机器学习算法和图像空间分辨率对土地覆盖分类影响：以大荔县为例

抖音热门 2025年09月01日 02:33 1 admin

阅读全文链接：http://ijabe.net/article/doi/10.25165/j.ijabe.20251801.7252

机器学习算法和图像空间分辨率对土地覆盖分类的影响：以陕西省大荔县为例

施宇1,2，靳宁3*，毋冰艳1,2，王泽坤4，王仕稳2，于强2,5

(1. 西北农林科技大学资源环境学院，杨凌712100，陕西，中国；

2. 西北农林科技大学水保所黄土高原土壤侵蚀与旱地农业国家重点实验室，杨凌712100，陕西，中国；

3. 山西能源学院，晋中030600，山西，中国；

4. 美国奥本大学机械工程系，阿拉巴马州36849，美国；

5. 中国科学院地理科学与资源研究所陆地水循环及地表过程院重点实验室，北京100101，中国)

摘要：土地利用和土地覆盖（LULC）随着全球人口的快速增长、经济发展和农业活动的扩张发生了剧烈变化。然而，基于卫星数据的分类算法和图像分辨率的不确定性仍需充分研究，特别是在耕地制图中。该研究通过比较四种常用的机器学习算法在LULC制图中的分类表现，探讨了不同模型和空间分辨率图像对分类结果的影响，并与现有的全球土地覆盖数据集进行了对比，以评估最优模型的可靠性。

研究结果表明，随机森林（RF）分类器在总体精度（OA）和Kappa系数方面表现优于支持向量机（SVM）、决策树和人工神经网络（ANN），分别为81.99%和0.78。然而，SVM和ANN在水体类和未利用土地类的分类精度上表现更好。基于五种不同分辨率图像（30 m、16 m、10 m、8 m和2 m）的研究表明，RF模型的分类精度随分辨率的增加呈现先升高后下降的趋势，特别是在8米分辨率的图像上，RF的OA达到82.54%，Kappa系数为0.78，表现最佳，同时在耕地类的OA也达到了0.88。其中地形是决定不同分辨率图像分类性能的主要因素。在使用验证数据对RF算法分类的10 m和30 m分辨率图像进行精度评估时，我们得到了分别为93.59%和94.59%的总体精度，这明显优于基于全球土地覆盖数据集的分类精度。该研究的结果强调了分类算法和图像分辨率是土地制图不确定性的重要来源。获得可靠的土地覆盖制图需要使用适当的分类算法和相匹配的空间分辨率图像。该研究的结果有将助于国家土地监测系统和生态气候机理模型的开发。

关键词：遥感数据；不同空间分辨率；机器学习；土地覆盖；随机森林；不确定性

DOI: 10.25165/j.ijabe.20251801.7252

引用信息: Shi Y, Jin N, Wu B Y, Wang Z K, Wang S W, Yu Q. Effects of machine learning models and spatial resolution on land cover classification accuracy in Dali County, Shaanxi, China. Int J Agric & Biol Eng, 2025; 18(1): 245–259.

Effects of machine learning models and spatial resolution on land cover classification accuracy in Dali County, Shaanxi, China

Yu Shi1,2, Ning Jin3*, Bingyan Wu1,2, Zekun Wang4 , Shiwen Wang2, Qiang Yu2,5

(1. College of Natural Resources and Environment, Northwest A&F University, Yangling 712100, Shaanxi, China;

2. State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Northwest A&F University, Yangling 712100, Shaanxi, China;

3. Department of Resources and Environment, Shanxi Institute of Energy, Jinzhong 030600, Shanxi, China;

4. Department of Mechanical Engineering, Auburn University, AL 36849, USA;

5. Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China)

Abstract: Land use and land cover (LULC) has undergone drastic changes with the rapid growth of the global population, economic development, and the expansion of agricultural activities. However, the uncertainty of classification algorithms and image resolution based on satellite data for land cover mapping, particularly cropland cover mapping, needs to be investigated sufficiently. In this study, the influence of different spatial-resolution images on classification results was explored by comparing the differences between four machine learning algorithms for LULC mapping. The classification results of this model were also compared with existing global land cover datasets to determine whether the model was capable of producing reliable results. According to the results of this study, the random forest (RF) classifier outperformed the support vector machine (SVM), decision tree (DT), and artificial neural network (ANN) with an overall accuracy (OA) and kappa coefficient of 81.99% and 0.78, respectively. However, SVM and ANN showed greater accuracy on the water class and unused land class, respectively. With increasing spatial resolution, RF’s accuracy increased initially and then decreased when classifying images with five different spatial resolutions (30 m, 16 m, 10 m, 8 m, and 2 m). In particular, with an OA of 82.54% and a kappa coefficient of 0.78, RF performed the best on images with 8 m resolution. Additionally, the RF-based image with 8 m resolution produced a higher OA of 0.88 for cropland. Topography is the main factor that determines the classification performance of different-resolution images. The classification accuracies of RF10 m and RF30 m (10 m and 30 m resolution images, respectively, using RF) were higher (OAs of 93.59% and 94.59%, respectively) than those of the global land cover dataset (LC10 m and LC30 m, land cover images with 10 m and 30 m resolution, respectively), whose high-resolution images showed more details of the land cover. The results of this study highlight that classification algorithms and image resolution are the sources of uncertainty for land mapping. Obtaining reliable land cover mapping requires the use of appropriate classification algorithms and spatial resolution. With the seresults, it will be possible to develop a national land monitoring system and basic ecological climate models using LULC.

Keywords: satellite data, different spatial resolutions, machine learning, land use, land cover, random forest, uncertainty

DOI: 10.25165/j.ijabe.20251801.7252

Citation: Shi Y, Jin N, Wu B Y, Wang Z K, Wang S W, Yu Q. Effects of machine learning models and spatial resolution on land cover classification accuracy in Dali County, Shaanxi, China. Int J Agric & Biol Eng, 2025; 18(1): 245–259.