基于随机森林算法的近地表气温遥感反演研究

doi:10.3724/SP.J.1047.2017.00390

摘要/Abstract

摘要：

近地表气温是城市热环境的重要表征,是改变和影响城区气候的重要因素。为获得空间上连续的近地表气温,本文以北京市为研究区,利用Landsat5/TM数据计算分别得到地表温度、归一化植被指数、改进的归一化差异水体指数、地表反照率、不透水面盖度,并结合气象站点气温和高程作为输入参数建立随机森林模型反演近地表气温。结果表明,随机森林反演的近地表气温平均绝对误差（MAE）为0.80 ℃,均方根误差（RMSE）为1.06 ℃,与传统多元线性气温回归方法相比,平均绝对误差（MAE）和均方根误差（RMSE）分别提高0.06 ℃和0.09 ℃。研究表明,利用随机森林模型反演近地表气温是可行的,并且具有一定的优越性。此外,对随机森林模型的输入参数进行重要性分析,地表温度对气温反演模型的影响最大,其次为高程。

关键词: 遥感, 地表温度, 近地表气温, 随机森林

Abstract:

Near-surface air temperature is an important symbol of urban thermal environment, which is also an important factor affecting and changing the climate of the city. The data of near-surface air temperature is often in absence because the number of meteorological stations is few. In order to obtain spatial continuous near surface air temperature data, this study takes Beijing city as the research area, using Landsat5/TM data to retrieve land surface temperature, normalized difference vegetation index, modified normalized difference water index, albedo and impervious surface cover. These are combined with the meteorological station temperature and DEM as the input parameters into random forest regression model to retrieve near surface air temperature. In this study, land surface temperature was retrieved by single-channel algorithm which was proposed by Jiménez-Muoz in 2003. The imperious surface cover was calculated by the linear spectral unmixing method and Vegetation-Impervious surface-Soil (VIS) model. The random forest is one of the most effective methods of classification and it runs by constructing multiple decision tree while training and outputting the class. This study uses the R language which is a free software environment for statistical computing and graphics to achieve random forest. The results show that the random forest method has good applicability in the near surface temperature retrieval. The mean absolute error (MAE) and root mean square error (RMSE) of the random forest method are 0.80 and 1.07, respectively. Compared with the ordinary regression model, the MAE and (RMSE) accuracy increased by 0.06 and 0.09. Using R language to analyze the importance of variables, land surface temperature has the greatest influence on the results. The increase in Mean Square Error of land surface temperature is 14% and the increase in node purity of land surface temperature is 241.36%.

Key words: remote sensing, land surface temperature, near surface air temperature, random forest

白琳, 徐永明, 何苗, 李宁. 基于随机森林算法的近地表气温遥感反演研究[J]. 地球信息科学学报, 2017, 19(3): 390-397.DOI:10.3724/SP.J.1047.2017.00390

BAI Lin,XU Yongming,HE Miao,LI Ning. Remote Sensing Inversion of Near Surface Air Temperature Based on Random Forest[J]. Journal of Geo-information Science, 2017, 19(3): 390-397.DOI:10.3724/SP.J.1047.2017.00390

图/表 8

图1

图2

图3

表1

参考文献 26

[1]	Mao K B, Tang H J, Wang X F, et al.Near-surface air temperature estimation from ASTER data based on neural network algorithm[J]. International Journal of Remote Sensing, 2008,29(20):6021-6028. doi: 10.1080/01431160802192160
[2]	齐述华,王军邦,张庆员,等.利用MODIS遥感影像获取近地层气温的方法研究[J].遥感学报,2005,9(5):570-575. doi: 10.3321/j.issn:1007-4619.2005.05.009
	[Qi S H, Wang J B, Zhang Q Y, et al.Study on the estimation of air temperature from MODIS data[J]. Journal of Remote Sensing, 2005,9(5):570-575. ] doi: 10.3321/j.issn:1007-4619.2005.05.009
[3]	祝善友,张桂欣.近地表气温遥感反演研究进展[J].地球科学进展,2011,26(7):724-730. doi: 10.11867/j.issn.1001-8166.2011.07.0724
	[Zhu S Y, Zhang G X.Progress in near surface air temperature retrieved by remote sensing technonlgy[J]. Advances in Earth Science, 2011,26(7):724-730. ] doi: 10.11867/j.issn.1001-8166.2011.07.0724
[4]	曲培青,施润和,刘剋,等.基于遥感和BP人工神经网络的城乡气象站点划分分析[J].地球信息科学学报,2010,12(5):726-732.
	[Qu P Q, Shi R H, Liu K, et al.Discrimination of urban and rural meteorological stations based on remote sensing and BP artificial neural network[J]. Journal of Geo-Information Science, 2010,12(5):726-732. ]
[5]	徐永明,覃志豪,万洪秀. 热红外遥感反演近地层气温的研究进展[J].国土资源遥感,2011(1):9-14. doi: 10.6046/gtzyyg.2011.01.02
	[Xu Y M, Qin Z H, Wan H X.Advances in the study of near surface air surface air temperature retrieval from thermal remote sensing[J]. Remote Sensing for Land & Resources, 2011,1:9-14. ] doi: 10.6046/gtzyyg.2011.01.02
[6]	Zhao C, Nan Z, Cheng G.Methods for modelling of temporal and spatial distribution of air temperature at landscape scale in the southern Qilian mountains, China[J]. Ecological Modelling, 2005,189(s1-2):209-220. doi: 10.1016/j.ecolmodel.2005.03.016
[7]	Cresswell M P, Morse A P, Thomson M C,et al.Estimating surface air temperatures from Meteosat land surface temperatures using an empirical solar zenith angle model[J]. International Journal of Remote Sensing,1999,20(6):1125-1132. doi: 10.1080/014311699212885
[8]	曲培青,施润和,刘朝顺,等. 基于MODIS地表参数产品和地理数据的近地层气温估算方法评价——以安徽省为例[J].国土资源遥感,2011(4):78-82. doi: 10.6046/gtzyyg.2011.04.15
	[Qu P Q, Shi R H, Liu C S, et al.The evaluation of MODIS data and geographic data for estimating near surface air temperature: A case of Anhui Pravince[J]. Remote Sensing for Land & Resources, 2011,4:78-82. ] doi: 10.6046/gtzyyg.2011.04.15
[9]	Prihodko L, Goward S N.Estimation of air temperature from remotely sensed surface observations[J]. Remote Sensing of Environment, 1997,60(3):335-346. doi: 10.1016/S0034-4257(96)00216-7
[10]	Vancutsem C, Ceccato P, Dinku T, et al.Evaluation of MODIS land surface temperature data to estimate air temperature in different ecosystems over Africa[J]. Remote Sensing of Environment, 2010,4(2):449-465. doi: 10.1016/j.rse.2009.10.002
[11]	Stisen S, Sandholt I, Norgaard A, et al.Estimation of diurnal air temperature using MSG SEVIRI data in West Africa[J]. Remote Sensing of Environment, 2007,10(2):262-274. doi: 10.1016/j.rse.2007.02.025
[12]	徐永明,覃志豪,沈艳.基于MODIS数据的长江三角洲地区近地表气温遥感反演[J].农业工程学报,2011,27(9):63-68. doi: 10.3969/j.issn.1002-6819.2011.09.013
	[Xu Y M, Qin Z H, Shen Y.Estimation of near surface air temperature from MODIS data in the Yangtze River Delta[J]. Transactions of the CSAE, 2011,27(9):63-68. ] doi: 10.3969/j.issn.1002-6819.2011.09.013
[13]	Sun Y, Wang J F, Zhang R H, et al.Air temperature retrieval from remote sensing data based on thermodynamics[J]. Theoretical & Applied Climatology, 2005,80(1):37-48. doi: 10.1007/s00704-004-0079-y
[14]	Breiman L. Random Forests[J]. Machine Learning,2001,45(1):5-32. doi: 10.1023/A:1010933404324
[15]	方匡南,吴见彬,朱建平,等.随机森林方法研究综述[J].统计与信息论坛,2011,26(3):32-38. doi: 10.3969/j.issn.1007-3116.2011.03.006
	[Fang K N, Wu J B, Zhu J P, et al.A review of technologies on random forests[J]. Statistics&Information Forum, 2011,26(3):32-38. ] doi: 10.3969/j.issn.1007-3116.2011.03.006
[16]	张雷,王琳琳,张旭东,等.随机森林算法基本思想及其在生态学中的应用——以云南松分布模拟为例[J].生态学报,2014,34(3):650-659. doi: 10.5846/stxb201306031292
	[Zhang L, Wang L L, Zhang X D, et al.The basic principle of random forest and its applications in ecology: A case study of Pinus yunnanensis[J]. Acta Ecologica Sinica, 2014,34(3):650-659. ] doi: 10.5846/stxb201306031292
[17]	田绍鸿,张显峰.采用随机森林法的天绘数据干旱区城市土地覆盖分类[J].国土资源遥感,2016,28(1):43-49. doi: 10.6046/gtzyyg.2016.01.07
	[Tian S H, Zhang X F.Random forest classification of land cover information of urban areas in arid regions based on TH-1 data[J]. Remote Sensing for Land and Resources, 2016,28(1):43-49. ] doi: 10.6046/gtzyyg.2016.01.07
[18]	Gleason C J, Im J.Forest biomass estimation from airborne LiDAR data using machine learning approaches[J]. Remote Sensing of Environment, 2012,125(125):80-91. doi: 10.1016/j.rse.2012.07.006
[19]	Jiménez-Muñoz J C, Sobrino J A. A generalized single-channel method for retrieving land surface temperature from remote sensing data[J]. Journal of Geophysical Research, 2003,108(D22):2015-2023. doi: 10.1029/2003JD003480
[20]	邹春城,张友水,黄欢欢.福州市城市不透水面景观指数与城市热环境关系分析[J].地球信息科学学报,2014,16(3):490-498. doi: 10.3724/SP.J.1047.2014.00490
	[Zou C C, Zhang Y S, Huang H H.Impacts of impervious surface area and landscape metrics on urban heat environment in Fuzhou City, China[J]. Journal of Geo-InformationScience, 2014,16(3):490-498. ] doi: 10.3724/SP.J.1047.2014.00490
[21]	Chester L Arnold Jr, C James Gibbons. Impervious surface coverage: The emergence of a key environmental indicator[J]. Journal of the American Planning Association, 1996,62(2):243-258.
[22]	Ridd M K.Exploring a V-I-S (Vegetation-Impervious Surface-Soil) model for urban ecosystem analysis through remote sensing[J]. International Journal of Remote Sensing, 1995,16(12):2165-2185. doi: 10.1080/01431169508954549
[23]	徐永明,刘勇洪.基于TM影像的北京市热环境及其与不透水面的关系研究[J].生态环境学报,2013,22(4):639-643. doi: 10.3969/j.issn.1674-5906.2013.04.015
	[Xu Y M, Liu Y H.Study on the thermal environment and its relationship with impervious surface in Beijing city using TM image[J]. Ecology and Environmental Sciences, 2013,22(4):639-643. ] doi: 10.3969/j.issn.1674-5906.2013.04.015
[24]	Rouse J W, Haas R H, Schell J A,et al.Monitoring vegetation systems in the great plains with Erts[A]. Third Earth Resources Technology Satellite-1 Symposium[C]. 1974:309-317.
[25]	徐涵秋. 利用改进的归一化差异水体指数(MNDWI)提取水体信息的研究[J].遥感学报,2005,9(5):589-595. doi: 10.3321/j.issn:1007-4619.2005.05.012
	[Xu H Q.A study on information extraction of water body with the modified normalized difference water index(MNDWI)[J]. Journal of Remote Sensing, 2005,9(5):589-595. ] doi: 10.3321/j.issn:1007-4619.2005.05.012
[26]	阿布都瓦斯提·吾拉木,秦其明.基于辐射模拟反演ETM+数据宽波段反照率[J].北京大学学报:自然科学版,2007,43(4):474-483. doi: 10.3321/j.issn:0479-8023.2007.04.007
	[Ghulam A, Qin Q M.Calculation of ETM+ broadband albedos by radiative simulations[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2007,43(4):474-483. ] doi: 10.3321/j.issn:0479-8023.2007.04.007

自变量	方程	参考文献
NDVI	$NDVI = (NIR - R) (NIR + R)$	文献[24]
MNDWI	$MNDWI = (Green - MIR) (Green + MIR)$	文献[25]
Albedo	$αshort = 0.356 α 1 + 0.13 α 3 + 0.373 α 4 + 0.085 α 5 + 0.072 α 7$	文献[26]

	精度平均减少值/%	节点不纯度平均减少值
LST	14.28	241.36
Altitude	12.82	213.30
NDVI	3.43	60.15
MNDWI	3.77	82.89
Albedo	4.82	61.09
ISC	2.50	42.74