Prediction of Land Surface Temperature under Cloudy Conditions using Microwave Remote Sensing and ANN

Land surface temperature (LST) is an important variable in climate, hydrologic, ecological, biophysical and biochemical studies (Mildrexler et al., 2011). The most effective way to obtain LST measurements is through satellites. Presently, LST from moderate resolution imaging spectroradiometer (MODIS) sensor is applied in various fields due to its high spatial and temporal availability over the globe, but quite difficult to provide observations in cloudy conditions. This study evolves of prediction of LST under clear and cloudy conditions using microwave vegetation indices (MVIs), elevation, latitude, longitude and Julian day as inputs employing an artificial neural network (ANN) model. MVIs can be obtained even under cloudy condition, since microwave radiation has an ability to penetrate through clouds. In this study LST and MVIs data of the year 2010 for the Cauvery basin on a daily basis were obtained from MODIS and advanced microwave scanning radiometer (AMSR-E) sensors of aqua satellite respectively. Separate ANN models were trained and tested for the grid cells for which both LST and MVI were available. The performance of the models was evaluated based on standard evaluation measures. The best performing model was used to predict LST where MVIs were available. Results revealed that predictions of LST using ANN are in good agreement with the observed values. The ANN approach presented in this study promises to be useful for predicting LST using satellite observations even in cloudy conditions. (C) 2015 The Authors. Published by Elsevier B.V.

Aerosol-cloud-precipitation interactions in marine stratocumulus clouds

Marine stratocumulus clouds are generally optically thick and shallow, exerting a net cooling influence on climate. Changes in atmospheric aerosol levels alter cloud microphysics (e.g., droplet size) and cloud macrophysics (e.g., liquid water path, cloud thickness), thereby affecting cloud albedo and Earth’s radiative balance. To understand the aerosol-cloud-precipitation interactions and to explore the dynamical effects, three-dimensional large-eddy simulations (LES) with detailed bin-resolved microphysics are performed to explore the diurnal variation of marine stratocumulus clouds under different aerosol levels and environmental conditions. It is shown that the marine stratocumulus cloud albedo is sensitive to aerosol perturbation under clean background conditions, and to environmental conditions such as large-scale divergence rate and free tropospheric humidity.

Based on the in-situ Eastern Pacific Emitted Aerosol Cloud Experiment (E-PEACE) during Jul. and Aug. 2011, and A-Train satellite observation of 589 individual ship tracks during Jun. 2006-Dec. 2009, an analysis of cloud albedo responses in ship tracks is presented. It is found that the albedo response in ship tracks depends on the mesoscale cloud structure, the free tropospheric humidity, and cloud top height. Under closed cell structure (i.e., cloud cells ringed by a perimeter of clear air), with sufficiently dry air above cloud tops and/or higher cloud top heights, the cloud albedo can become lower in ship tracks. Based on the satellite data, nearly 25% of ship tracks exhibited a decreased albedo. The cloud macrophysical responses are crucial in determining both the strength and the sign of the cloud albedo response to aerosols.

To understand the aerosol indirect effects on global marine warm clouds, multisensory satellite observations, including CloudSat, MODIS, CALIPSO, AMSR-E, ECMWF, CERES, and NCEP, have been applied to study the sensitivity of cloud properties to aerosol levels and to large scale environmental conditions. With an estimate of anthropogenic aerosol fraction, the global aerosol indirect radiative forcing has been assessed.

As the coupling among aerosol, cloud, precipitation, and meteorological conditions in the marine boundary layer is complex, the integration of LES modeling, in-situ aircraft measurements, and global multisensory satellite data analyses improves our understanding of this complex system.

近海面气象参数的反演及应用研究

全球气候与人类活动密切相关，气候研究一直是现代科学家关注的重点问题之一。过去海面气象参数的研究主要借助于浮标和站点的观测，但是它们稀少的观测资料极大地限制了海面气象参数的研究。现在借助卫星观测大大地提高我们的认识能力，卫星可以对全球海洋进行连续观测，获取长期大范围的海洋气象资料，为全面深入地了解大洋甚至全球大气活动提供可能。 本文的工作就是利用卫星资料进行月平均和实时的近海面气象参数的反演及应用研究。利用近十八年SSM/I和AVHRR卫星资料与实测资料进行结合，建立神经网络（ANN）模型反演近海面月平均气温和湿度，与实测资料相比气温的均方根差为0.87 ℃，相关系数为 0.99，相对湿度的均方根差为3.73%，相关性为0.65。利用同步物理方法从TOVS资料反演中国海区上空1000mb到10mb之间的温湿廓线，再利用神经网络方法和基于Bowen比的方法从温湿廓线的结果反演出近海面处的实时气温和露温参数，取得了比较合理的结果，气温和露温结果的均方根差分别是1.85K和2.59K（与实测数据相比）。利用2005年1月的AMSR-E亮温资料对实时气象参数反演进行探讨，分析AMSR-E的各个探测通道与海表温度、近海面气温、湿度和风速等参数的相关性，把12个通道分为四种情况并在每种情况下分别进行试验，选择最合适的组合通道并进行气象参数的反演，结果与TAO资料进行比较，海表温度的均方根差是0.55℃，近海面气温的均方根差是0.74℃，海面湿度的均方根差是3.24%，海面风速的均方根差是1.11m/s。目前，与其它结果相比该结果的精度是最好的。 把以上反演得到的近海面气象参数结果应用于海气界面热通量的计算，以更好地研究海气相互作用。分别采用神经网络和Bulk公式两种方法计算月平均潜热和感热通量，结果与GSSTF2资料进行比较，Bulk方法反演的感热和潜热的均方根差分别为9.05±4.6 W/m2和23.7±4.0 W/m2，ANN模型得到的分别是7.54±3.0 W/m2和20.1±3.2 W/m2，，结果表明ANN模型得到的结果明显的好于Bulk公式。ANN模型反演的全球潜热和感热结果在空间分布上与GSSTF2的吻合很好，但在极大值区，ANN模型得到的结果偏小。把AMSR-E亮温资料反演得到的实时的海表温度、近海面气温、湿度和风速四个参数结合ANN模型，计算太平洋赤道地区的实时感热和潜热通量。 对反演得到的约18年近海面月平均气温进行年变化分析,得知近18年里气温呈现上升趋势。近海面气温在反映气候异常信息上有与海表温度相似的表达性，1989冷事件年，全球平均海面气温明显偏低（14.25℃），1998暖事件年，平均海面气温值最大（14.57℃ ）。用经验正交函数（EOF）和经验模态分解（EMD）分析近海面气温的距平变化，得到的前三个主要模态解释了84%的总体变化，EOF1解释了76.1%的变化，主要表达了太阳辐射引起的年周期气温变化；EOF2解释了4.6%的变化，主要解释ENSO对气温异常的作用；EOF3（3.3%）的空间模态呈现了很多有趣的现象，比如气温正异常主要表现在北半球的高纬度区，南极附近高低起伏的气温异常可以作为南极绕极流的一个证据等。

海上低空大气波导的遥感反演及数值模拟研究

由于电磁波传播技术的广泛应用，大气环境成为电磁波传播研究的一个重要领域，尤其是天气变化多端的对流层。对流层大气波导是对流层大气的一种异常结构。大气波导可以使通信电路的相互干扰问题变得复杂，既可能干扰其他系统,又可能形成另外的系统，可以使探测系统产生超视距探测、雷达盲区等反常传播问题。然而通信系统或雷达探测系统遭受大气波导影响的时机具有很大的偶然性，它完全取决于对流层大气的特点。假如能够掌握大气波导的变化规律，准确而及时地预测、预报其出现的时间和地域，结合频率、射线波束、天线仰角和发生功率的选择，就能更好地发挥探测及通信设备的作用。因此，根据实际大气环境，分析、预测和预报大气波导，对于评估大气环境对电磁波传播和探测系统性能具有重要的理论意义和实际价值，尤其是在军事领域。然而在海洋大气环境中，特别是在广阔的海洋上，传统的海洋测量手段，如浮标、船只，获取的海洋水文数据稀少，探空数据更少，无法满足现代海洋大气环境监测所需的大范围观测信息，即使获取数据也很难保证数据点时间和空间的有效性；同时海洋大气边界层日变化缓慢，发生的大气波导持续时间长，比较稳定，对于海上舰船通讯等急需开展大气波导环境区域研究。我国是海洋大国，海岸线狭长，更需要开展沿海区域及近海对流层大气波导研究，特别是边界层内低空大气波导环境预测、预报研究。 目前国内大气波导的研究大多数情况下只注重单站波导发生情况，其代表的时空有效性是有限的，而在实际应用中，人们更加关注波导存在区域，因为波导区域决定了电磁波传输范围。本文针对我国近海区域大气波导不同的形成机理和类型，分别进行了研究。对于蒸发波导，本文利用卫星遥感获取海水表面温度、海面风场、气温、相对湿度等资料，建立卫星遥感资料与蒸发波导高度之间的诊断模式，研究大气波导反演算法和预测模型，首次分析了蒸发波导高度的空间不均匀性。对于低空表面波导和悬空波导，利用中尺度数值模式MM5对大气波导进行了系统的研究，填补了国内中尺度大气波导数值模拟研究的空白，同时选取典型的天气过程，进行了中尺度数值模拟个例研究，分析了大气波导形成机理。 本文主要工作如下： 一、结合P-J模式，利用AMSR-E卫星数据用两种神经网络方法反演了热带海域的蒸发波导高度，并进行了比较，两种方法得到的与浮标实测参数计算得到的蒸发波导高度之间的相关系数相当，都为0.82左右，均方根差后者比前者小，分别为2.64米和1.89米。神经网络直接反演蒸发波导高度方法要优于间接反演蒸发波导高度的方法。利用AMSR-E卫星数据直接反演了南海海域的蒸发波导高度，可以清楚看出蒸发波导高度的空间分布的不均匀性，为研究蒸发波导的环境特性奠定基础。 二、以2005年6月2日出现在黄海海域的大气波导为例，设计了三种数值试验，利用MM5模式对大气波导进行了24h数值模拟研究，包括是否加入常规探空资料进行格点分析同化，垂直分层的多寡，粗、细网格模拟比较分析。通过研究得到了以下结论：（1）利用常规资料进行格点同化对海上大气波导影响较小，对陆地影响较大；（2）垂直分层对大气波导特征参数影响明显，层数较多描述大气波导比较合适；（3）对于海上大气波导，细网格刻画大气波导特征参数比较细致，但粗网格的模拟已经足够进行大气波导分析。 三、以NCEP再分析资料为背景场,利用MM5模式，对2005年6月2日至4日出现在黄海海域的大气波导和2008年5月10日12时到11日12时由‘威玛逊’台风引起的大气波导分别进行了数值模拟研究。通过研究发现：模拟的修正折射率廓线与探空数据得到的廓线基本吻合，但是模拟的较强波导强度和高度普遍比实测的要小得多；同时给出了大气波导参数的演变过程，分析了两次大气波导形成的天气过程，说明了模拟的大气波导区域是可靠的和可信的； MM5模式能够模拟出特定天气条件下低空较强大气波导三维空间变化过程，可以为定量地描述大气波导特征提供理论和试验依据。

基于微波图像的辽东湾海冰典型要素信息提取

业务化海冰遥感监测的要素主要包括海冰类型、海冰厚度和海冰范围（冰缘线）。本文针对辽东湾海冰的特点，结合星载合成孔径雷达（SAR）、机载微波辐射计（ABMR）和星载微波辐射计遥感图像观测辽东湾海冰，发展海冰要素探测算法，提取海冰类型、海冰厚度和冰缘线信息。 海冰类型SAR信息提取方面，首先分析了双极化ENVISAT ASAR数据不同极化方式的海冰探测能力。然后利用SAR图像，结合同步TM数据和航空照片，开展辽东湾地区不同类型海冰的电磁特性研究。研究结果表明，SAR图像可较好地区分沿岸固定冰、平整冰和碎冰堆积区，但在探测初生冰方面并不可靠，探测结果与海冰生长阶段以及海冰周围环境条件有关。根据SAR图像中海冰类型的分析，将PCNN神经网络用于海冰SAR图像的分割和海冰分类，并对PCNN做了简化和改进。经SAR图像分类结果测试，简化和改进后的PCNN可较好地区分SAR图像中的海冰类型。通过分析了PCNN网络各参数对SAR图像分割结果的影响，指出了各参数的取值范围，并在此基础上建立了基于PCNN神经网络的海冰SAR图像半自动分类判读系统。 海冰厚度机载微波辐射计信息提取方面，推导了非相干模式的ABMR海冰厚度反演模型，并首次得到了模型中高阶亮温辐射项的计算表达式。通过对模型的分析，指出ABMR只能测量一定范围内的海冰厚度。其中，最大海冰探测厚度不仅决定于ABMR的波长，而且还受到仪器精度的限制；最小海冰探测厚度只与仪器的波长有关。在此基础上，定量计算了我国几种常用波长ABMR的海冰厚度探测范围。并结合辽东湾海冰冰情等级，指出适合不同海冰冰情等级的ABMR的选择。最后分析了海冰厚度反演模型的影响因素和模型的适应范围。 海冰边缘线星载微波辐射计信息提取方面，首次将PSSM算法用于AMSR数据，提取辽东湾海冰边缘线信息，可得到空间分辨率为2.5km、重复周期为1d的辽东湾海冰边缘线信息。通过提取结果与Jason-1高度计和SAR探测海冰边缘线的比较，验证了PSSM方法探测辽东湾海冰边缘线的有效性。 论文的最后对全文做了总结，并提出和讨论了需要进一步开展的工作。

The effects of variation in snow properties on snow mass estimation using the Chang algorithm

Estimating snow mass at continental scales is difficult but important for understanding landatmosphere interactions, biogeochemical cycles and Northern latitudes’ hydrology. Remote sensing provides the only consistent global observations, but the uncertainty in measurements is poorly understood. Existing techniques for the remote sensing of snow mass are based on the Chang algorithm, which relates the absorption of Earth-emitted microwave radiation by a snow layer to the snow mass within the layer. The absorption also depends on other factors such as the snow grain size and density, which are assumed and fixed within the algorithm. We examine the assumptions, compare them to field measurements made at the NASA Cold Land Processes Experiment (CLPX) Colorado field site in 2002–3, and evaluate the consequences of deviation and variability for snow mass retrieval. The accuracy of the emission model used to devise the algorithm also has an impact on its accuracy, so we test this with the CLPX measurements of snow properties against SSM/I and AMSR-E satellite measurements.

The effects of variation in snow properties on passive microwave snow mass estimation

Estimating snow mass at continental scales is difficult, but important for understanding land-atmosphere interactions, biogeochemical cycles and the hydrology of the Northern latitudes. Remote sensing provides the only consistent global observations, butwith unknown errors. Wetest the theoretical performance of the Chang algorithm for estimating snow mass from passive microwave measurements using the Helsinki University of Technology (HUT) snow microwave emission model. The algorithm's dependence upon assumptions of fixed and uniform snow density and grainsize is determined, and measurements of these properties made at the Cold Land Processes Experiment (CLPX) Colorado field site in 2002–2003 used to quantify the retrieval errors caused by differences between the algorithm assumptions and measurements. Deviation from the Chang algorithm snow density and grainsize assumptions gives rise to an error of a factor of between two and three in calculating snow mass. The possibility that the algorithm performsmore accurately over large areas than at points is tested by simulating emission from a 25 km diameter area of snow with a distribution of properties derived from the snow pitmeasurements, using the Chang algorithm to calculate mean snow-mass from the simulated emission. The snowmass estimation froma site exhibiting the heterogeneity of the CLPX Colorado site proves onlymarginally different than that from a similarly-simulated homogeneous site. The estimation accuracy predictions are tested using the CLPX field measurements of snow mass, and simultaneous SSM/I and AMSR-E measurements.

A statistical model for sea surface diurnal warming driven by numercial weather predictions fluxes and winds

A statistical model is derived relating the diurnal variation of sea surface temperature (SST) to the net surface heat flux and surface wind speed from a numerical weather prediction (NWP) model. The model is derived using fluxes and winds from the European Centre for Medium-Range Weather Forecasting (ECMWF) NWP model and SSTs from the Spinning Enhanced Visible and Infrared Imager (SEVIRI). In the model, diurnal warming has a linear dependence on the net surface heat flux integrated since (approximately) dawn and an inverse quadratic dependence on the maximum of the surface wind speed in the same period. The model coefficients are found by matching, for a given integrated heat flux, the frequency distributions of the maximum wind speed and the observed warming. Diurnal cooling, where it occurs, is modelled as proportional to the integrated heat flux divided by the heat capacity of the seasonal mixed layer. The model reproduces the statistics (mean, standard deviation, and 95-percentile) of the diurnal variation of SST seen by SEVIRI and reproduces the geographical pattern of mean warming seen by the Advanced Microwave Scanning Radiometer (AMSR-E). We use the functional dependencies in the statistical model to test the behaviour of two physical model of diurnal warming that display contrasting systematic errors.

Global snow mass measurements and the effect of stratigraphic detail on inversion of microwave brightness temperatures

Snow provides large seasonal storage of freshwater, and information about the distribution of snow mass as Snow Water Equivalent (SWE) is important for hydrological planning and detecting climate change impacts. Large regional disagreements remain between estimates from reanalyses, remote sensing and modelling. Assimilating passive microwave information improves SWE estimates in many regions but the assimilation must account for how microwave scattering depends on snow stratigraphy. Physical snow models can estimate snow stratigraphy, but users must consider the computational expense of model complexity versus acceptable errors. Using data from the National Aeronautics and Space Administration Cold Land Processes Experiment (NASA CLPX) and the Helsinki University of Technology (HUT) microwave emission model of layered snowpacks, it is shown that simulations of the brightness temperature difference between 19 GHz and 37 GHz vertically polarised microwaves are consistent with Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E) and Special Sensor Microwave Imager (SSM/I) retrievals once known stratigraphic information is used. Simulated brightness temperature differences for an individual snow profile depend on the provided stratigraphic detail. Relative to a profile defined at the 10 cm resolution of density and temperature measurements, the error introduced by simplification to a single layer of average properties increases approximately linearly with snow mass. If this brightness temperature error is converted into SWE using a traditional retrieval method then it is equivalent to ±13 mm SWE (7% of total) at a depth of 100 cm. This error is reduced to ±5.6 mm SWE (3 % of total) for a two-layer model.

(Table 1) Iceberg characteristics and start and end date and location of iceberg buoys in the Weddell Sea

The drift of 52 icebergs tagged with GPS buoys in the Weddell Sea since 1999 has been investigated with respect to prevalent drift tracks, sea ice/iceberg interaction, and freshwater fluxes. Buoys were deployed on small- to medium-sized icebergs (edge lengths ? 5 km) in the southwestern and eastern Weddell Sea. The basin-scale iceberg drift of this size class was established. In the western Weddell Sea, icebergs followed a northward course with little deviation and mean daily drift rates up to 9.5 ± 7.3 km/d. To the west of 40°W the drift of iceberg and sea ice was coherent. In the highly consolidated perennial sea ice cover of 95% the sea ice exerted a steering influence on the icebergs and was thus responsible for the coherence of the drift tracks. The northward drift of buoys to the east of 40°W was interrupted by large deviations due to the passage of low-pressure systems. Mean daily drift rates in this area were 11.5 ± 7.2 km/d. A lower threshold of 86% sea ice concentration for coherent sea ice/iceberg movement was determined by examining the sea ice concentration derived from Special Sensor Microwave Imager (SSM/I) and Advanced Microwave Scanning Radiometer for EOS (AMSR-E) satellite data. The length scale of coherent movement was estimated to be at least 200 km, about half the value found for the Arctic Ocean but twice as large as previously suggested. The freshwater fluxes estimated from three iceberg export scenarios deduced from the iceberg drift pattern were highly variable. Assuming a transit time in the Weddell Sea of 1 year, the iceberg meltwater input of 31 Gt which is about a third of the basal meltwater input from the Filchner Ronne Ice Shelf but spreads across the entire Weddell Sea. Iceberg meltwater export of 14.2 × 103 m3 s?1, if all icebergs are exported, is in the lower range of freshwater export by sea ice.

Daily polynya area and ice production in the North Water polynya region

The North Water (NOW) Polynya is a regularly-forming area of open-water and thin-ice, located between northwestern Greenland and Ellesmere Island (Canada) at the northern tip of Baffin Bay. Due to its large spatial extent, it is of high importance for a variety of physical and biological processes, especially in wintertime. Here, we present a long-term remote sensing study for the winter seasons 1978/1979 to 2014/2015. Polynya characteristics are inferred from (1) sea ice concentrations and brightness temperatures from passive microwave satellite sensors (Advanced Microwave Scanning Radiometer (AMSR-E and AMSR2), Scanning Multichannel Microwave Radiometer (SMMR), Special Sensor Microwave Imager/Sounder (SSM/I-SSMIS)) and (2) thin-ice thickness distributions, which are calculated using MODIS ice-surface temperatures and European Center for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis data in a 1D thermodynamic energy-balance model. Daily ice production rates are retrieved for each winter season from 2002/2003 to 2014/2015, assuming that all heat loss at the ice surface is balanced by ice growth. Two different cloud-cover correction schemes are applied on daily polynya area and ice production values to account for cloud gaps in the MODIS composites. Our results indicate that the NOW polynya experienced significant seasonal changes over the last three decades considering the overall frequency of polynya occurrences, as well as their spatial extent. In the 1980s, there were prolonged periods of a more or less closed ice cover in northern Baffin Bay in winter. This changed towards an average opening on more than 85% of the days between November and March during the last decade. Noticeably, the sea ice cover in the NOW polynya region shows signs of a later-appearing fall freeze-up, starting in the late 1990s. Different methods to obtain daily polynya area using passive microwave AMSR-E/AMSR2 data and SSM/I-SSMIS data were applied. A comparison with MODIS data (thin-ice thickness < 20 cm) shows that the wintertime polynya area estimates derived by MODIS are about 30 to 40% higher than those derived using the polynya signature simulation method (PSSM) with AMSR-E data. In turn, the difference in polynya area between PSSM and a sea ice concentration (SIC) threshold of 70% is fairly low (approximately 10%) when applied to AMSR-E data. For the coarse-resolution SSM/I-SSMIS data, this difference is much larger, particularly in November and December. Instead of a sea ice concentration threshold, the PSSM method should be used for SSM/I-SSMIS data. Depending on the type of cloud-cover correction, the calculated ice production based on MODIS data reaches an average value of 264.4 ± 65.1 km**3 to 275.7 ± 67.4 km**3 (2002/2003 to 2014/2015) and shows a high interannual variability. Our achieved long-term results underline the major importance of the NOW polynya considering its influence on Arctic ice production and associated atmosphere/ocean processes.

Comparisons of observed ice properties and Envisat Advanced Synthetic Aperture Radar (ASAR) data during Nathanial B. Palmer cruise NBP09-01 and Oden cruise OSO08/09

Envisat Advanced Synthetic Aperture Radar (ASAR) Wide Swath Mode (WSM) images are used to derive C-band HH-polarization normalized radar cross sections (NRCS). These are compared with ice-core analysis and visual ship-based observations of snow and ice properties observed according to the Antarctic Sea Ice Processes and Climate (ASPeCt) protocol during two International Polar Year summer cruises (Oden 2008 and Palmer 2009) in West Antarctica. Thick first-year (TFY) and multi-year (MY) ice were the dominant ice types. The NRCS value ranges between -16.3 ± 1.1 and -7.6 ± 1.0 dB for TFY ice, and is -12.6 ± 1.3 dB for MY ice; for TFY ice, NRCS values increase from ~-15 dB to -9 dB from December/January to mid-February. In situ and ASPeCt observations are not, however, detailed enough to interpret the observed NRCS change over time. Co-located Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E) vertically polarized 37 GHz brightness temperatures (TB37V), 7 day and 1 day averages as well as the TB37V difference between ascending and descending AMSR-E overpasses suggest the low NRCS values (-15 dB) are associated with snowmelt being still in progress, while the change towards higher NRCS values (-9dB) is caused by commencement of melt-refreeze cycles after about mid-January.

(Table 1) Ice thickness at lake Nam Co, Tibet, during 2008-2009

Lake ice change is one of the sensitive indicators of regional and global climate change. Different sources of data are used in monitoring lake ice phenology nowadays. Visible and Near Infrared bands of imagery (VNIR) are well suited for the observation of freshwater ice change, for example data from AVHRR and MODIS. Active and passive microwave data are also used for the observation of lake ice, e.g., from satellite altimetry and radiometry, backscattering coefficient from QuickSCAT, brightness temperature (Tb) from SSM/I, SMMR, and AMSR-E. Most of the studies are about lake ice cover phenology, while few studies focus on lake ice thickness. For example, Hall et al. using 5 GHz (6 cm) radiometer data showed a good relationship between Tb and ice thickness. Kang et al. found the seasonal evolution of Tb at 10.65 GHz and 18.7 GHz from AMSR-E to be strongly influenced by ice thickness. Many studies on lake ice phenology have been carried out since the 1970s in cold regions, especially in Canada, the USA, Europe, the Arctic, and Antarctica. However, on the Tibetan Plateau, very little research has focused on lake ice-cover change; only a small number of published papers on Qinghai Lake ice observations. The main goal of this study is to investigate the change in lake ice phenology at Nam Co on the Tibetan Plateau using MODIS and AMSR-E data (monitoring the date of freeze onset, the formation of stable ice cover, first appearance of water, and the complete disappearance of ice) during the period 2000-2009.