Object-Based Land Use and Land Cover Change Detection in Multi Temporal Remote-Sensing Images

The changing of land use and land cover (LULC) are both affected by climate and human activity and affect climate, biological diversity, and human well-being. Accurate and timely information about the LULC pattern and change is crucial for land management decision-making, ecosystem monitoring, and urban planning, especially in developing economies undergoing industrialization, urbanization, and globalization. Biodiversity degradation and urban expansion in eastern China are research hot-spots. However, the influence of LULC changes on the region remains largely unexplored. Here, an object-based and multi-temporal image analysis approach was developed to detect how LULC changes during 1985-2015 in the Tiaoxi watershed (Zhejiang province, eastern China) using Landsat TM and OLI data. The main objective of this study is to improve the accuracy of unsupervised change detection from object-based and multi-temporal images. To this end, a total of seven LULC maps are generated with multi-temporal images. A random stratified sample design was used for assessing change detection accuracy. The proposed method achieved an overall accuracy of 91.86%, 92.14%, 92.00%, and 93.86% for 2000, 2005, 2010, and 2015, respectively. Nevertheless, the proposed method, in conjunction with object-oriented and multi-temporal satellite images, offers a robust and flexible approach to LULC changes mapping that helps with emergency response and government management. Urbanization and agriculture efficiency are the main reasons for LULC changes in the region. We anticipate that this freely available data will improve the modeling for surface forcing, provide evidence of changes in LULC, and inform water-management decision-making.


Introduction
Land use and land cover (LULC) changes are important ingredients in global envi- 23 ronmental change [1]. Land cover is an essential climate variable that impacts numerous 24 environmental processes and patterns ranging from albedo and hence climate to zoo- 25 geographic distributions and hence patterns of biodiversity. Such changes are usually 26 caused by human activities (e.g., deforestation, urbanization, agriculture intensification, 27 overgrazing, and subsequent land degradation), however, natural factors can also con- 28 tribute to these changes [2]. People's responses to economic opportunities, as mediated 29 by institutional factors, drive land-cover changes [3]. The rapid development of the 30 economy requires LULC information for the efficient management of the environment 31 and living conditions. Therefore, the time series of legacy land use maps are needed for 32 the quantification of changes [4]. The phenological information for vegetation derived 33 from multi-seasonal imagery is very useful for mapping tree species [5,6], forest cover 34 [7,8], crop types [9,10], bush encroachment [11], grassland [12], and LULC changes 35 [13,14]. 36 Nowadays, the amount and availability of multi-temporal images are experiencing 37 a fast increase. This is due to the increasing number of space missions, the increases 38 in data temporal resolution, as well as free accessible data policy adopted for missions 39 The traditional change detection method is pixel-based image analysis (PBIA). It 48 detects the occurrence of changes based on the comparison of pixels without considera-49 tion of spatial context or membership to real-world objects. Object-based image analysis 50 (OBIA), which operates at the scale of real-world objects rather than pixels, offers a 51 means of analyzing Earth observation (EO) data in a realistic context and integrating 52 associated ancillary information to support real-world applications [19]. Object-based 53 image mapping reduces noise in the original image (i.e., erroneous pixel values, often re-54 ferred to as the "salt and pepper" effect) to characterize the features of interest effectively 55 [20], and these can exploit landscape features to increase the accuracy and usability of 56 EO-derived products [19]. The advantage of object-based classification is that it groups 57 neighboring pixels into meaningful areas according to their spatial and spectral [21]. 58 According to the most recent studies, OBIA methods have been more effective and 59 reliable than the traditional PBIA methods for image processing [22][23][24][25][26].

60
China's urbanization process has followed a unique course and transformed the 61 country since the early 1980s. Tremendous LULC changes have occurred in many coastal 62 regions of China such as the Yangtze River Delta region [27] and the Pearl River Delta 63 region [28]. Taihu Lake is the third-largest freshwater lake in China and serves as a 64 drinking water source for 30 million residents. It is also the region with the most rapid 65 economic development and the most intense land-use change. Tiaoxi River is one of 66 the main rivers connected to Taihu lake and contributes >60% of the source water [29].

67
The ecological environment of the Tiaoxi River basin has been seriously disturbed by 68 anthropogenic activities. The region is more representative of the eastern part of China.  Remote sensing (RS) and Geographic Information System (GIS) are two effective 73 tools for detecting and analyzing land cover and its changes over a certain period through 74 integrating spatial and temporal windows of the study area. To determine changes over 75 time, land cover maps for several different years are needed and resultant analysis helps 76 the respective administrator to understand the current landscape along with changing 77 patterns [30]. It also helps to understand and evaluate past management decisions as 78 well as predict possible effects of their current decisions before their implementation 79 [31]. The objective of this research was to utilize GIS and RS applications to find out   Farmland, orchard, tea garden, surface waters, and urban areas occupy the rest of the 99 watershed [32]. As one of the most active areas in China's economy, the agricultural 100 production structure of this region has gradually changed, and the planting of cash crops 101 has gradually replaced grain crops. In this study, land uses were categorized into the   (Table 1). To distinguish the main crops from the tea garden, multi-temporal images 125 were selected, including spring, summer, autumn, and winter [34]. The digital elevation   The Normalized Difference Vegetation Index (NDVI) [35], the Normalized Difference Water Index (NDWI) [34], and the Normalized Difference Built-up Index (NDBI) [36] are the most commonly used indexes for detection vegetation, water, and built-up. The NDVI, MNDWI, and NDBI are expressed as follows:

Materials and Methods
Where Green is a green band such as TM band 2 and OLI band 3, Red is a red band such 132 as TM band 3 and OLI band 4, Nir is a near-infrared band such as TM band 4 and OLI 133 band 5, and Mir is a mid-infrared band such as TM band 5 and OLI band 6.

Accuracy assessment 151
To know where and when LULC changes occur, the primary source for reference 152 data is Landsat images themselves [39]. High spatial resolution images from Google

153
Earth and ground-truth data can help manual interpretation of the land cover classes [40].

154
Their high spatial resolution helps determine LULC changes at longer time intervals.

155
Ground-truth data is used to evaluate how well the classification represents the real 156 world. A random stratified sample design was used for assessing change detection 157 accuracy [41].  Table 2, Table 3, Table 4, and   Table 7 and Figure 4, respectively.     In contrast, the garden-plot area increased seven-fold, from 218.7 km 2 to 1,555 km 2 .

Spatial distribution of LULC changes 218
The spatial analysis distinguished the same six time periods as found in the tem-

251
The causes of LULC changes in the Tiaoxi watershed included urbanization, re-252 gional population growth, and agriculture efficiency improvement.