基于循环神经网络的土地利用变化与驱动力分析——以山东日照为例

Analysis on Land Use Change and Driving Forces Based on Recurrent Neural Networks ——Taking Rizhao City in Shandong Province as an Example

摘要:

为深入探讨日照市土地利用/覆盖变化的规律，本文利用循环神经网络模型对其2019—2023年期间土地利用/覆盖进行遥感影像分类，并基于分类结果研究分析了多种驱动因子与土地利用/覆盖变化之间的关系。结果表明，循环神经网络在处理遥感影像数据方面表现出色，5年的测试总体精度均超过92%，Kappa系数均超过0.84；并且在驱动力分析方面，城镇化率、生产总值和一般公共预算支出因素对日照市土地利用/覆盖变化有显著影响，改变土地资源的分配和使用。研究结果不仅揭示了日照市城市化和经济增长对土地利用/覆盖的深远影响，也为未来的土地管理和规划提供了有价值的参考。

Abstract:

In order to deeply study the law of land use/land cover change in Rizhao city, by using the model of recurrent neural network, land use/land cover from 2019 to 2023 has been classified. Based on the classification results, the relationship between various driving factors and land use/cover change have been studied and analyzed. It is showed that the cyclic neural network performs well in processing remote sensing image data. The overall accuracy of the 5 year testing has exceeded 92%, and the Kappa coefficient is over 0.84. Moreover, in terms of driving force analysis, urbanization rate, GDP and general public bud get expenditure factors have a significant impact on land use/land cover change in Rizhao city. It has changed the distribution and use of land resources. The research results not only reveal the farreaching influence of urbanization and economic growth on land use/land cover in Rizhao city, but also provide valuable references for land management and planning in the future.