本期要文(可全文下载)

基于马尔可夫链和最大后验准则的模拟静止气象卫星数据地表组分温度反演

展开
  • 1. 中国科学院地理科学与资源研究所,北京100101;
    2. 中国科学院水利部成都山地灾害与环境研究所,成都610041
赵伟(1984-),男,江西上高人,博士,助理研究员,主要从事定量遥感反演方法研究。E-mail:zhaow@imde.ac.cn

收稿日期: 2012-11-29

  修回日期: 2013-02-02

  网络出版日期: 2013-06-17

基金资助

中国科学院重点部署项目(KZZD-EW-08-01);中国科学院-国家外专局国际合作创新团队项目(KZZD-EW-TZ-06);中国科学院“百人计划”项目;国家自然科学基金项目(41271433)联合资助。

Component Temperature Estimation from Simulated Geostationary Meteorological Satellite Data Based on MAP Criterion and Markov Models

Expand
  • 1. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;
    2. Institute of Mountain Hazards and Environment, CAS, Chengdu 610041, China

Received date: 2012-11-29

  Revised date: 2013-02-02

  Online published: 2013-06-17

摘要

地表组分温度比像元混合温度具有更强的物理意义和实用价值,是定量遥感反演的一个重要研究方向。本文以马尔可夫链和最大后验准则地表温度尺度转换方法,结合静止气象卫星数据高时间分辨率的特点,通过模拟静止气象卫星数据地表组分温度反演进行分析和验证。在研究过程中,地面被简化为由植被和土壤两组分组成,同时假设邻近像元的植被和土壤组分温度相同。鉴此,本文通过模拟构建20×20像元大小的静止气象卫星混合像元图像,并对各像元各时刻温度添加均值为0标准差为2K的随机误差,最终应用所提算法估算各像元各时刻的植被和土壤组分温度大小。精度分析结果表明,该算法能够较为精确地反演植被和土壤组分温度,且误差基本控制在2K以内。此外,本文还进一步讨论了算法的适用性及其对混合像元温度误差、植被覆盖度误差,以及邻近像元植被覆盖度变化范围的敏感度。分析结果再次证明,该方法对混合像元温度误差和植被覆盖度误差都具有较低的敏感性,在最大温度误差条件(均值为1.8K,标准差为5K)和最大植被覆盖度误差(均值为0.18,标准差为0.2)的条件下,各组分温度的估算精度分别能控制在3K和2K以内,满足精度要求。但是,由于组分温度初值的确定方法,对所计算窗口内植被覆盖度变化范围有较强的敏感性,反演结果与植被覆盖度变化范围相关,要求窗口内植被覆盖度变化范围足够大才能满足初值估算的精度要求。

本文引用格式

赵伟, 李爱农, 李召良 . 基于马尔可夫链和最大后验准则的模拟静止气象卫星数据地表组分温度反演[J]. 地球信息科学学报, 2013 , 15(3) : 422 -430 . DOI: 10.3724/SP.J.1047.2013.00422

Abstract

Land surface temperature (LST) is one of the key parameters in the physics of land surface processes, and satellite observation is the major way to derive regional LST distribution. However, the pixel observed by satellite is usually a mixture of different land cover types and the temperature of each type seems more meaningful land useful than the mixed pixel temperature. Therefore, component temperature estimation is an important research field. To take use of the rich temporal information imbedded in the geostationary satellite data, a practical approach was proposed to derive the component temperature based on MAP criterion and Markov models in this study. The approach was on the basis of mixed pixel temperature theory and under the assumption that there were only two components in the mixed pixel (vegetation and soil) and the nearby pixels had the same temperature for each component. The method was applied to simulated geostationary satellite data, and the component temperature estimation results showed good agreement with the true values. The RMSEs of vegetation and soil temperature were within 2K for most cases. In addition, the sensitivity analysis of the method to the mixed pixel temperature estimation error, the vegetation fraction cover (FVC) estimation error and its range of the pixels considered in the estimation were conducted. The results indicated that the pixel temperature error and the FVC error showed little effect on the estimation results while the FVC range played an important role in the estimation because of its determinant function on the first guess of the component temperature.

参考文献

[1] Balick L K,Hutchinson B A. Directional thermal infraredexitance distributions from a leafless deciduous forest[J].IEEE Transactions on Geoscience and Remote Sensing,1997(24):693-698.

[2] Kimes D S,Kirchener J A. Directional radiometric measurementsof row-crop temperatures[J]. International Journalof Remote Sensing, 1983(4):299-311.

[3] Lagouarde J P, Kerr Y H,Brunet Y. An experimental studyof angular effects on surface temperature for various plantcanopies and bare soils[J]. Agricultural and Forest Meteorology,1995(77):167-190.

[4] 王锦地,李小文,孙晓敏,等.用热辐射方向性模型反演非同温像元组分温度[J].中国科学E辑,2000(30):54-60.

[5] Li Z L, Stoll M P, Zhang R H, et al. On the separate retrievalof soil and vegetation temperature from ATSR2 data[J].Science in China (Series D), 2001,44(2):97-111.

[6] Zhang R H, Su H B, Li Z L, et al. The potential informationin the temperature difference between shadow and sunlitof surfaces and a new way of retrieving the soil moisture[J]. Science in China (Series D), 2001,44(2):112-123

[7] Shi Y L. Thermal infrared inverse model for componenttemperatures of mixed pixels[J]. International Journal ofRemote Sensing, 2011,32(8):2297-2309

[8] 陈良富,庄家礼,徐希孺,等.非同温混合像元热辐射有效比辐射率概念及其验证[J].科学通报,2000,45(1):22-28.

[9] 徐希孺,范闻捷,陈良富.开放的复杂目标热辐射特性的矩阵表达式[J].中国科学D辑,2001,3l(12):1046-1051.

[10] 刘强,陈良富,柳钦火,等.作物冠层的热红外辐射传输模型[J].遥感学报,2003,7(3):161-167.

[11] Becker F,Li Z L. Temperature independent spectral indicesin thermal infrared bands[J]. Remote Sensing of Environment,1990(32):17-33.

[12] Palluconi F, Kahle A B, Hoover G, et al. The spectralemissivity of prairie and pasture grasses at Konza prairie[J]. Kansas, In Proc. Am. Meteorol, Soc. Symp. on theFirst ISLSCP Field Experiment (FIFE), Anaheim, CA,Am. Meteorol. Soc., Boston, MA.1990,7-9:77-78.

[13] Kallel A, Ottlé C,LeHégarat-Mascle S. Surface temperaturedownscaling from multi-resolution instruments basedon MAP criterion and Markov models[C] 3rd InternationalSymposium; Recent advances in Quantitative remotesensing, 26 September-1 October Torrent, Spain, 2010,50-55.

[14] Kallel A, Ottlé C, LeHégarat-Mascle S, et al. Surface temperaturedownscaling from multiresolution instrumentsbased on Markov models[J]. IEEE Transaction on Geoscienceand Remote Sensing, 2012, doi:10.1109/TGRS.2012.2207461

[15] Jiang G M, Li Z L,Nerry F. Land surface emissivity retrievalfrom combined mid-infrared and thermal infrareddata of MSG-SEVERI[J]. Remote Sensing of Environment,2006(105):326-340.

文章导航

/