Validation of a Simple MODIS Land Surface Temperature-Based Model for Potential Evapotranspiration (PET) using Long-Term Global Dataset

A new land surface temperature fusion strategy based on cumulative distribution function matching and multiresolution Kalman filtering

Land surface temperature (LST) is a critical parameter in land surface process. The Moderate Resolution Imaging Spectroradiometer (MODIS) can be used to generate various LST data products, and these data have been widely applied in many studies. Unfortunately, cloud contamination brings about numerous missing or abnormal values, which negatively affect the application of LST data. To reconstruct missing values and improve data quality, this paper proposes an integrated method for reconstructing LST data under two conditions: Clear sky and cloudy sky. For the clear-sky condition, the MODIS eight-day LST (MOD11A2) product is used to be interpolated into the low-quality daily LST dataset using the harmonic analysis of time series (HANTS) algorithm. And then the linear regression algorithm is implemented on the original good-quality pixels of the MODIS daily LST (MOD11A1) product. After that, seamless processing on the reconstructed low-quality daily LST dataset is carried using the Poisson image editing method, and finally the high-quality daily LST dataset under clear-sky condition are then obtained. For the cloudy-sky conditions, the revised neighboring-pixel (NP) algorithm that originates from the surface energy balance theory is used to reconstruct the real LST data. To evaluate the reconstruction performance under clear-sky condition, a simulated dataset is generated from simulated missing pixels with good quality that are randomly chosen from 98 available LST images in the year 2010. Meanwhile, the real LST measurements collected from ground sites are used to assess the reconstructed results under cloudy-sky condition. Satisfactory validation results show that the proposed integrated method effectively reconstructs the missing information and low-quality pixels caused by cloud cover and other factors. The filled data can seamlessly preserve the temporal and spatial consistence of the daily LST data, which do promote the practical utility of the MODIS LST product.

Northern high latitudes have experienced pronounced warming throughout the last decades, with particularly high temperatures during winter and spring. Due to Arctic Amplification, the Arctic region is warming four times faster than anywhere else. Permafrost, a crucial component of arctic ecosystems, is particularly sensitive to increasing air temperatures and changes in the snow regime. In the last decade, satellite-derived land surface temperature (LST) products combined with snow cover information and land cover data have been increasingly used for permafrost modelling. For example, the CryoGrid community model, a ground thermal model, is used within the frame of the ESA Permafrost Climate Change Initiative (CCI) project to produce permafrost extent maps on a hemispheric scale. These maps and permafrost modelling outputs are based on Moderate Resolution Imaging Spectroradiometer (MODIS) LST data. A drawback is that MODIS LST products have only been available since 2001, which prevents differentiating multi-decadal climate trends from decadal-scale climate oscillations.To leverage the historic Advanced Very High-Resolution Radiometer (AVHRR) sensors series, a new pan-Arctic LST dataset has been developed using EUMETSAT’s AVHRR Fundamental Data Record (FDR). The pan-Arctic AVHRR LST product covers a period from 1981 to 2021 and has a spatial resolution of approximately 4 km. It incorporates snow cover information derived from fractional snow cover and snow water equivalent data, allowing for accurate emissivity and temperature retrievals over snow and ice. To obtain AVHRR LST data at a spatial resolution similar to the MODIS LST dataset (~ 1 km) and allow for intercomparison of the permafrost modelling outputs, the AVHRR pan-Arctic LST dataset is downscaled to a spatial resolution of 1 km. Recent advances in spatiotemporal fusion and super-resolution models offer new solutions to downscale thermal infrared (TIR) data, allowing obtaining LST data at a high spatial and temporal resolution. Guided super-resolution (SR) is another downscaling strategy that only relies on a low-resolution source and a high-resolution guide. It returns a high-resolution version of the source. In the case of the AVHRR LST downscaling, the guide comprises information derived from land cover, elevation models, and canopy height data. Downscaling results of the pan-Arctic LST dataset based on guided deep anisotropic diffusion for the region of the Yamal Peninsula (Siberia) and along the Alaska Highway in the Yukon (Canada) showed promising results. The downscaling methodology demonstrated its potential for capturing the complexities of typical permafrost landscapes.

Having a good knowledge of the uncertainty in the land surface temperature (LST) product will help to encourage its use in a wide number of applications, including urban heat islands, geothermal detection, and surface energy balance. Landsat 9 was launched on 27 September 2021 and provides an LST product, which is generated by the radiative transfer equation algorithm and has a spatial resolution of 30 m. In this article, we evaluated the performance of the Landsat 9 LST product by using a temperature-based (T-based) method and cross-validation. The T-based validation results showed that the average bias at the surface radiation budget network and baseline surface radiation network sites was 0.24 K and that the corresponding root mean square error (RMSE) was 3.42 K. The Landsat 9 LST product was in good agreement with the Landsat 7/8 LSTs, with an average bias of 0.25/0.08 K, an RMSE of 0.51/1.04 K, and a mean absolute error of 0.38/0.64 K. The comparable performance of the Landsat 7/8/9 LST products can be explained by the consistent LST retrieval algorithm. The absolute differences in the LST between Landsat 9 LST and MOD11 (MOD21) LST images were between 0.01 (0.65) and 2.50 K (1.76 K), whereas the RMSE values were between 1.40 (1.80) and 3.65 K (3.26 K). The specific heat capacity and thermal inertia of the different land surface covers can explain the significant biases. The above evaluation results are consistent with the initial performance testing of thermal infrared sensor-2 (TIRS-2) by the National Aeronautics and Space Administration and the U.S. Geological Survey. Although the released Landsat 9 LST product showed good performance in the preliminary evaluation, the split-window algorithm may be a better option for Landsat 9 LST retrieval, as the TIRS-2 data addressed stray light incursion. Since there are no official validation results that have been published, this article provides a third-party performance evaluation of the Landsat 9 LST product and will benefit research fields that require Landsat series LST products.

Accuracy assessment of land surface temperature (LST) products is critical to facilitate their use in various studies. As an alternative method for assessing the accuracy of LST products, a new satellite LST product is compared with a heritage LST product to validate and determine the uncertainties in the satellite-derived LST approach. In this study, we propose a method for the intercomparison of the Meteosat second generation-spinning enhanced visible and infrared imager (MSG-SEVIRI) and Terra/Aqua-moderate resolution imaging spectroradiometer (MODIS) LST products. The intercomparison was performed by verifying the collocation in space, temporal concurrence, viewing geometry alignment, and spatial homogeneity between the two LST products. The discrepancies between the SEVIRI and MODIS LST products were investigated over different seasons, times of day, and surface types. SEVIRI LST values are generally higher than MODIS LST values, with positive biases during daytime (approximately 2–4 K) and nighttime (approximately 1–2 K). Significant variability of the daytime LST discrepancies with season, time of day, and surface type is observed. Compared with the daytime LST discrepancies, the nighttime LST discrepancies are less dependent on season, time of day, and surface type.

Global Water Balance According to the Penman Approach*

Reconstruction of daytime land surface temperatures under cloud-covered conditions using integrated MODIS/Terra land products and MSG geostationary satellite data

In this study, to improve the accuracy of land surface temperature (LST) products over barren surfaces, we present an operational algorithm to retrieve the LST from Moderate-Resolution Imaging Spectroradiometer (MODIS) thermal infrared data using physically retrieved emissivity products. The LST algorithm involved two steps. First, the emissivity in the two MODIS split-window (SW) channels was estimated using the vegetation cover method, with the bare soil component emissivity derived from the ASTER global emissivity data set. Then, the LST was retrieved using a modified generalized SW algorithm. This algorithm was implemented in the MUlti-source data SYnergized Quantitative (MuSyQ) remote sensing product system. The MuSyQ MODIS LST product and the Collection 6 MODIS LST product (MxD11_L2) were compared and validated using ground measurements collected from four barren surface sites in Northwest China during the Heihe Watershed Allied Telemetry Experimental Research (HiWATER) experiment from June 2012 to December 2015. In total, 2268 and 2715 clear-sky samples were used in the validation for Terra and Aqua, respectively. The evaluation results indicate that the MuSyQ LST products provide better accuracy than the C6 MxD11 product during both daytime and nighttime at all four sites. For the daytime results, the LST is underestimated by the C6 MxD11 products at all four sites, with a mean bias of -1.78 and -2.86 K and a mean root-mean-square error (RMSE) of 3.16 and 3.94 K for Terra and Aqua, respectively, whereas the mean biases of the MuSyQ LST products are within 1 K, with a mean bias of -0.26 and -1.03 K and a mean RMSE of 2.45 and 2.71 K for Terra and Aqua, respectively. For the nighttime results, the LST is also underestimated by the C6 MxD11 products at all four sites, with a mean bias of -1.60 and -1.26 K and a mean RMSE of 1.93 and 1.60 K for Terra and Aqua, respectively, whereas the mean biases of the MuSyQ LST products are 0.16 and 0.58 K and the mean RMSEs are 1.12 and 1.25 K for Terra and Aqua, respectively. The results indicate that the underestimation of the C6 MxD11 LST product at all four sites mainly results from the overestimation of the emissivities in MODIS bands 31 and 32. This study demonstrates that physically retrieved emissivity products are a useful source for LST retrieval over barren surfaces and can be used to improve the accuracy of global LST products.

In this article, we present empirical models for estimating daily mean air temperature (Ta) in the Hurd Peninsula of Livingston Island (Antarctica) using Moderate Resolution Imaging Spectroradiometer (MODIS) Land Surface Temperature (LST) data and spatiotemporal variables. The models were obtained and validated using the daily mean Ta from three Spanish in situ meteorological stations (AEMET stations), Juan Carlos I (JCI), Johnsons Glacier (JG), and Hurd Glacier (HG), and three stations in our team’s monitoring sites, Incinerador (INC), Reina Sofía (SOF), and Collado Ramos (CR), as well as daytime and nighttime Terra-MODIS LST and Aqua-MODIS LST data between 2000 and 2016. Two types of multiple linear regression (MLR) models were obtained: models for each individual station (for JCI, INC, SOF, and CR—not for JG and HG due to a lack of data) and global models using all stations. In the study period, the JCI and INC stations were relocated, so we analyzed the data from both locations separately (JCI1 and JCI2; INC1 and INC2). In general, the best individual Ta models were obtained using daytime Terra LST data, the best results for CR being followed by JCI2, SOF, and INC2 (R2 = 0.5–0.7 and RSE = 2 °C). Model cross validation (CV) yielded results similar to those of the models (for the daytime Terra LST data: R2CV = 0.4–0.6, RMSECV = 2.5–2.7 °C, and bias = −0.1 to 0.1 °C). The best global Ta model was also obtained using daytime Terra LST data (R2 = 0.6 and RSE = 2 °C; in its validation: R2CV = 0.5, RMSECV = 3, and bias = −0.03), along with the significant (p < 0.05) variables: linear time (t) and two time harmonics (sine-cosine), distance to the coast (d), slope (s), curvature (c), and hour of LST observation (H). Ta and LST data were carefully corrected and filtered, respectively, prior to its analysis and comparison. The analysis of the Ta time series revealed different cooling/warming trends in the locations, indicating a complex climatic variability at a spatial scale in the Hurd Peninsula. The variation of Ta in each station was obtained by the Locally Weighted Regression (LOESS) method. LST data that was not “good quality” usually underestimated Ta and were filtered, which drastically reduced the LST data (<5% of the studied days). Despite the shortage of “good” MODIS LST data in these cold environments, all months were represented in the final dataset, demonstrating that the MODIS LST data, through the models obtained in this article, are useful for estimating long-term trends in Ta and generating mean Ta maps at a global level (1 km2 spatial resolution) in the Hurd Peninsula of Livingston Island.

The assessment of satellite-derived land surface temperature (LST) data is essential to ensure their high quality for climate applications and research. This study intercompared seven LST products (i.e., ATSR_3, MODISA, MODIST, SLSTRA, SLSTRB, SEVIR2 and SEVIR4) of the European Space Agency’s (ESA) LST Climate Change Initiative (LST_cci) project, which are retrieved for polar and geostationary orbit satellites, and three operational LST products: NASA’s MODIS MOD11/MYD11 LST and ESA’s AATSR LST. All data were re-gridded on to a common spatial grid of 0.05° and matched for concurrent overpasses within 5 min. The matched data were analysed over Europe and Africa for monthly and seasonally aggregated median differences and studied for their dependence on land cover class and satellite viewing geometry. For most of the data sets, the results showed an overall agreement within ±2 K for median differences and robust standard deviation (RSD). A seasonal variation of median differences between polar and geostationary orbit sensor data was observed over Europe, which showed higher differences in summer and lower in winter. Over all land cover classes, NASA’s operational MODIS LST products were about 2 K colder than the LST_cci data sets. No seasonal differences were observed for the different land covers, but larger median differences between data sets were seen over bare soil land cover classes. Regarding the viewing geometry, an asymmetric increase of differences with respect to nadir view was observed for day-time data, which is mainly caused by shadow effects. For night-time data, these differences were symmetric and considerably smaller. Overall, despite the differences in the LST retrieval algorithms of the intercompared data sets, a good consistency between the LST_cci data sets was determined.

This study analyzed the scaling problem of land surface temperature (LST) data retrieved with the Temperature Emissivity Separation (TES) algorithm. We compiled a remotely sensed dataset that included Thermal Airborne Hyperspectral Imager (TASI) and satellite-based Advanced Spaceborne Thermal Emission Reflection (ASTER) data, which were acquired simultaneously. This dataset provided the range of spatial heterogeneities of land surface necessary for the study, which was quantified by the dispersion variance. The LST scaling problem was studied by comparing the remotely sensed LST products in two ways. First, the LST products calculated in the distributed method and the lumped method were compared. Second, the airborne and satellite-based LST products derived from the TES algorithm were compared. Four upscaling methods of LST were used in the process. A scaling correction methodology was developed based on the comparisons. The results showed that the scaling effect could be as large as 0.8 when the spatial resolution of the TASI LST data was coarse. The scaling effect increases quickly with the spatial resolution until it reaches the characteristic scale of the landscape and is positively correlated with the spatial heterogeneity. The first two upscaling methods denoted as Methods 1–2 can upscale the LST more effectively when compared with the other two scaling methods (Methods 3–4). The scaling effect for the ASTER data is not notable. The comparison between the TASI and ASTER data showed that they were highly consistent, with a root mean square error (RMSE) of approximately 0.88 K, when the pixels were relatively homogeneous. When the spatial heterogeneity was significant, the RMSE was as large as 2.68 K The scaling correction methodology provided resolution-invariant results with scaling effects of less than 0.5 K.

Does quality control matter? Surface urban heat island intensity variations estimated by satellite-derived land surface temperature products

Land surface temperature (LST) is a key parameter in ecological and farm environment studies. The study area is located in Zhangye of Gansu province, mainly was covered by crops and desert. To retrieve LST from ASTER thermal infrared (TIR), split window algorithm was used. Surface emissivity and atmospheric transmittance was estimated previously. To evaluate the estimated result, the ASTER and MODIS LST production was collected and compared in both visual method and spatial distributions of LST profiles derived from typical transects. The maps showed that the general distribution tendency of ASTER LST was consistent with MODIS LST data and corresponded to the NDVI image in an inverse fashion. To gain an insight into the negative relationship between LST and NDVI, empirical statistics was conducted and the results showed that there was a strong negative relationship between LST and NDVI (R2=0.508). Further, the mean temperature and standard deviation of each land cover types for two standard LST productions and LST estimated in our method were collected to make a comparison. For the three LST data, the sequence of temperature values for land use/land cover (LULC) from high to low was same: sand, desert, impervious, vegetation and water. However, ASTER LST retrieval in our method was lower than the other two LST data. It may be caused by the estimated parameters or the coarse resolution of MODIS. In our study, a relative comparison approach was adopted to verify the result, which proved LST images retrieved from only two ASTER thermal channels using our developed algorithms were reliable and easily realized.

Fine spatial and temporal resolution land surface temperature (LST) data are of great importance for various researches and applications. Spatio-temporal fusion provides an important solution to obtain fine spatio-temporal resolution LST. For example, 100-m, daily LST data can be created by fusing 1-km, daily Moderate Resolution Imaging Spectroradiometer (MODIS) LST with 100-m, 16-day Landsat LST data. However, the quality of MODIS LST products has been decreasing noticeably in recent years, which has a great impact on fusion accuracy. To address this issue, this paper proposes to use Visible Infrared Imaging Radiometer Suite (VIIRS) LST to replace MODIS LST in spatio-temporal fusion. Meanwhile, to cope with the data discrepancy caused by the large difference in overpass time between VIIRS LST and Landsat LST, a spatio-temporal fusion method based on the Restormer (RES-STF) is proposed. Specifically, to effectively model the differences between the 2 types of data, RES-STF uses Transformer modules in Restormer, which combines the advantages of convolutional neural networks (CNN) and Transformer to effectively capture both local and global context in images. In addition, the calculation of self-attention is re-designed by concatenating CNN to increase the efficiency of feature extraction. Experimental results on 3 areas validated the effectiveness of RES-STF, which outperforms one non-deep learning- and 3 deep learning-based spatio-temporal fusion methods. Moreover, compared to MODIS LST, VIIRS LST data contain richer spatial texture information, leading to more accurate fusion results, with both RMSE and MAE reduced by about 0.5 K.

Normalization of the temporal effect on the MODIS land surface temperature product using random forest regression