Search

Corn and soybeans play pivotal roles in the agricultural landscape of the United States, and accurately delineating their cultivation areas is indispensable for ensuring food security and addressing hunger-related challenges. Traditional methods for crop mapping are both labor-intensive and time-consuming. Fortunately, the advent of high-resolution imagery, exemplified by Sentinel-2A (S2A), has opened avenues for precise identification of these crops at a field scale, with the added advantage of cloud computing. This paper presents an innovative algorithm designed for large-scale mapping of corn and soybean planting areas on the Google Cloud Engine, drawing inspiration from symmetrical theory. The proposed methodology encompasses several sequential steps. First, S2A data undergo processing incorporating phenological information and spectral characteristics. Subsequently, texture features derived from the grayscale matrix are synergistically integrated with spectral features in the first step. To enhance algorithmic efficiency, the third step involves a feature importance analysis, facilitating the retention of influential bands while eliminating redundant features. The ensuing phase employs three base classifiers for feature training, and the final result maps are generated through a collective voting mechanism based on the classification results from the three classifiers. Validation of the proposed algorithm was conducted in two distinct research areas: Ford in Illinois and White in Indiana, showcasing its commendable classification capabilities for these crops. The experiments underscore the potential of this method for large-scale mapping of crop areas through the integration of cloud computing and high-resolution imagery. [ABSTRACT FROM AUTHOR]

This paper presents an R-based approach to mapping dynamics of the flooded areas in the Inner Niger Delta (IND), Mali, using time series analysis of Landsat 8–9 satellite images. As the largest inland wetland in West Africa, the habitats of IND offers high potential for biodiversity of the flood-dependent eco systems. IND is one of the most productive areas in West Africa. Mapping flooded areas based on satellite images enables to provide strategies for land management and rice planting and modelling vegetation types of IND. Our approach is based on using libraries of R programming language for processing six Landsat images, and each image was taken on November from 2013 to 2022. By capturing spatial and temporal structures of the satellite images on 2013, 2015, 2018, 2020, 2021 and 2022, the remote sensing data are combined to yield estimates of landscape dynamics that is temporally coherent, while helping to analyse fluctuations of spatial extent in fluvial wetlands caused by the hydrological processes of seasonal flooding. Further, by allowing packages of R to support image processing, an approach to mapping vegetation by NDVI, SAVI and EVI indices and visualising changes in distribution of different land cover classes over time is realised. In this context, processing Earth observation data by advanced scripting tools of R language provides new insights into complex interlace of climate-hydrological processes and vegetation responses. Our study contributes to the sustainable management of natural resources and improving knowledge on the functioning of IND ecosystems in Mali, West Africa. [ABSTRACT FROM AUTHOR]

Aiming at few-shot classification in the field of hyperspectral remote sensing images, this paper proposes a classification method based on cross-scene adaptive learning. First, based on the unsupervised domain adaptive technology, cross-scene knowledge transfer learning is carried out to reduce the differences between source scene and target scene. At the same time, depthwise over-parameterized convolution is used in the deep embedding model to improve the convergence speed and feature extraction ability. Second, two symmetrical subnetworks are designed in the model to further reduce the differences between source scene and target scene. Then, Manhattan distance is learned in the Manhattan metric space in order to reduce the computational cost of the model. Finally, the weighted K-nearest neighbor is introduced for classification, in which the weighted Manhattan metric distance is assigned to the clustered samples to improve the processing ability to the imbalanced hyperspectral image data. The effectiveness of the proposed algorithm is verified on the Pavia and Indiana hyperspectral dataset. The overall classification accuracy is 90.90% and 65.01%. Compared with six other kinds of hyperspectral image classification methods, the proposed cross-scene method has better classification accuracy. [ABSTRACT FROM AUTHOR]