摘要
传统地面调查手段虽在马铃薯种植面积的估算研究中被广泛应用,然而其时效性与精确度上的局限性日益凸显。在谷歌地球引擎(GEE)平台对典型地表覆盖物遥感反射特性的研究,经归一化处理后采用随机森林算法实现了马铃薯种植区与非作物区域的精准分割。构建了混淆矩阵,并进行了精度评估与对比分析。结果显示:依托GEE平台进行的吉林省马铃薯种植区分类,分类精度达90.5%,优于传统方法。Kappa系数为0.912,整体图像分类精度良好。本研究成功实现了马铃薯种植区域的大范围精准识别与提取,在实际应用中具有广泛价值与深远意义。
Although traditional ground survey methods are widely used in the estimation of potato planting areas,their limitations in terms of timeliness and accuracy are becoming increasingly prominent.This paper studied remote sensing reflection characteristics of typical surface coverings based on the Google Earth Engine(GEE)platform.After normalization,the paper used a random forest algorithm to achieve accurate segmentation of potato planting areas and non-crop areas,constructed a confusion matrix,and performed accuracy evaluation and comparative analysis.The results show that the classification accuracy of potato planting areas in Jilin Province based on the GEE platform reaches 90.5%,which is better than traditional methods.The Kappa coefficient is 0.912,indicating good overall image classification accuracy.This paper successfully achieves large-scale accurate identification and extraction of potato planting areas,reflecting its broad value and profound significance in practical applications.
作者
周铠文
张凤
周恒毅
ZHOU Kaiwen;ZHANG Feng;ZHOU Hengyi(College of Geographic Science and Tourism,Jilin Normal University,Siping,Jilin 136000,China)
出处
《北京测绘》
2025年第1期90-96,共7页
Beijing Surveying and Mapping
基金
国家级大学生创新创业训练计划(202310203023)
吉林省教育厅科学技术研究项目(JJKH20230502KJ)
吉林省科技发展计划(20230508029RC)。
