In-flight validation and recovery of water surface temperature with Landsat-5 thermal infrared data using an automated high-altitude lake validation site at Lake Tahoe

S.J. Hook; G. Chander; J. A. Barsi; R.E. Alley; A. Abtahi; Frank Don Palluconi; B. L. Markham; R.C. Richards; S.G. Schladow; D. L. Helder

doi:10.1109/tgrs.2004.839092

USGS · 70026363

In-flight validation and recovery of water surface temperature with Landsat-5 thermal infrared data using an automated high-altitude lake validation site at Lake Tahoe

S.J. Hook·G. Chander·J. A. Barsi·R.E. Alley·A. Abtahi·Frank Don Palluconi·B. L. Markham·R.C. Richards·S.G. Schladow·D. L. Helder

View original source ↗Export metadata ↓

Abstract

The absolute radiometric accuracy of the thermal infrared band (B6) of the Thematic Mapper (TM) instrument on the Landsat-5 (L5) satellite was assessed over a period of approximately four years using data from the Lake Tahoe automated validation site (California-Nevada). The Lake Tahoe site was established in July 1999, and measurements of the skin and bulk temperature have been made approximately every 2 min from four permanently moored buoys since mid-1999. Assessment involved using a radiative transfer model to propagate surface skin temperature measurements made at the time of the L5 overpass to predict the at-sensor radiance. The predicted radiance was then convolved with the L5B6 system response function to obtain the predicted L5B6 radiance, which was then compared with the radiance measured by L5B6. Twenty-four cloud-free scenes acquired between 1999 and 2003 were used in the analysis with scene temperatures ranging between 4/spl deg/C and 22/spl deg/C. The results indicate L5B6 had a radiance bias of 2.5% (1.6/spl deg/C) in late 1999, which gradually decreased to 0.8% (0.5/spl deg/C) in mid-2002. Since that time, the bias has remained positive (predicted minus measured) and between 0.3% (0.2/spl deg/C) and 1.4% (0.9/spl deg/C). The cause for the cold bias (L5 radiances are lower than expected) is unresolved, but likely related to changes in instrument temperature associated with changes in instrument usage. The in situ data were then used to develop algorithms to recover the skin and bulk temperature of the water by regressing the L5B6 radiance and the National Center for Environmental Prediction (NCEP) total column water data to either the skin or bulk temperature. Use of the NCEP data provides an alternative approach to the split-window approach used with instruments that have two thermal infrared bands. The results indicate the surface skin and bulk temperature can be recovered with a standard error of 0.6/spl deg/C. This error is larger than errors obtained with other instruments due, in part, to the calibration bias. L5 provides the only long-duration high spatial resolution thermal infrared measurements of the land surface. If these data are to be used effectively in studies designed to monitor change, it is essential to continue to monitor instrument performance in-flight and develop quantitative algorithms for recovering surface temperature.

Explore related subjects

IEEE Transactions on Geoscience and Remote Sensing Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

S.J. Hook, G. Chander, J. A. Barsi, R.E. Alley, A. Abtahi, Frank Don Palluconi, B. L. Markham, R.C. Richards, S.G. Schladow, D. L. Helder. 2004. In-flight validation and recovery of water surface temperature with Landsat-5 thermal infrared data using an automated high-altitude lake validation site at Lake Tahoe. https://doi.org/10.1109/tgrs.2004.839092

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2026

Across-track tilt concept for sun glint mitigation in future Earth-observing missions

Sun glint is a major limitation for spaceborne optical remote sensing of aquatic environments, particularly in tropical and subtropical regions. By dominating the at-sensor radiance, sun glint complicates atmospheric correction and reduces the quality of aquatic reflectance products derived across Earth observation missions. This study presents a machine learning framework to quantify sun glint probability and evaluates the across-track tilt observational concept as a potential sun glint mitigation approach for satellite remote sensing missions. More than 14,000 cloud-free Landsat 8 and 9 scenes were analyzed to quantify glint occurrence as a function of sun-sensor geometry, latitude, and season. Logistic regression and machine learning models identified solar zenith angle (SZA), wind speed, and viewing zenith angle (VZA) as the primary predictors of sun glint probability, achieving >98% classification accuracy. Sun glint contamination was found to be negligible for SZA > 40° but frequent at SZA <35°, particularly in forward-scattering geometries. Introducing a 3° across-track westward tilt reduces the proportion of glint-contaminated pixels by ~21% in tropical and subtropical regions, while larger tilts of 6°, 9°, and 12° yield reductions of 43%, 59%, and 74%, respectively. These findings underscore the value of incorporating glint mitigation strategies in future Earth observing mission designs. A modest across-track tilt, combined with post-acquisition glint detection and correction, offers a practical pathway to increasing the availability of high-quality aquatic observations for monitoring inland and coastal water quality and ecosystems.

IEEE Transactions on Geoscience and Remote Sensing↗

USGS2024

Novel approach for ranking DEMs: Copernicus DEM improves one arc second open global topography

We present a practical approach to intercompare a range of candidate digital elevation models (DEMs) based on predefined criteria and a statistically sound ranking approach. The presented approach integrates the randomized complete block design (RCBD) into a novel framework for DEM comparison. The method presented provides a flexible, statistically sound, and customizable tool for evaluating the quality of any raster—in this case, a DEM—by means of a ranking approach, which takes into account a confidence level and can use both quantitative and qualitative criteria. The users can design their own criteria for the quality evaluation in relation to their specific needs. The application of the RCBD method to rank six 1′′ global DEMs, considering a wide set of study sites, covering different morphological and land cover settings, highlights the potentialities of the approach. We used a suite of criteria relating to the differences in the elevation, slope, and roughness distributions compared to reference DEMs aggregated from 1- to 5-m light detection and ranging (LiDAR)-derived DEMs. Results confirmed the significant superiority of Copernicus DEM (CopDEM) 1′′ and its derivative forests and buildings removed DEM (FABDEM) as the overall best 1′′ global DEMs. They are slightly better than Advanced Land Observing Satellite (ALOS) and clearly outperform NASADEM and SRTM, which are, in turn, much better than advanced spaceborne thermal emission and reflection radiometer (ASTER).

IEEE Transactions on Geoscience and Remote Sensing↗

USGS2024

Ecological dissimilarity matters more than geographical distance when predicting land surface indicators using machine learning

Supervised training techniques, such as those used in machine learning, use generally large sets of in situ data to train models that can, in turn, be used to make predictions (or prediction maps) about the Earth’s surface in times or places where no in situ data exist. The purpose of the present study is to investigate, using a very large set of in situ data from across the western United States (U.S.), the conditions under which training data from a different geographic region where predictions are desired may be substituted. To do this, we train models using in situ data from level IV ecoregions and test how well these models predict surface conditions in different ecoregions. We characterize the difference between the possible pairs of ecoregion in terms of geographical (centroid-to-centroid) distance and “ecological dissimilarity.” Ecological dissimilarity between pairs of ecoregions is defined in two ways: 1) as the Euclidean distance in multivariate space defined by in situ indicators designed for monitoring purposes and 2) in terms of the difference in temporal behavior from model- and remote sensing-derived datasets. Although, overall, prediction error increases with geographical distance between training and testing ecoregions, our results indicate that ecological dissimilarity can be used to predict the error expected from a model trained with data from one ecoregion when applied in a different ecoregion.

IEEE Transactions on Geoscience and Remote Sensing↗