Retrieval Of Biophysical Variables Research Articles

Biophysical Variable Retrieval of Silage Maize with Gaussian Process Regression and Hyperparameter Optimization Algorithms

Quantification of vegetation biophysical variables such as leaf area index (LAI), fractional vegetation cover (fCover), and biomass are among the key factors across hydrological, agricultural, and irrigation management studies. The present study proposes a kernel-based machine learning algorithm capable of performing adaptive and nonlinear data fitting so as to generate a suitable, accurate, and robust algorithm for spatio-temporal estimation of the three mentioned variables using Sentinel-2 images. To this aim, Gaussian process regression (GPR)–particle swarm optimization (PSO), GPR–genetic algorithm (GA), GPR–tabu search (TS), and GPR–simulated annealing (SA) hyperparameter-optimized algorithms were developed and compared against kernel-based machine learning regression algorithms and artificial neural network (ANN) and random forest (RF) algorithms. The accuracy of the proposed algorithms was assessed using digital hemispherical photography (DHP) data and destructive measurements performed during the growing season of silage maize in agricultural fields of Ghale-Nou, southern Tehran, Iran, in the summer of 2019. The results on biophysical variables against validation data showed that the developed GPR-PSO algorithm outperformed other algorithms under study in terms of robustness and accuracy (0.917, 0.931, 0.882 using R2 and 0.627, 0.078, and 1.99 using RMSE in LAI, fCover, and biomass of Sentinel-2 20 m, respectively). GPR-PSO also possesses the unique ability to generate pixel-based uncertainty maps (confidence level) for prediction purposes (i.e., estimated uncertainty level <0.7 in LAI, fCover, and biomass, for 96%, 98%, and 71% of the total study area, respectively). Altogether, GPR-PSO appears to be the most suitable option for mapping biophysical variables at the local scale using Sentinel-2 images.

Local interpretation of machine learning models in remote sensing with SHAP: the case of global climate constraints on photosynthesis phenology

ABSTRACT Data-driven models using machine learning have been widely used in remote-sensing applications such as the retrieval of biophysical variables and land cover classification. However, these models behave as a ‘black box’, meaning that the relationships between the input and predicted variables are hard to interpret. Recent regression models that downscale sun-induced fluorescence (SIF) with MODIS and weather variables are an example. The impact of weather variables on the predicted SIF in these models is unknown. The explanation of such weather–SIF relationships would aid in the understanding of climate-related constraints on photosynthesis phenology since SIF is a proxy of gross primary productivity. Here, we used SHapley Additive exPlanations (SHAP) – a novel technique based on game theory – for explaining the contribution of input variables to the individual predictions in a machine learning model. We explored the capabilities of this technique with a weather–SIF model. The regression model predicted ESA-TROPOSIF measurements from ERA5-Land air temperature, shortwave radiation, and vapour-pressure-deficit (VPD) data. The SHAP values of the model were estimated at the start and end of the growing season for the entire globe. These values depicted the global constraints of the three climate variables on the photosynthetically active season and confirmed existing knowledge on the limiting factors of terrestrial photosynthesis with unprecedented spatial detail. Radiation was the limiting factor in tropical rainforest and VPD constrained the start and end of the growing season in tropical dryland ecosystems. In extra-tropical regions, temperature was the main limiting factor during the start of the growing season, but both temperature and radiation constrained photosynthesis at the end of the growing season. This technique may help future remote sensing studies that require the use of non-interpretable machine-learning regression models and explain how input variables contribute to the model prediction in a spatiotemporally explicit manner.

Quantifying uncertainty in high resolution biophysical variable retrieval with machine learning

The estimation of biophysical variables is at the core of remote sensing science, allowing a close monitoring of crops and forests. Deriving temporally resolved and spatially explicit maps of parameters of interest has been the subject of intense research. However, deriving products from optical sensors is typically hampered by cloud contamination and the trade-off between spatial and temporal resolutions. In this work we rely on the HIghly Scalable Temporal Adaptive Reflectance Fusion Model (HISTARFM) algorithm to generate long gap-free time series of Landsat surface reflectance data by fusing MODIS and Landsat reflectances. An artificial neural network is trained on PROSAIL inversion to predict monthly biophysical variables at 30 m spatial resolution with associated, realistic uncertainty bars. We emphasize the need for a more thorough analysis of uncertainty, and propose a general and scalable approach to combine both epistemic and aleatoric uncertainties by exploiting Monte Carlo (MC) dropout techniques from the trained artificial network and the propagation of HISTARFM uncertainties through the model, respectively. A model recalibration was performed in order to provide reliable uncertainties. We provide new high resolution products of several key variables to quantify the terrestrial biosphere: Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), Canopy Water Content (CWC) and Fractional Vegetation Cover (FVC) are at 30 m Landsat spatial resolution and over large continental areas. Two study areas are considered: the large heterogeneous but moderately cloud covered contiguous United States, and the homogeneous but largely cloud covered Amazonia. The produced vegetation products largely agree with the test dataset (R = 0.90, RMSE = 0.80 m2/m2 and ME = 0.12 m2/m2 for LAI, and R = 0.98, RMSE = 0.07 and ME = 0.01 for FAPAR) providing low error and high accuracy. Additionally, the validation considers a thorough comparison with operational and largely validated medium resolution products, such as the Moderate-Resolution Imaging Spectroradiometer (MODIS) and Copernicus Global Land Service. Our products presented a good agreement and consistency with both MODIS (R = 0.84 and R = 0.85 for LAI and FAPAR, respectively) and Copernicus (R = 0.92 and R = 0.91 for LAI and FAPAR, respectively). To foster a wider adoption and reproducibility of the methodology we provide an application in GEE and source code at:https://github.com/IPL-UV/ee_BioNet/

Assessing the Potential of Fully Polarimetric Mono- and Bistatic SAR Acquisitions in L-Band for Crop and Soil Monitoring

Theoretical studies have shown that the use of simultaneous mono- and bistatic synthetic aperture radar (SAR) data could be beneficial to agriculture and soil moisture monitoring. This study makes use of extensive ground-truth measurements and synchronous high-resolution fully polarimetric mono- and bistatic airborne SAR data in L-band to assess and compare the sensitivity of mono- and multistatic systems to the maize canopy row structure and biophysical variables, as well as to soil moisture and surface roughness in both vegetated and bare fields. The effect of the row structure of maize crops is assessed through the impact of the orientation of the planting rows relative to the sensor beam on microwave scattering measurements. The results of this analysis suggest that the row orientation of maize crops has a significant influence on both mono- and bistatic scattering measurements in both copolarizations, and especially, in HH, while the cross polarizations are not affected. Furthermore, the study also shows through a linear regression analysis that bistatic data, even with a very small bistatic baseline, can provide valuable additional information for maize crop biophysical variable retrieval, which however does not appear to be the case for soil moisture retrieval over bare soils.

The influence of surface canopy water on the relationship between L-band backscatter and biophysical variables in agricultural monitoring

The presence of surface water on the canopy affects radar backscatter. However, its influence on the relationship between radar backscatter and crop biophysical parameters has not been investigated. The aim of this study was to quantify the influence of surface canopy water (SCW) on the relationship between L-band radar backscatter and biophysical variables of interest in agricultural monitoring. In this study, we investigated the effect of SCW on the relationship between co- and cross-polarized radar backscatter, cross ratios (VH/VV and HV/HH), and radar vegetation index (RVI) and dry biomass, vegetation water content (VWC), plant height and leaf area index (LAI). In addition, the effect of SCW on estimated vegetation optical depth (VOD) and its relationship with internal VWC was investigated. The analysis was based on data collected during a field experiment in Florida, USA in 2018. A corn field was scanned with a truck-mounted, fully polarimetric, L-band radar along with continuous monitoring of SCW (dew, interception) and soil moisture every 15min for 58 days. In addition, pre-dawn destructive sampling was conducted to measure internal vegetation water content and dry biomass. Results showed that the presence of SCW can increase the radar backscatter up to 2 dB and this effect was lower for cross ratios (CRs) and RVI. The Spearman's rank correlations between radar observables and biophysical parameters were, on average, 0.2 higher for dry vegetation compared to wet vegetation. The estimated VOD from wet vegetation was generally higher than those from dry vegetation, which led to different fitting parameter (so-called b) values in the linear fit between VOD and VWC. The results presented here underscore the importance of considering the influence of SCW on the retrieval of biophysical variables of interest in agricultural monitoring. In particular, they highlight the importance of overpass time, and the impact that daily patterns in dew and interception can have on the retrieval of biophysical variables of interest.

Sino–EU Earth Observation Data to Support the Monitoring and Management of Agricultural Resources

Novel approaches and algorithms to estimate crop physiological processes from Earth Observation (EO) data are essential to develop more sustainable management practices in agricultural systems. Within this context, this paper presents the results of different research activities carried out within the ESA-MOST Dragon 4 programme. The paper encompasses two research avenues: (a) the retrieval of biophysical variables of crops and yield prediction; and (b) food security related to different crop management strategies. Concerning the retrieval of variables, results show that LAI, derived by radiative transfer model (RTM) inversion, when assimilated into a crop growth model (i.e., SAFY) provides a way to assess yields with a higher accuracy with respect to open loop model runs: 1.14 t·ha−1 vs 4.42 t·ha−1 RMSE for assimilation and open loop, respectively. Concerning food security, results show that different pathogens could be detected by remote sensing satellite data. A k coefficient higher than 0.84 was achieved for yellow rust, thus assuring a monitoring accuracy, and for the diseased samples k was higher than 0.87. Concerning permanent crops, neural network (NN) algorithms allow classification of the Pseudomonas syringae pathogen on kiwi orchards with an overall accuracy higher than 91%.

A Kernel-Driven BRDF Approach to Correct Airborne Hyperspectral Imagery over Forested Areas with Rugged Topography

Airborne hyper-spectral imaging has been proven to be an efficient means to provide new insights for the retrieval of biophysical variables. However, quantitative estimates of unbiased information derived from airborne hyperspectral measurements primarily require a correction of the anisotropic scattering properties of the land surface depicted by the bidirectional reflectance distribution function (BRDF). Hitherto, angular BRDF correction methods rarely combined viewing-illumination geometry and topographic information to achieve a comprehensive understanding and quantification of the BRDF effects. This is in particular the case for forested areas, frequently underlaid by rugged topography. This paper describes a method to correct the BRDF effects of airborne hyperspectral imagery over forested areas overlying rugged topography, referred in the reminder of the paper as rugged topography-BRDF (RT-BRDF) correction. The local viewing and illumination geometry are calculated for each pixel based on the characteristics of the airborne scanner and the local topography, and these two variables are used to adapt the Ross-Thick-Maignan and Li-Transit-Reciprocal kernels in the case of rugged topography. The new BRDF model is fitted to the anisotropy of multi-line airborne hyperspectral data. The number of pixels is set at 35,000 in this study, based on a stratified random sampling method to ensure a comprehensive coverage of the viewing and illumination angles and to minimize the fitting error of the BRDF model for all bands. Based on multi-line airborne hyperspectral data acquired with the Chinese Academy of Forestry’s LiDAR, CCD, and Hyperspectral system (CAF-LiCHy) in the Pu’er region (China), the results applying the RT-BRDF correction are compared with results from current empirical (C, and sun-canopy-sensor (SCS) adds C (SCS+C)) and semi-physical (SCS) topographic correction methods. Both quantitative assessment and visual inspection indicate that RT-BRDF, C, and SCS+C correction methods all reduce the topographic effects. However, the RT-BRDF method appears more efficient in reducing the variability in reflectance of overlapping areas in multiple flight-lines, with the advantage of reducing the BRDF effects caused by the combination of wide field of view (FOV) airborne scanner, rugged topography, and varying solar illumination angle over long flight time. Specifically, the average decrease in coefficient of variation (CV) is 3% and 3.5% for coniferous forest and broadleaved forest, respectively. This improvement is particularly marked in the near infrared (NIR) region (i.e., >750 nm). This finding opens new possible applications of airborne hyperspectral surveys over large areas.

Retrieval of crop biophysical parameters from Sentinel-2 remote sensing imagery

The red-edge bands place the recently available multispectral Sentinel-2 imagery at an advantage over other multispectral sensors, and hypothetically offer improved crop biophysical variable retrieval accuracy. In this study, Sentinel-2 data was tested for its ability to estimate winter wheat leaf area index (LAI), leaf chlorophyll content (LCC) and canopy chlorophyll content (CCC). Artificial neural network (ANN) and look-up table (LUT) (based on PROSAIL simulations) and vegetation index (VI) methods were applied to retrieve biophysical parameters, and compared with the biophysical processor module embedded in the Sentinel Application Platform (SNAP) software. Based on a set of in situ measurements (62 samples) and near-synchronous Sentinel-2 images, the inversion approaches were applied and validated. The results showed that: 1) Sentinel-2 red-edge bands improved the retrievals of chlorophyll / LAI compared to traditional VIs; 2) the red-edge VIs outperformed other approaches; and 3) the SNAP biophysical processor obtained comparable accuracies of LAI and CCC estimation compared to the ANN and LUT approaches, giving R2 values above 0.5 with relatively low RMSE (1.53 m2/m2 for LAI, and 148.58 μg/cm2 for CCC). We recommend VI retrieval approach for small region with ground measurements, whereas where ground data is not available, SNAP is applicable for versatile and rapid winter wheat parameter estimation (though results need to be evaluated alongside the provided quality indicators). Summarizing, the results demonstrate the suitability of Sentinel-2 data, especially its red-edge bands, for crop biophysical variables retrieval. Future studies will need to make comparisons across canopy types to better assess the capability of the SNAP biophysical processor.

Global Estimation of Biophysical Variables from Google Earth Engine Platform

This paper proposes a processing chain for the derivation of global Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), Fraction Vegetation Cover (FVC), and Canopy water content (CWC) maps from 15-years of MODIS data exploiting the capabilities of the Google Earth Engine (GEE) cloud platform. The retrieval chain is based on a hybrid method inverting the PROSAIL radiative transfer model (RTM) with Random forests (RF) regression. A major feature of this work is the implementation of a retrieval chain exploiting the GEE capabilities using global and climate data records (CDR) of both MODIS surface reflectance and LAI/FAPAR datasets allowing the global estimation of biophysical variables at unprecedented timeliness. We combine a massive global compilation of leaf trait measurements (TRY), which is the baseline for more realistic leaf parametrization for the considered RTM, with large amounts of remote sensing data ingested by GEE. Moreover, the proposed retrieval chain includes the estimation of both FVC and CWC, which are not operationally produced for the MODIS sensor. The derived global estimates are validated over the BELMANIP2.1 sites network by means of an inter-comparison with the MODIS LAI/FAPAR product available in GEE. Overall, the retrieval chain exhibits great consistency with the reference MODIS product (R2 = 0.87, RMSE = 0.54 m2/m2 and ME = 0.03 m2/m2 in the case of LAI, and R2 = 0.92, RMSE = 0.09 and ME = 0.05 in the case of FAPAR). The analysis of the results by land cover type shows the lowest correlations between our retrievals and the MODIS reference estimates (R2 = 0.42 and R2 = 0.41 for LAI and FAPAR, respectively) for evergreen broadleaf forests. These discrepancies could be attributed mainly to different product definitions according to the literature. The provided results proof that GEE is a suitable high performance processing tool for global biophysical variable retrieval for a wide range of applications.

Hyperspectral dimensionality reduction for biophysical variable statistical retrieval

Current and upcoming airborne and spaceborne imaging spectrometers lead to vast hyperspectral data streams. This scenario calls for automated and optimized spectral dimensionality reduction techniques to enable fast and efficient hyperspectral data processing, such as inferring vegetation properties. In preparation of next generation biophysical variable retrieval methods applicable to hyperspectral data, we present the evaluation of 11 dimensionality reduction (DR) methods in combination with advanced machine learning regression algorithms (MLRAs) for statistical variable retrieval. Two unique hyperspectral datasets were analyzed on the predictive power of DR+MLRA methods to retrieve leaf area index (LAI): (1) a simulated PROSAIL reflectance data (2101 bands), and (2) a field dataset from airborne HyMap data (125 bands). For the majority of MLRAs, applying first a DR method leads to superior retrieval accuracies and substantial gains in processing speed as opposed to using all bands into the regression algorithm. This was especially noticeable for the PROSAIL dataset: in the most extreme case, using the classical linear regression (LR), validation results RCV2 (RMSECV) improved from 0.06 (12.23) without a DR method to 0.93 (0.53) when combining it with a best performing DR method (i.e., CCA or OPLS). However, these DR methods no longer excelled when applied to noisy or real sensor data such as HyMap. Then the combination of kernel CCA (KCCA) with LR, or a classical PCA and PLS with a MLRA showed more robust performances (RCV2 of 0.93). Gaussian processes regression (GPR) uncertainty estimates revealed that LAI maps as trained in combination with a DR method can lead to lower uncertainties, as opposed to using all HyMap bands. The obtained results demonstrated that, in general, biophysical variable retrieval from hyperspectral data can largely benefit from dimensionality reduction in both accuracy and computational efficiency.

Modeling Canopy Reflectance Over Sloping Terrain Based on Path Length Correction

Sloping terrain induces distortion of canopy reflectance (CR), and the retrieval of biophysical variables from remote sensing data needs to account for topographic effects. We developed a 1-D model (the path length correction (PLC)-based model) for simulating CR over sloping terrain. The effects of sloping terrain on single-order and diffuse scatterings are accounted for by PLC and modification of the fraction of incoming diffuse irradiance, respectively. The PLC model was validated via both Monte Carlo and remote sensing image simulations. The comparison with the Monte Carlo simulation revealed that the PLC model can capture the pattern of slope-induced reflectance distortion with high accuracy (red band: $R^{2} = 0.88$ ; root-mean-square error (RMSE) = 0.0045; relative RMSE (RRMSE) = 15%; near infrared response (NIR) band: $R^{2} = 0.79$ ; RMSE = 0.041; RRMSE = 16%). The comparison of the PLC-simulated results with remote sensing observations acquired by the Landsat8-OLI sensor revealed an accuracy similar to that with the Monte Carlo simulation (red band: $R^{2} = 0.83$ ; RMSE = 0.0053; RRMSE = 13%; NIR band: $R^{2} = 0.77$ ; RMSE = 0.023; RRMSE = 8%). To further validate the PLC model, we used it to implement topographic normalization; the results showed a large reduction in topographic effects after normalization, which implied that the PLC model captures reflectance variations caused by terrain. The PLC model provides a promising tool to improve the simulation of CR and the retrieval of biophysical variables over mountainous regions.

A regression approach to the mapping of bio-physical characteristics of surface sediment using in situ and airborne hyperspectral acquisitions

Remote sensing has been successfully utilized to distinguish and quantify sediment properties in the intertidal environment. Classification approaches of imagery are popular and powerful yet can lead to site- and case-specific results. Such specificity creates challenges for temporal studies. Thus, this paper investigates the use of regression models to quantify sediment properties instead of classifying them. Two regression approaches, namely multiple regression (MR) and support vector regression (SVR), are used in this study for the retrieval of bio-physical variables of intertidal surface sediment of the IJzermonding, a Belgian nature reserve. In the regression analysis, mud content, chlorophyll a concentration, organic matter content, and soil moisture are estimated using radiometric variables of two airborne sensors, namely airborne hyperspectral sensor (AHS) and airborne prism experiment (APEX) and and using field hyperspectral acquisitions by analytical spectral device (ASD). The performance of the two regression approaches is best for the estimation of moisture content. SVR attains the highest accuracy without feature reduction while MR achieves good results when feature reduction is carried out. Sediment property maps are successfully obtained using the models and hyperspectral imagery where SVR used with all bands achieves the best performance. The study also involves the extraction of weights identifying the contribution of each band of the images in the quantification of each sediment property when MR and principal component analysis are used.

Active Learning Methods for Efficient Hybrid Biophysical Variable Retrieval

Kernel-based machine learning regression algorithms (MLRAs) are potentially powerful methods for being implemented into operational biophysical variable retrieval schemes. However, they face difficulties in coping with large training datasets. With the increasing amount of optical remote sensing data made available for analysis and the possibility of using a large amount of simulated data from radiative transfer models (RTMs) to train kernel MLRAs, efficient data reduction techniques will need to be implemented. Active learning (AL) methods enable to select the most informative samples in a dataset. This letter introduces six AL methods for achieving optimized biophysical variable estimation with a manageable training dataset, and their implementation into a Matlab-based MLRA toolbox for semi-automatic use. The AL methods were analyzed on their efficiency of improving the estimation accuracy of leaf area index and chlorophyll content based on PROSAIL simulations. Each of the implemented methods outperformed random sampling, improving retrieval accuracy with lower sampling rates. Practically, AL methods open opportunities to feed advanced MLRAs with RTM-generated training data for development of operational retrieval models.

Evaluation of Six Algorithms to Monitor Wheat Leaf Nitrogen Concentration

The rapid and non-destructive monitoring of the canopy leaf nitrogen concentration (LNC) in crops is important for precise nitrogen (N) management. Nowadays, there is an urgent need to identify next-generation bio-physical variable retrieval algorithms that can be incorporated into an operational processing chain for hyperspectral satellite missions. We assessed six retrieval algorithms for estimating LNC from canopy reflectance of winter wheat in eight field experiments. These experiments represented variations in the N application rates, planting densities, ecological sites and cultivars and yielded a total of 821 samples from various places in Jiangsu, China over nine consecutive years. Based on the reflectance spectra and their first derivatives, six methods using different numbers of wavelengths were applied to construct predictive models for estimating wheat LNC, including continuum removal (CR), vegetation indices (VIs), stepwise multiple linear regression (SMLR), partial least squares regression (PLSR), artificial neural networks (ANNs), and support vector machines (SVMs). To assess the performance of these six methods, we provided a systematic evaluation of the estimation accuracies using the six metrics that were the coefficients of determination for the calibration (R2C) and validation (R2V) sets, the root mean square errors of prediction (RMSEP) for the calibration and validation sets, the ratio of prediction to deviation (RPD), the computational efficiency (CE) and the complexity level (CL). The following results were obtained: (1) For the VIs method, SAVI(R1200, R705) produced a more accurate estimation of the LNC than other indices, with R²C, R²V, RMSEP, RPD and CE values of 0.844, 0.795, 0.384, 2.005 and 0.10 min, respectively; (2) For the SMLR, PLSR, ANNs and SVMs methods, the SVMs using the first derivative canopy spectra (SVM-FDS) offered the best accuracy in terms of R²C, R²V, RMSEP, RPD, and CE, at 0.96, 0.78, 0.37, 2.02, and 21.17, respectively; (3) The PLSR-FDS, ANN-OS and SVM-FDS methods yield similar accuracies if the CE and CL are not considered, however, ANNs and SVMs performed better on calibration set than the validation set which indicate that we should take more caution with the two methods for over-fitting. Except PLS method, the performance for most methods did not enhance when the spectrum were operated by the first derivative. Moreover, the evaluation of the robustness demonstrates that SVM method may be better suited than the other methods to cope with potential confounding factors for most varieties, ecological site and growth stage; (4) The prediction accuracy was found to be higher when more wavelengths were used, though at the cost of a lower CE. The findings are of interest to the remote sensing community for the development of improved inversion schemes for hyperspectral applications concerning other types of vegetation. The examples provided in this paper may also serve to illustrate the advantages and shortcomings of empirical hyperspectral models for mapping important vegetation biophysical properties of other crops.

An Emulator Toolbox to Approximate Radiative Transfer Models with Statistical Learning

Physically-based radiative transfer models (RTMs) help in understanding the processes occurring on the Earth’s surface and their interactions with vegetation and atmosphere. When it comes to studying vegetation properties, RTMs allows us to study light interception by plant canopies and are used in the retrieval of biophysical variables through model inversion. However, advanced RTMs can take a long computational time, which makes them unfeasible in many real applications. To overcome this problem, it has been proposed to substitute RTMs through so-called emulators. Emulators are statistical models that approximate the functioning of RTMs. Emulators are advantageous in real practice because of the computational efficiency and excellent accuracy and flexibility for extrapolation. We hereby present an “Emulator toolbox” that enables analysing multi-output machine learning regression algorithms (MO-MLRAs) on their ability to approximate an RTM. The toolbox is included in the free-access ARTMO’s MATLAB suite for parameter retrieval and model inversion and currently contains both linear and non-linear MO-MLRAs, namely partial least squares regression (PLSR), kernel ridge regression (KRR) and neural networks (NN). These MO-MLRAs have been evaluated on their precision and speed to approximate the soil vegetation atmosphere transfer model SCOPE (Soil Canopy Observation, Photochemistry and Energy balance). SCOPE generates, amongst others, sun-induced chlorophyll fluorescence as the output signal. KRR and NN were evaluated as capable of reconstructing fluorescence spectra with great precision. Relative errors fell below 0.5% when trained with 500 or more samples using cross-validation and principal component analysis to alleviate the underdetermination problem. Moreover, NN reconstructed fluorescence spectra about 50-times faster and KRR about 800-times faster than SCOPE. The Emulator toolbox is foreseen to open new opportunities in the use of advanced RTMs, in which both consistent physical assumptions and data-driven machine learning algorithms live together.

Estimating winter wheat (Triticum aestivum) LAI and leaf chlorophyll content from canopy reflectance data by integrating agronomic prior knowledge with the PROSAIL model

Leaf area index (LAI) and leaf chlorophyll content (LCC) are major considerations in management decisions, agricultural planning, and policy-making. When a radiative transfer model (RTM) was used to retrieve these biophysical variables from remote-sensing data, the ill-posed problem was unavoidable. In this study, we focused on the use of agronomic prior knowledge (APK), constructing the relationship between LAI and LCC, to restrict and mitigate the ill-posed inversion results. For this purpose, the inversion results obtained using the SAILH+PROSPECT (PROSAIL) canopy reflectance model alone (no agronomic prior knowledge, NAPK) and those linked with APK were compared. The results showed that LAI inversion had high accuracy. The validation results of the root mean square error (RMSE) between measured and estimated LAI were 0.74 and 0.69 for NAPK and APK, respectively. Compared with NAPK, APK improved LCC estimation; the corresponding RMSE values of NAPK and APK were 13.36 µg cm–2 and 9.35 µg cm–2, respectively. Our analysis confirms the operational potential of PROSAIL model inversion for the retrieval of biophysical variables by integrating APK.

Multi-temporal, multi-sensor retrieval of terrestrial vegetation properties from spectral–directional radiometric data

The synergy of a time series of optical satellite observations from a variety of sensors can be exploited to improve the retrieval of biophysical variables. Information from different sensors may assist in the variable retrieval by limiting potential ambiguities. This involves observations at different spatial, spectral, temporal and angular resolutions, etc. Furthermore, using timely data is of much importance for vegetation monitoring in environmental modeling. While the other necessary variables for such models can be collected daily (e.g. meteorological variables), the temporal resolution of optical sensors (high to intermediate spatial resolutions) does not allow having temporally frequent products of vegetation characteristics due to the revisit time of the sensors and cloud coverage. A multi-temporal, multi-sensor approach applied to a temporal sequence of radiometric data acquired by different sensors can improve mapping and monitoring of vegetation state variables over time. Even when no observations are available due to cloudiness or orbital configuration, the prior retrievals are taken.The study provides a prototype proof of concept for a multi-temporal, multi-sensor approach to retrieve vegetation state variables using data collected by different imaging spectral-radiometers over time. Focus is given to the retrieval of LAI, fCover and chlorophyll content over the agricultural test site in Barrax, Spain. The approach is evaluated over a limited number of remotely sensed data acquired during a short period in the Sentinel-3 Experiment (SEN3EXP 2009) field campaign. A variety of satellite observations including CHRIS-Proba, Landsat TM and ASTER are integrated and combined by means of the multi-temporal, multi-sensor approach and through inversion of a coupled surface-atmosphere radiative transfer model. We applied the iterative Bayesian model inversion in which retrievals of the current observation are incorporated as prior information to the successor observation.This paper presents an overview of the results and challenges in utilizing the multi-temporal, multi-sensor approach and the Bayesian inversion technique in the retrieval of terrestrial vegetation properties. Overall, the accuracy obtained with data acquired by multiple sensors over time was higher than when using a single sensor. LAI and fCover were retrieved with RMSE= 0.7 (m2/m2) and 0.1 respectively (multiple sensors), while RMSE= 1.09 (m2/m2) and 0.15 respectively, when using data acquired by a single sensor.

Monitoring of Irrigation Schemes by Remote Sensing: Phenology versus Retrieval of Biophysical Variables

The appraisal of crop water requirements (CWR) is crucial for the management of water resources, especially in arid and semi-arid regions where irrigation represents the largest consumer of water, such as the Doukkala area, western Morocco. Simple and (semi) empirical approaches have been applied to estimate CWR: the first one is called Kc-NDVI method, based on the correlation between the Normalized Difference Vegetation Index (NDVI) and the crop coefficient (Kc); the second one is the analytical approach based on the direct application of the Penman-Monteith equation with reflectance-based estimates of canopy biophysical variables, such as surface albedo (r), leaf area index (LAI) and crop height (hc). A time series of high spatial resolution RapidEye (REIS), SPOT4 (HRVIR1) and Landsat 8 (OLI) images acquired during the 2012/2013 agricultural season has been used to assess the spatial and temporal variability of crop evapotranspiration ETc and biophysical variables. The validation using the dual crop coefficient approach (Kcb) showed that the satellite-based estimates of daily ETc were in good agreement with ground-based ETc, i.e., R2 = 0.75 and RMSE = 0.79 versus R2 = 0.73 and RMSE = 0.89 for the Kc-NDVI, respective of the analytical approach. The assessment of irrigation performance in terms of adequacy between water requirements and allocations showed that CWR were much larger than allocated surface water for the entire area, with this difference being small at the beginning of the growing season. Even smaller differences were observed between surface water allocations and Irrigation Water Requirements (IWR) throughout the irrigation season. Finally, surface water allocations were rather close to Net Irrigation Water Requirements (NIWR).

Evaluating the capabilities of Sentinel-2 for quantitative estimation of biophysical variables in vegetation

The red edge position (REP) in the vegetation spectral reflectance is a surrogate measure of vegetation chlorophyll content, and hence can be used to monitor the health and function of vegetation. The Multi-Spectral Instrument (MSI) aboard the future ESA Sentinel-2 (S-2) satellite will provide the opportunity for estimation of the REP at much higher spatial resolution (20m) than has been previously possible with spaceborne sensors such as Medium Resolution Imaging Spectrometer (MERIS) aboard ENVISAT. This study aims to evaluate the potential of S-2 MSI sensor for estimation of canopy chlorophyll content, leaf area index (LAI) and leaf chlorophyll concentration (LCC) using data from multiple field campaigns. Included in the assessed field campaigns are results from SEN3Exp in Barrax, Spain composed of 35 elementary sampling units (ESUs) of LCC and LAI which have been assessed for correlation with simulated MSI data using a CASI airborne imaging spectrometer. Analysis also presents results from SicilyS2EVAL, a campaign consisting of 25 ESUs in Sicily, Italy supported by a simultaneous Specim Aisa-Eagle data acquisition. In addition, these results were compared to outputs from the PROSAIL model for similar values of biophysical variables in the ESUs. The paper in turn assessed the scope of S-2 for retrieval of biophysical variables using these combined datasets through investigating the performance of the relevant Vegetation Indices (VIs) as well as presenting the novel Inverted Red-Edge Chlorophyll Index (IRECI) and Sentinel-2 Red-Edge Position (S2REP). Results indicated significant relationships between both canopy chlorophyll content and LAI for simulated MSI data using IRECI or the Normalised Difference Vegetation Index (NDVI) while S2REP and the MERIS Terrestrial Chlorophyll Index (MTCI) were found to have the strongest correlation for retrieval of LCC.

Derivation of biophysical variables from Earth observation data: validation and statistical measures

Evaluation is an essential step of model development. However, there is a missing definition of appropriate validation strategies, needed to guarantee reproducibility and generalizability of modeling results. Also, there is a lack of a generally agreed set of 'optimal' statistical measure(s) to assess model accuracy. The objective of the present study is to provide for remote sensing practitioners (i.e., non-statisticians) guidance for model validation strategies and to propose an optimal set of statistical measures for the quantitative assessment of model performance in the context of vegetation biophysical variable retrieval from Earth observation (EO) data. For these purposes, main terms and concepts were reviewed. Then, validation strategies were tested on a polynomial regression model and discussed. Moreover, a literature review was carried out, summarizing the statistical measures used to evaluate model performances. Supported by some exemplary datasets, these measures were calculated and their meanings discussed in view of several model validation criteria. From the results, we recommend to further exploit cross-validation and bootstrapping strategies to guarantee the development/validation of reliable models. An 'optimal' statistic set is suggested, including root mean square error (RMSE), coefficient of determination (R 2 ), slope and intercept of Theil-Sen regression, relative RMSE, and Nash-Sutcliffe efficiency index. A wide acceptance and use of these statistics should enable a better intercomparison of scientific results, urgently needed in times of increasing model development activities that are carried out with respect to upcoming EO missions.