基于Copula函数的地表温度空间降尺度研究

Copula函数能够用于构建粗尺度数据与细尺度数据间的二维联合分布实现降水、土壤湿度空间降尺度。为了探讨Copula函数在地表温度空间降尺度研究中的适用性,选取黑河中游荒漠绿洲区作为研究区,以2012年7月10日ASTER、MODIS地表温度影像作为数据源,开展基于Copula函数的地表温度空间降尺度研究,并利用地面实测地表温度数据对降尺度结果进行验证。结果表明:基于Copula函数的空间降尺度方法能够较好地刻画出细尺度像元的温度值,但是不易捕捉到地表温度突变区域的细节信息;能够显著地提高热红外遥感影像反演得到的地表温度数据的精度,MAE和RMSE分别从2.99K、3.87K减小至1.51K、2.36K。     

基于TsHARP模型和STITFM算法的地表温度影像融合研究

基于MODIS地表温度产品和Landsat ETM+影像,提出采用将TsHARP(Thermal sHARPening)模型和STITFM(Spatio-Temporal Integrated Temperature Fusion Model)算法相结合的方法CTsSTITFM进行地表温度数据的融合。先利用TsHARP方法对相邻t_1和t_2时刻的1km MODIS地表温度数据降尺度为250m空间分辨率地表温度,再将降尺度结果输入STITFM模型进行影像融合,最终生成t_2时刻30m空间分辨率的地表温度数据。结果表明:该方法比与单独采用STITFM算法的模拟结果精度有所提高,在默认参数设置下,融合影像的地表温度与Landsat ETM+数据反演地表温度值相比,均方根误差(RMSE)小于1.33K。通过对CTsSTITFM融合方法的参数中窗口大小的调节发现,随窗口不断增大,在所选区域融合效果表现出一定的规律性,合理的窗口大小设置有助于提升融合效果。     

基于MODIS和AMSR-E遥感数据的土壤水分降尺度研究

微波传感器获得的土壤水分产品空间分辨率一般都很粗,而流域尺度上的研究需要中高分辨率的土壤水分数据。用MODIS逐日地表温度产品MOD11A1和逐日地表反射率产品MOD09GA构建温度-植被指数特征空间,并计算得到TVDI(Temperature Vegetation Dryness Index)指数,它与土壤水分呈负相关关系,能够反映土壤水分的空间分布模式,但并不是真实的土壤水分值。在AMSR-E像元尺度上求得TVDI与土壤水分的负相关系数,进而对VUA AMSR-E土壤水分产品进行降尺度计算得到0.01°分辨率的真实土壤水分值。经NAFE06(The National Airborne Field Experiment 2006)试验地面采样数据验证,降尺度后的土壤水分均方根误差平均值为6.1%。     

利用随机森林和地理加权回归的黄河源GIMMS3gNDVI降尺度方法

针对现有AVHRR、SPOTVGT、MODIS产品难以构建长时序、高时空分辨率NDVI数据集的问题,提出了利用随机森林和地理加权回归模型对GIMMS_3g NDVI进行降尺度的方法。基于“关系尺度不变”假设,从不同空间分辨率和数据源角度将其空间分辨率从8 km提高至250 m,并利用MODIS数据进行精度评价。结果表明：降尺度数据的空间分辨率有较大提升,能真实反映源区内NDVI空间分布特征;降尺度数据与MODIS数据具有较好的一致性,除500 m分辨率下的RF降尺度外,其他降尺度结果的绝对误差≤0.1的比例达到70%;东南部高植被覆盖区的降尺度效果要优于西北部;RF模型在体现局部细节纹理特征方面更具优势;解释变量的不同组合会影响降尺度的精度;降尺度模型尺度的改变对RF模型降尺度结果影响较大。     

AMSR-E亮温数据与MODIS陆表分类产品结合反演全球陆表温度

AMSR-E被动微波传感器获取的亮温数据与MODIS陆表分类产品(MOD12)相结合,将全球陆表分为16类,并假设每种类型的地表在各个被动微波通道具有较一致的发射率,在此基础上针对每种陆表类型分别建立了陆表温度反演算法。在算法的建立过程中,为了避免混合像元以及冻土、积雪发射率不确定性带来的影响,仅对单一地表类型占90%以上以及MODIS陆表温度产品高于273K的被动微波像元进行回归。同时,考虑到降雨对回归结果的影响,在数据选择中加入了降雨判识,在被动微波亮温数据中除去了降雨像元。利用上述算法,用2004年1～10月的全球部分地区AMSR-E数据在MODIS陆表分类产品的基础上对每种地表类型分别进行了陆表温度反演,并与MODIS陆表温度产品进行对比,结果显示相关性较好,均方根误差为2～4 K。     

基于被动微波与时空联合算法的云下像元LST重建

地表温度作为衡量地球表面水热平衡的关键参数,具有两大时空分布特征:第一,空间分布一致性,即属性相近的像元地表温度与其地表亮温间的相关关系相对稳定;第二,时间序列周期性,且同一地区时间越接近地表温度值越相似。基于这两大特征将空间统计模型与时间序列滤波相结合,提出了用于云下像元地表温度重建的时空联合算法。以2008年MODIS地表温度产品为研究对象,采用Landsat TM数据和AMSR_E地表亮温数据重建中国9个省份的地表温度值,并与基于MODIS地表分类产品的多通道统计模型重建结果进行对比。实验结果表明,所提算法实用性强,能有效实现大面积复杂下垫面区域的地表温度重建;平均重建误差约为1.2 K,相较于基于下垫面分类的多通道统计模型下降了76%,算法精度明显提高。
     

基于缨帽变换分析地表温度变化

利用缨帽变换提取土壤亮度指数、绿度植被指数、湿度指数等地表参数,利用模型提取归一化植被指数NDVI、比值植被指数RVI、修改型土壤调整指数MSAVI等植被指数和水体指数MNDWI,利用Artis单窗算法估算热红外波段像元尺度地表温度,将地表温度的影响因素作为BP神经网络输入估算30m空间分辨率的亚像元地表温度,分析1989~2006年桂林城区土地利用变化、缨帽变换特征分量变化、植被参数变化、水体指数变化对地表温度的影响机理。     

Landsat-8与ZY-3卫星影像降尺度地表温度反演及分析

地表温度(land surface temperature,LST)在表征地表能量转换和气候方面具有很重要的作用。目前获取的高空间分辨率遥感影像,通常没有热红外波段,而能获取的热红外影像,空间分辨率往往不够。针对时间分辨率和空间分辨率的矛盾问题,提出了一种利用降尺度进行地表温度反演的方法。该方法利用Landsat-8卫星影像的热红外波段进行地表温度LSTOLI,30m反演;根据不同地物的光谱特性,选取代表城市热特性的地物端元,对ZY-3多光谱影像大气校正后,进行混合像元分解,获取每个像元内不同端元的丰度,利用端元的平均温度建模,估算地表温度,得到高空间分辨率的地表温度LSTZY-3,5.8m。为了验证估算结果的准确性,将LSTZY-3,5.8m升尺度为30mLSTZY-3,30m,与LSTOLI,30m进行对比。结果表明,LSTZY-3,30m和LSTOLI,30m具有很高的一致性,精度较高。这种方法可以作为一种实用的估算地表温度的方法。     

FY3A/MERSI地表温度反演

MERSI是我国第二代极轨气象卫星上的重要传感器,可获取高空间分辨率和高时间分辨率的对地观测影像。为使Jimènez-Mu珘nozSobrino算法更适用于FY3A/MERSI传感器通道特性,更新了大气函数的估算系数,并引入观测角度因子,以获取更为精确像元间更为平滑的地表温度。用MODTRAN4模拟验证该算法精度,得引入角度因子后反演精度显著提升,所有角度下平均误差为-0.6±2.2K。用实测的敦煌戈壁地表温度和MODIS地表温度产品评价MERSI反演结果,显示MERSI地表温度的空间分布准确,结果精度也较高。与实测温度对比,平均误差为1.74K,均方根误差小于1.9K。研究区域与MODIS地表温度间差异平均为2.6307K。虽然会受云检测精度和观测亮温偏高的影响,由MERSI反演的高精度地表温度在相关科研和业务方面仍然具有极好的应用前景。     

MOD09A1数据产品中缺失条带的插补方法

MODIS地表反射率产品(MOD09A1)是MODIS系列化数据产品中一项重要而基础性的产品。在实际应用中发现该数据产品中仍然继承了原始数据固有的条带缺失问题,且随着数据投影转换,缺失数据在表现为条带状的同时,表现出新的分布特点,以往的插补方法不再适用。利用MOD09A1数据集中描述MODIS数据获取和处理质量的QC数据逐一确定单个缺失像元的准确位置,采用其8邻域内的非缺失像元均值对缺失像元进行插补,在对非条带信息不产生影响的前提下实现了MOD09A1缺失条带的去除,从而确保了该产品数据的质量。选用不同年份不同天数的3景数据进行处理,并将模拟的条带缺失数据采用本文方法处理的结果与真实数据比较,以及将同一条带缺失数据采用不同方法处理的结果比较,结果显示本文方法对于MOD09A1数据条带去除优于以往的方法,并具有普适性和可靠性。
     

Improving the disaggregation of MODIS land surface temperatures in an urban environment: a statistical downscaling approach using high-resolution emissivity

Spatially and temporally dense land surface temperature (LST) data are necessary to capture the high variability of the urban thermal environment. Sensors on board satellites with high revisit time cannot provide adequately detailed spatial information; thus, the downscaling of LST is recognized as being an important and inevitable intermediate process. In this paper, improvement in the downscaled LST accuracy is investigated, employing the statistical downscaling methodology in an urban setting. A new approach is proposed, where thermal radiances are disaggregated using multiple regression analysis and are then combined with emissivity values derived from a high-resolution image classification. Predictors include reflectance values, built-up and vegetation indices, and topographic data. Surface classification is performed utilizing machine learning techniques and fusing Sentinel-2 imagery with ancillary data. Thermal data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor are downscaled from their original resolution to 100 m in the city of Athens, Greece. Validation of sharpened temperatures is performed using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) surface temperature product and in-situ measurements. It is demonstrated that the proposed downscaling framework using ridge regression has the potential to produce reliable, high temporal LST estimates with an average error of fewer than 2 K, while consistently having a better accuracy than the reference, single-predictor downscaling of the MODIS LST product.     

New refinements and validation of the MODIS Land-Surface Temperature/Emissivity products

This paper discusses the lessons learned from analysis of the Moderate Resolution Imaging Spectroradiometer (MODIS) Land-Surface Temperature/Emissivity (LST) products in the current (V4) and previous versions, and presents eight new refinements for V5 product generation executive code (PGE16) and the test results with real Terra and Aqua MODIS data. The major refinements include considering surface elevation when using the MODIS cloudmask product, removal of temporal averaging in the 1 km daily level-3 LST product, removal of cloud-contaminated LSTs in level-3 LST products, and the refinements for the day/night LST algorithm. These refinements significantly improved the spatial coverage of LSTs, especially in highland regions, and the accuracy and stability of the MODIS LST products. Comparisons between V5 LSTs and in-situ values in 47 clear-sky cases (in the LST range from − 10 °C to 58 °C and atmospheric column water vapor range from 0.4 to 3.5 cm) indicate that the accuracy of the MODIS LST product is better than 1 K in most cases (39 out of 47) and the root of mean squares of differences is less than 0.7 K for all 47 cases or 0.5 K for all but the 8 cases apparently with heavy aerosol loadings. Emissivities retrieved by the day/night algorithm are well compared to the surface emissivity spectra measured by a sun-shadow method in two field campaigns. The time series of V5 MODIS LST product over two sites (Lake Tahoe in California and Namco lake in Tibet) in 2003 are evaluated, showing that the quantity and quality of MODIS LST products depend on clear-sky conditions because of the inherent limitation of the thermal infrared remote sensing.     

Evaluation and improvement of the MODIS land surface temperature/emissivity products using ground‐based measurements at a semi‐desert site on the western Tibetan Plateau

Current MODerate‐resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST, surface skin temperature)/emissivity products are evaluated and improvements are investigated. The ground‐based measurements of LST at Gaize (32.30° N, 84.06° E, 4420 m) on the western Tibetan Plateau from January 2001 to December 2002 agree well (mean and standard deviation of differences of 0.27 K and 0.84 K) with the 1‐km Version 004 (V4) Terra MODIS LST product (MOD11A1) generated by the split‐window algorithm. Spectral emissivities measured from surface soil samples collected at and around the Gaize site are in close agreement with the landcover‐based emissivities in bands 31 and 32 used by the split‐window algorithm. The LSTs in the V4 MODIS LST/emissivity products (MYD11B1 for Aqua and MOD11B1 for Terra) from the day/night LST algorithm are higher by 1–1.7 K (standard deviation around 0.6 K) in comparisons to the 5‐km grid aggregated values of the LSTs in the 1‐km products, which is consistent with the results of a comparison of emissivities. On average, the emissivity in MYD11B1 (MOD11B1) is 0.0107 (0.0167) less than the ground‐based measurements, which is equivalent to a 0.64 K (1.25 K) overestimation of LST around the average value of 285 K. Knowledge obtained from the evaluation of MODIS LST/emissivity retrievals provides useful information for the improvement of the MODIS LST day/night algorithm. Improved performance of the refined (V5) day/night algorithm was demonstrated with the Terra MODIS data in May–June 2004.     

结合参量统计与时空融合的土壤水分降尺度方法

土壤水分的降尺度研究为解决被动微波产品的粗分辨率问题,更好地服务于流域小尺度应用提供了技术手段。以美国俄克拉荷马州为研究区域,基于SMAP土壤水分产品和MODIS产品等多种辅助数据,在地表分类数据的支持下,结合参量统计降尺度和时空融合降尺度发展了一种土壤水分混合降尺度方法,并利用SMAP 9 km产品和站点实测数据对降尺度效果进行了评估。结果表明：混合降尺度方法可以得到细节丰富、空间覆盖完整的降尺度结果。相较于参量统计或时空融合两种单一降尺度而言,混合降尺度结果的空间分布与SMAP 9 km真实产品最为相似,并且混合降尺度结果与站点的整体时序精度最高,在不同地表分类下的时序精度也优于单一方法的降尺度结果。由此证明结合参量统计与时空融合的降尺度方法是可行的。     

Spatial Downscaling Land Surface Temperature based on Copula

A Copula is used to construct a bivariate distribution describing the relation between coarse\|scale and fine\|scale rainfall or soil moisture.This distribution is then used to downscale rainfall or soil moisture.In order to explore the feasibility of spatial downscaling Land Surface Temperature (LST)based on Copula,we implemented LST downscaling based on Copula and ASTER LST and MODIS LST products at Yingke oasis\|desert area in the middle streams of the Heihe River Basin.The downscaled LST was calibrated by the ground observations from HiWATER\|MUSOEXE experiment.The results show that the downscaling method based on Copula is able to achieve the LST downscaling in general,but the method can’t obtain the fine\|scale LST correctly at the interface between oasis and desert.The accuracy of LST obtained from thermal infrared satellite image was improved significantly by the method.The MAE and RMSE in LST are reduced from 2.99 K,and 3.89 K to 1.5 1K,and 2.36 K,respectively.     

Downscaling AVHRR land surface temperatures for improved surface urban heat island intensity estimation

Surface urban heat island (SUHI) is a phenomenon of both high spatial and temporal variability. In this context, studying and monitoring the SUHIs of urban areas through the satellite remote sensing technology, requires land surface temperature (LST) image data from satellite-borne thermal sensors of high spatial resolution as well as temporal resolution. However, due to technical constrains, satellite-borne thermal sensors yield a trade-off between their spatial and temporal resolution; a high spatial resolution is associated with a low temporal resolution and vice versa. To resolve this drawback, we applied in this study four downscaling techniques using different scaling factors to downscale 1-km LST image data provided by the Advanced Very High Resolution Radiometer (AVHRR) sensor, given that AVHRR can offer the highest temporal resolution currently available. The city of Athens in Greece was used as the application site. Downscaled 120-m AVHRR LSTs simulated by the downscaling techniques, were then used for SUHI intensity estimation based on LST differences observed between the main urban land covers of Athens and the city's rural background. For the needs of the study, land cover information for Athens was obtained from the Corine Land Cover (CLC) 2000 database for Greece. Validation of the downscaled 120-m AVHRR LSTs as well of the retrieved SUHI intensities was performed by comparative analysis with time-coincident observations of 120-m LST and SUHI intensities generated from the band 6 of the Thermal Mapper (TM) sensor onboard the Landsat 5 platform. The spatial pattern of the downscaled AVHRR LST was found to be visually improved when compared to that of the original AVHRR LST and to resemble more that of TM6 LST. Statistical results indicated that, when compared to 120-m TM6 LST, the root mean square error (RMSE) in 120-m AVHRR LST generated by the downscaling techniques ranged from 4.9 to 5.3 °C. However, the accuracy in SUHI intensity was found to have significantly improved, with a RMSE value decreasing from 2.4 °C when the original AVHRR LST was utilized, down to 0.94 °C in case that downscaling was applied.     

Evaluation of LST downscaling algorithms on seasonal thermal data in humid subtropical regions of India

Thermal image downscaling algorithms use a unique relationship between land surface temperature (LST) and vegetation indices (e.g. normalized difference vegetation index (NDVI)). The LST–NDVI correlation and regression parameters vary in different seasons depending on land-use practices. Such relationships are dynamic in humid subtropical regions due to inter-seasonal changes in biophysical parameters. The present study evaluates three downscaling algorithms, namely disaggregation of radiometric surface temperature (DisTrad), sharpening thermal imagery (TsHARP), and local model using seasonal (25 February 2010, 14 April 2010, and 26 October 2011) thermal images. The aggregated Landsat LST of 960 m resolution is downscaled to 480, 360, 240, and 120 m using DisTrad, TsHARP, and the local model and validated with aggregated Landsat LSTs of a similar resolution. The results illustrate that the seasonal variability of the LST–NDVI relationship affects the accuracy of the downscaling model. For example, the accuracy of all algorithms is higher for the growing seasons (February and October) unlike the harvesting season (April). The root mean square error of the downscaled LST increases from 480 to 120 m spatial resolution in all seasons. The models are least suitable in water body and dry-river bed sand areas. However, the downscaling accuracy is higher for NDVI > 0.3. The present study is useful to understand the applicability of the downscaling models in seasonally varied landscapes and different NDVI ranges.     

Towards deterministic downscaling of SMOS soil moisture using MODIS derived soil evaporative efficiency

A deterministic approach for downscaling ～ 40 km resolution Soil Moisture and Ocean Salinity (SMOS) observations is developed from 1 km resolution MODerate resolution Imaging Spectroradiometer (MODIS) data. To account for the lower soil moisture sensitivity of MODIS surface temperature compared to that of L-band brightness temperature, the disaggregation scale is fixed to 10 times the spatial resolution of MODIS thermal data (10 km). Four different analytic downscaling relationships are derived from MODIS and physically-based model predictions of soil evaporative efficiency. The four downscaling algorithms differ with regards to i) the assumed relationship (linear or nonlinear) between soil evaporative efficiency and near-surface soil moisture, and ii) the scale at which soil parameters are available (40 km or 10 km). The 1 km resolution airborne L-band brightness temperature from the National Airborne Field Experiment 2006 (NAFE'06) are used to generate a time series of eleven clear sky 40 km by 60 km near-surface soil moisture observations to represent SMOS pixels across the three-week experiment. The overall root mean square difference between downscaled and observed soil moisture varies between 1.4% v/v and 1.8% v/v depending on the downscaling algorithm used, with soil moisture values ranging from 0 to 15% v/v. The accuracy and robustness of the downscaling algorithms are discussed in terms of their assumptions and applicability to SMOS.     

Preliminary applications of a land surface temperature retrieval method to IASI and AIRS data

Land surface temperature (LST) is one of the key state variables for many applications. This article aims to apply our previously developed LST retrieval method to infrared atmospheric sounding interferometer (IASI) and atmospheric infrared sounder (AIRS) data. On the basis of the opposite characteristics of the atmospheric spectral absorption and surface spectral emissivity, a ‘downwelling radiance residual index’ (DRRI) has been recalled and improved to obtain LST and emissivity. To construct an efficient DRRI, an automatic channel selection procedure has been proposed, and 11 groups of channels have been selected within the range 800–1000 cm^?1. The DRRI has been tested with IASI and AIRS data. For the IASI data, the radiosonde data have been used to correct for atmospheric effects and to retrieve LST, while the atmospheric profiles retrieved from AIRS data have been used to perform the atmospheric corrections and subsequently to estimate LST from AIRS data. The differences between IASI- and Moderate Resolution Imaging Spectroradiometer (MODIS)-derived LSTs are no more than 2 K, while the differences between AIRS- and MODIS-derived LSTs are less than 5 K. Even though an exceptionally problematic value occurred (–12.89 K), the overall differences between AIRS-estimated LST and the AIRS L2 LST product are no more than 5 K. Although the IASI-derived LST is more accurate than the AIRS-derived one, the convenient retrieval of AIRS atmospheric profile made this method more applicable. Limitations and uncertainties in retrieving LST using the DRRI method are also discussed.