[Show abstract][Hide abstract] ABSTRACT: In all generality, retrieval methods dedicated to the quantification of terrestrial biophysical variables can be categorized into three main domains: 1) parametric regression, 2) non-parametric regression, and 3) physically-based methods. For the last few years, we have made significant advances in all these domains, including the development of software to automate these methods. It eventually led to a scientific software package ARTMO (Automated Radiative Transfer Models Operator) that embodies multiple toolboxes and a suite of leaf and canopy radiative transfer models (RTMs); http://ipl.uv.es/artmo/. The following toolboxes enable to fully exploit the three retrieval domains:
Parametric methods refer to the use of regression models through spectral indices. The ‘Spectral Indices’ toolbox allows systematic calculation of all possible band combinations of a sensor according to the formulation of an established index. The prediction efficiency of each index can be automatically evaluated against in-situ data or input data coming from a RTM by using a fitting curve (e.g. linear, exponential, power). Options to add noise, to control calibration/validation partitioning and various goodness-of-fit measures to assess the performances (e.g., r2, RMSE) are provided. The best performing regression model can subsequently be applied to an imagery, which leads to instantaneous mapping of the targeted biophysical variable.
Non-parametric methods refer to the use of machine learning regression algorithms (MLRA). The ‘MLRA’ toolbox encompasses a collection of linear and non-linear nonparametric regression techniques such as partial least squares (PLS), neural networks (NN), support vector regression (SVR), kernel ridge regression (KRR) and Gaussian processes regression (GPR) and others. Depending on the chosen MLRA, multi-output is possible (PLS, NN, KRR) or associated uncertainty estimates are delivered (GPR). This toolbox is designed in a similar way as the Spectral Indices toolbox; with the same type of calibration and validation options and goodness-of-fit measures provided. The best performing MLRA model can subsequently be applied to an imagery which leads to instantaneous mapping of the targeted biophysical variable(s).
Physically-based methods refer to the inversion of Lookup-table (LUT)-based RTMs through cost functions. This method is considered a physically-sound and robust to retrieve biophysical variables, but regularization strategies are required to mitigate the drawback of ill-posedness. The ‘Inversion’ toolbox encompasses a collection of more than 60 cost functions, originating from three different mathematical families, being: information measures, M-estimates and minimum contrast methods. Various regularization options can be introduced in the inversion, being: adding noise, multiple solutions, and data normalizing. Simultaneous retrieval of multiple variables is possible. Additional uncertainty estimates can be provided in the form of standard deviation and residuals. The best assessed inversion strategy can subsequently be applied to an imagery, which leads to mapping of the targeted biophysical variable(s).
In this work, all these methods were evaluated by using Simulated Sentinel-2 data against ground-based validation data from the ESA campaign SPARC (Barrax, Spain). Results will be presented for leaf area index (LAI) retrieval. Apart from retrieval accuracy also processing speed was analyzed. This work will close with consolidated guidelines towards powerful retrieval methods that are implementable in operational processing chains.
4th International Symposium on Recent Advances in Quantitative Remote Sensing: RAQRS'IV, Valencia, Spain; 09/2014
[Show abstract][Hide abstract] ABSTRACT: The FLuorescence EXplorer (FLEX), a candidate mission for ESA’s Earth Explorer 8, will be the first space mission optimized for estimation of terrestrial vegetation fluorescence at a global scale. The mission is proposed to fly in tandem with ESA’s Copernicus Sentinel-3 satellite. On board FLEX, the Fluorescence Imaging Sensor (FLORIS), will measure the radiance between 500 and 800 nm with a bandwidth between 0.1 nm and 2 nm, providing images with a 150 km swath and 300 m pixel size. This information will improve the methods for the estimation of classical biophysical parameters, as well as the introduction of fluorescence-related products, e.g., fluorescence yield. These products will allow a more thorough study of vegetation physiological status such as actual photosynthetic activity. Eventually, this information will improve our understanding of the way carbon moves between plants and the atmosphere and how it affects the carbon and water cycles.
Several scientific and industrial studies have been initiated by ESA to establish scientific benchmarks for the FLEX mission. One of the scientific studies is the Photosynthesis Study that considers the potential of fluorescence for quantifying photosynthesis, vegetation health, and stress status. To this end, the study involves development of a soil-vegetation-atmosphere-transfer (SVAT) model to quantitatively link fluorescence to photosynthesis. This model will eventually facilitate fluorescence retrievals from space, e.g., through inversion against FLEX observations. The SCOPE (Soil-Canopy Observation, Photosynthesis and Energy Balance) model was selected as baseline model and has been improved and extended with leaf biochemical sub-models. To facilitate the usability of SCOPE, the model has been subsequently integrated into the GUI framework of the ARTMO (Automated Radiative Transfer Models Operator) software package, hereafter referred to as ‘A-SCOPE’. Essentially, A SCOPE allows the user to: (1) configure and run SCOPE through interfaces; (2) simulate and store a massive quantity of spectra based on a look-up table (LUT); (3) plot groups of simulated spectra or fluxes with color gradients as a function of input variables; (4) export simulated spectra and associated meta-data to a text file for further processing.
Because SCOPE is fundamentally designed as an energy budget model, its large number of input variables currently makes it less suitable to be implemented into an operational processing scheme. Model simplification is required. A global sensitivity analysis (GSA) has been conducted to quantify the relative importance of each input parameter to model outputs. Considering that both surface reflectance and vegetation fluorescence emission will be FLEX key level-1 products, the GSA identified the following driving variables for these outputs: leaf area index, chlorophyll content, vegetation height, maximum carboxylation capacity, Ball-Berry stomatal conductance parameter, solar zenith angle, dry matter content, leaf water content, leaf thickness, and senescent material. Non-influential variables were: Extinction coefficient for Vcmax, leaf width, azimuth difference, and observation zenith angle. By setting the least influential variables to fixed values, the SCOPE model can be considerably simplified. The simplified model will be subsequently integrated into a FLEX End-to-end simulator that enables simulation of scenes as if generated by FLEX and development of inversion strategies.
4th International Symposium on Recent Advances in Quantitative Remote Sensing: RAQRS'IV, Valencia, Spain; 09/2014
[Show abstract][Hide abstract] ABSTRACT: !!!The Report will become available again as soon as it has officially been accepted and released for the public by our sponsors.!!!
Since conventional air pollution monitoring stations provide coarse-scale information on exposure to pollutants, only , a growing interest in monitoring and modeling urban air pollution to obtain information with a higher spatial resolution is apparent. One of the possibilities of street-scale monitoring is biomonitoring of urban vegetation. With increasing traffic intensity, leaves act as a natural sink for particulate matter (PM) (Kardel et al., 2011; Maher et al., 2008), and can even enrich in nitrogen (Laffray et al., 2010) or heavy metals, such as lead (Gajic et al., 2009). Besides deposition, retention and even enrichment of trace elements and metals, leaves are exposed to a whole range of traffic-induced gaseous pollutants such as nitrogen oxides (NOx), carbon monoxide (CO), carbon dioxide (CO2), sulfur dioxide (SO2) which have an impact on their physiological behavior.
Biomonitoring of natural vegetation allows the acquisition of well-defined samples at an affordable cost and allows the determination of air pollution at different time-scales. It reflects longer-term changes of environmental quality, because plant leaves accumulate pollution over months, or even years for evergreen species. Pollutants absorbed by vegetation can also be fixed into the plant system. By phytoremediation, i.e. the use of plants to mitigate pollutant concentrations in contaminated soils, water, or air, several tree species can be used to detoxify urban air affected by a high traffic load (Kvesitadze et al., 2006). Another advantage of a biomonitoring approach is the high spatial resolution that can be obtained.
[Show abstract][Hide abstract] ABSTRACT: Spatially distributed chlorophyll content of urban vegetation provides an important indicator of a plant's health status, which might depend on the habitat quality of the specific urban environment. Recent advances in optical remote sensing led to improved methodologies to monitor vegetation properties. The hyperspectral index NAOC (Normalized Area Over reflectance Curve) is one of these new tools that can be used for mapping chlorophyll content. As part of the BIOHYPE project, we present the work done to quantify vegetation chlorophyll content over the city of Valencia (Spain) based on chlorophyll measurements of four representative tree species: the London plane tree (Platanus x. acerifolia), the Canarian date palm (Phoenix canariensis), the European nettle tree (Celtisaustralis) and the white mulberry (Morus alba). Measurements were acquired during the summer of 2011, in a field campaign in which for 320 leaf samples, chlorophyll content was measured both in the laboratory and by using a SPAD-502 chlorophyll meter. Both methods were correlated (R2> 0.86), using best fit power type functions. During the field campaign an aircraft with a CASI (Compact Airborne Spectral Imager) hyperspectral sensor onboard overflew the city obtaining imagery with a spatial resolution of ~1 m suitable to identify individual urban trees. From the CASI data the NAOC index was calculated and linked with the laboratory chlorophyll content measurements. This led to a detailed chlorophyll content map with a RMSE of 15 μg cm-2. Chlorophyll map analysis at the individual crown level suggests the applicability to identify trees with lowered chlorophyll content due to a suboptimal habitat quality.
5th International Workshop on RS of Vegetation Fluorescence, Paris; 04/2014
[Show abstract][Hide abstract] ABSTRACT: A new atmospheric correction algorithm is proposed to support data analysis from the ESA's 8th Earth Explorer Fluorescence EXplorer (FLEX) candidate mission. The Fluorescence Imaging Spectrometer (FLORIS) on board FLEX covers, with a very high spectral resolution, a narrow spectral range, from 500 nm to 780 nm, ideal for vegetation fluorescence detection but insufficient for atmospheric characterization. For this reason, FLEX is planned as a tandem mission with Sentinel-3 (S3). Therefore, to perform the FLEX atmospheric correction, atmospheric parameters such as aerosol optical properties and water vapour content will be estimated from S3 data. Once the atmospheric state has been characterized, a second step deals with the retrieval of surface apparent reflectance, i.e. the surface reflectance modified by the fluorescence radiance emission. The first part of this paper is dedicated to the description of the method, summarising the main steps in the atmospheric characterization and in the succeeding surface apparent reflectance retrieval. In the second part of the paper, different databases have been simulated covering a wide range of atmospheric and surface reflectance properties to show accuracy obtained with the methodology proposed, especially over O2 absorption band spectral regions. The validation task is developed by comparing apparent reflectance retrieved from the atmospheric correction algorithm and those obtained using atmospheric parameters defined in the database creation. In addition, to demonstrate that accuracy obtained from the atmospheric correction is enough to provide a precise chlorophyll fluorescence retrieval, a first fluorescence estimation have been performed for all the cases covered by the simulated databases.
5th International Workshop on Remote Sensing of Vegetation proceedings; 04/2014
[Show abstract][Hide abstract] ABSTRACT: A diierential absorption technique is used in the 940 nm to re-trieve the columnar water vapour content. Geospatial co-registration between SENTINEL-3 and FLORIS Level-1 data is necesary to be performed as rst step. In addition, spectral calibra-tion of FLORIS data and cross calibration between both products is essential before starting with the atmospheric correction process. can be provided by MODTRAN5 (or can be approximated by a straight line) can be approximated by polynomial interpolation, using selected key points over entire spectral range., "MERIS/AATSR synergy algorithms for cloud screening, aerosol retrieal, and atmospheric correction," Tech. Rep., 2009., "An integrated model of soil-canopy spectral radiances, photosynthesis, uorescence, temperature and energy balance," Biogeosciences, vol. 6, no. 12, pp. 3109–3129, 2009. -Sentinel-3 level 1 lgorithms theorethical baseline document-part 2, " Optical products[SY-24], vol. Level 1c ARBD, no. Ref.:S3-DD-TAF-SY-0062. BIBLIOGRAPHY (Eq.2) (Eq.3) (Eq.4) () -2 -1 -1 0.2 mWm sr nm s F ε ≤ Integrated values at canopy level are the ones required by models. for instantaneous observations. 300 m original spatial resolution. Includes as subproducts carotenoids / chlorophyll ratio and violaxanthin / zeaxan-thin ratio, responsible for regulated energy dissipation. Accounts for the fraction of light absorbed by non-photochemical pigments (anthocyanin). Ratio between energy emitted as uorescence versus energy absorbed by chlorophyll-a. Accounts for actual chlorophyll speciic absorption. Actual electron current after charge separation at PSII, also accounts for instantaneous surface temperature eeects. Deened as "actual photosynthesis / potential photosynthesis" Deened at Level-2, but recommended usage as Level-3 product m FLEX is planned as a tandem mission with Copernicus' mission Sen-tinel-3 (S3). S-3 instruments are necessary to extract information about the atmosphere and to perform an accurate atmospheric co-rrection of the acquired images. S3's Ocean and Land Colour Imaging spectrometer (OLCI), cove-ring from 400 nm to 1020 nm, and the Sea and Land Surface Tem-perature Radiometer (SLSTR), covering from visible to Thermal Infrared (TIR) with its dual view, will provide the information needed to characterize the atmosphere. FLEX SENTINEL-3
5th ESA's Fluorescence Workshop, Paris (FRANCE); 03/2014
[Show abstract][Hide abstract] ABSTRACT: A detailed description of FLEX END-TO-END Simulator (E2ES) scientific modules, i.e. the Scene Generator (SG) and the Level-2 Retrieval (L2R) modules is presented in this paper. On one hand, the SG offers the possibility to simulate a wide range of realistic scenarios for FLEX/Sentinel-3 (S3) tandem mission by coupling two radiative transfer codes, at soil level and at atmospheric level. On the other hand, the L2R contains a set of algorithms able to perform the atmospheric correction and the fluorescence retrieval only using Top Of Atmosphere (TOA) radiances as input. In addition, L2R provides as output not only fluorescence radiance, but also a list of biophysical parameters such as Chlorophyll Content (Chl.) and Leaf Area Index (LAI) among others. In the literature, many SGs for optic passive missions only offer a collection of images predefined off-line. User simulation capabilities are then restricted to choose between one of those predefined images. However, FLEX SG allows the user to define its own scene configuring atmospheric and surface properties. In addition, the L2R does not require the usage of any auxiliary input data, which makes this module autonomous.
5th ESA's fluorescence workshop, Paris (FRACE); 03/2014
[Show abstract][Hide abstract] ABSTRACT: Spatially distributed chlorophyll content of urban vegetation provides an important indicator of a plant's health status, which might depend on the habitat quality of the specific urban environment. Recent advances in optical remote sensing led to improved methodologies to monitor vegetation properties. The hyperspectral index NAOC (Normalized Area Over reflectance Curve) is one of these new tools that can be used for mapping chlorophyll content. In this paper we present the work done to quantify vegetation chlorophyll content over the city of Valencia (Spain) based on chlorophyll measurements of four representative tree species: the London plane tree (Platanus x. acerifolia), the Canarian date palm (Phoenix canariensis), the European nettle tree (Celtis australis) and the white mulberry (Morus alba). Measurements were acquired during the summer of 2011, in a field campaign in which for 320 leaf samples, chlorophyll content was measured both in the laboratory and by using a SPAD-502 chlorophyll meter. Both methods were correlated (R2 > 0.86), using best fit power type functions. During the field campaign an aircraft with a CASI (Compact Airborne Spectral Imager) hyperspectral sensor onboard overflew the city obtaining imagery with a spatial resolution of ∼1 m suitable to identify individual urban trees. From the CASI data the NAOC index was calculated and linked with the laboratory chlorophyll content measurements. This led to a detailed chlorophyll content map with a RMSE of 15 μg cm−2. Chlorophyll map analysis at the individual crown level suggests the applicability to identify trees with lowered chlorophyll content due to a suboptimal habitat quality.
[Show abstract][Hide abstract] ABSTRACT: Passive steady-state chlorophyll fluorescence (Fs) provides a direct diagnosis of the functional status of vegetation photosynthesis. With the prospect of mapping Fs using remote sensing techniques, field measurements are mandatory to understand to which extent Fs allows detecting plant stress in different environments. Trees of four common species in Valencia were classified in either a low or a high local traffic exposure class based on their leaf magnetic value. Upward and downward hyperspectral fluorescence yield (FY) and indices based on the two Fs peaks (at 687 and 741 nm) were calculated. FY indices of P. canariensis and P. x acerifolia were significantly different between the two traffic exposure classes defined, but not for C. australis nor M. alba. While chlorophyll content could not indicate the difference between low and high traffic exposure, the FY(687)/FY(741) peak ratio increased significantly (p < 0.05) for both leaf sides for the higher traffic exposure class.
[Show abstract][Hide abstract] ABSTRACT: Biophysical parameters such as leaf chlorophyll content (LCC) and leaf area index (LAI) are standard vegetation products that can be retrieved from Earth observation imagery. This paper introduces a new machine learning regression algorithms (MLRAs) toolbox into the scientific Automated Radiative Transfer Models Operator (ARTMO) software package. ARTMO facilitates retrieval of biophysical parameters from remote observations in a MATLAB graphical user interface (GUI) environment. The MLRA toolbox enables analyzing the predictive power of various MLRAs in a semiautomatic and systematic manner, and applying a selected MLRA to multispectral or hyperspectral imagery for mapping applications. It contains both linear and nonlinear state-of-the-art regression algorithms, in particular linear feature extraction via principal component regression (PCR), partial least squares regression (PLSR), decision trees (DTs), neural networks (NNs), kernel ridge regression (KRR), and Gaussian processes regression (GPR). The performance of multiple implemented regression strategies has been evaluated against the SPARC dataset (Barrax, Spain) and simulated Sentinel-2 (8 bands), CHRIS (62 bands) and HyMap (125 bands) observations. In general, nonlinear regression algorithms (NN, KRR, and GPR) outperformed linear techniques (PCR and PLSR) in terms of accuracy, bias, and robustness. Most robust results along gradients of training/validation partitioning and noise variance were obtained by KRR while GPR delivered most accurate estimations. We applied a GPR model to a hyperspectral HyMap flightline to map LCC and LAI. We exploited the associated uncertainty intervals to gain insight in the per-pixel performance of the model.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 01/2014; · 2.87 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Terrestrial gross primary production (GPP) is an important parameter to explore and quantify carbon fixation by plant ecosystems at various scales. Remote sensing (RS) offers a unique possibility to investigate GPP in a spatially explicit fashion; however, budgeting of terrestrial carbon cycles based on this approach still remains uncertain. To improve calculations, spatio-temporal variability of GPP must be investigated in more detail on local and regional scales. The overarching goal of this study is to enhance our knowledge on how environmentally induced changes of photosynthetic light-use efficiency (LUE) are linked with optical RS parameters. Diurnal courses of sun-induced fluorescence yield (FSyield) and the photochemical reflectance index of corn were derived from high-resolution spectrometric measurements and their potential as proxies for LUE was investigated. GPP was modeled using Monteith's LUE-concept and optical-based GPP and LUE values were compared with synoptically acquired eddy covaria
Global Change Biology 10/2013; 16(1):171-186. · 8.22 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Inversion of radiative transfer models (RTM) using a lookup-table (LUT) approach against satellite reflectance data can lead to concurrent retrievals of biophysical parameters such as leaf chlorophyll content (Chl) and leaf area index (LAI), but optimization strategies are not consolidated yet. ESA's upcoming satellites Sentinel-2 (S2) and Sentinel-3 (S3) aim to ensure continuity of old generation satellite sensors by providing superspectral images of high spatial and temporal resolution. This unprecedented data availability leads to an urgent need for developing robust, accurate, and operational retrieval methods. For three simulated Sentinel settings (S2-10 m: 4 bands, S2-20 m: 8 bands and S3-OLCI: 19 bands) various optimization strategies in LUT-based RTM inversion have been evaluated, being the role of i) added noise, ii) multiple best solutions, iii) combined parameters (Chl ×LAI), and iv) applied cost functions. By inverting the PROSAIL model and using data from the ESA-led field campaign SPARC (Barrax, Spain), it was demonstrated that introducing noise and opting for multiple best solutions in the inversion considerably improved retrievals. However, the widely used RMSE was not the best performing cost function. Three families of alternative cost functions were applied here: information measures, minimum contrast, and M-estimates. We found that so-called "Power divergence measure", "Trigonometric", and spectral measure with "Contrast function K(x) = -log(x) + x", yielded more accurate results, although this also depended on the biophysical parameter. Particularly, when simultaneous retrieval of multiple biophysical parameters is the objective then "Contrast function K(x) = -log(x) + x" provided most consistent optimized estimates of leaf Chl, LAI and canopy Chl across the different Sentinel configurations (relative RMSE: 24-29 %).
IEEE Transactions on Geoscience and Remote Sensing 01/2013; · 3.47 Impact Factor