Land Surface Emissivity Estimation Research Articles

A global historical twice-daily (daytime and nighttime) land surface temperature dataset produced by Advanced Very High Resolution Radiometer observations from 1981 to 2021

Abstract. Land surface temperature (LST) is a key variable for monitoring and evaluating global long-term climate change. However, existing satellite-based twice-daily LST products only date back to 2000, which makes it difficult to obtain robust long-term temperature variations. In this study, we developed the first global historical twice-daily LST dataset (GT-LST), with a spatial resolution of 0.05∘, using Advanced Very High Resolution Radiometer (AVHRR) Level-1b Global Area Coverage (GAC) data from 1981 to 2021. The GT-LST product was generated using four main processes: (1) GAC data reading, calibration, and preprocessing using open-source Python libraries; (2) cloud detection using the AVHRR-Phase I algorithm; (3) land surface emissivity estimation using an improved method considering annual land cover changes; (4) LST retrieval based on a nonlinear generalized split-window algorithm. Validation with in situ measurements from Surface Radiation Budget (SURFRAD) sites and Baseline Surface Radiation Network sites showed that the overall root-mean-square errors (RMSEs) of GT-LST varied from 1.6 to 4.0 K, and nighttime LSTs were typically better than daytime LSTs. Intercomparison with the Moderate Resolution Imaging Spectroradiometer LST products (MYD11A1 and MYD21A1) revealed that the overall root-mean-square difference (RMSD) was approximately 3.0 K. Compared with MYD11A1 LST, GT-LST was overestimated, and relatively large RMSDs were obtained during the daytime, spring, and summer, whereas the significantly smaller positive bias was obtained between GT-LST and MYD21A1 LST. Furthermore, we compared our newly generated dataset with a global AVHRR daytime LST product at the selected measurements of SURFRAD sites (i.e., measurements of these two satellite datasets were valid), which revealed similar accuracies for the two datasets. However, GT-LST can additionally provide nighttime LST, which can be combined with daytime observations estimating relatively accurate monthly mean LST, with an RMSE of 2.7 K. Finally, we compared GT-LST with a regional twice-daily AVHRR LST product over continental Africa in different seasons, with RMSDs ranging from 2.1 to 4.3 K. Considering these advantages, the proposed dataset provides a better data source for a range of research applications. GT-LST is freely available at https://doi.org/10.5281/zenodo.7113080 (1981–2000) (Li et al., 2022a), https://doi.org/10.5281/zenodo.7134158 (2001–2005) (Li et al., 2022b), and https://doi.org/10.5281/zenodo.7813607 (2006–2021) (J. H. Li et al., 2023).

Estimation and Validation of Land Surface Temperature by using Remote Sensing & GIS for Chittoor District, Andhra Pradesh

Land Surface Temperature (LST) quantification is needed in various applications like temporal analysis, identification of global warming, land use or land cover, water management, soil moisture estimation and natural disasters. The objective of this study is estimation as well as validation of temperature data at 14 Automatic Weather Stations (AWS) in Chittoor District of Andhra Pradesh with LST extracted by using remote sensing as well as Geographic Information System (GIS). Satellite data considered for estimation purpose is LANDSAT 8. Sensor data used for assessment of LST are OLI (Operational Land Imager) and TIR (Thermal Infrared). Thermal band contains spectral bands of 10 and 11 were considered for evaluating LST independently by using algorithm called Mono Window Algorithm (MWA). Land Surface Emissivity (LSE) is the vital parameter for calculating LST. The LSE estimation requires NDVI (Normalized Difference Vegetation Index) which is computed by using Band 4 (visible Red band) and band 5 (Near-Infra Red band) spectral radiance bands. Thermal band images having wavelength 11.2 µm and 12.5 µm of 30th May, 2015 and 21st October, 2015 were processed for the analysis of LST. Later on validation of estimated LST through in-suite temperature data obtained from 14 AWS stations in Chittoor district was carried out. The end results showed that, the LST retrieved by using proposed method achieved 5 per cent greater correlation coefficient (r) compared to LST retrieved by using existing method which is based on band 10.

Validation of Landsat land surface temperature product in the conterminous United States using in situ measurements from SURFRAD, ARM, and NDBC sites

ABSTRACT Since 1982, Landsat series of satellite sensors continuously acquired thermal infrared images of the Earth’s land surface. In this study, Landsat 5, 7, and 8 land surface temperature (LST) products in the conterminous United States from 2009 to 2019 were validated using in situ measurements collected at 6 SURFRAD (Surface Radiation Budget Network) sites, 6 ARM (Atmospheric Radiation Measurement) sites, and 9 NDBC (National Data Buoy Center) sites. The results indicate that a relatively consistent performance among Landsat 5, 7, and 8 LST products is obtained for most sites due to the consistent LST retrieval algorithm in conjunction with the same atmospheric compensation and land surface emissivity (LSE) correction methods for Landsat 5, 7, and 8 sensors. Large bias and root mean square error (RMSE) of Landsat LST product are obtained at some vegetated sites due to incorrect LSE estimation where LSE is invariant with the increasing of normalized difference vegetation index (NDVI). Except for the sites with incorrect LSE estimation, a mean bias (RMSE) of the differences between Landsat LST and in situ LST is 1.0 K (2.1 K) over snow-free land surfaces, −1.1 K (1.6 K) over snow surfaces, and −0.3 K (1.1 K) over water surfaces.

An Improved Linear Spectral Emissivity Constraint Method for Temperature and Emissivity Separation Using Hyperspectral Thermal Infrared Data.

The linear spectral emissivity constraint (LSEC) method has been proposed to separate temperature and emissivity in hyperspectral thermal infrared data with an assumption that land surface emissivity (LSE) can be described by an equal interval piecewise linear function. This paper combines a pre-estimate shape method with the LSEC method to provide an initial-shape estimation of LSE which will create a new piecewise scheme for land surface temperature (LST) and LSE separation. This new scheme is designated as the pre-estimate shape (PES)-LSEC method. Comparisons with the LSEC method using simulated data sets show that the PES-LSEC method has better performance in terms of accuracy for both LSE and LST. With an at-ground error of 0.5 K, the root-mean-square errors (RMSEs) of LST and LSE are 0.07 K and 0.0045, respectively, and with the scale factor of moisture profile 0.8 and 1.2, the RMSEs of LST are 1.11 K and 1.14 K, respectively. The RMSEs of LSE in each channel are mostly below 0.02 and 0.04, respectively, which are better than for the LSEC method. In situ experimental data are adopted to validate our method: The results show that RMSE of LST is 0.9 K and the mean value of LSE accuracy is 0.01. The PES-LSEC method with fewer segments achieves better accuracy than that of LSEC and preserves most of the crest and trough information of emissivity.

Effects of prediction accuracy of the proportion of vegetation cover on land surface emissivity and temperature using the NDVI threshold method

Predicting land surface energy budgets requires precise information of land surface emissivity (LSE) and land surface temperature (LST). LST is one of the essential climate variables as well as an important parameter in the physics of land surface processes at local and global scales, while LSE is an indicator of the material composition. Despite the fact that there are numerous publications on methods and algorithms for computing LST and LSE using remotely sensed data, accurate prediction of these variables is still a challenging task. Among the existing approaches for calculating LSE and LST, particular attention has been paid to the normalised difference vegetation index threshold method (NDVITHM), especially for agriculture and forest ecosystems. To apply NDVITHM, knowledge of the proportion of vegetation cover (PV) is essential. The objective of this study is to investigate the effect of the prediction accuracy of the PV on the estimation of LSE and LST when using NDVITHM. In August 2015, a field campaign was carried out in mixed temperate forest of the Bavarian Forest National Park, in southeastern Germany, coinciding with a Landsat-8 overpass. The PV was measured in the field for 37 plots. Four different vegetation indices, as well as artificial neural network approaches, were used to estimate PV and to compute LSE and LST. The results showed that the prediction accuracy of PV improved using an artificial neural network (R2CV = 0.64, RMSECV = 0.05) over classic vegetation indices (R2CV = 0.42, RMSECV = 0.06). The results of this study also revealed that variation in the accuracy of the estimated PV affected calculation results of the LSE. In addition, our findings revealed that, though LST depends on LSE, other parameters should also be taken into account when predicting LST, as more accurate LSE results did not increase the prediction accuracy of LST.

Spatiotemporal analysis of urban heat island intensification in the city of Minneapolis-St. Paul and Chicago metropolitan areas using Landsat data from 1984 to 2016

Most major cities worldwide are affected Urban Heat Islands – a condition of relatively higher temperatures being observed in one area compared to another that can be caused by a decrease in greenspace. One of the major reasons attributed to this increase in the warming of urban landscapes is the decrease in green space. This concept has received a lot of attention due to the destruction of vegetation for urban development and has prompted long-term spatial-temporal studies of Urban Heat Islands to understanding local climates. The objective of this study is to use Landsat data to examine the temporal intensification of UHIs and their variability from 1984–2016 for the cities of Chicago and Minneapolis-St. Paul. Landsat L4-5 TM), L7 ETM+), OLI and TIRS from 1984 to 2016 was used to examine land surface temperature (LST). Firstly, we converted the digital number (DN) to spectral radiance (L) and to temperature in Kelvin and from kelvin to Celsius and a conversion from Radiance to Top of the Atmosphere Reflectance and estimation of land surface emissivity. Finally, LST was estimated and Urban Heat Island retrieval and anomalies computed to help examine inconsistencies in our data. Our analysis showed year-to-year fluctuations in surface temperature, intensification of UHIs for both metro areas. Using a defined deductive index to identify environmentally critical areas, estimates of UHIs based on LST showed that both metropolitan areas are UHIs with LST > µ + 0.5 × δ. Higher intensification values were observed in 1988 and 2010 for Chicago and 1984, 1999 and 2016 for Minneapolis-St. Paul from analysis. While both areas have the similar climatic conditions, our analysis show differences in UHIs intensification as observed in their urban growth patterns. Chicago experiences a higher UHI intensity compared to Minneapolis-St. Paul and this could be explained by higher number of tall buildings than Minneapolis-St. Paul.

A semi-empirical approach for the estimation of land-surface emissivity from satellite data based on spectral index fusion using ensemble regression

ABSTRACTLand Surface Emissivity (LSE) is a key parameter in the thermal remote sensing, with several important applications, most notably in Land Surface Temperature (LST) estimation. This paper presents a semi-empirical method of LSE estimation from remote sensing data based on a fusion of spectral indices using the ensemble regression methods. The performance of the proposed method for Moderate Resolution Imaging Spectroradiometer (MODIS), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) data was evaluated and compared with other semi-empirical methods developed for these sensors. The proposed method was designed in four stages. In the first stage, the reflectance of non-thermal bands and emissivity of thermal bands were simulated for different classes using the ASTER spectral library and the spectral response function of each sensor. In the second stage, the dataset to be used for the training of ensemble regression was arranged by calculating a number of spectral indices, which constitute the feature space along with non-thermal bands. In the third stage, the regression between emissivity of thermal bands of each sensor and the features extracted in the second stage was derived by the use of bagging, boosting and Random Forest (RF) regression methods. In the final stage Using Normalized Difference Vegetation Index (NDVI) values, the image was categorized into three classes including vegetation, non-vegetation and mixture areas using conditions NDVI > 0.5, NDVI < 0.2 and 0.2 ≤ NDVI ≤ 0.5, respectively. The non-vegetation class was then categorized to soil, rock, and man-made classes using land use map. The spectral indices of these classes were then calculated, and the corresponding model trained in the third stage was used to estimate the LSE for that band. The results of LSE estimations were compared with the standard product of each sensor. Due to the lack of standard product for Landsat-8, the ASTER product was used as a substitute. For better analysis, the proposed method was also evaluated with other semi-empirical methods developed for MODIS, ASTER and OLI/TIRS sensors. This evaluation showed that the lowest Root Mean Square Error (RMSE) values for OLI/TIRS bands 10 and 11 are 0.0070 and 0.0075 obtained, respectively, by bagging and RF regression methods. For ASTER bands 13 and 14, the lowest RMSE values of 0.0078 and 0.0077 are both obtained by RF regression. For MODIS bands 31 and 32, the lowest RMSE values are 0.0053 and 0.0049 and obtained by boosting method. A comparison between the proposed method and other semi-empirical methods provided for these sensors demonstrated the ability of the method to improve the RMSE by up to 0.5%. Regarding the higher accuracy and applicability of the proposed method, it can serve as an effective and efficient means of estimating LSE using remote sensing data.

Estimation of Consistent Global Microwave Land Surface Emissivity from AMSR-E and AMSR2 Observations

AbstractAccurate estimation of passive microwave land surface emissivity (LSE) is crucial for numerical weather prediction model data assimilation, for microwave retrievals of land precipitation and atmospheric profiles, and for a better understanding of land surface and subsurface characteristics. In this study, global instantaneous LSE is estimated for a 9-yr period from the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) and for a 5-yr period from the Advanced Microwave Scanning Radiometer 2 (AMSR2) sensors. Estimates of LSE from both sensors were obtained by using an updated algorithm that minimizes the discrepancy between the differences in penetration depths from microwave and infrared remote sensing observations. Concurrent ancillary datasets such as skin temperature from the Moderate Resolution Imaging Spectroradiometer (MODIS) and profiles of air temperature and humidity from the Atmospheric Infrared Sounder are used. The latest collection 6 of MODIS skin temperature is used for the LSE estimation, and the differences between collections 6 and 5 are also comprehensively assessed. Analyses reveal that the differences between these two versions of infrared-based skin temperatures could lead to approximately a 0.015 difference in passive microwave LSE values, especially in arid regions. The comparison of global mean LSE features from the combined use of AMSR-E and AMSR2 with an independent product—Tool to Estimate Land Surface Emissivity from Microwave to Submillimeter Waves (TELSEM2)—shows spatial pattern correlations of order 0.92 at all frequencies. However, there are considerable differences in magnitude between these two LSE estimates, possibly because of differences in incidence angles, frequencies, observation times, and ancillary datasets.

Fusion Methods for Land Surface Emissivity and Temperature Retrieval of the Landsat Data Continuity Mission Data

In this paper, a new approach based on two fusion schemes is proposed to overcome the uncertainties in land surface emissivity (LSE) estimation and, consequently, land surface temperature (LST) retrieval. The fusion schemes are called image-based weighted methods and knowledge-based weighted methods, in which each of them includes two LSE estimation methods. The effectiveness of the two proposed fusion schemes is empirically tested over two scenes of Landsat-8 (known as Landsat Data Continuity Mission) data sets, and the obtained LSEs by individual and proposed methods were compared to the LSE product of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) by image-based and class-based cross-comparison. In both scenes, the adjusted normalized emissivity method (ANEM) and NDVI-based emissivity method (NBEM) provide appropriate results among five individual methods. In contrast, weighted to median (WMED) achieves superior results among the proposed fusion methods for both scenes. In addition, the root-mean-square error (rmse) values of LSE obtained by ANEM and WMED are 1.48% and 0.87%, which lead to 1.25 K and 0.73 K errors in the LST retrieval by the single-channel algorithm in the first scene, respectively. For the second scene, the error values of NBEM and WMED are 1.10% and 0.52%, which lead to 0.93 K and 0.44 K errors in the LST, respectively. Moreover, the error ranges and rmse of cross-comparison for the obtained LSE in the proposed methods were remarkably decreased. Also, in this research, for LST cross-comparison, an alternative scaling method based on LST products of the Moderate Resolution Imaging Spectroradiometer was proposed. The LST validation results demonstrated that the proposed methods provide better estimates in terms of three accuracy measures in both examined data sets.

Effect of emissivity uncertainty on surface temperature retrieval over urban areas: Investigations based on spectral libraries

Land surface emissivity (LSE) is a prerequisite for retrieving land surface temperature (LST) through single channel methods. According to error model, a 0.01 (1%) uncertainty of LSE may result in a 0.5K error in LST under a moderate condition, while an obvious error (approximately 1K) is possible under a warmer and less humid situation. Significant emissivity variations are presented among the anthropogenic materials in three spectral libraries, which raise a critical question that whether urban LSE can be estimated accurately to meet the needs for LST retrieval. Methods widely used for urban LSE estimation are investigated, including the classification-based method, the spectral-index based method, and the linear spectral mixture model (LSMM). Results indicate that the classification-based method may not be effectively applicable for urban LSE estimation, due mainly to the insignificant relation between the short-wave multispectral reflectance and the long-wave thermal emissivity shown by the spectra. Compared with the classification-based method, the LSMM shows relatively more accurate predictions, whereas, the performance of the LSMM largely depends on the determination of endmembers. Obvious uncertainties in LSE estimation likely appear if endmembers are determined improperly. Increasing the spectra for endmembers is a practical and beneficial means for LSMM when there is not a priori knowledge, which emphasizes the necessity of building a comprehensive spectral library of urban materials. Furthermore, the LST retrieval from a single channel of Landsat 8 is more challenging as compared with the retrieval from the channels of its predecessors—Landsat 4/5/7.

An examination of methods for estimating land surface microwave emissivity

AbstractLand surface emissivity is a critical variable for the passive microwave‐based remote sensing of the land and atmosphere. Driven by the Global Precipitation Measurement mission, we implemented and evaluated a variety of approaches for quantitative estimation of land surface emissivity and its variability, within a well‐defined common framework. These approaches fall into three classes: physical modeling, statistical modeling, and a hybrid of physical and statistical modeling. Every approach is subject to evaluation against retrieved emissivity over a large area in the Southern Great Plains for a period of 2 years. Physical modeling, based on two radiative transfer models coupled to a land surface modeling framework, produced reasonable estimates, with channel‐ and polarization‐dependent errors. Calibration of these models with historical data substantially improved their performance at lower frequencies. The statistical method was tested with five different regression models, and each of them consistently outperformed physical models by about 50%. The best statistical model had an average error of 0.9–2.1%. These statistical models were slightly improved when empirical orthogonal function analysis was incorporated in the emissivity data. The hybrid approach produced errors between the uncalibrated and calibrated physical model errors. In addition to their predictive performance, other aspects of each approach's strengths and weaknesses are discussed.

Modeling the effective emissivity of the urban canopy using sky view factor

Surface emissivity is a critical parameter for studying city-, meso-, and micro-scale climate and energy balance. The emissivity of complex surfaces e.g. a forest or an urban canopy is an effective surface property since it depends on both surface materials and geometry. This study presents a novel methodology for estimating effective emissivity using sky view factor retrieved from airborne Lidar data, building GIS data, and land use and land cover classification data. First, a high correlation between the effective emissivity retrieved from ASTER TIR bands 10–14 and the sky view factor was observed (r2=0.93, 0.99, 0.99, 0.97, 0.97). When the sky view factor decreases, the effective emissivity tends to increase, which is mainly due to multiple scattering (cavity effect), thus increases the effective emissivity. A simplified model which assumes that reflection and scattering only occurs within a single pixel was developed. Results showed that the correlations between the modeled and the spectral (band) emissivity retrieved from the ASTER multispectral TIR data (five spectral bands) are high (r2=0.93, 0.99, 0.98, 0.93, 0.97), and with low RMSE (0.019, 0.016, 0.012, 0.003 and 0.004 from band 10–14 respectively). The emissivity derived from this simplified model, however, tends to be overestimated in band 10–12. Thus, a refined urban emissivity model based on sky view factor (UEM-SVF) which considers the scattering and reflection from adjacent pixels was developed in this study. Results show a good agreement with ASTER spectral (band) emissivity: r2=0.90, 0.98, 0.96, 0.94 and 0.96, and very low RMSE (0.006, 0.003, 0.004, 0.002 and 0.004). This study illustrates that the UEM-SVF can be useful for estimation of land surface emissivity of complex surfaces, and can further be used for accurate land surface temperature retrieval.

THE ESTIMATION OF LAND SURFACE EMISSIVITY FOR LANDSAT TM6

Land surface emissivity (LSE) is an essential parameter for land surface temperature (LST) retrieval from thermal remote sensing data. Up till now, three methods have been proposed for LST retrieval from TM6 data, namely, atmospheric correction, mono-window algorithm and single channel algorithm, which all require LSE as a parameter. In this paper the authors have first reviewed the three methods and then dealt emphatically with the estimation of this parameter. The method was applied to Lingxian area of Shangdong Province in North China Plain, the most important agricultural area in China, for LSE estimation and LST retrieval. The result shows that the method can yield a reasonable estimation of thermal variation of that area.

Surface emissivity retrieval from Digital Airborne Imaging Spectrometer data

A study has been carried out on the most recent algorithms for the estimation of land surface emissivity (ε) using high‐resolution data (Digital Airborne Imaging Spectrometer, DAIS) over the Rhine Valley (France) and Castilla La Mancha (Spain). Three published methods have been applied for extracting absolute spectral emissivity information from images recorded during the DAISEX experiment in 1999. They are NDVI Thresholds Method (NDVITHM), Normalized Emissivity Method (NEM) and Temperature/Emissivity Separation (TES). These lather two methods were originally designed to work over geological surfaces. Five methods have been used for extracting relative spectral emissivity. They are temperature‐independent spectral indices method (TISI), reference channel method (REF), emissivity normalization method (NOR), emissivity re‐normalization method (RE), and alpha emissivity method (ALPHA), respectively. NDVITHM and NEM give the same absolute emissivity values with differences between 1% and 0.2% depending on the thermal channel considered, while NDVITHM and TES give the greatest differences, around 2%. The comparison with in situ values shows that NDVITHM gives the best results for vegetation plots, while NEM gives the best ones for bare soil and water plots. The TISI and NOR relative methods give the same relative emissivity values within less than 0.4%, in accordance with Becker and Li's [1990] conclusion as regards the superiority of these methods compared to the others.

A Comparative Study of Land Surface Emissivity Retrieval from NOAA Data

A comparative study has been carried out on the most recent algorithms for the estimation of land surface emissivity (ϵ) using Advanced Very High Resolution Radiometer (AVHRR) data. Three of the algorithms are based on the Temperature-Independent Spectral Indices (TISI) concept using atmospherically corrected channels 3, 4, and/or 5, namely: (1) TISI BL, (2) TS-RAM, and (3) Δday. The fourth is a simplified method based on the estimation of ϵ using atmospherically corrected data in the visible and near-infrared channels, called Normalized Difference Vegetation Index (NDVI) Thresholds Method (NDVI THM). This method integrates a wide spectral data set of bare soil reflectivity measurements in the 0.4–14-μm band and uses different approaches in function of the NDVI value. All methods have been applied to the Iberian Peninsula using AVHRR/National Oceanic and Atmospheric Administration (NOAA-14) data during March 17th, 1997. In terms of emissivity, the results show that the difference between the NDVI THM method and the other methods is always positive, with a bias of less than 0.010 and a root mean square (rms) error of less than 0.010 when compared to the TISI BL method, a bias of 0.018 and an rms error of 0.020 when compared to the Δday method, and a bias of 0.025 and an rms error of 0.012 when compared to the TS-RAM method. In terms of land surface temperature, the NDVI THM method shows a bias of less than −0.4 K and an rms error of less than 1.0 K when compared to TISI BL, a bias of −1.0 K and a rms error of 1.3 K when compared to Δday, and a bias of −1.1 K and an rms error of 0.5 K when compared to TS-RAM. In conclusion, although the TISI BL is a precise method, it needs the AVHRR channel 3 for its application, which is not always available, for example, as in the AVHRR archives provided in the frame of the Pathfinder AVHRR Land project (more than 18 years of data). In this case we have shown that the NDVI THM method shows promising results and can be applied to obtain land surface temperature and emissivity from NOAA data without losing accuracy.