XÁC ĐỊNH PHẠM VI CHE PHỦ CỦA CÂY DỪA NƯỚC TẠI KHU DỰ TRỮ SINH QUYỂN CÙ LAO CHÀM-HỘI AN BẰNG DỮ LIỆU SENTINEL-2A

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Vũ Khắc Đặng

Ngôn ngữ: vie

Ký hiệu phân loại:

Thông tin xuất bản: Tạp chí Khoa học - Trường Đại học Sư phạm Hà Nội: Khoa học Tự nhiên, 2023

Mô tả vật lý: tr.181

Bộ sưu tập: Báo, Tạp chí

ID: 341573

Thêm vào giỏ Liên kết toàn văn

Dừa nước phát triển ở cửa sông Thu Bồn. Dừa nước góp phần bảo tồn hệ sinh thái tự nhiên, đa dạng sinh học của Khu dự trữ sinh quyển thế giới Cù Lao Chàm - Hội An, cũng như phát triển kinh tế - xã hội của thành phố Hội An. Tuy nhiên, trong những năm vừa qua phạm vi che phủ của nó bị giảm dần do các hoạt động kinh tế - xã hội của con người. Nghiên cứu này sử dụng tất cả các kênh ảnh gốc của vệ tinh Sentinel-2A nhờ khả năng tăng cường độ phân giải của thuật toán Sen2Res và kết hợp với một số chỉ số phổ được tính toán thông qua tỉ số kênh để phân loại lớp phủ mặt đất và dừa nước đã được tách ra khỏi các loại lớp phủ mặt đất khác bằng tiếp cận phân loại ảnh dựa trên đối tượng với thuật toán Máy vectơ hỗ trợ (SVMs) do sự hiệu quả và tính linh hoạt của thuật toán này. Phân tích thành phần chính (PCA) đã được áp dụng cho tất cả các kênh ảnh để loại bỏ các thành phần dư thừa và nhiễu trong dữ liệu trước khi thực hiện phân loại. Bản đồ thu được cho thấy dừa nước phân bố chủ yếu ở tả ngạn sông Thu Bồn với tổng diện tích là 147 ha vào năm 2019. Việc kiểm chứng được thực hiện bằng cách so sánh kết quả phân loại với các mẫu ngẫu nhiên lấy thông qua giải đoán mắt thường từ ảnh Google Earth có độ phân giải cao. Việc đánh giá sai số cho thấy độ chính xác tổng thể đạt 88,67% và hệ số Kappa là 0,84. Mặc dù còn một số sai khác liên quan tới độ phân giải của ảnh Sentinel-2, nhưng các kết quả thu được như vậy có thể hỗ trợ các cơ quan chức năng trong việc ra quyết định thông qua việc cung cấp bản đồ để đề xuất các biện pháp bảo tồn cây dừa nước ở khu vực ven biển này.The mangrove palm grows at the Thu Bon river’s mouth. Mangrove palm has contributed to the conservation of natural ecosystems, and the biological diversity of Cu Lao Cham – Hoi An Biosphere Reserve, as well as the socio-economic development of Hoi An city. However, in recent year, its coverage has been gradually reduced due to human socio-economic activities. This study has used not only the raw bands of the Sentinel-2A image but also some spectral indices have been calculated via the band ratio for classifying land covers. The principal component analysis (PCA) was applied to these all bands to eliminate redundant and noise contributions in data. Mangrove palm was separated from other land cover categories with an object-based image classification approach using Support Vector Machines (SVMs) algorithm because of its high efficiency, performance, and flexibility. The obtained map indicates that mangrove palm principally distributes on the left bank of Thu Bon River with a total area of 147 ha by 2019. The validation, was realized by comparing the classification results with random samples through the visual interpretation of Google Earth high resolution image. The error assessment shows that the overall accuracy is about 88.67% and the Kappa coefficient reaches 0.84. Although there are some differences related to the spatial resolution of Sentinel-2 images, such obtained results may support authorities in decision-making by providing maps able to suggest necessary responses for the conservation of mangrove palms this coastal area.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR