A Novel Efficient Method for Land Cover Classification in Fragmented Agricultural Landscapes Using Sentinel Satellite Imagery

Abstract

Updated and accurate land cover maps are essential and crucial for sustainable crop production and efficient land management. However, accurate and efficient land cover mapping is still a challenge for agricultural regions with complicated landscapes. This study proposed a novel spectral-phenological based land cover classification (SPLC) method to identify the land cover for fragmented agricultural landscapes, with less requirement of ground truth data. The SPLC method integrated a pixel-based support vector machine (SVM) algorithm for cropland and various non-cropland classification, and a phenology-based automatic decision tree algorithm for identification of various crop types. It was then tested and applied in two typical case areas (i.e., Jiyuan in the upstream and Yonglian in the downstream) of Hetao Irrigation District (Hetao) in the upper Yellow River basin (YRB), northwest China. The field survey sampling data and the regional visual interpretation maps were jointly used to evaluate the accuracy of land cover classification. Results indicated that stable phenological rules can be established for crop identification even with complex planting patterns, and the SPLC method performed well in land cover mapping in case areas. Four high-accuracy land cover maps were produced for Jiyuan in 2020 and 2021, Yonglian in 2021, and Hetao in 2021. The overall accuracies (OA) can reach 0.90–0.94 based on evaluation with abundant ground truth data, and land cover maps agreed well with the visual interpretation maps in space. Overall, the case application validated the applicability and efficiency of the SPLC method in land cover mapping for regions with fragmented agricultural landscapes, and also implied the potential use in other similar regions.

Anisotropic Weighted Total Variation Feature Fusion Network for Remote Sensing Image Denoising

“…Our network consists of four parts: The AWTV-Net module, four SOSB modules, three AuEncoder modules, and three FB modules. The detailed architecture of AWTVF 2 Net is shown in Figure 4.…”

“…The red arrow in Figure 4 represents a convolution layer, used to restore the dimension of the denoised images. AWTVF 2 Net has a powerful feature extraction capability, working on multi-scale feature maps. This nature of AWTVF 2 Net preserves some critical details of the restored remote sensing images.…”

“…Remote sensing imagery has been widely used in various fields, including land use and land cover mapping [1][2][3], forest fire recognition [4], military target reconnaissance [5], and so on. However, due to the irreducible influences of imaging equipment and the processes of image compression, transmission, and storage, the collected images will be distorted by random noise [6].…”

“…Extensive experiments on synthetic noisy remote sensing images and real noisy remote sensing images were conducted, in order to verify the effectiveness of the proposed AWTVF 2 Net. Numerical metrics and visual comparisons indicate that our proposed anisotropic weighted total variation feature fusion network has significantly superior performance over the State-of-The-Art (SoTA) algorithms for remote sensing image noise removal.…”

“…We present the neural network settings and our method's implementation process in Section 4. Section 5 provides the comparative results of our comprehensive experiments, illustrating the competitive effectiveness of the proposed AWTVF 2 Net. Section 6 discusses the limitations of this study and the differences among all the compared methods.…”