J. C. Jiménez-Muñoz

Error sources on the land surface temperature retrieved from thermal infrared single channel remote sensing data

In this paper, a theoretical study complementary to others given in the literature about the errors committed on the land surface temperature retrieved from the radiative transfer equation in the thermal infrared region by remote sensing techniques has been analysed. For this purpose, the MODTRAN 3.5 code has been used in order to simulate different conditions and evaluate the influence of several parameters on the land surface temperature accuracy: atmospheric correction, noise of the sensor, land surface emissivity, aerosols and other gaseous absorbers, angular effects, wavelength uncertainty, full‐width half‐maximum of the sensor and band‐pass effects. The results show that the most important error source is due to atmospheric effects, which leads to an error on surface temperature between 0.2 K and 0.7 K, and land surface emissivity uncertainty, which leads to an error on surface temperature between 0.2 and 0.4 K. Hence, assuming typical uncertainties for remote sensing measurements, a total error for land surface temperature between 0.3 K and 0.8 K has been found, so it is difficult to achieve an accuracy lower than these values unless more accurate in situ values for emissivity and atmospheric parameters are available.

Split-Window Coefficients for Land Surface Temperature Retrieval From Low-Resolution Thermal Infrared Sensors

In this letter, we provide a complete set of split-window coefficients that can be used to retrieve land surface temperature (LST) from thermal infrared sensors onboard the most popular remote-sensing satellites: ERS-ATSR2, ENVISAT-AATSR, Terra/Aqua-MODIS, NOAA series-AVHRR, METOP-AVHRR3, GOES series-IMAGER, and MSG1/MSG2-SEVIRI. The coefficients have been obtained by minimization from an extensive simulated database constructed from MODTRAN radiative transfer code calculations, emissivity spectra extracted from spectral libraries, and spectral response functions of the thermal bands considered. This letter also analyzes the magnitude of the error on the LST retrieval and the contribution to the error of the different uncertainties. Results are summarized in a lookup table useful for scientists interested on land surface retrievals at global scale, thereby facilitating and homogenizing the task of retrieving this parameter from different common sensors.

Evaluation of the DART 3D model in the thermal domain using satellite/airborne imagery and ground-based measurements

This work provides an evaluation of the discrete anisotropy radiative transfer (DART) three-dimensional (3D) model in assessing the simulation of directional brightness temperatures (T b) at both sensor and surface levels. Satellite imagery acquired with the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), airborne imagery acquired with the Airborne Hyperspectral Scanner (AHS) sensor and ground-based measurements collected over an agricultural area were used to evaluate the DART model at nadir views. Directional radiometric temperatures measured with a goniometric system at ground level were also used to evaluate modelling results at different view angles. The DART model was evaluated over three homogeneous plots: bare soil (BS), green grass (GG) and sand (NS). The results show good agreement between the simulations and the satellite, airborne and ground-based measurements, with root mean square errors (RMSEs) less than 2.0 K. However, three major discrepancies were found: (1) differences greater than 4.0 K over BS when comparing DART and ASTER, attributed to turbulence-induced temperature fluctuations, (2) higher differences in sensor-level than in surface-level comparisons when using AHS due to thermal heterogeneity of the selected regions of interest in the image and also to differences in atmospheric correction performed over the imagery and the correction included in the DART model, especially for bands located in the lowest atmospheric transmissivity regions and (3) RMSEs greater than 2.0 K when comparing DART results and ground measurements over the NS plot, due to the strong emissivity correction in the 8.0–9.0 μm bands, where the measured emissivity was below 0.75. Despite these discrepancies, we show that the DART model is a useful tool for simulating remotely sensed thermal images over different landscapes. Finally, new versions of this model are continuously being released to solve technical problems and improve the simulation results.

Mapping wild pear trees Pyrus bourgaeana in Mediterranean forest using high-resolution QuickBird satellite imagery

Recent advances in spatial and spectral resolution of satellite imagery as well as in processing techniques are opening new possibilities of fine-scale vegetation analysis with interesting applications in natural resource management. Here we present the main results of a study carried out in Sierra Morena, Cordoba (southern Spain), aimed at assessing the potential of remote-sensing techniques to discriminate and map individual wild pear trees (Pyrus bourgaeana) in Mediterranean open woodland dominated by Quercus ilex. We used high spatial resolution (2.4 m multispectral/0.6 m panchromatic) QuickBird satellite imagery obtained during the summer of 2008. Given the size and features of wild pear tree crowns, we applied an atmospheric correction method, Fast Line-of-Sight Atmospheric Analysis of Spectral Hypercube (FLAASH), and six different fusion ‘pan-sharpening’ methods (wavelet ‘à trous’ weighted transform, colour normalized (CN), Gram–Schmidt (GS), hue–saturation–intensity (HSI) colour transformation, multidirection–multiresolution (MDMR), and principal component (PC)), to determine which procedure provides the best results. Finally, we assessed the potential of supervised classification techniques (maximum likelihood) to discriminate and map individual wild pear trees scattered over the Mediterranean open woodland.

Global Atmospheric Profiles from Reanalysis Information GAPRI: a new database for earth surface temperature retrieval

This paper presents the Global Atmospheric Profiles derived from Reanalysis Information (GAPRI) database, which was designed for earth surface temperature retrieval. GAPRI is a comprehensive compilation of selected atmospheric vertical profiles at global scale which can be used for radiative transfer simulation in order to obtain generalized algorithms to estimate land surface temperature (LST). GAPRI includes information on geopotential height, atmospheric pressure, air temperature, and relative humidity derived from the European Centre for Medium-Range Weather Forecasts Re-Analysis data from year 2011. The atmospheric profiles are structured for 29 vertical levels and extracted from a global spatial grid of about 0.75° × 0.75° latitude–longitude with a temporal resolution of 6 hours. The selection method is based in the extraction of clear sky profiles over different atmospheric weather conditions such as tropical, mid-latitude summer, subarctic, and arctic, while also considering sea and land areas and day- and night-time conditions. The GAPRI database was validated by comparing land and sea surface temperature values derived from it to those obtained using other existing atmospheric profile databases and in situ measurements. Moreover, GAPRI was also compared to previous radiosonde atmospheric profiles using simulated split-window algorithms. Results show good agreement between GAPRI and previous atmospheric databases, thus demonstrating the potential of GAPRI for studies related to forward simulations in the thermal infrared range. GAPRI is a freely available database that can be modified according to the user’s needs and local atmospheric conditions.

Estimation of the Spatially Distributed Surface Energy Budget for AgriSAR 2006, Part I: Remote Sensing Model Intercomparison

A number of energy balance models of variable complexity that use remotely sensed boundary conditions for producing spatially distributed maps of surface fluxes have been proposed. Validation typically involves comparing model output to flux tower observations at a handful of sites, and hence there is no way of evaluating the reliability of model output for the remaining pixels comprising a scene. To assess the uncertainty in flux estimation over a remote sensing scene requires one to conduct pixel-by-pixel comparisons of the output. The objective of this paper is to assess whether the simplifications made in a simple model lead to erroneous predictions or deviations from a more complex model and under which circumstances these deviations most likely occur. Two models, the S-SEBI and TSEB algorithms, which have potential for operationally monitoring ET with satellite data are described and a spatial inter-comparison is made. Comparisons of the spatially distributed flux maps from the two models are made using remotely sensed imagery collected over an agricultural test site in Northern Germany. With respect to model output for radiative and conductive fluxes no major differences are noted. Results for turbulent flux exchange demonstrate that under relatively dry conditions and over tall crops model output differs significantly. The overall conclusion is that under unstressed conditions and over homogeneous landcover a simple index model is adequate for determining the spatially distributed energy budget.

Estimation of the Spatially Distributed Surface Energy Budget for AgriSAR 2006, Part II: Integration of Remote Sensing and Hydrologic Modeling

In most hydrologic modeling studies, the hypothesis is made that an improvement in the modeled soil moisture leads to an improvement in the modeled surface energy balance. The objective of this paper is to assess whether this hypothesis is true. The study was performed over the winter wheat fields in the AgriSAR 2006 domain. Remotely sensed soil moisture values and latent heat fluxes were used, in combination with in situ observations. First, the land cover and saturated subsurface flow parameters were estimated using the in situ observations. A spatially distributed model simulation was then performed, for which the Brooks-Corey parameters were derived from a soil texture map, and of which the results were validated using the remote sensing data. The remotely sensed soil moisture values were then used to optimize the Brooks-Corey parameters. As expected, a better performance with respect to the soil moisture estimation was obtained. However, this did not improve the latent heat flux estimates. This can be explained by the consumption of water from the deeper soil layers by the vegetation. The overall conclusion is that, under conditions where evapotranspiration is limited by energy and not by the soil moisture content, surface soil moisture values alone are not sufficient for the optimization of hydrologic model results. More data sets are needed for this purpose.

Fluorescence estimation in the framework of the CEFLES2 campaign

Chlorophyll fluorescence (ChF) is a relevant indicator of the actual plant physiological status. In this article different methods to measure ChF from remote sensing are evaluated: the Fraunhofer Line Discrimination (FLD), the Fluorescence Radiative Method (FRM) and the improved Fraunhofer Line Discrimination (iFLD). The three methods have been applied to data acquired in the framework of the CarboEurope, FLEX and Sentinel-2 (CEFLES2) campaign in Les Landes, France in September 2007. Comparing with in situ measurements, the results indicate that the methods that provide the best results are the FLD and the iFLD with root mean square errors (RMSEs) of 0.4 and 0.5 mW m−2 sr−1 nm−1, respectively, while the FRM provides an error of 0.8 mW m−2 sr−1 nm−1.

The role of emissivity during the cooling of a body: an experimental design for a laboratory classroom

Mechanisms of heat transfer and Newtons law of cooling are introduced in the first physics and biophysics courses for a number of university science majors. Several papers have commented on the derivation of the exponential decay and validity of this law. However, the description of the phenomena is traditionally described without consideration of basic factors that contribute to the cooling rate of a body. One of these key factors is the emissivity of the body, which requires specific instrumentation to be measured. In particular, we present in this paper an experiment to record the cooling temperatures of an avian egg by means of a thermal camera. The objective is to comment on the dependence of the cooling process on emissivity, and then propose a methodology for estimating the emissivity of the cooling object. The method can be applied a priori to other bodies and is suitable for a biophysics laboratory classroom in higher education.

Global vegetation monitoring through multitemporal analysis of pathfinder AVHRR land database

We have applied a Land Surface Temperature algorithm to the whole Pathfinder AVHRR Land (PAL) database, aiming at studying the evolution of the vegetation at a global scale. The Land Surface Temperature parameter, along with NDVI, will allow retrieving vegetation changes between July 1981 and September 2001. We have also built a classification which takes into account both vegetation variations and thermal patterns, from NDVI and Air Temperature at 2 meters height data. This classification allows differentiating areas which present close vegetation changes throughout the year, but totally different climates, as for example in mountainous and semiarid regions. The main quality of this classification is that it does not need any a priori information on the encountered vegetation, and thus can evolve from year to year. Through the 20 years of data, the evolution of Land Surface Temperature shows to be strongly affected by orbital drift and satellite changes. This will require an adequate correction to allow deeper study. On the other hand, NDVI does not show this trend, but aerosol absorption from Mount Pinatubos eruption in June 1991 seems to corrupt temporarily the data in the northern hemisphere.