuzhen Town of Puyang County have completed relocation but have only undergone limited autonomous reclamation without unified reclamation efforts.
The Three-Year Plan for resident relocation and reclamation has seen slow progress, with a farmland reclamation rate of only 17%. The Three-Year Plan (2018–2020) for resident relocation and reclamation involved four provincial-level cities (Zhengzhou, Kaifeng, Xinxiang, and Puyang) and eight counties/districts (Changyuan County, Zhongmu County, Xiangfu District, Fengqiu County, Yuanyang County, Puyang County, Fan County, and Taiqian County), covering thirty-four towns and two hundred seven villages, all located in flood-prone areas. As of August 2023, out of the 207 villages within the relocation scope, 153 villages (74%) have completed house demolition, while 54 villages (26%) remain undemolished. Among the relocated villages, only 35 villages (17%) have completed reclamation, with 118 villages (83%) yet to be reclaimed. Regarding resident relocation, progress has been better in Changyuan County, Xiangfu District, Fengqiu County, and Fan County, where relocation rates exceed 90%. In contrast, Zhongmu County has seen the worst performance, with none of its three villages having been demolished. For farmland reclamation, Changyuan County leads with a reclamation rate of 77%, followed by Yuanyang County at 61%. However, none of the 17 relocated villages in Xiangfu District have been reclaimed, and the reclamation rates in Puyang County, Fan County, and Taiqian County are extremely low, all below 5%, resulting in extensive idle land that has not been utilized, leading to a decline in regional ecological environment quality.

4.4. Spatial Distribution of Relocation and Reclamation and Land Transfer Status

Field surveys revealed that differences in the scale of relocated villages, the intensity of relocation implementation, and financing channels have led to significant variations in the progress of relocation and reclamation across various districts and counties in the floodplain area (Figure 10).

4.4.1. Significant Differences in the Progress of Relocation and Reclamation Among Various Districts and Counties

The relocation and reclamation efforts have progressed relatively smoothly in Fan County, Changyuan City, and Lankao County, with over 94% of villages completing demolition. Changyuan City and Lankao County have nearly completed farmland reclamation, where the post-demolition land reclamation is mainly organized by the government, resulting in standardized farmland. In Fengqiu County and Xiangfu District, 80% of the old villages have been demolished and the land leveled, but the proportion of reclaimed land is relatively low, with only a few villages being uniformly planned and cultivated into regular farmland. Puyang County, Taiqian County, and Yuanyang County have achieved a demolition completion rate of around 50%. Yuanyang County has a farmland reclamation rate of about 30%, whereas the old villages in Puyang County and Taiqian County remain largely abandoned after demolition, urgently requiring promotion of land transfer. In Zhongmu County, the scale of the villages being demolished is quite large, and similarly, the scale of the resettlement area construction is also substantial. This has led to significant difficulties in mobilizing residents to move, as well as challenges in securing funding. These factors have impacted the overall progress of relocation and reclamation.

4.4.2. Analysis of Constraining Factors for Relocation and Reclamation

For the villages slated for relocation under the two pilot batches and the Three-Year Plan, 23% have yet to be demolished. Field research has identified several key factors constraining the progress of old village relocation. Firstly, the original production land of the relocated population remains unchanged, and after relocation, some resettlement areas are far from the farmland, increasing the farming radius and making agricultural activities inconvenient. Additionally, some districts and counties face difficulties in advancing resettlement area construction alongside industrial development and employment transfer, making it challenging for relocated residents, particularly the working-age population, to secure stable employment near their new homes. Furthermore, the planning and design of housing in resettlement areas, including accompanying education, healthcare, and municipal infrastructure, have not yet reached the expected levels desired by residents. Lastly, in some villages, compensation payments for the demolition of old houses have not been made, placing a heavy financial burden on the residents, which leads to delays in the demolition process.
Most of the relocated villages in the floodplain area have not completed farmland reclamation, making land transfer significantly more difficult. The main reasons for this challenge are twofold: (1) The farmland in the floodplain area is relatively remote. After residents relocate, the transportation costs for returning to farm these lands are high. Coupled with the historically low development levels and poor infrastructure in the floodplain area, it is challenging for land contractors to carry out large-scale farming. As a result, some villages remain abandoned after demolition without organized farmland reclamation efforts. (2) The development and utilization of land in the Yellow River floodplain must consider regional ecological protection requirements. Agricultural practices are subject to policy limitations, which restrict planting methods. Additionally, the yields from grain crops are low while the costs are high, significantly reducing the enthusiasm of land contractors to reclaim and cultivate the farmland.
We propose three recommendations to better complete the entire process of relocation and reclamation. First, strengthen the coordinated development of resettlement area construction and industrial layout, synchronously promoting the transfer and employment of residents to ensure they can achieve smooth employment close to their new homes. Second, further improve the supporting infrastructure in resettlement areas, including education, healthcare, and municipal services, to meet the expectations of residents. Third, the government should promptly fulfill the compensation for the demolition of old houses to reduce the financial burden on residents, thereby facilitating the progress of old house demolition.

4.5. Advanced Methodology of This Study

The monitoring approach for resident relocation and reclamation proposed in this study represents a further development and application of research on land use type identification and land use change monitoring. Targeting the relocation and reclamation projects in the Yellow River floodplain area, we have developed a recognition framework for three typical states. Based on the generation of high-resolution shortwave infrared bands, this framework utilizes EASI and kNDVI to monitor land use status throughout the entire process of relocation and reclamation. Moreover, we have expanded the recognition objects from pixels to village scale, allowing our research results to provide a more direct understanding of land reclamation at the administrative division level. This serves timely monitoring and management by the government.
More importantly, the technical framework proposed in this study serves as a demonstration for specific land use change monitoring scenarios. The band downscaling fitting technique utilized in this study effectively compensates for the spectral resolution deficiencies of images that have high spatial resolution, significantly expanding the application scenarios of high-resolution data. By integrating remote sensing indices with village-scale land use status identification, this approach greatly reduces the cost of manual interpretation for tasks covering large research areas. Therefore, the research framework, based on the aforementioned contributions, is not only applicable to monitoring tasks for relocation and reclamation but also suitable for any task fundamentally involving land use change monitoring, especially those requiring high spatial resolution. This includes applications such as mine tailings pond restoration projects, ecological migration projects, and more.

5. Conclusions

This paper integrated multiple remote sensing indices with high-resolution imagery to propose a rapid identification method for the progress status of rural resident relocation and reclamation. The innovation in remote sensing value was reflected in the following two aspects:
  • The study improved upon the existing Artificial Surface Index (ASI) by incorporating the Simple Soil Index (SSI), which helped suppress bare soil confusion within construction land. This addressed the issue of distinguishing between idle or temporarily unseeded land and construction land in rural areas.
  • By combining EASI and kNDVI, the study proposed a remote sensing identification method tailored to data characteristics and application scenarios for monitoring the progress of village relocation and reclamation. Applying multiple remote sensing indices to evaluate the effectiveness of relocation and reclamation policies significantly expanded the application scenarios of high-resolution remote sensing data in studies related to resident relocation and reclamation.
The aforementioned technical methods were applied in a study conducted in the Henan floodplain area, and the results demonstrated the following outcomes:
  • The study found that EASI and kNDVI had high separability in identifying the land cover conditions (construction land and farmland) at different stages of resident relocation and reclamation. The overall accuracy of remote sensing identification reached 93.5%. The multiple remote sensing index method combined with optimal threshold determination provided an effective approach for high-resolution land cover status identification.
  • The implementation of resident relocation in the Henan floodplain area achieved relatively good results, with a completion rate of 77% for the two pilot batches and the Three-Year Plan. However, the reclamation rate of farmland after the demolition of old villages was only 23%, and the difficulty of land transfer was significant, with actual progress far behind the requirements of the pilot plans.
  • Due to the spectral and textural similarities, precise identification between construction land and bare soil in rural areas, as well as between uncultivated but vegetated land and reclaimed farmland, posed challenges. For future work, the study proposed introducing multi-temporal remote sensing images to investigate the combination of deep learning and remote sensing indices at different times, aiming to improve the extraction of land cover information from high-resolution remote sensing images. The quantitative impact of climate and socioeconomic factors on the progress of relocation and reclamation projects is also an area we will focus on in the future.

Author Contributions

Conceptualization, H.H. and Y.W.; methodology, H.H., Y.W. and W.Z.; validation, H.H., C.Y. and Y.T.; investigation, H.H., Y.W. and W.Z.; resources, H.H. and Y.T.; writing—original draft preparation, H.H. and Y.W.; writing—review and editing, C.Y. and Y.T.; visualization, H.H. and Y.W.; supervision, C.Y. and Y.T.; funding acquisition, H.H. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Major Project of Chinese High-resolution Earth Observation System, grant number 00-Y30B01-9001-22/23.

Data Availability Statement

The data presented in this study are available on reasonable request from the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
EASIEnhanced Artificial Surface Index
kNDVIkernel Normalized Difference Vegetation Index
SWIRshort-wave infrared
SSISimple Soil Index
WSFWorld Settlement Footprint
POIPoint of Interest

References

  1. Wan, Z. Research on Ecological Management of the Lower Yellow River Beach Area. Yellow River 2018, 40, 1. [Google Scholar]
  2. National Development and Reform Commission. Planning for Relocation of Residents in the Yellow River Bund Areas of Henan Province; National Development and Reform Commission: Beijing, China, 2018.
  3. Notice on Accelerating the Relocation of Poverty Alleviation and the Relocation of Residents in the Yellow River Bund Areas to Demolish the Old and Reclaim the Land; Henan Provincial Department of Natural Resources; Henan Provincial Development and Reform Commission: Zhengzhou, China, 2020.
  4. Xu, X.; Li, B.; Liu, X.; Li, X.; Shi, Q. Mapping annual global land cover changes at a 30 m resolution from 2000 to 2015. Natl. Remote Sens. Bull. 2021, 25, 1896–1916. [Google Scholar]
  5. Xu, H.; Xiao, X.; Qin, Y.; Qiao, Z.; Long, S.; Tang, X.; Liu, L. Annual Maps of Built-Up Land in Guangdong from 1991 to 2020 Based on Landsat Images, Phenology, Deep Learning Algorithms, and Google Earth Engine. Remote Sens. 2022, 14, 3562. [Google Scholar] [CrossRef]
  6. Gong, J.; Liu, C.; Huang, X. Advances in Urban Information Extraction from High-Resolution Remote Sensing Imagery. Sci. China Earth Sci. 2020, 63, 463–475. [Google Scholar] [CrossRef]
  7. Li, J.; Huang, X.; Tu, L.; Zhang, T.; Wang, L. A Review of Building Detection from Very High Resolution Optical Remote Sensing Images. GISci. Remote Sens. 2022, 59, 1199–1225. [Google Scholar] [CrossRef]
  8. Liu, H.; Gong, P. 21st century daily seamless data cube reconstruction and seasonal to annual land cover and land use dynamics mapping-iMap (China)1.0. Natl. Remote Sens. Bull. 2021, 25, 126–147. [Google Scholar] [CrossRef]
  9. Ali, M.; Hasim, A.; Abidin, M. Monitoring the Built-up Area Transformation Using Urban Index and Normalized Difference Built-up Index Analysis. Int. J. Eng. Trans. B Appl. 2019, 32, 647–653. [Google Scholar]
  10. Zhang, Y.; Guo, Y.; Tang, R.; Tang, Z. Progress and trends of application of hyperspectral remote sensing in plant diversity research. Natl. Remote Sens. Bull. 2023, 27, 2467–2483. [Google Scholar] [CrossRef]
  11. Huang, X.; Zhang, L.; Zhu, T. Building Change Detection from Multitemporal High-Resolution Remotely Sensed Images Based on a Morphological Building Index. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2014, 7, 105–115. [Google Scholar] [CrossRef]
  12. Shuai, C.; Sha, J.; Lin, J.; Ji, J.; Zhou, Z.; Gao, S. Spatial difference of the relationship between remote sensing index and land surface temperature under different underlying surfaces. J. Geo-Inf. Sci. 2018, 20, 1657–1666. [Google Scholar]
  13. Zha, Y.; Gao, J.; Ni, S. Use of Normalized Difference Built-up Index in Automatically Mapping Urban Areas from TM Imagery. Int. J. Remote Sens. 2003, 24, 583–594. [Google Scholar] [CrossRef]
  14. Bouhennache, R.; Bouden, T.; Taleb-Ahmed, A.; Cheddad, A. A New Spectral Index for the Extraction of Built-up Land Features from Landsat 8 Satellite Imagery. Geocarto Int. 2019, 34, 1531–1551. [Google Scholar] [CrossRef]
  15. Gluch, R.; Quattrochi, A.; Luvall, J. A multi-scale approach to urban thermal analysis. Remote Sens. Environ. 2006, 104, 123–132. [Google Scholar] [CrossRef]
  16. Powell, S.; Cohen, W.; Yang, Z.; Pierce, J.D.; Alberti, M. Quantification of impervious surface in the Snohomish Water Resources Inventory Area of western Washington from 1972–2006. Remote Sens. Environ. 2008, 112, 1895–1908. [Google Scholar] [CrossRef]
  17. Xu, H. Spatiotemporal dynamics of the bare soil cover in the Hetian basinal area of County Changting, China, during the past 35 years. Acta Ecol. Sin. 2013, 33, 2946–2953. [Google Scholar]
  18. Zhao, Y.; Zhu, Z. ASI: An Artificial Surface Index for Landsat 8 Imagery. Int. J. Appl. Earth Obs. Geoinf. 2022, 107, 102703. [Google Scholar] [CrossRef]
  19. Zhu, W.; Yuan, C.; Tian, Y.; Wang, Y.; Li, L.; Hu, C. A New High-Resolution Rural Built-Up Land Extraction Method Based on Artificial Surface Index with Short-Wave Infrared Downscaling. Remote Sens. 2024, 16, 1126. [Google Scholar] [CrossRef]
  20. Deng, J.; Wang, K.; Shen, Z.; Xu, H. Decision tree algorithm of automatically extracting farmland information from SPOT-5images based on characteristic bands. Trans. Chin. Soc. Agric. Eng. 2004, 20, 145–148. [Google Scholar]
  21. Camps-Valls, G.; Campos-Taberner, M.; Moreno-Martinez, A.; Walther, S.; Duveiller, G.; Cescatti, A.; Mahecha, M.D.; Muñoz-Marí, J.; García-Haro, F.J.; Guanter, L.; et al. A Unified Vegetation Index for Quantifying the Terrestrial Biosphere. Sci. Adv. 2021, 7, eabc7447. [Google Scholar] [CrossRef]
  22. Wang, Q.; Moreno-Martínez, Á.; Muñoz-Marí, J.; Campos-Taberner, M.; Camps-Valls, G. Estimation of Vegetation Traits with Kernel NDVI. ISPRS J. Photogramm. Remote Sens. 2023, 195, 408–417. [Google Scholar] [CrossRef]
  23. Yang, L.; Driscol, J.; Sarigai, S.; Wu, Q.; Chen, H.; Lippitt, C.D. Google Earth Engine and Artificial Intelligence (AI): A Comprehensive Review. Remote Sens. 2022, 14, 3253. [Google Scholar] [CrossRef]
  24. Tian, Y.; Chen, H.; Song, Q.; Zheng, K. A Novel Index for Impervious Surface Area Mapping: Development and Validation. Remote Sens. 2018, 10, 1521. [Google Scholar] [CrossRef]
  25. World Settlement Footprint 2015 (WSF—DLR-EOC). 2020. Available online: https://www.esa.int/ESA_Multimedia/Videos/2021/11/World_Settlement_Footprint (accessed on 1 March 2022).
  26. Xue, B.; Zhao, B.; Li, J. Evaluation and enhancement methods of POI data quality in the context of geographic big data. Acta Geogr. Sin. 2023, 78, 1290–1303. [Google Scholar]
  27. Pan, X.; Zhu, X.; Yang, Y.; Cao, C.; Zhang, X.; Shan, L. Applicability of Downscaling Land Surface Temperature by Using Normalized Difference Sand Index. Sci. Rep. 2018, 8, 9530. [Google Scholar] [CrossRef] [PubMed]
  28. Njuki, S.; Mannaerts, C.; Su, Z. An Improved Approach for Downscaling Coarse-Resolution Thermal Data by Minimizing the Spatial Averaging Biases in Random Forest. Remote Sens. 2020, 12, 3507. [Google Scholar] [CrossRef]
  29. Zhao, W.; Duan, S.; Li, A.; Yin, G. A Practical Method for Reducing Terrain Effect on Land Surface Temperature Using Random Forest Regression. Remote Sens. Environ. 2019, 221, 635–649. [Google Scholar] [CrossRef]
  30. Xu, S. Research on Spatial Downscaling Method of Land Surface Temperature and Its Application; Aerospace Information Research Institute Chinese Academy of Sciences: Beijing, China, 2021. [Google Scholar]
  31. Quan, J.; Zhan, W.; Chen, Y.; Liu, W. Downscaling remotely sensed land surface temperatures: A comparison of typical methods. Natl. Remote Sens. Bull. 2013, 17, 361–373+374–387. [Google Scholar]
  32. Wang, L.; Niu, X.; Wei, B.; Chen, L. Study on Quality Evaluation Methods for Remotely Sensed Images Fusion. Bull. Surv. Mapp. 2015, 77–79. [Google Scholar] [CrossRef]
  33. Li, X.; Chen, X.; Gu, T.; Li, M.; Xu, F. Comparison Study of Fusion Methods of GF-1 PAN and Landsat 8 Muti-spectral Images. J. East China Univ. Technol. (Nat. Sci.) 2017, 40, 376–380+385. [Google Scholar]
Land 14 00401 g001
Figure 1. Satellite and UAV images of different stages in the process of village relocation and reclamation.
Figure 1. Satellite and UAV images of different stages in the process of village relocation and reclamation.
Land 14 00401 g001
Land 14 00401 g002
Figure 2. Schematic diagram of relocation areas in the Yellow River floodplain region ((a) China; (b) part of the Yellow River floodplain area; (c) the study area of this research).
Figure 2. Schematic diagram of relocation areas in the Yellow River floodplain region ((a) China; (b) part of the Yellow River floodplain area; (c) the study area of this research).
Land 14 00401 g002
Land 14 00401 g003
Figure 3. Schematic diagram of the 2015 human activity data and village boundary extraction results.
Figure 3. Schematic diagram of the 2015 human activity data and village boundary extraction results.
Land 14 00401 g003
Land 14 00401 g005
Figure 5. Short-wave infrared image (Fengqiu County): (a) is from Landsat-8 OLI, and (b) is the fitted GaoFen-2 short-wave infrared image.
Figure 5. Short-wave infrared image (Fengqiu County): (a) is from Landsat-8 OLI, and (b) is the fitted GaoFen-2 short-wave infrared image.
Land 14 00401 g005
Land 14 00401 g006
Figure 6. Optimal thresholds for extracting construction land and vegetation using EASI and kNDVI.
Figure 6. Optimal thresholds for extracting construction land and vegetation using EASI and kNDVI.
Land 14 00401 g006
Land 14 00401 g007
Figure 7. Results of relocation and reclamation status identification and overall progress statistics ((a) identification results of village relocation and reclamation status; (b) remote sensing imagery in 2015 (before relocation and reclamation); (c) remote sensing imagery in 2023 (after relocation and reclamation); (d) the proportion of villages undergoing relocation and reclamation).
Figure 7. Results of relocation and reclamation status identification and overall progress statistics ((a) identification results of village relocation and reclamation status; (b) remote sensing imagery in 2015 (before relocation and reclamation); (c) remote sensing imagery in 2023 (after relocation and reclamation); (d) the proportion of villages undergoing relocation and reclamation).
Land 14 00401 g007
Land 14 00401 g008
Figure 8. Comparison of drone imagery from field surveys showing (a) idle land, (b) resident-initiated reclamation, and (c) fully reclaimed land.
Figure 8. Comparison of drone imagery from field surveys showing (a) idle land, (b) resident-initiated reclamation, and (c) fully reclaimed land.
Land 14 00401 g008
Land 14 00401 g009
Figure 9. Schematic diagram of the implementation effects of relocation and reclamation projects based on EASI and kNDVI.
Figure 9. Schematic diagram of the implementation effects of relocation and reclamation projects based on EASI and kNDVI.
Land 14 00401 g009
Land 14 00401 g010
Figure 10. Spatial distribution of relocation and reclamation progress of residential areas in Henan floodplain area.
Figure 10. Spatial distribution of relocation and reclamation progress of residential areas in Henan floodplain area.
Land 14 00401 g010
Table 1. Contrast and information entropy of short-wave infrared images before and after downscaling for three villages.
Table 1. Contrast and information entropy of short-wave infrared images before and after downscaling for three villages.
VillageBandLandsat-8 OLIGaoFen-2
CONHCONH
NanwangzhuangSWIR19385.584.7816,672.464.95
SWIR28286.085.0415,805.285.11
QinghejiSWIR116,695.524.7117,663.115.07
SWIR215,316.94.6916,500.045.2
SanheSWIR114,770.594.6819,399.144.9
SWIR213,809.394.9318,566.055.06
Table 2. Remote sensing identification results of relocation and reclamation status for villages.
Table 2. Remote sensing identification results of relocation and reclamation status for villages.
ProjectNumber of Villages to Be RelocatedNot RelocatedRelocated
TotalUnreclaimed or Semi-ReclaimedReclaimed
Land Use TypeConstruction LandBare Land or Sparse VegetationFarmland
First batch14014212
Second batch25223149
Three-Year Plan2075415311835
Total2465619013456
Table 3. Confusion matrix for EASI identification of construction land.
Table 3. Confusion matrix for EASI identification of construction land.
CategoryPredicted Construction LandPredicted Non-Construction Land
True Construction Land2775
True Non-Construction Land2395
Overall Accuracy (OA): 93.00%, Kappa: 0.82
Commission ErrorOmission ErrorProducer’s AccuracyUser’s Accuracy
Construction Land7.67%1.77%98.23%92.33%
Non-Construction Land5.00%19.49%80.51%95.00%
Table 4. Remote sensing identification accuracyof relocation and reclamation status for villages.
Table 4. Remote sensing identification accuracyof relocation and reclamation status for villages.
Pilot/PlanningRelocation StatusReclamation Status
Remote SensingActualRemote SensingActual
First Batch Pilot, Inspected 13 Villages13131011
Remote Sensing Average Precision: 96.15%100%92.3%
Second Batch Pilot, Inspected 17 Villages151542
Remote Sensing Average Precision: 94.12%100%88.23%
Three-Year Plan, Inspected 52 Villages47452524
Remote Sensing Average Precision: 93.26%96.15%90.38%
Overall Identification Precision: 93.5%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Huang, H.; Wang, Y.; Yuan, C.; Zhu, W.; Tian, Y. Monitoring and Analysis of Relocation and Reclamation of Residential Areas Based on Multiple Remote Sensing Indices. Land 2025, 14, 401. https://doi.org/10.3390/land14020401

AMA Style

Huang H, Wang Y, Yuan C, Zhu W, Tian Y. Monitoring and Analysis of Relocation and Reclamation of Residential Areas Based on Multiple Remote Sensing Indices. Land. 2025; 14(2):401. https://doi.org/10.3390/land14020401

Chicago/Turabian Style

Huang, Huiping, Yingqi Wang, Chao Yuan, Wenlu Zhu, and Yichen Tian. 2025. "Monitoring and Analysis of Relocation and Reclamation of Residential Areas Based on Multiple Remote Sensing Indices" Land 14, no. 2: 401. https://doi.org/10.3390/land14020401

APA Style

Huang, H., Wang, Y., Yuan, C., Zhu, W., & Tian, Y. (2025). Monitoring and Analysis of Relocation and Reclamation of Residential Areas Based on Multiple Remote Sensing Indices. Land, 14(2), 401. https://doi.org/10.3390/land14020401

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop