A non-regression-coefficients method of sea surface temperature retrieval from space

For several years now NOAA/NESDIS have derived an operational global sea surface temperature (SST) product from the AVHRR instrument on the NOAA satellites. This is done using the MCSST and CPSST algorithms which contain coefficients that are determined from a regression analysis of satellite data against in situ surface data. The current algorithms are used to provide global SST data without taking into account the latitude, climate or location of the satellite data, although the CPSST coefficients do have a weak dependence on the satellite brightness temperatures. Because of this global application the current SST algorithms have inherent errors due to local climate influences. In this paper a new SST algorithm is developed that does not rely on regression analysis to derive its coefficients. By using the spatial variation of the brightness temperatures in a small area (50 km by 50 km) it is possible to derive the appropriate coefficients to use in the algorithm. The SST field can thus be derived at any location without need for prior determination of the algorithm coefficients. In a simulation study, data from twenty-five radiosonde ascents-arc use with an atmospheric transmission model to derive a range of atmospheric transmittances and satellite brightness temperatures. Coincident AVHRR data and ship data are used to assess the accuracy of the new algorithm. The various dependencies of the terms in the SST algorithm are investigated. As with the MCSST and CPSST algorithms, the new method has largest errors when applied in situations of abnormal atmospheric structure. The improvement over the MCSST product may initially be only marginal, but with the advent of the more precise data from the Along Track Scanning Radiometer (ATSR) a more accurate global SST product may be possible.     

Atmospheric effects on SMMR and SSM/1 37 GHz polarization difference over the Sahel

Abstract

The difference of vertically and horizontally polarized brightness temperatures (referred to here as the polarization difference, δT) observed at 37GHz frequency of the scanning multi-channel microwave radiometer (SMMR) on board the Nimbus-7 satellite and special sensor microwave imager (SSM/I) on board the DMSP-F8 satellite could provide useful information about land surface change within the span of these global observations, November 1978 to August 1987 for SMMR and July 1987 to present for SSM/I. The atmospheric effects on the δT are studied over two 2-5° by 2-5° regions within the Sahel and Sudan zones or Africa from January 1985 to December 1986 through radiative transfer analysis using surface temperature, atmospheric water vapour and cloud optical thickness developed under the International Satellite Cloud Climatology Project (ISCCP). The atmospheric effects are also studied using surface observations of air temperature and vapour pressure at Niamey (13-5° N, 2-2° E) for the period January 1979 to December 1990. It is found that atmospheric effects alone cannot explain the observed temporal variation of δT, although the atmosphere introduces important modulations on the observed seasonal variations of δT due to rather significant seasonal variation of precipitable water vapour. Therefore, these δT data should be corrected for atmospheric effects before any quantitative analysis of land surface change over the Sahel and Sudan zones. The entire global data set from December 1978 to December 1990 has been archived for unrestricted distribution and use.     

Estimating surface air temperatures,from Meteosat land surface temperatures,using an empirical solar zenith angle model

Temperature values derived from Meteosat are an indication of emitted long-wave radiation, and are not a true indication of ambient air temperature. The authors believe that Solar Zenith Angle (SZA) can be used as a proxy for solar energy reaching the ground surface, and its subsequent effects upon the land surface temperature detected by Meteosat. Raw satellite temperatures often overestimate the actual screen temperature during the day, and underestimate at night. By using a statistical model which relates Meteosat and WMO screen temperature deviations, and SZA values, it has been possible to generate a correction algorithm which minimizes these differences. The algorithm generates a new proxy value, being a simulated ambient (screen) air temperature. The algorithms achieve an accuracy of within 3 C for over 70% of the Meteosat temperatures processed. The operational use of this algorithm requires only the raw Meteosat temperature value, and the SZA. Such temperature corrections are useful for a wide range of environmental monitoring applications. An example is in the field of vector-borne disease modelling which requires proxies for temperature across large regions, and where more conventional meteorological stations are inadequate.     

Nearshore sea temperature variability off Rottnest Island (Western Australia) derived from satellite data

While satellite‐derived sea‐surface temperatures (SSTs) have been widely used in open ocean monitoring, they have rarely been applied to the nearshore region. In this study, full‐resolution Advanced Very High Resolution Radiometer (AVHRR) satellite data were used to derive a sea temperature climatology for three nearshore sites at Rottnest Island (Western Australia) for the 7‐year period 1995 to 2001. Because of the proximity to the land, image geolocation and careful pixel selection were crucial. The mean annual cycle shows the influence of both air–sea heat flux in the shallow waters and the seasonal strengthening of the tropical Leeuwin Current. Self‐recording temperature loggers were installed on near‐surface buoys for a few months in 2001 to assess how reliably the satellite temperatures could be used in such a dynamic nearshore environment. About 80% of the individual satellite temperatures were within ±0.5°C of the logger measurements (at the same time of day) and over 95% were within ±1°C. Occasionally, however, the satellite temperatures differed from the in situ measurements by more than 2°C and possible reasons for this are discussed. The monthly mean satellite temperatures were generally within 0.3°C of the monthly logger averages, which may be taken as a practical reliability limit for the climatology.     

Combining thermal inertia and a diurnal temperature difference cycle model to estimate thermal inertia from MSG-SEVIRI data

Thermal inertia is an important parameter in geological and agricultural applications. In this study, we present a method that combines models of thermal inertia and the diurnal temperature difference cycle to estimate the thermal inertia from Meteosat Second Generation Spinning Enhanced Visible and Infrared Imager (MSG-SEVIRI) data. This method can directly derive thermal inertia from MSG-SEVIRI brightness temperatures without the need to include the land surface temperature and emissivity. Two important parameters (the time of the maximum temperature and the diurnal temperature difference) that were input into the thermal inertia model were obtained by fitting the diurnal temperature difference cycle model to the diurnal cycle of land surface temperatures. The spatial distribution of thermal inertia shows that high thermal inertia values occur over vegetated areas, whereas low thermal inertia values occur over bare areas. The uncertainty in thermal inertia is investigated in terms of the uncertainties in the surface albedo, the time of the maximum temperature, and the diurnal temperature difference. The results indicate that the uncertainty in thermal inertia over vegetated areas is greater than that over bare areas. The consistency of the thermal inertia model is evaluated by analysing the difference in thermal inertia values on two consecutive days. The root mean square error of the thermal inertia differences under nearly identical surface and atmospheric conditions on two consecutive days is considered to be the error of the thermal inertia model.     

Estimating subpixel surface temperatures and energy fluxes from the vegetation index-radiometric temperature relationship

Routine (i.e., daily to weekly) monitoring of surface energy fluxes, particularly evapotranspiration (ET), using satellite observations of radiometric surface temperature has not been feasible at high pixel resolution (i.e., ∼10¹-10² m) because of the low frequency in satellite coverage over the region of interest (i.e., approximately every 2 weeks). Cloud cover further reduces the number of useable observations of surface conditions resulting in high-resolution satellite imagery of a region typically being available once a month, which is not very useful for routine ET monitoring. Radiometric surface temperature observations at ∼1- to 5-km pixel resolution are available multiple times per day from several weather satellites. However, this spatial resolution is too coarse for estimating ET from individual agricultural fields or for defining variations in ET due to land cover changes. Satellite data in the visible and near-infrared wavelengths, used for computing vegetation indices, are available at resolutions an order of magnitude smaller than in the thermal-infrared, and hence provide higher resolution information on vegetation cover conditions. A number of studies have exploited the relationship between vegetation indices and radiometric surface temperature for estimating model parameters used in computing spatially distributed fluxes and available moisture. In this paper, the vegetation index-radiometric surface temperature relationship is utilized in a disaggregation procedure for estimating subpixel variation in surface temperature with aircraft imagery collected over the US Southern Great Plains. The disaggregated surface temperatures estimated by this procedure are compared to actual observations at this subpixel resolution. In addition, a remote sensing-based energy balance model is used to compare output using actual versus estimated surface temperatures over a range of pixel resolutions. From these comparisons, the utility of the surface temperature disaggregation technique appears to be most useful for estimating subpixel surface temperatures at resolutions corresponding to length scales defining agricultural field boundaries across the landscape.     

Synthesis of passive microwave and radar altimeter data for estimating accumulation rates of dry polar snow

In this paper, we compare dry-snow extinction coefficients derived from satellite radar altimeter data with brightness temperature data from passive microwave measurements over a portion of the East Antarctic plateau. The comparison between the extinction coefficients and the brightness temperatures shows a strong negative correlation, where the correlation coefficients ranged from –0·87 to –0·95. The large-scale trend shows that the extinction coefficient of the dry polar snow decreases with increasing surface elevation, while the average brightness temperature increases with surface elevation. Our analysis shows that the observed trends are related to geographical variations in scattering coefficient of snow, which, in turn, are controlled by variations in surface temperature and snow accumulation rate. By combining informa.tion present in the extinction coefficient and brightness temperature datasets, we develop a simple semi-empirical model that can be used to obtain accumulation rate estimates of dry polar snow.     

Lake surface temperature estimation using NOAA satellite APT data†

Abstract

The water surface temperature of Lake Okeechobee under clear sky conditions was investigated in this study through the use of Automatic Picture Transmission (APT) thermal infrared data from the National Oceanic and Atmospheric Administration (NOAA) satellite Advanced Very High Resolution Radiometers (AVHRR). Cloud-free data were acquired during the months of May and December 1989. An image processing approach and regression model were used to convert the grey-level digital numbers of the APT imagery into temperature values. A third-order polynomial equation was successfully used to convert the digital numbers into temperature values with a high coefficient of determination (r²=0·98). The satellite-derived temperatures for Lake Okeechobee were correlated with ground-truth temperature measurements (r² = 0·74). The results of this study indicate that APT data are capable of being converted into lake surface temperatures with acceptable accuracy. Results of paired Mest and variogram analyses indicate that the temperature uniformity of the lake varied with the season, space, and time of day     

A dynamic global cloud layer for virtual globes

We describe a technique to merge multiple environmental satellite data sets for an hourly updated, near real-time global depiction of cloud cover for virtual globe applications. A global thermal infrared composite obtained from merged geostationary- (GEO) and low-Earth-orbiting (LEO) satellite data is processed to depict clear and cloudy areas in a visually intuitive fashion. This GEO-plus-LEO imagery merging is complicated by the fact that each individual satellite observes a single ‘snapshot’ of the cloud patterns, each taken at different times, whereas the underlying clouds themselves are constantly moving and evolving. For the cloudy areas, the brightness and transparency are approximated based upon the cloud top temperature relative to the local radiometric surface temperatures (corrected for surface emissivity variations) at the time of the satellite observation. The technique clearly defines and represents mid- to high-level clouds over both land and ocean. Due to their proximity to the Earth's surface, low-level clouds such as stratocumulus and stratus clouds will be poorly represented with the current technique, since warmer temperatures in this case do not correspond to higher cloud transparency. Overcoming this problem requires the introduction of multispectral channel combinations.     

Determination of land surface spectral reflectances using Meteosat and NOAAA/AVHRR Shortwave Channel Data

Abstract

A method to derive surface spectral reflectances from currently available Meteosat geostationary and NOAA/AVHRR polar orbiting satellite data is described. Broadband reflectance was derived from Meteosat measurements while NOAA/AVHRR vegetation index provided a spectral weighting which enabled the spectral reflectances on either side of 0-7 μm to be estimated. The method takes into account satellite calibrations, viewing geometry, and correction of some atmospheric effects. Conversion from narrow-band to broadband reflectances is discussed. The method was applied to a month of data to obtain the surface spectral reflectances of Africa which are compared with some data sets used by climate modellers, in order to assess them and to monitor their seasonal and interannual changes on a global scale.     

Derivation of pan-Arctic soil decomposition rate constant,heterotrophic respiration and NEE using AMSR-E and MODIS data

An approach was developed for regional assessment and monitoring of land-atmosphere carbon dioxide (CO₂) exchange, soil heterotrophic respiration (R _h), and vegetation productivity of Arctic tundra using global satellite remote sensing at optical and microwave wavelengths. C- and X-band brightness temperatures were used from the Advanced Microwave Scanning Radiometer Earth Observing System (AMSR-E) to extract surface wetness and temperature, and MODerate Resolution Imaging Spectroradiometer (MODIS) data were used to derive land cover, Leaf Area Index (LAI), and Net Primary Production (NPP) information. Calibration and validation activities involve comparisons between satellite remote sensing and tundra CO₂ eddy flux towers, and hydroecological process model simulations. Analyses of spatial and temporal anomalies and environmental drivers of land-atmosphere net CO₂ exchange at weekly and annual time steps were conducted. Surface soil moisture and temperature, as detected from satellite remote-sensing observations, were found to be major drivers for spatial and temporal patterns of tundra net ecosystem CO₂ exchange and photosynthetic and respiration processes. Satellite microwave measurements are capable of capturing seasonal variations and regional patterns in tundra soil heterotrophic respiration and CO₂ exchange, while the ability to extract spatial patterns at the scale of surface heterogeneity is limited by the coarse spatial scale of the satellite remote-sensing footprint. The microwave-derived surface temperature and soil moisture were used to estimate net ecosystem carbon exchange (NEE) at the boreal-Arctic region. These were validated using flux tower sites data. Existing satellite-based measurements of vegetation structure (i.e. LAI) and productivity (i.e. Gross Primary Production (GPP) and NPP) from the Aqua/Terra MODIS with the AMSR-E-derived land-surface temperature and soil moisture were used and integrated. Spatially explicit estimates of NEE for the pan-Arctic region at daily, weekly and annual intervals were derived. Comparative analysis of satellite data-derived NEE with measurements from CO₂ eddy flux tower sites and the BIOME-BGC model were carried out and good agreement was found. The comparative analysis is statistically significant with high regression (i.e. R ²?=?0.965), especially in the R _h calculation and the overall NEE regression is 0.478. The results also indicate that the carbon cycle response to climate change is nonlinear and is strongly coupled to Arctic surface hydrology.     

A sampling strategy for satellite sensor-based assessments of the urban heat-island bias

A data sampling strategy was developed for the use of a satellite sensor-based methodology to estimate the urban heat-island temperature bias associated with climate observation stations. NOAA-AVHRR observations at a grid scale of 1 km 2 1 km were analysed on a local (3 km 2 3 km) and regional (41 km 2 41 km) basis centred on the climate observation stations of interest. The grid cells of the regional sample were evaluated and only those designated as rural were used in further analysis. Local and regional differences in the normalized difference vegetation index and radiant surface temperature were used to estimate the urban heat-island bias associated with the climate observation stations. These values were compared to a population-based methodology and a satellite sensor should read satellite sensor/station-derived methodology that required locations of known rural observation stations associated with the local observation station of interest. Generally, the heat-island bias estimates provided by the methodology that relied solely on satellite sensor data were similar to the other methodologies. The datasets used to identify the regional grid cells as rural are available on a global basis as are the vegetation index and radiant surface temperature data. Thus, the satellite sensor-based methodology developed may be uniformly applicable on a global basis.     

Land surface temperature and emissivity estimation from passive sensor data: Theory and practice-current trends

Abstract

Land surface temperature (LST) and emissivity for large areas can only be derived from surface-leaving radiation measured by satellite sensors. These measurements represent the integrated effect of the surface and are, thus, for many applications, superior to point measurements on the ground, e.g. in Earth's radiation budget and climate change detection. Over the years, a substantial amount of research was dedicated to the estimation of LST and emissivity from passive sensor data. This article provides the theoretical basis and gives an overview of the current status of this research. Sensors operating in the visible, infrared and microwave range onboard various meteorological satellites are considered, e.g. Meteosat-MVIRI, NOAA-AVHRR, ERS-ATSR, Terra-MODIS, Terra-ASTER and DMSP-SSM/I. Atmospheric effects on measured brightness temperatures are described and atmospheric corrections using radiative transfer models (RTM) are explained. The substitution of RTM with neural networks (NN) for faster forward calculations is also discussed. The methods reviewed for LST estimation are the single-channel method, the split-window techniques (SWT), and the multi-angle method, and, for emissivity estimation, the normalized emissivity method (NEM), the thermal infrared spectral indices (TISI) method, the spectral ratio method, alpha residuals, normalized difference vegetation index (NDVI )-based methods, classification-based emissivity and the temperature emissivity separation (TES) algorithm.     

卷积神经网络提取风云影像土壤湿度

目的 时空分辨率较高的土壤湿度数据对于生产实践和科学研究具有重要意义。以国产的风云气象卫星为数据源,利用卷积神经网络自主学习输入变量间深层关联的优势,获取高质量土壤湿度数据,为科学研究和生产实践服务。方法 首先构建了一个土壤湿度提取卷积神经网络（soil moisture convolutional neural network,SMCNN）,SMCNN由温度子网络和土壤湿度子网络构成,每个子网络均包含特征提取器和编码器。特征提取器用于为每个像素生成一个特征向量,其中温度子网络的特征提取器由11个卷积层组成,湿度子网络的特征提取器由9个卷积层组成,卷积层均使用1×1的卷积核。编码器用于将提取到的特征拟合为目标变量。两个子网络均使用平均方差作为损失函数。使用随机梯度下降算法对模型进行训练,最后利用训练好的模型提取区域土壤湿度数据。结果 选择宁夏回族自治区为实验区,利用获取的2016-2019年风云3D影像和相应地面站点数据作为实验数据,选择线性回归模型、BP（back propagation）神经网络模型作为对比模型开展数据实验,选择均方根误差作为评价指标。实验结果表明,SMCNN的均方根误差为0.006 7,优于对比模型,SMCNN模型在从风云影像中提取土壤湿度方面具有优势。结论 本文利用卷积神经网络分别构建用于反演地表温度和土壤湿度的子网络,再组成一个完整的土壤湿度反演网络结构,从风云3D数据中获取数值精度、时空分辨率均较高的土壤湿度数据,满足了科学研究和生产实践对大范围高精度土壤湿度数据的需求。     

Zenith angle effects in multichannel infrared sea surface remote sensing

Several approaches to infrared multichannel sea surface temperature retrievals propose using a universal set of constants. It is shown that the single-channel multiangle technique (i.e., GOES and NOAA) and the multichannel single-angle technique (NOAA-n) are similar concepts with a common derivation from radiative transfer theory. It is also shown that the linear correlation factor between surface temperature minus satellite temperature in one channel versus the difference in satellite temperatures in two channels is not independent of the difference in satellite sensed equivalent blackbody temperature. The 3.7-μm, 11-μm, and 12-μm channels on the NOAA-n AVHRR can be used in combination to compute atmospheric transmissivity and average atmospheric temperature, but a better combination would be substituting three 0.5-μm-wide channels centered on 11.25 μm, 11.75 μm, and 12.25 μm. A triple window multispectral scanner in the 11–12.5 μm region allows determination of diffuse surface reflectance which can bias sea surface temperatures $\text{?0.4 K±0.3 K}$.     

On the Earth's surface energy exchange determination from ERS satellite ATSR data. Part I: Long-wave radiation

This is the first in a series of papers which addresses the determination of the Earth's surface energy exchange using data from the Along-Track Scanning Radiometer (ATSR). This paper focuses on long-wave radiation from sea and land surfaces and a technique is proposed for the derivation of land surface temperature (LST) and land surface emissivity retrieval using ATSR data in a new simultaneous split-window method. Two points regarding net long-wave radiation are also considered. Firstly, over land and sea, differences in several previously published are discussed. Secondly, over sea, the effect on the net longwave radiation of using sea surface skin temperatures, which can be derived accurately from satellite thermal band data, as input to the empirical formulae is compared to the use of bulk water temperature taken from in situ measurements. Finally, a new formula is developed for the calculation of net long-wave radiation at the surface. The equivalent sky temperature, T, is used and the results agree sky with those obtained using the Oberhuber formula. Both of these formulae are useful for the calculation of net long-wave radiation over wet areas such as sea, with a high relative humidity. Initial tests of the formulae were carried out using ATSR sea surface temperatures (SSTs) and LSTs in the United Kingdom. The formulae were also tested using TOGA-TAO data and ATSR sea surface temperatures over the Pacific Ocean. From our results, the net long-wave radiation showed the magnitude and spatial variability to be (15-35 W m 2). The absolute difference of net long-wave radiation by using ATSR SST and TOGA SST is around 3 W m 2 for most areas, but the maximum difference is up to 7W m 2. The relative difference is more than 10% and up to 30%.     

Accuracy of in-flight calibration of the water-vapour channel of a satellite radiometer by radiance calculation

Abstract

The accuracy of a calibration of a water-vapour channel of a radiometer on board a satellite by using a radiative transfer code with radiosonde profiles was examined with the aid of the calibrated WV channel data of the Visible and Thermal Infrared Radiometer (VTIR) on the MOS-1 satellite. It has been verified that the assumption that a linear extrapolation for the relative humidity in the upper troposphere was reasonable for estimating the radiances in the water-vapour channel by a comparison of the brightness temperatures of the channel measured by the calibrated VTIR with those calculated by the radiative transfer code. An error in water-vapour distribution or radiosonde data considerably affected the estimated brightness temperature by the radiance calculation in the water vapour absorption band. The accuracy of the calibration of the water-vapour channel by the radiance calculation was with the root-mean-square (rms) temperature difference of 3–96°C.     

Relationship between AVHRR surface temperature and NDVI in Arctic tundra ecosystems

A negative, linear relationship between thermal emissions and a spectral vegetation index has been demonstrated for numerous mid‐latitude ecosystems. In this study, it is hypothesized that the relationship between surface temperature and the normalized difference vegetation index (NDVI) will be linear, but positive in Arctic tundra ecosystems due to the contrast between warm vegetation and the cold soil/moss background. This hypothesis is tested using Advanced Very High Resolution Radiometer (AVHRR) data collected over the North Slope of Alaska on three days during the summer of 1999. Results of the study generally provide support for this hypothesis. However, a consistent relationship observed across two contrasting physiographic provinces on one study day was shown to change the following day and could not be readily explained by differences in satellite zenith angle or observed air temperature. Surface temperatures are shown to respond directly to spatial and temporal variations in air temperature.     

Modelling the diurnal variations of urban heat islands with multi-source satellite data

Examination of the diurnal variations in surface urban heat islands (UHIs) has been hindered by incompatible spatial and temporal resolutions of satellite data. In this study, a diurnal temperature cycle genetic algorithm (DTC-GA) approach was used to generate the hourly 1 km land-surface temperature (LST) by integrating multi-source satellite data. Diurnal variations of the UHI in ‘ideal’ weather conditions in the city of Beijing were examined. Results show that the DTC-GA approach was applicable for generating the hourly 1 km LSTs. In the summer diurnal cycle, the city experienced a weak UHI effect in the early morning and a significant UHI effect from morning to night. In the diurnal cycles of the other seasons, the city showed transitions between a significant UHI effect and weak UHI or urban heat sink effects. In all diurnal cycles, daytime UHIs varied significantly but night-time UHIs were stable. Heating/cooling rates, surface energy balance, and local land use and land cover contributed to the diurnal variations in UHI. Partial analysis shows that diurnal temperature range had the most significant influence on UHI, while strong negative correlations were found between UHI signature and urban and rural differences in the normalized difference vegetation index, albedo, and normalized difference water index. Different contributions of surface characteristics suggest that various strategies should be used to mitigate the UHI effect in different seasons.     

Measurement of land surface temperature from oblique angle airborne thermal camera observations

ABSTRACT

This paper presents a novel airborne remote sensing method using thermal imaging to directly georeference and calculate Earth surface temperature with a high spatiotemporal resolution. A tethered balloon is used to elevate an uncooled thermal camera in the field. When deployed, images with oblique view angles of the surrounding Earth surface are collected. Images recorded from a field environmental monitoring campaign in a northern Canadian mining facility are processed with open source software, and it is shown that they successfully represent the diurnal and spatial surface temperature variations within the facility. Furthermore, in comparison to MODerate resolution Imaging Spectroradiometer (MODIS) satellite images, the approach results in a median absolute error of 0.64 K, with a bias and Root Mean Square Error (RMSE) of 0.5 K and 5.45 K, respectively.