Land Surface Temperature Fields Research Articles

Combining Landsat 8 and Sentinel-2 Data in Google Earth Engine to Derive Higher Resolution Land Surface Temperature Maps in Urban Environment

Thermal infrared (TIR) satellite imagery collected by multispectral scanners is important to map land surface temperature on a global scale. However, the TIR spectral bands are typically available in coarser spatial resolution than other multispectral bands of shorter wavelengths. Therefore, the spatial resolution of the derived land surface temperature (LST) is limited to around 100 m. This constrains the applications of such thermal satellite sensors in which finer detail of LST spatial pattern is relevant, especially in an urban environment where the land cover structure is complex. Among the missions deployed on the Earth’s orbit, NASA’s TIRS sensor onboard Landsat 8 and Landsat 9, and ASTER onboard Terra provide the highest spatial resolution of the thermal band. On the other hand, ESA’s Sentinel-2 multispectral imagery is collected at a higher spatial resolution of 10 m with a 5-day temporal resolution, but scanning in the TIR band is not available. This study makes use of the known relationship between LST and land cover metrics, such as the normalized difference vegetation index (NDVI), built-up index (NDBI), and water index (NDWI). We define a multiple linear regression model based on the spectral indices and LST derived from Landsat 8 data to inform the same model in which the equivalent spectral indices derived from Sentinel-2 are used to predict LST at 10 m resolution. Results of this approach are demonstrated in a case study for Košice city, Slovakia, where the multiple linear model based on Landsat 8 data achieved an R2 of 0.642. The correlation between the observed Landsat 8 LST and predicted LST from Sentinel-2 aggregated to the same resolution as the observed LST was high (r = 0.91). Despite the imperfections of the downscaling model, the derived LST at 10 m resolution provides a better perception of the LST field that can be easily associated with land cover features present in urban environment. The LST downscaling approach was implemented into Google Earth Engine. It provides a user-friendly online application that can be used for any city or urban region for generating a more realistic spatial pattern of LST than can be directly observed by contemporary Earth observation satellites. The tool aids in urban decision making and planning on how to mitigate overheating of cities to improve the life quality of their citizens.

Physically constrained spatiotemporal modeling: generating clear-sky constructions of land surface temperature from sparse, remotely sensed satellite data

Satellite remote-sensing is used to collect important atmospheric and geophysical data at various spatial resolutions, providing insight into spatiotemporal surface and climate variability globally. These observations are often plagued with missing spatial and temporal information of Earth's surface due to (1) cloud cover at the time of a satellite passing and (2) infrequent passing of polar-orbiting satellites. While many methods are available to model missing data in space and time, in the case of land surface temperature (LST) from thermal infrared remote sensing, these approaches generally ignore the temporal pattern called the ‘diurnal cycle’ which physically constrains temperatures to peak in the early afternoon and reach a minimum at sunrise. In order to infill an LST dataset, we parameterize the diurnal cycle into a functional form with unknown spatiotemporal parameters. Using multiresolution spatial basis functions, we estimate these parameters from sparse satellite observations to reconstruct an LST field with continuous spatial and temporal distributions. These estimations may then be used to better inform scientists of spatiotemporal thermal patterns over relatively complex domains. The methodology is demonstrated using data collected by MODIS on NASA's Aqua and Terra satellites over both Houston, TX and Phoenix, AZ USA.

Study of Temperature Heterogeneities at Sub-Kilometric Scales and Influence on Surface–Atmosphere Energy Interactions

The retrieval of land surface temperature (LST) from remote sensing techniques has been studied and validated during the past 40 years, leading to important improvements. Accurate LST values are currently obtained through measurements using medium resolution thermal infrared (TIR) sensors. However, the most recent review reports demonstrated that the future TIR LST products need to obtain reliable temperature values at a high spatial resolution (100 m or higher) to study temperature variations between different elements in a heterogeneous kilometric area. The launch of high-resolution TIR sensors in the near future requires studies of the temporal evolution and spatial heterogeneities of the elements in a mixed region. The present study analyzes the LST in a sub-kilometric highly heterogeneous area, combining the use of LST products from high-resolution TIR orbiting sensors with the LST maps created from a TIR camera onboard an unmanned aerial vehicle (UAV). The aim is to estimate the LST variability in a heterogeneous area containing different surfaces (roads, buildings, and grass), observed from different TIR sensors at different spatial resolutions, covering from the meter to the kilometer scales. Several results showed that variations in the LST up to 18 °C were identified with the UAV-TIR camera, and significant differences were also present in the LST products obtained from simultaneous overpasses of high-resolution satellite TIR sensors. A second objective of the study, due to the availability of the high-resolution LST fields, was to explore the thermal advection between different elements and determine if it correlates with the surface energy budget in the same area, thus indicating that this process is of importance for heterogeneous terrains at these scales. This paper also highlights the relevance of the UAV-TIR camera flight for future studies since it is not commonly used in TIR remote sensing but has substantial potential advantages.

A Methodology to Simulate LST Directional Effects Based on Parametric Models and Landscape Properties

The correction of directional effects on satellite-retrieved land surface temperature (LST) is of high relevance for a proper interpretation of spatial and temporal features contained in LST fields. This study presents a methodology to correct such directional effects in an operational setting. This methodology relies on parametric models, which are computationally efficient and require few input information, making them particularly appropriate for operational use. The models are calibrated with LST data collocated in time and space from MODIS (Aqua and Terra) and SEVIRI (Meteosat), for an area covering the entire SEVIRI disk and encompassing the full year of 2011. Past studies showed that such models are prone to overfitting, especially when there are discrepancies between the LSTs that are not related to the viewing geometry (e.g., emissivity, atmospheric correction). To reduce such effects, pixels with similar characteristics are first grouped by means of a cluster analysis. The models’ calibration is then performed on each one of the selected clusters. The derived coefficients reflect the expected impact of vegetation and topography on the anisotropy of LST. Furthermore, when tested with independent data, the calibrated models are shown to maintain the capability of representing the angular dependency of the differences between LST derived from polar-orbiter (MODIS) and geostationary (Meteosat, GOES and Himawari) satellites. The methodology presented here is currently being used to estimate the deviation of LST products with respect to what would be obtained for a reference view angle (e.g., nadir), therefore contributing to the harmonization of LST products.

Hierarchical Bayesian space-time estimation of monthly maximum and minimum surface air temperature

Surface air temperature (SAT) is a critical metric that is used to assess regional warming and cooling patterns, and maximum and minimum SATs are required to evaluate the model predictions of climate extremes. Since station SAT data are irregularly distributed, land surface temperature (LST) values derived from Moderate Resolution Imaging Spectroradiometer (MODIS) data are used to estimate regional SAT by using linear regression methods. The deviations between SAT and LST are largely dependent on space and time, which hampers the estimation of linear regression, especially for the maximum SAT. To obtain accurate regional SAT estimates, a three-stage hierarchical Bayesian (HB) model is proposed that incorporates the MODIS LSTs as model covariates and specifies the deviations with structured dependence of MODIS LST fields. Sampling of model parameters and estimation of SAT values are implemented under the Bayesian paradigm using a Markov Chain Monte Carlo algorithm. Sensitivity analyses involving various model configurations and running processes are discussed to help build a robust HB model. The model's performance is evaluated using station measurements that are not used in the modeling process, with RMSEs of 2.15 K (0.75%) and 1.97 K (0.73%) for monthly maximum and minimum SATs, respectively. The evaluation indicates that HB modeling is an effective method to estimate SAT from MODIS LST. The verified HB model with the covariate inputs of both MODIS daytime and nighttime LSTs is used to reproduce monthly maximum and minimum SATs that are spatially continuous over the Qinghai province in Northwestern China for 2003–2011. From the comparison between MODIS LST and HB-estimated SAT, it is found that the spatial structure and warming patterns of LST and SAT show significant distinctions, implying that they cannot be substituted for one another when assessing the regional warming trends. The spatial heterogeneity of HB model estimation is able to provide thorough insights into regional SAT status changes that could otherwise be biased by station deployment.

Spatiotemporal multiresolution modeling to infill missing areal data and enhance the temporal frequency of infrared satellite images

High‐resolution environmental observations from satellite remote sensing are useful for gaining insight into climate variability over large regions of the earth. However, data products derived from infrared remote sensing can have large amounts of missing data due to periodic cloud cover obscuring the earth's surface. In addition, these products are commonly derived from instruments aboard polar‐orbiting satellites, which make relatively infrequent passes over each region of the planet, leaving large blocks of data that are missing in time. The goal of this research is to overcome these issues by spatially and temporally infilling temperature data using satellite observations. To accomplish this, we propose novel statistical modeling techniques to account for dissonance between the areal spatial support of the data and the continuous temporal domain. In addition, we develop a novel multiresolution approach that accommodates the large spatial dimensionality of high‐resolution satellite images. The result of our work is a computationally efficient methodology that provides the means to infill satellite data products both spatially and temporally. The methodology is demonstrated using land surface temperature fields from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument onboard NASA's Terra and Aqua satellites, over Harris County, Texas, USA.

Downscaling land surface temperatures at regional scales with random forest regression

Abstract Environmental monitoring with satellite data is facilitated by frequent observations at a fine spatial scale. As land surface temperature (LST) is one environmental key variable, we implemented a random forest (RF) regression approach to increase the spatial resolution of LST maps from ~ 1 km, routinely available in daily repetition from the Moderate Resolution Imaging Spectroradiometer (MODIS), to ~ 250 m. LST was downscaled based on its relationship to topographic variables derived from digital elevation data of the Shuttle Radar Topography Mission (SRTM), land cover data (MODIS product MCD12Q1), and surface reflectances in the visible red and near-infrared, which both are provided with the MODIS/Terra daily product MOD09GQ at ~ 250 m resolution. The approach was tested for a complex landscape in the Eastern Mediterranean, the Jordan River Region, with LST fields from aggregated Landsat-7 ETM + and MODIS (MODIS/Terra LST product MOD11A1) data; as reference at the finer scale, we used Landsat-7 derived LST data. For the ideal-case scenario with both degraded and reference values from the same sensor (Landsat-7 ETM +), root mean square errors (RMSE) of downscaled LSTs ranged from 1.02 K to 1.43 K for six different acquisition dates. When compared to the widely-adopted and in parallel applied TsHARP sharpening method, that is based on the relationship between NDVI and LST, downscaling accuracy with RF improved up to 19%. Applying the RF approach to MODIS LST products yielded RMSEs from 1.41 K to 1.92 K, whereas the TsHARP method and also a uniform disaggregation by resampling provided only slightly worse results. For the real MODIS LST product, downscaling with RF was affected by lower thermal contrasts in the image data that hindered an adequate training to reproduce temperature variations at the finer scale of ~ 250 m. In this context, we assume the LST product of the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument (as a successor of MODIS onboard the SUOMI NPP platform) to be a better candidate for downscaling, as it provides a spatial resolution of ~ 750 m. One advantage of the RF approach is that predictor datasets can easily be adapted to data availability. In an extended RF approach with all Landsat spectral bands, downscaling results for formerly aggregated Landsat data improved distinctly and now ranged from 0.98 K to 1.33 K. This approach is also promising for the downscaling of real MODIS or VIIRS LST data as it may be combined with already available reflective data fusion models that are able to blend Landsat data and sensor data with a coarse spatial resolution (given spectral bandwidths corresponding to Landsat) to generate temporally dense synthetic Landsat time series.

Estimation of land surface temperature from atmospherically corrected LANDSAT TM image using 6S and NCEP global reanalysis product

Water vapour is the most variable constituent in the atmosphere which is responsible for serious noise in the optical satellite images. This research is focused on the vertical distribution of water vapour and deducing its possible effects on the atmospheric correction process. The vertical distribution of precipitable water vapour, water vapour mixing ratio with geopotential height and pressure were estimated through the weather research and forecasting (WRF) model by downscaling the National Center for Environmental Prediction (NCEP) global reanalysis product. In addition, the most widely used LANDSAT TM satellite image has been used for this assessment. The WRF model was applied with three domains centred on a LANDSAT captured image over the area. The 6S atmospheric correction code was utilised for viewing the effect of precipitable water vapour on satellite image correction. The analysis was conducted on two pressure levels (1,000 and 100 hPa) representing the troposphere and stratosphere, respectively. The validation of the atmospheric correction has been performed by estimating the land surface temperature (LST) over the Walnut Creek region and its comparison with the Soil Moisture Experiments in 2002 (SMEX02) LST field validation datasets. The overall analyses indicate a higher accuracy of LST repossession with 100 hPa corrected image.

Land surface temperature from multiple geostationary satellites

This article provides a description of a land surface temperature (LST) data set generated (and provided in near-real-time or offline) based on infrared data from sensors onboard different geostationary (GEO) satellites: Meteosat Second Generation (MSG), Geostationary Operational Environmental Satellite (GOES), and Multifunction Transport Satellite (MTSAT). Given the different characteristics of the imagers onboard each GEO platform, different algorithmic methodologies for the retrieval of LST are presented and implemented – namely the Generalized Split-Window (GSW) algorithm and the Dual Algorithm (DA) in its mono- and dual-channel forms – using semi-empirical functions that relate LST to top-of-atmosphere brightness temperatures in infrared window channels. The assumptions and physics underlying each methodology, as well as the uncertainties of LST estimates, are discussed. The formulations are trained using a data set of radiative transfer simulations for a wide range of atmospheric and surface conditions. The performance of each algorithm is then assessed by comparing its output against an independent set of simulations, suggesting that product uncertainties range from 2°C (for GSW and the two-channel algorithm) to 4°C (for the one-channel algorithm). Finally, LST retrievals from different GEO satellites are merged into a single field. In overlapping areas, the average discrepancies between LST products derived from GOES and from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard MSG are within 1°C during night-time, but may reach 3°C during daytime. Over those areas, the merged LST field is obtained as a weighted average of available LST retrievals for the same time slot, taking into account the respective error bar.

A daily merged MODIS Aqua–Terra land surface temperature data set for the conterminous United States

A major shortcoming of any remotely-sensed land surface temperature (LST) dataset is the lack of observations for cloud-covered areas. A method is presented that uses the Moderate Resolution Imaging Spectroradiometer (MODIS) flying on the Terra platform to fill in spatial gaps in the Aqua MODIS LST dataset over the conterminous United States (CONUS) and limited adjacent areas. Over this domain, data are available for only about 50% of all times and pixels for each of the two MODIS sensors. Coverage is highest in summer and lowest in winter, with major regional variations.The relative close temporal proximity (~3h) of the Aqua and Terra overpasses provides an opportunity to combine information from the two data sources, which can reduce the data loss, most of which we assume is cloud-related. We applied the approach to create a ‘merged’ data set that supplements existing Aqua and Terra daytime and nighttime LST products. We used Terra LST data to fill gaps in Aqua data, resulting in a data set tied to the ~1:30AM/PM overpass times, so that the resulting data closely approximate daily minimum and maximum LST values. In order to use Terra LST observations to supplement Aqua data, an adjustment was applied to account for the different overpass times of the two platforms. Terra's 10:30AM overpass usually senses a cooler surface than does Aqua with its 1:30PM overpass. Conversely, for nighttime overpasses, Terra typically measures a warmer surface at 10:30PM than does Aqua at 1:30AM. Our approach was to determine, by season, mean Aqua and Terra LST values on the CONUS grid, based on data from a multi-year (2003–2008) period. Adding the mean Aqua-Terra LST differences for the respective season and time of day to a daily gridded Terra LST field removes the mean offset related to overpass time, resulting in LST values that can then be used to fill Aqua LST data gaps. Using independent offsets for each grid cell and season provides a first-order accounting for factors such as land cover, elevation, terrain slope and aspect, latitude, season and snow cover, which control the diurnal cycle of LST. For the six-year period, the merged data set increases data coverage by 24% and 30% for daytime and nighttime overpasses, respectively, relative to the Aqua LST product alone. The CONUS data set is a potentially valuable tool for weather and climate studies in which high spatial and temporal coverage are desired.

Spatial-temporal variation of the land surface temperature field and present-day tectonic activity

This study attempts to acquire information on tectonic activity in western China from land surface temperature ( LST) field data. On the basis of the established relationship between heat and strain, we analyzed the LST distribution in western China using the satellite data product MODIS/Terra. Our results show that: 1. There are departures from annual changes of LST in some areas, and that these changes are associated with the activity of some active tectonic zones. 2. When annual-change background values caused by climate factors are removed, the long-period component ( LST LOW) of temperature residual (Δ T) of the LST is able to serve as an indicator for tectonic activity. We have found that a major earthquake can produce different effects on the LST fields of surrounding areas. These effects are characterized by both rises and drops in temperature. For example, there was a noteworthy temperature decline associated with the Sumatran M9 earthquake of 2004 in the Bayan Har-Songpan block of central Tibetan Plateau. 3. On the other hand, the LST field of a single area may respond differently to major shocks occurring in different areas in the regions surrounding China. For instance, the Kunlun M 8.1 event made the LST on the Longmen Mountains fault zone increase, whereas the Zaisan Lake M 7.9 quake of 2003, and the Sumatran M 9 event of 2004, caused decreases in the same area’s LST. 4. The variations of land surface temperature ( LST) over time are different in different tectonic areas. These phenomena may provide clues for the study of tectonic deformation processes. On the basis of these phenomena, we use a combination of temperature data obtained at varied depths, regional seismicity and strain results obtained with GPS measurements, to test the information related to tectonic activity derived from variations of the LST field, and discuss its implications to the creation of models of regional tectonic deformation.

Spatio-temporal exploratory analysis of urban surface temperature field in Shanghai, China

The integrated application of remote sensing, geographic information system and quantitative analytical modeling can provide scientific and effective methods for monitoring and studying urban heat island, based on land surface temperature (LST) retrieved from thermal infrared channel data of sensors. In this paper, LST is retrieved from Landsat TM6 and ETM + 6 data of Shanghai central city in 1989, 1997, 2000 and 2002, by using the mono-window algorithm. Based on the data, global and local spatial autocorrelation analysis, and geostatistical methods are adopted to quantitatively describe the characteristics of spatial heterogeneity and temporal evolution of land surface thermal landscape at different scales and periods in Shanghai central city, by utilizing exploratory spatial data analysis. Results show that LST field in Shanghai central city tends to fragmentize and complicate with the development of Shanghai, and its global spatial difference becomes greater gradually. The spatial variance pattern of the change of LST field from 1997 to 2002 indicates that the dynamic change of LST presents a tendency of increase in circularity. LST declines distinctly in the districts of Puxi and Pudong near and inside the inner ring road, while it rises obviously outside the central city and near the out ring road. The extrema of temporal change in LST field have a characteristic of spatial clustering. Besides, as the city of Shanghai expands in a circular pattern as a whole, the directional difference of dynamic change of urban surface thermal landscape exists but is not very obvious.