Image interpretation and field investigation combined methods
Human-computer interaction interpretation
The information extraction method in this project was mainly man-machine interactive interpretation to ensure the accuracy of information extraction. Man-machine interactive interpretation is directly on the computer screen, with advanced remote sensing image processing software directly delineated, outlining the high standard of farmland boundaries(Blaschke et al., 2010). Image-based object identification was generally approached in one of two ways. Firstly, region-growing techniques can be employed to adjacent group pixels with similar spectral values into individual objects (Aplin et al., 2008).
The computer screen achieves information extraction can be arbitrarily enlarged or reduced, accurately determining the location or recourse to the border (Zlinsinky et al., 2017).
The "image object" is the primary theoretical unit of object-based image analysis, and it falls within the realm of inquiry midway between the technologically detectable and the practically plausible (Lang, 2008). The knowledge of the human expert who commands the computer to identify specific objects is key to determining the particular objects on the image(Blaschke et al., 2014). Combine the process of image segmentation with categorization based on prior experience. In addition, picture segmentation simplifies the substance of images by reducing the amount of detail present in them. Once evaluated and deemed significant, the picture sections formed by segmentation become image objects (Hussein et al., 2018; Hay et al., 2001). It adopts the combination of route survey and point observation, where the key areas of the field survey are mainly the areas with a low level of interpretation and unclear spots and where only remotely sensed imagery is available, objects can be identified based on (spatial) patches of spectrally similar pixels (Aplin et al., 2008).
Following a segmentation approach, several authors (Piazza et al., 2016; Yan et al., 2007) have claimed that image segmentation is intuitively attractive and that human vision typically divides images into homogenous areas first and then characterizes those areas more carefully. Baoding County, Hebei province, in the North of China, was first processed and analyzed. Later, Dongguan county, Guangdong province, in South China, was interpreted using similar techniques(Fig. 2). By correctly segmenting an image into meaningful items of the land surface and land cover, it is suggested that more significant, logical, and intuitive outcomes would be achieved (Blaschke, 2000; Aplin et al., 2018).
Remote sensing images carry a broad amount of spatial information because they carry spatial information for larger geographical areas. Traditionally, remote sensing was used as a data-generating tool before the raster, or vector GIS's analytical capabilities were applied, making the transition from multi-spectral to thematic data a one-way route (Blaschke et al., 2000; Ruddick et al., 2005).
High-standard farmland was identified using a farmland vector shapefile as a layout for the detection (Fig. 2). Once the target had been identified, the classification process was conducted to delineate the study's area of interest. Upon completion of the interest area, the identification process of image features follows to determine the circumstances of the farmland in the interest area. Furthermore, suspicious and irregular features identified were recorded on a physical form for a statistical compilation to determine the extent of irregularities and features of interest detected during the investigation (Table 4).
Table 4
Remote sensing interpretation of high-standard farmland construction and utilization monitoring
Suspected pattern type
|
Data type and timing
|
Area
|
Interpretation sign
|
Sample image
|
Occupation of arable land
|
GF-2 (2017.2.15)
|
Dongguan 113.987143
22.937671
|
The image shows dark green color, continuous dense, texture features obvious, and more uniform shades of color in the plain area showing occupation on arable land
|
|
Destruction of arable land
|
GF-2 (2017.2.15)
|
Dongguan
113.650157
23.033214
|
The image was grey, a more uniform color, flaky, and continuous. The land has been destroyed and is no longer fertile for cultivation
|
|
Usage conversion of arable land
|
GF-2 (2017.3.7)
|
Baoding
115.346923
38.895033
|
Image color and farmland have larger color differences, the outline of a clear regularity. Land use has been converted
|
|
Unutilized arable land
|
GF-2 (2017.3.7)
|
Baoding
115.165049
38.619324
|
Differences in texture and farmland are large with shades, dark green spots scattered, no plate or stripline
|
|
According to Lizarazo (2014), the quality of the information that may be retrieved from remotely sensed data is determined mainly by the accuracy assessments performed.
The image feature must be properly identified and interpreted, and categorized into accurate homogeneity. The features identified during image analysis correctly classify the identified targets, among which was unutilized land, especially in Hebei province (Table 4). Land conversions were also detected, destructions to farmland were observed, and the presence of buildings on farmland was observed. Differences in texture, color, and shape directed the classification process, and suspected features were noted and then identified.
Detected farmland destructions and field survey
After the preliminary remote sensing interpretation is compiled, a field investigation should be conducted to verify the initial translation results (Tangao et al., 2010). Field investigation and verification give a detailed survey and understanding of the high standards of farmland in the study area. They can be conducted on the whole to verify the reliability of the computer information extraction results and the plaques. It helps the researcher to add missing information and modify inaccurate information.
Thus, farmland conversions were identified and verified in Hebei province Baoding Area interpreted image and ground verification image (Fig. 3).
Field survey methods and requirements: Before the field verification work, formulate a field verification route based on the indoor interpretation situation, select the field verification points, and conduct field verification work on this basis. Field survey spot total requirements should be equal to the interpretation of the total number of suspected patches. The targeted areas and ground verification in Hebei province show that illegal farmland occupation and conversions that were identified and registered were mainly due to domestic purposes. The presence of a domestic compound in the middle of the farmland shows that the local community's residential needs necessitated the conversion (Fig. 3a). (Fig. 3b), the occupation detected was informal residential shacks, chicken scoops, and domestic buildings, which were once again constructed on the farmland. Furthermore, the building identified were for domestic purposes and on a minimal scale (Fig. 3c). The damage observed was not extensive and did not pose a great environmental risk.
There is a different and contrasting phenomenon (Fig. 4) compared to Hebei province (Fig. 3), where farmland detected was mostly domestically motivated.
The south of china in the targeted areas indicated that detections registered were rather commercial and on a larger scale. As shown in (Fig. 4a), the farmland has been transformed into high-rise residential apartments which cover a large area of the farmland. The number of apartment blocks counted was 7 in total, as can also be observed in the image. (Fig. 4b) the farmland has been turned factory, and high-rise walls were installed to block access. Farmland has been destroyed for sand mining, an indication of a boom in the construction industry, which has negative environmental implications in the long run (Fig. 4c). The farmland was still being utilized in Guangdong provinces (Fig. 4d), and very advanced methods were deployed to maximize the available remaining arable farmland. The build-up area encroaching on the farmland was one of the regular findings, as more build-up structures were discovered in the areas that were classified as high-standard farmland (Fig. 5).
Whereby yellow represents the farmland build-up structures in the classified image, green represents high-standard farmland, and blue for farmland irrigation, which has been reduced. The conversion of farmland into other economic activities such as residential apartment blocks, office complexes, industrial warehouses, and factories was observed.
Farmland Land use and Land cover classification map of the GF-2 image in Guangdong provinces (Fig. 5b). The red color represents bare soil, blue for water, green for cultivated farmland, yellow for build-up areas, and black for roads. Arable farming land has been destroyed and is now used to supply sand for construction (Fig. 5b). Red shows that part of the farmland can no longer be used for cultivation due to the destruction, the topsoil has been removed, and the area is completely exposed.
A statistical compilation of farmland destruction
Hebei province farmland destruction statistical summary
North China's economy is relatively traditional; therefore, the evidence shows that the destruction occurred due to domestic activities rather than commercial purposes. Therefore most of the arable land is still available and can be classified as high-standard Farmland. In Mancheng County farmland construction projects, only 0.3464% of the total area of 45876.53 M2 has been taken up and transformed into residential shakes on the farm, and illegal activities destroyed about 0.7%. In Baoding New Urban Jiangcheng basic farmland finishing, only about 1.3889% was taken up from an area of 208476.33 M2.
Many small towns in Hebei province showed little evidence of destruction. In Mancheng full town and the South Korean village, the basic farmland finishing project was destroyed from an area of 43089.68 M2, and only 0.3072% was found not to meet the requirements, which is significantly low and not considered high risk. Dingzhou City, Zao Zaozhen high village, and other (21) villages, the high-standard basic farmland construction soil were found to have the highest destruction in Hebei province at 13.2513% from an area measuring 6475338.35 M2. A significant number of the villages were moderate to low impact destruction, meaning most of the farms assessed complied with the regulations.
Object-based image analysis (OBIA) accuracy was considerably higher, particularly for the Buildings, Roads/Railroad, and agriculture classes (MacFaden et al., 2012). This was attributed to the addition of high-quality thematic vector datasets for these characteristics. Outlining challenges, especially in Hebei province, satellite images showed evidence of destruction to farmland. However, the identified objects and features required ground visits to establish the current land use land cover change properly. Illegal farmland destructions were classified into four categories: Taken up, Destroyed, Converted, and Shortage. (Table 5) gives a statistical compilation of how the north of china's high-standard farmland was detected from various villages and towns within the targeted study area.
Table 5
Baoding high-standard farmland statistical usage summary results (unit: square meters)
Name
|
Area
|
Status
|
Problem area
|
Problem area %
|
Mancheng County 100,000 mu of high-standard basic farmland construction projects
|
64942737.93
|
Taken up
|
224931.13
|
0.3464%
|
Destroyed
|
|
|
Converted
|
28964.02
|
0.0446%
|
Shortage
|
45876.53
|
0.0706%
|
Baoding New Urban Jiangcheng basic farmland finishing
|
15010404.35
|
Taken up
|
208476.33
|
1.3889%
|
Destroyed
|
|
|
Converted
|
5005.94
|
0.0333%
|
Shortage
|
2879.27
|
0.0192%
|
Mancheng full town, South Korea village basic farmland finishing project
|
14025787.40
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
43089.68
|
0.3072%
|
Dingzhou City, Zao Zaozhen high on the village and other (21) village high-standard basic farmland construction soil
|
48865681.15
|
Taken up
|
6475338.35
|
13.2513%
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
28208.73
|
0.0577%
|
Dingzhou City breeze shops and stay in the town of high standards of basic farmland construction land improvement project
|
25337988.80
|
Taken up
|
1429.21
|
0.0056%
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
|
|
Baoding North District Han Zhuang Township, Hundred House Township, East Jin Zhuang basic farmland finishing
|
4073182.92
|
Taken up
|
6781.70
|
0.1665%
|
Destroyed
|
|
|
Converted
|
483440.34
|
11.8689%
|
Shortage
|
85401.1300
|
2.0967%
|
Qing yuan County, South Camp, Shao Zhuang village basic farmland finishing
|
4736968.87
|
Taken up
|
71777.05
|
1.5153%
|
Destroyed
|
|
|
Converted
|
3120.57
|
0.0659%
|
Shortage
|
|
|
Guangdong provinces farmland destruction statistical summary
The illegal establishment of high-standard farmland is derived from economic interests. Many people around the world in recent years have preferred to live in urban areas due to the economic benefits that come with especially people in the developing world (PAI, 2011), and this has led to a decrease in available agricultural land, pushing most people to overcrowding conditions and illegal land occupation.
Table 6
Dongguan area high-standard farmland usage statistical results (unit: square meters)
Summary
|
Area
|
Status
|
Problem map area
|
Problem area ratio
|
Dalingshan Town, Dongguan City, 2012 high-standard basic farmland construction projects
|
1312668.6757
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
31001.2869
|
2.3617%
|
Dongguan Dao Zhen 2012 high-standard basic farmland construction projects
|
1502003.3790
|
Taken up
|
41137.6656
|
2.7389%
|
Destroyed
|
|
|
Converted
|
15686.7426
|
1.0444%
|
Shortage
|
|
|
Shenzhen Guangming North land development and consolidation of the arable land project (03 − 2,03–7)
|
1211744.0877
|
Taken up
|
|
|
Destroyed
|
7172.3728
|
0.5919%
|
Converted
|
|
|
Shortage
|
237841.9601
|
19.6281%
|
Shenzhen Guangming North land development and consolidation of the arable land project (03–11)
|
38267.1159
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
15559.5069
|
40.6603%
|
Shenzhen Guangming North land development and finishing supplementary farmland project second phase
|
254356.1829
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
3833.1275
|
1.5070%
|
Shenzhen Guangming North land development and consolidation of supplementary farmland project (03–5,8)
|
342331.6370
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
shortage
|
26441.6698
|
7.7240%
|
Shenzhen Guangming North land development and consolidation of the arable land project (03-3b)
|
126771.5926
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
38785.3248
|
30.5946%
|
Shenzhen Guangming North basic farmland improvement project(03–9)
|
40695.0706
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
|
|
Shenzhen Guangming North land development and consolidation of supplementary farmland project (03-3c, 4)
|
346375.5651
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
|
|
Shenzhen Guangming North land development and consolidation of arable land projects(03-3a)
|
705654.5862
|
Taken up
|
|
|
Destroyed
|
|
|
Converted
|
|
|
shortage
|
71205.4286
|
10.0907%
|
Changping Town, Dongguan City, 2012 high-standard basic farmland construction projects
|
2171211.7689
|
Taken up
|
7410.4357
|
0.3413%
|
Destroyed
|
3395.2241
|
0.1564%
|
Converted
|
14113.6866
|
0.6500%
|
Shortage
|
146159.6922
|
6.7317%
|
Dongguan Dongkeng 2012 high-standard basic farmland construction project
|
690920.9237
|
Taken up
|
3279.3598
|
0.4746%
|
Destroyed
|
5009.1106
|
0.7250%
|
Converted
|
|
|
Shortage
|
|
|
Shatian Town, Dongguan City, 2012 high-standard basic farmland construction projects
|
1307123.3494
|
Taken up
|
35088.3795
|
2.6844%
|
Destroyed
|
|
|
Converted
|
|
|
Shortage
|
|
|
For example, in the more developed economy of Guangdong province, mostly illegal activities are driven by economic interests, and a higher percentage of land is fast transformed for economic activities like the construction of real estate on previously classified farmland. In Dalingshan Town, Dongguan City, of a 31001.2869 M2, 2.3617% was destroyed, which is far less destruction in comparison to other towns in Guangdong provinces, Shenzhen Guangming Northland development, and consolidation of arable land project, from an area of 237841.9601 M2, 19.6281% was found to been destroyed which is quite a significant loss of arable land. However, Shenzhen Guangming Northland development and consolidation of the arable land project (03-3b), 38785.3248 M2, 30.5946% loss of arable land Shenzhen Guangming Northland development and consolidation of arable land project, 15559.5069 M2, 40.6603% of arable land destroyed or transformed into residential properties. There is evidence of up to 40% loss of land in Shenzhen because the city has transformed from a rural agricultural economy to becoming the technological hub of China. Evidence from (Table 6) is a statistical summary of farmland destruction in different selected towns and villages in Guangdong province.
Arable land availability in China
By the early 1960s, 36.42% of the land had been converted into agricultural land, and China saw an increase in agricultural land due to friendly central government policies aimed at achieving food security for the country. The trend continued to rise steadily; in 1977, there was 43.86% already available. Rapid urbanization and the expanding needs of China's industrial and agricultural economies have strained the country's limited land resources. It can be observed that from the mid-90s, China's agricultural land expansion was slowing down (Fig. 6), and it had only increased by 1%, which at its highest record in 2009 was 56.13%. From 2009 it can be observed that land was being lost slowly. By 2020 only 56.06 of land could be classified as Agricultural land in China. The central government will acknowledge this threat. In response, "red lines" have been established to prevent industrial development on farms and to preserve natural forests, wetlands, and endangered species habitats from human development.
As a result, different climatic conditions determine the soil type and topography. Precipitation and temperature (Fig. 7) determine the type of agricultural activities, such as the kinds of crops and methods that farmers cultivate. It is important to note that China is a large, vast country with contrasting climatic conditions. Wheat, barley, and other grains are the most widely grown crops in north China because of their arid continental environment. A temperate continental climate and a temperate monsoon climate predominate in the Hebei province area. The average summer temperature is 20°C, while the winter temperature is below 0°C. Temperature fluctuations mark four seasons. July is the wettest month, while January is the coldest month. There is not much precipitation throughout the year, and the seasonal distribution is unequal, with summer being the season with the most precipitation. In China, there are several different climate types. From the southeast to the northwest, the precipitation lessens. Southeast coastal cities often get around 2,000 mm. In regions along the Huai River further north, the annual rainfall drops to around 880 mm. Annual rainfall along the Yellow River's northeastern tributaries is about 600 mm. Typically, the rainy season begins in June and lasts until August. Therefore, summers are dryer, and winters are wetter. Understanding climate trends is crucial for high-standard farmlands to maintain high output.
China's Precipitation image shows that the southern provinces receive a great amount of precipitation. However, the results of this research show that the Guangdong province is losing more land due to rapid industrial development. China's temperature indicates the Guangdong provinces region is warmer compared to the northern region of Hebei, which indicates better favorable farming conditions in the south. However, the south is vulnerable to climatic-induced natural disasters to its location. China's arable land from 2010 to 2020 shows a steady decline(Fig. 8), exacerbated by industrialization.
The concerning trend reveals China's rapidly declining arable land availability (Fig. 8). In ten years, the arable percentage has reduced from 56.13% in 2010 to 56.06% in 2020. There is a threat to high-standard farmland in China considering the increase of industrialization in the Guangdong provinces regions for domestic purposes in the north. Regardless of the circumstances and the causes thereof, it is clearly evident that arable land is under threat which is significant for three reasons high population pressure, industrialization, climate, and topography. As seen in (Table 6), most land losses occur in Guangdong provinces, where farmland is converted into economic generation uses such as warehouses, residential blocks, and factories. The losses are caused by industrial expansion and demand.
Furthermore, arable land has been converted into other economic and domestic activities, which further puts pressure on the already scarce arable land resources, as shown in (Figs. 3 and 4). Furthermore, a significant amount of arable land (Li et al., 2018) has been transferred to cooperatives seconded by firms and, thirdly, into other economic activities. However, what is significant is that agricultural land has been transferred into other sectors of the economy, which threatens the sustainability and availability of arable land.
In order to convert underutilized farmland, land with low productivity, or property classed as waste in rural areas. Farmers should be able to sell or lease their land to those with higher agricultural standards. Since the 1980s, the transfer of rural land has been legal, and it has since garnered the backing of the federal government. The Land Contract Law of China formalized the land transfer procedure in 2002. From 2007 to 2017, the percentage of arable land transferred increased dramatically, from 5 to 36.5% (Li et al., 2018).