西藏青稞和小麦种植空间分布数据集（2022）

青稞和小麦是西藏最重要的作物种植类型。然而，目前还没有西藏青稞和小麦种植空间分布数据，极大地限制了西藏农业种植管理工作。本项目为了克服西藏地区作物样本数量和质量不足，以及缺少高精度大尺度分类制图算法的问题，提出了一种 Mamba 模型和对比学习相结合的农作物时序分类模型（SCPMamba，Supervised Contrastive learning with Prototype memory in Mamba）。其中， Mamba模型主要用于提取不同农作物的时序趋势特征，而对比学习主要用于增强分类模型的空间迁移性。项目进一步对西藏青稞和小麦主要产区进行了制图，得到了较高精度的农作物制图成果，分类的总体精度OA为0.903， F1分数为0.902，Kappa 系数为 0.829。该数据产品将为西藏的农业种植管理提供重要的参考。

Highland barley and wheat are the two most important cultivated crop types in Tibet. However, no spatial distribution data of highland barley and wheat planting in Tibet is currently available, which greatly hinders agricultural planting management efforts in the region. To address the issues of insufficient quantity and quality of crop samples, as well as the lack of high-precision large-scale classification and mapping algorithms in Tibet, this study proposes a crop temporal classification model (SCPMamba, Supervised Contrastive Learning with Prototype Memory in Mamba) that integrates the Mamba model and contrastive learning. Specifically, the Mamba model is primarily employed to extract temporal trend features of different crops, while contrastive learning is used to enhance the spatial transferability of the classification model. Furthermore, this study conducted mapping for the major producing areas of highland barley and wheat in Tibet, and obtained high-precision crop mapping results. The overall classification accuracy (OA) reached 0.903, the F1-score was 0.902, and the Kappa coefficient was 0.829. This dataset product will provide critical references for agricultural planting management in Tibet.

搜集汇总

背景与挑战

背景概述

该数据集提供了2022年西藏青稞和小麦种植的空间分布信息，空间分辨率为10m至100m，数据大小为37.64 MB。数据集采用了一种结合Mamba模型和对比学习的农作物时序分类模型（SCPMamba），分类总体精度OA为0.903，F1分数为0.902，Kappa系数为0.829，为西藏农业种植管理提供了重要参考。