摘要
冠层结构与植被覆盖特征共同影响城市森林的降温效应,探究二者降温潜力对改善城市热环境具有重要意义。于夏季日间08:00—18:00采集了5个时间段的无人机热红外和实测地面温度数据,并分别利用无人机激光雷达和多光谱数据提取冠层结构参数与植被指数,以评估不同时段冠层结构与植被覆盖特征对降温效应的贡献。为实现城市森林对夏季日间降温效应的准确预测,筛选最优特征组合,并对比多种机器学习模型的降温预测潜力。研究发现:(1)不同时间段的降温主导因子存在显著差异。高度指标在10:00—12:00时段表现出最强预测能力(0.27<R^(2)<0.35),而覆盖度与开放度指标、林分结构及异质性指标在16:00—18:00时段预测性能最优(0.43<R^(2)<0.70)。此外,植被指数与降温之间存在显著负相关关系;(2)结合冠层结构与植被指数的模型在夏季降温预测中表现更优(0.47<R^(2)<0.66),比采用最优单一变量模型的R^(2)值提高了2%~14%,均方根误差(RMSE,E RMS)值降低了0.02~0.23。基于最优特征变量的机器学习模型由大到小表现为随机森林、支持向量机、多元线性回归、最小k近邻;(3)预测结果表明,研究区域高温时段集中于10:00—14:00(平均气温25.96℃,标准差3.59℃)。研究揭示了冠层结构与植被指数对夏季日间降温的影响及预测潜力,为夏季高温背景下的城市森林管理优化提供了数据支持。
Canopy structure and vegetation coverage characteristics jointly affect the cooling effect of urban forests,and exploring their cooling potential is of great significance for improving the urban thermal environment.From 08:00 to 18:00 in summer daytime,UAV thermal infrared and measured ground temperature data were collected in 5 time periods,and canopy structure parameters and vegetation indices were extracted using UAV lidar and multispectral data,respectively,to evaluate the contribution of canopy structure and vegetation coverage characteristics to the cooling effect in different time periods.To accurately predict the daytime cooling effect of urban forests in summer,the optimal feature combination was screened,and the cooling prediction potential of multiple machine learning models was compared.The results showed that:(1)There were significant differences in the dominant cooling factors in different time periods.The height index showed the strongest predictive ability(0.27<R^(2)<0.35)during 10:00-12:00,while the coverage and openness index,stand structure and heterogeneity index showed the best predictive performance(0.43<R^(2)<0.70)during 16:00-18:00.In addition,there was a significant negative correlation between vegetation index and cooling.(2)The model combining canopy structure and vegetation index performed better in summer cooling prediction(0.47<R^(2)<0.66),with the R^(2)value increased by 2%-14%and the root mean square error(E RMS)value decreased by 0.02-0.23 compared with the optimal single variable model.The machine learning models based on the optimal feature variables were ranked from best to worst as Random Forest,Support Vector Machine,Multiple Linear Regression,and k-Nearest Neighbor.(3)The prediction results showed that the high temperature period in the study area was concentrated in 10:00-14:00(average temperature 25.96℃,standard deviation 3.59℃).This study revealed the influence and prediction potential of canopy structure and vegetation index on daytime cooling in summer,and provided data support for the optimization of urban forest management under the background of summer high temperature.
作者
王蕾
马小婷
张心语
丁梦思
姚允龙
Wang Lei;Ma Xiaoting;Zhang Xinyu;Ding Mengsi;Yao Yunlong(Northeast Forestry University,Harbin 150040,P.R.China;Key Laboratory of Germplasm Resources Development of Cold-Region Landscape Plants and Landscape Ecological Restoration,Heilongjiang Province(Northeast Forestry University))
出处
《东北林业大学学报》
北大核心
2026年第3期115-124,共10页
Journal of Northeast Forestry University
基金
国家自然科学基金项目(42171246)。
关键词
城市森林
冠层结构
植被指数
降温潜力
预测模型
空地数据
Urban forest
Canopy structure
Vegetation index
Cooling potential
Predictive model
Airborne and ground data
