Inversion Of Radiative Transfer Model Research Articles

FIREMAP: Cloud-based software to automate the estimation of wildfire-induced ecological impacts and recovery processes using remote sensing techniques

The formulation and planning of integrated fire management strategies must be strengthened by decision support systems about fire-induced ecological impacts and ecosystem recovery processes, particularly in the context of extreme wildfire events that challenge land management initiatives. Wildfire data collection and analysis through remote sensing earth observations is of utmost importance for this purpose. However, the needs of land managers are not always met because the exploitation of the full potential of remote sensing techniques requires a high level of technical expertise. In addition, data acquisition and storage, database management, networking, and computing requirements may present technical difficulties. Here, we present FIREMAP software, which leverages the potential of Google Earth Engine (GEE) cloud-based platform, an intuitive graphical user interface (GUI), and the European Forest Fire Information System (EFFIS) wildfire database for wildfire analyses through remote sensing techniques and data collections. FIREMAP software allows automatic computing of (i) machine learning-based burned area (BA) detection algorithms to facilitate the mapping of (historical) fire perimeters, (ii) fire severity spectral indices, and (iii) post-fire recovery trajectories through the inversion of physically-based radiative transfer models. We introduce (i) the FIREMAP platform architecture and the GUI, (ii) the implementation of well-established algorithms for wildfire science and management in GEE, (iii) the validation of the algorithm implementation in fifteen case-study wildfires across the western Mediterranean Basin, and (iv) the near-future and long-term planned expansion of FIREMAP features.

Hybrid inversion of radiative transfer models based on topographically corrected Landsat surface reflectance improves leaf area index and aboveground biomass retrievals of grassland on the hilly Loess Plateau

ABSTRACT Accurate monitoring of the leaf area index (LAI) and aboveground biomass (AGB) using remote sensing at a fine scale is crucial for understanding the spatial heterogeneity of vegetation structure in mountainous ecosystems. Understanding discrepancies in various retrieval strategies considering topographic effects or not is necessary to improve LAI and AGB estimations over mountainous areas. In this study, the performances of the look-up table method (LUT) using radiative transfer model (RTM), machine learning algorithms (MLAs), and hybrid RTM integrating RTM and MLAs based on Landsat surface reflectance (SR) before and after topographic correction were compared and analyzed. The results show that topographic correction improves the accuracies of retrieval methods involving RTM more significantly than the MLAs, meanwhile, it reduces the performance variability of different MLAs. Based on the topographically corrected Landsat SR, the random forest (RF) combined with RTM improves the retrieval accuracy of RTM-based LUT by 7.7% for LAI and 13.8% for AGB, and reduces the simulation error of MLA by 15.1% for LAI and 20.1% for AGB. Compared with available remote sensing products, the hybrid RTM based on Landsat SR with topographic correction has better feasibility to capture LAI and AGB variation at 30 m scale over mountainous areas.

Hybrid retrieval of grass biophysical variables based-on radiative transfer, active learning and regression methods using Sentinel-2 data in Marakele National Park

Biophysical variables such as leaf area index (LAI) and leaf chlorophyll content (LCC) are cited as essential biodiversity variables. A comprehensive comparison and integration of retrieval methods is needed for the estimation of biophysical variables such as LAI and LCC over a multispecies grass canopy. This study tested an assortment of five potentially robust, nonparametric regression methods (NPRMs) for inversion of radiative transfer model (RTM) to retrieve grass LAI and LCC in the Marakele National Park (MNP) of South Africa. The NPRMs used were, namely (i) Partial least squares regression (PLSR), (ii) Principle components regression (PCR), (iii) Kernel ridge regression (KRR), (iv) Random forest regression (RFR), and (v) K-nearest neighbours regression (KNNR). Furthermore, the study attempted to constrain the inversion process by using active learning (AL) techniques which ensured the selection of informative samples from a large pool of RTM simulations. Results show the most accurate grass LAI and LCC retrievals had lower relative root mean squared errors (RRMSEs) of 39.87% and 16.58% respectively. These findings have significant implications for the development of transferable rangeland monitoring systems in protected mountainous regions.

Pixel-level regression for UAV hyperspectral images: Deep learning-based quantitative inverse of wheat stripe rust disease index

Previous research on utilizing unmanned aerial vehicle (UAV) remote sensing imagery for plant disease detection has predominantly focused on the qualitative identification of healthy and infected plants. Notably, pixel-level regression analysis for the quantification of the wheat stripe rust disease index (DI) using hyperspectral imaging data and deep learning methods is still lacking. Traditionally, quantitative inversion has been achieved by employing radiative transfer model inversion techniques or a combination of vegetation indices and machine learning methodologies. This investigation presents an end-to-end, pixel-level quantitative regression methodology, underpinned by deep learning techniques. This methodology carries substantial importance not only for the accurate assessment of disease distribution maps, but also for an array of common quantitative regression challenges within agricultural systems. For example, the approach can be utilized for the regression inversion of continuous phenotypes, including crop yield and plant height. In this study, 1,560 local wheat varieties (lines) from Henan Province were selected as experimental subjects, resulting in a wide-ranging gradient of DI. Hyperspectral images at a height of 100 m were obtained based on UAVs at different stages of infection. This work utilized a deep learning semantic segmentation method with continuous loss functions such as Laplacian loss to achieve pixel-level regression and end-to-end quantitative inversion of the DI. The performance of models with different loss functions, model architectures and datasets was compared. The optimal results were achieved using the Laplacian + MSE loss function combined with the HRNet_W18 algorithm model, yielding a test set R2 value of 0.875 and an MSE of 0.0129. Incorporating a PSA module further improved the outcomes, resulting in an R2 value of 0.880 and an MSE of 0.0123. Modeling with a limited number of feature indices (e.g., six feature indices) reduced the model recognition performance to 0.829 compared to full-band modeling. These findings suggest that full-band, end-to-end modeling based on deep learning algorithms can lead to superior inversion outcomes and streamline data analysis steps. The insights from this research hold relevance for high-throughput crop phenotyping, plant disease and pest monitoring, and quantitative yield assessment.

Early detection of pine shoot beetle attack using vertical profile of plant traits through UAV-based hyperspectral, thermal, and lidar data fusion

Pine shoot beetle (PSB) is one of the most damaging forest insects of Yunnan pine plantations in southwest China. However, the subtle symptoms of heterogeneous tree crowns make it difficult to accurately detect at early stage of PSB attack. Here, we evaluated the potential of a combination of plant traits (PTs) and vegetation indices (VIs) to distinguish different levels of tree damage by integrating hyperspectral, thermal imagery, and light detection and ranging (lidar) data based on unmanned airborne vehicle (UAV) systems. A voxelization method was used to fuse hyperspectral reflectance, temperature, and lidar point cloud data. Subsequently, PTs such as pigments (i.e., chlorophyll, carotenoid, and anthocyanin contents) were retrieved from a radiative transfer model inversion, and structural, fluorescence, and thermal traits, as well as VIs, were derived from the fusion data. We developed a novel analytical approach using random forest (RF) algorithm with predictors from different spatial distributions (i.e., horizontal directions, vertical layers, and vertical clusters) to compare the performance of tree severity classification. The results showed that the difference of both PTs and VIs in vertical layers between different severity levels are more than that in horizontal directions. The performance of RF model with predictors of the vertical layers (OA = 74 %, kappa = 0.65) was better than that using the predictive variables in horizontal directions (OA = 69 % and kappa = 0.60) of tree crowns. Using the vertical clustering features RF model increased the accuracy (OA = 78 % and kappa = 0.70), especially for slightly and moderately damaged trees, with improvements of 10 % and 12 %, respectively. Among all variables analyzed, chlorophylls were the most important predictor, followed by photochemical reflectance index and structural traits. Our work demonstrates the effectiveness of using fused UAV-based multi-sensor data for early detection of PSB attack, and can be applied other potential forest diseases and insect monitoring.

Estimation of grain protein content in commercial bread and durum wheat fields via traits inverted by radiative transfer modelling from Sentinel-2 timeseries

Wheat (Triticum spp.) is crucial to food security. Grain protein content (GPC) is key to its nutritional and economic value and is controlled by genetic and agronomic factors, soil properties and weather. GPC prediction from remote sensing could reduce nitrogen (N) losses, help management decisions, and improve profit. However, GPC prediction is complex because multiple plant traits influence GPC and their effects change through the growing season. Traits with known physiological links to GPC, which can be retrieved from imaging spectroscopy, include leaf area index (LAI), chlorophyll (Ca+b), and stress indicators. Further inspection of these and other traits retrieved from satellite data can advance research relevant to precision agriculture. Sentinel-2 (S2) timeseries (TS) were acquired for 6,355 ha of commercial dryland bread (T. aestivum) and durum (T. durum) wheat fields in south-east Australia through two consecutive years with dissimilar rainfall. Wheat growers provided ∼ 92,000 GPC data points from harvester-mounted protein monitors. For each, Ca+b, leaf dry matter, leaf water content (Cw) and LAI were retrieved from the S2 images by radiative transfer model inversion. A gradient boosted machine learning algorithm was applied to analyse these traits’ importance to GPC and to predict GPC in 30% of samples unseen by the algorithm in training. The strongest relationships between predicted and observed GPC (R2 = 0.86, RMSE = 0.56 %), in a model built from five S2 images across a season, were better than those from single-date hyperspectral (HS). In severe water stress, LAI was the main predictor of GPC early in the season, but this switched to Cw later. Trait importance was more evenly distributed in milder conditions. S2 TS had a clear accuracy advantage over single-date S2 and HS, especially in benign conditions, emphasising the potential of S2 TS for large-scale GPC monitoring.

Insights from field phenotyping improve satellite remote sensing based in-season estimation of winter wheat growth and phenology

Timely knowledge of phenological development and crop growth is pivotal for evidence-based decision making in agriculture. We propose a near real-time approach combining radiative transfer model inversion with physiological and phenological priors from multi-year field phenotyping. Our approach allows to retrieve Green Leaf Area Index (GLAI), Canopy Chlorophyll Content (CCC) and hence Leaf Chlorophyll Content (Cab) from Sentinel-2 optical satellite imagery to quantify winter wheat growth conditions in a physiologically sound way. Phenological macro stages are based on accumulated growing degree day thresholds obtained from multi-year field phenotyping covering more than 2400 ratings from roughly 300 winter wheat varieties and reflect important physiological transitions. These include the transition from vegetative to reproductive growth and the onset of flowering, which is important information for agricultural decision support. Validation against a large data set of on-farm trials in Switzerland collected in 2019 and 2022 revealed high accuracy of our approach that produced spatio-temporally consistent results. Phenological macro stages were predicted for 970 Sentinel-2 observations reaching a weighted F1-score of 0.96. Sentinel-2 derived GLAI and CCC explained between 77 to 84% and between 79 to 84% of the variability in in-situ measurements, respectively. Here, the incorporation of phenological priors clearly increased trait retrieval accuracy. Besides, this work highlights that physiological priors, e.g., obtained by field phenotyping, can help enhancing landscape scale observations and hold potential to advance the retrieval of remotely sensed vegetation traits and in-season phenology.

PROSPECT-GPR: Exploring spectral associations among vegetation traits in wavelength selection for leaf mass per area and water contents

Leaf mass per area (LMA) and equivalent water thickness (EWT) are key indicators providing information on plant growth status and agricultural management, and their retrieval is commonly done through radiative transfer models (RTMs) such as the PROSPECT model. However, the PROSPECT model is frequently hampered by the ill-posed problem as a consequence of measurement and model uncertainties. Here, we propose a wavelength selection method to improve the inversion of EWT and LMA by integrating PROSPECT with a machine learning algorithm (Gaussian process regression (GPR); PROSPECT-GPR for short). The GPR model conducted sorting of wavelengths and the PROSPECT-D was used to determine the optimal number of characteristic wavelengths. The results demonstrated that the estimation of EWT (R2 = 0.80; RMSE = 0.0021) and LMA (R2 = 0.71; RMSE = 0.0021) using the proposed wavelengths and PROSPECT inversion all exhibited superior accuracy in comparison with those from previous studies. The efficacy of PROSPECT-GPR in exploring the spectral linkage among vegetation traits was demonstrated by selecting wavelengths associated with leaf structure parameter N and EWT (1368 nm) that turn out to contribute to the estimation of LMA. The findings lay a strong foundation for understanding the spectral linkage among vegetation traits, and the proposed wavelength selection method provides valuable insights for selecting informative spectral wavelengths for RTMs inversion and designing future remote sensors.

Tree Species Diversity Mapping—Success Stories and Possible Ways Forward

The special issue “Tree species diversity mapping” presents research focused on the remote assessment of tree species diversity, using different sensor modalities and platforms. The special issue thereby recognizes that the continued loss of biodiversity poses a great challenge to humanity. Precise and regularly updated baseline information is urgently needed, which is difficult, using field inventories, especially on a large scale. On such scales, remote sensing methods excel. The work presented in the special issue demonstrates the great potential of Earth Observation (EO) for addressing knowledge gaps, as EO provides rich (spectral) information at high revisit frequencies and spatial resolutions. Many tree species can be distinguished well using optical data, in particular, when simultaneously leveraging both the spectral and temporal dimensions. A combination with other sensor modalities can further improve performance. EO approaches are, however, limited by the availability of high-quality reference information. This complicates the task as the collection of field data is labor and time-consuming. To mitigate this limiting factor, resources should be better shared amongst the community. The reliance on in situ data also highlights the need to focus research on the extraction of more permanent (i.e., species-inherent) properties. In this respect, we identify and discuss some inherent limitations of current approaches regarding tree species discrimination. To this end, we offer a more fundamental view on tree species classification based on physical principles. To provide both a summary of the special issue and some stimulating thoughts about possible future research directions, we structured the present communication into four parts. We first introduce the need for biodiversity information, followed by a summary of all 19 articles published within the special issue. The articles are ordered by the number of species investigated. Next, we provide a short summary of the main outputs. To stimulate further research and discussion within the scientific community, we conclude this communication by offering a more fundamental view on tree species classification based on EO data and its biophysical foundations. In particular, we purport that species can possibly be more robustly identified if we classify/identify them in the biophysical feature space and not in the spectral-temporal feature space. This involves the creation and inversion of so-called physically-based radiative transfer models (RTM), which take hyper/multispectral observations together with their observation geometry (as well as other priors), and project these into biophysical variables such as chlorophyll content and LAI etc. The perceived advantage of such an approach is that the generalizability (and scalability) of EO based classifications will increase, as the temporal trajectory of species in the biophysical parameter space is probably more robust compared to the sole analysis of spectral data, which—amongst other perturbing factors—also depend on site/time specific illumination geometry.

High resolution retrieval of leaf chlorophyll content over Himalayan pine forest using Visible/IR sensors mounted on UAV and radiative transfer model

Forests play an essential role towards net primary productivity, biological cycles and provide habitat to flora & fauna. To monitor key physiological activities in forest canopies such as photosynthesis, respiration, transpiration, spatially-explicit and precise information of the biochemical (biological) variables such as Leaf Chlorophyll Content (LCC) is required. While lookup-table (LUT)-based Radiative Transfer Model (RTM) inversion against optical remote sensing imagery is regarded as a physically sound and robust approach for retrieving biochemical and biophysical variables, regularization procedures are required to offset the problem of ill-posedness. To optimize the RTM inversion of LCC over a sub-tropical pine forest plantation, in the Western Himalaya, we investigated the role of: (1) cost functions (CFs), (2) added noise, and (3) multiple finest solutions in LUT inversion. Principal CFs were evaluated belonging to three categories: information measures, M-estimates, and minimal contrast approaches. The inversion approaches were applied to a LUT produced by the coupled leaf-canopy model known as PROSAIL RTM and tested in contrast field spectral data obtained from reflectance data derived from UAV (Unmanned Aerial Vehicle) images taken over the canopies of covered pine forests. The Bhattacharyya divergence, an information measure, outperformed all other CFs in LCC inversion, with R2 of 0.94, RMSE of 6.20 μg/cm2 and NRMSE of 12.27% during the validation. The optimized inversion strategy was subsequently applied to a UAV-acquired multispectral image at an 8.2 cm pixel resolution for detailed landscape forest LCC mapping. The associated residuals as provided by the LUT-based inversion provided insights in the spatial consistency of the LCC map.

Towards uninterrupted canopy-trait time-series: A Bayesian radiative transfer model inversion using multi-sourced satellite observations

Canopy traits play an important role in plant growth, carbon sequestration, and the land surface water and energy balance, and thus it is important to understand and monitor how these traits change over time. Canopy radiative transfer models, which use canopy traits to predict spectra, provide a powerful means to understand canopy trait variation and these models can be inverted, allowing us to use remotely sensed data to monitor canopy traits. However, a single inversion represents a snapshot in time for a single sensor. In this study, we used a Bayesian approach to assimilate snow-free observations from three different sensors (Landsat 8, Sentinel-2A and MODIS) over the course of 2019 by inverting the ProSAIL radiative transfer model, producing a robust estimate of six canopy traits and their joint uncertainties at each time point. Next, we used a state-space time series model to integrate observations from multiple sensors to capture variation in vegetation optical traits throughout the year. The results demonstrate that our model can capture seasonal variation of surface reflectance with less than 5% RMSE compared with observed reflectance. Moreover, linking observations in time narrows the uncertainty intervals compared to the posterior distributions for individual time points. We are also able to make statistical inferences on dates when no data is available, and provide reasonable and reliable time-series inversion results compared with LAI ground measurements.

Airborne hyperspectral imaging of cover crops through radiative transfer process-guided machine learning

Cover cropping between cash crop growing seasons is a multifunctional conservation practice. Timely and accurate monitoring of cover crop traits, notably aboveground biomass and nutrient content, is beneficial to agricultural stakeholders to improve management and understand outcomes. Currently, there is a scarcity of spatially and temporally resolved information for assessing cover crop growth. Remote sensing has a high potential to fill this need, but conventional empirical regression operated with coarse-resolution multispectral data has large uncertainties. Therefore, this study utilized airborne hyperspectral imaging techniques and developed new process-guided machine learning approaches (PGML) for cover crop monitoring. Specifically, we deployed an airborne hyperspectral system covering visible to shortwave-infrared wavelengths (400–2400 nm) to acquire high spatial (0.5 m) and spectral (3–5 nm) resolution reflectance over 23 cover crop fields across Central Illinois in March and April of 2021. Airborne hyperspectral surface reflectance with high spectral and spatial resolution can be well matched with field data to quantify cover crop traits. Furthermore, the PGML models were pre-trained by synthetic data from soil-vegetation radiative transfer modeling (one million records), and then fine-tuned with field data of cover crop biomass and nutrient content. Results show that airborne hyperspectral data with PGML can achieve high accuracy to predict cover crop aboveground biomass (R2 = 0.72, relative RMSE = 15.16%) and nitrogen content (R2 = 0.69, relative RMSE = 16.59%) through leave-one-field-out cross-validation. Unlike the pure data-driven approach (e.g., partial least-squares regression), PGML incorporated radiative transfer knowledge and obtained higher predictive performance with fewer field data. Meanwhile, with field data for model fine-tuning, PGML predicted biomass more accurately than the inversion of radiative transfer models. We also found that the red edge has a high contribution in quantifying aboveground biomass and nitrogen content, followed by green and shortwave spectra. This study demonstrated the first attempt of utilizing hyperspectral remote sensing to accurately quantify cover crop traits. We highlight the strength of PGML in exploiting sensing data to quantify ecosystem variables to advance agroecosystem monitoring for sustainable agricultural management.

Retrieval of crop biophysical-biochemical variables from airborne AVIRIS-NG data using hybrid inversion of PROSAIL-D

Sustainable agricultural growth and management reduces over-utilization of farm resources and squeezes risk of negative impacts on environment. Monitoring continuous crop growth and health under various conditions at different spatio-temporal resolutions is a key to assess yield stability, crop diversity, adaptability, mitigation for stress and response. The quantification of crop biophysical and biochemical variables spatially from remotely sensed data with help of spectroscopic methods provide a reliable discerning information in the context of crop foliar condition like leaf greenness, senescence, canopy density, crop growth, stress and eco-physiological processes. Airborne Visible/Infrared Imaging Spectrometer-Next Generation (AVIRIS-NG) airborne hyperspectral data offers high spatial and spectral resolution giving a unique advantage and opportunity to test retrievals of crop biophysical (BP)-biochemical (BC) variables under varying conditions over different types of crops. A hybrid inversion of leaf-canopy Radiative Transfer model – PROSAIL-D complemented with use of data driven nonlinear non-parametric methods which offer simplicity, fastness, reliability and competency is a powerful method to retrieve crop biophysical and biochemical variables. The Hyperspectral band (feature) selection is a computationally cost efficient method to overcome data redundancy in high dimensional correlated input spectral bands. The determination of optimum subset or combination of hyperspectral bands specific to retrieval of vegetation properties (including canopy effects) determined using feature selection algorithm for regression-based retrievals are active research topic unlike classification problems in which it is more common. A band selection algorithm based on Gaussian Processes Regression was used to choose most sensitive bands from in-situ biophysical-biochemical measurements and crop spectral signatures collected for analysis in two different agricultural regions: Raichur (Karnataka) and Anand (Gujarat) districts of India representing diverse landscapes, heterogeneous crop canopies and agro climatic settings. The retrieval algorithm for AVIRIS-NG image employed a decision tree ensemble algorithm Canonical Correlation Forests using the optimum subset of bands for retrieval of targeted crop variable. Validation of retrieved crop variables were done using in-situ ground observations collected over heterogeneous diverse crop landscape. The results showed chlorophyll (RMSE = 6.61 µg cm−2), equivalent water thickness (RMSE = 0.002 cm), leaf area index (RMSE = 0.35 m2/m−2) and dry matter (RMSE = 0.003 g cm−2), carotenoid (RMSE = 14.3 µg cm−2), anthocyanin (RMSE = 12.92 µg cm−2) retrieved with better accuracies. The results obtained in context of feature (band) selection approach for Radiative Transfer model inversion show overlapping of sensitive spectral bands of chlorophyll-ab, carotenoid and anthocyanin especially between narrow spectral range 480 to 560 nm as predominant reason for decreased accuracies of anthocyanin and carotenoid compared to other variable. This poses a limitation as well as opportunity for further research in signal separation in context of feature selection approach especially in context of broader spectral bandwidths. It also sets ground for further development of feature selection algorithms that use hybrid regression methods customized for crop specific traits.

RTM Inversion through Predictive Equations for Multi-Crop LAI Retrieval Using Sentinel-2 Images

Near-real-time, high-spatial-resolution leaf area index (LAI) maps would enable producers to monitor crop health and growth status, improving agricultural practices such as fertiliser and water management. LAI retrieval methods are numerous and can be divided into statistical and physically based methods. While statistical methods are generally subject to high site-specificity but possess high ease of implementation and use, physically based methods are more transferable, albeit more complex to use in operational settings. In addition, statistical methods need a large amount of data for calibration and subsequent validation, and this is only seldom feasible. Techniques based on predictive equations (PEphysical) represent a viable alternative, allowing the partial combination of statistical and physical methods merits while minimising their shortcomings. In this paper, predictive equation-based techniques were compared with four other methods: two radiative transfer model (RTM) inversion methods, one based on neural network (NNET) and one based on a look-up table (LUT), and two empirical methods (one using empirical models based on vegetation indices and in situ data and one based on empirical models found in the scientific literature). The methods were chosen based on common use. To evaluate the performance of the studied methods, the coefficient of determination (R2), root mean square error (RMSE), and normalised root mean square error (nRMSE, %) between the estimates and in situ LAI measurements were reported. The best PEphysical results, achieved by the OSAVI index (RMSE = 0.84 m2 m−2), provided better performance for LAI recovery than the NNET-based RTM inversions (0.86 m2 m−2) or the estimates made by LUT (0.94 m2 m−2). Furthermore, the best PEphysical produced accuracies comparable to the best empirical model (RMSE = 0.71 m2 m−2), calibrated through in situ data, and similar to the best literature model (RMSE = 0.76 m2 m−2). These results indicated that PEphysical can be used to recover LAI with transferability comparable to literature models.

Comparison of Physical-Based Models to Measure Forest Resilience to Fire as a Function of Burn Severity

We aimed to compare the potential of physical-based models (radiative transfer and pixel unmixing models) for evaluating the short-term resilience to fire of several shrubland communities as a function of their regenerative strategy and burn severity. The study site was located within the perimeter of a wildfire that occurred in summer 2017 in the northwestern Iberian Peninsula. A pre- and post-fire time series of Sentinel-2 satellite imagery was acquired to estimate fractional vegetation cover (FVC) from the (i) PROSAIL-D radiative transfer model inversion using the random forest algorithm, and (ii) multiple endmember spectral mixture analysis (MESMA). The FVC retrieval was validated throughout the time series by means of field data stratified by plant community type (i.e., regenerative strategy). The inversion of PROSAIL-D featured the highest overall fit for the entire time series (R2 > 0.75), followed by MESMA (R2 > 0.64). We estimated the resilience of shrubland communities in terms of FVC recovery using an impact-normalized resilience index and a linear model. High burn severity negatively influenced the short-term resilience of shrublands dominated by facultative seeder species. In contrast, shrublands dominated by resprouters reached pre-fire FVC values regardless of burn severity.

Challenging the link between functional and spectral diversity with radiative transfer modeling and data

In a context of accelerated human-induced biodiversity loss, remote sensing (RS) is emerging as a promising tool to map plant biodiversity from space. Proposed approaches often rely on the Spectral Variation Hypothesis (SVH), linking the heterogeneity of terrestrial vegetation to the variability of the spectroradiometric signals. Yet, due to observational limitations, the SVH has been insufficiently tested, remaining unclear which metrics, methods, and sensors could provide the most reliable estimates of plant biodiversity. Here we assessed the potential of RS to infer plant biodiversity using radiative transfer simulations and inversion. We focused specifically on “functional diversity,” which represents the spatial variability in plant functional traits. First, we simulated vegetation communities and evaluated the information content of different functional diversity metrics (FDMs) derived from their optical reflectance factors (R) or the corresponding vegetation “optical traits,” estimated via radiative transfer model inversion. Second, we assessed the effect of the spatial resolution, the spectral characteristics of the sensor, and signal noise on the relationships between FDMs derived from field and remote sensing datasets. Finally, we evaluated the plausibility of the simulations using Sentinel-2 (multispectral, 10 m pixel) and DESIS (hyperspectral, 30 m pixel) imagery acquired over sites of the Functional Significance of Forest Biodiversity in Europe (FunDivEUROPE) network. We demonstrate that functional diversity can be inferred both by reflectance and optical traits. However, not all the FDMs tested were suited for assessing plant functional diversity from RS. Rao's Q index, functional dispersion, and functional richness were the best-performing metrics. Furthermore, we demonstrated that spatial resolution is the most limiting RS feature. In agreement with simulations, Sentinel-2 imagery provided better estimates of plant diversity than DESIS, despite the coarser spectral resolution. However, Sentinel-2 offered inaccurate results at DESIS spatial resolution. Overall, our results identify the strengths and weaknesses of optical RS to monitor plant functional diversity. Future missions and biodiversity products should consider and benefit from the identified potentials and limitations of the SVH.

Error and uncertainty characterization of soil moisture and VOD retrievals obtained from L-band SMAP radiometer

L-band passive microwave remote sensing has evolved over the past decade to estimate soil moisture (SM) and Vegetation Optical Depth (VOD). Novel Radiative Transfer Model (RTM) schemes and model parameterizations are proposed to achieve this goal. In this work, we attempt to characterize errors and uncertainties that propagate from RTMs and their parameters while retrieving SM and VOD from Soil Moisture Active Passive (SMAP) data. Three RTMs are considered, including a zeroth order model (τ-ω model) and two first order models (First order RTM and 2-Stream Emission model (2S-EM)). Surface roughness (h; characterizes undulations on soil surface) and single scattering albedo (ω; accounts for the scattering of emissions due to vegetation structure) are the two parameters (of RTMs) considered. SM and VOD are retrieved concurrently using a multi-temporal RTM inversion scheme. Errors and uncertainty contributions are determined using the Analysis of Variance (ANOVA) approach. To assess the role of land cover conditions on error and uncertainties in retrievals, ten reference sites are chosen to represent various biomes.Initially, brightness temperatures are simulated by varying RTMs and their parameters to determine their sensitivity. The SM and VOD retrievals are compared with reference datasets, and the performance is found to be acceptable. Error decomposition analysis indicates RTMs and their parameters (h and ω) induce noticeable and significant error in SM and VOD retrievals. Errors contributions due to the above factors are more prominent in VOD retrievals compared to SM. However, the uncertainty contributions indicate minimal influence of RTMs and their parameters on SM retrievals. The ω parameter followed by the choice of RTM have significant contributions to the uncertainty in the VOD retrievals. The h parameter resulted in significant uncertainties in VOD retrievals under sparsely vegetated regions. The results are used to comment on the suitability of each of the three RTMs and the model parameterization design that could alleviate the issue of errors and uncertainties in concurrent retrievals of SM and VOD. These inferences shall contribute towards developing robust multi-parameter retrieval algorithms.

Effective GAI is best estimated from reflectance observations as compared to GAI and LAI: Demonstration for wheat and maize crops based on 3D radiative transfer simulations

The definition of LAI (Leaf Area Index) is important when deriving it from reflectance observation for model application and validation. Canopy reflectance and the corresponding quantities of LAI, PAI (Plant Area Index), GAI (Green Area Index) and effective GAI (GAI eff ) are first calculated using a 3D radiative transfer model (RTM) applied to 3D wheat and maize architecture models. A range of phenological stages, leaf optical properties, soil reflectance, canopy structure and sun directions is considered. Several retrieval methods are compared, including vegetation indices (VIs) combined with a semi-empirical model, and 1D or 3D RTM combined with a machine learning inversion approach. Results show that GAI eff is best estimated from remote sensing observations. The RTM inversion using a 3D model provides more accurate GAI eff estimates compared with VIs and the 1D PROSAIL model with RMSE = 0.33 for wheat and RMSE= 0.43 for maize. GAI eff offers the advantage to be easily accessible from ground measurements at the decametric resolution. It was therefore concluded that the most efficient retrieval approach would be to use machine learning algorithms trained over paired GAI eff and the corresponding canopy reflectance derived either from realistic 3D canopy models or from experimental measurements. • Wheat and maize canopy reflectance are simulated with realistic 3D model. • Effective GAI is best estimated from remote sensing observations. • 3D model provides the best estimation of effective GAI compared to 1D model and Vis.

A machine learning method trained by radiative transfer model inversion for generating seven global land and atmospheric estimates from VIIRS top-of-atmosphere observations

The Visible Infrared Imaging Radiometer Suite (VIIRS) has observed the Earth since 2011 and will continue for several decades. Unfortunately, few high-level land surface products have been produced and publicly released. The numerical inversion of radiative transfer model (RTM) has long been used for the retrieval of land surface and atmospheric variables from satellite data; however, it is computationally intensive. Some methods (e.g., using look-up tables) are efficient but are most suitable for estimating individual variable. Various machine learning (ML) models have been developed, mostly trained by either ground measurements that may not be adequate to represent different conditions, or RTM simulations that may not be realistic due to model uncertainty and unnecessary combinations of illumination-viewing geometries and variable values. In this study, we replaced the numerical inversion scheme by the ML algorithm that is trained using RTM inversion for estimating multiple variables simultaneously. Our training dataset for the ML models was generated by inverting the land and atmospheric variables from VIIRS top-of-atmosphere (TOA) reflectance data using a coupled land-surface–atmosphere RT model. Two multi-output ML algorithms were explored: backpropagation neural networks and random forest (RF) regression. The best-performing RF model was then applied to estimate seven land and atmospheric variables globally from VIIRS TOA data: the leaf area index (LAI), incident photosynthetically active radiation (PAR), fraction of absorbed photosynthetically active radiation (FAPAR), incident shortwave radiation (ISR), land surface albedo, land surface reflectance, and TOA albedo. The inversion results were validated using ground measurements at 54 sites and showed comparable accuracy to the numerical inversion scheme that is suitable for only small regions. An experiment was also conducted to test this method on a much larger region for detailed evaluation. Finally, seven global estimates in 2013 were experimentally produced. Inter-comparison with the existing satellite products showed that the VIIRS estimates had high consistency with the LAI and FAPAR products of the Global LAnd Surface Satellite (GLASS) products suite. Moreover, the VIIRS officially released the VNP43MA3 albedo and VNP09A1 surface reflectance products, the retrieved estimates also matched them well. It was also shown that simultaneous estimation of all seven variables was more accurate and efficient than estimating individual variable sequentially.

Multi-band bottom index: A novel approach for coastal environmental monitoring using hyperspectral data

Hyperspectral remote sensing has potential for improving shallow-water environmental monitoring, including benthic cover mapping and estimation of bathymetry and water optical properties. Detailed spectral information provided by hyperspectral sensors is particularly required to discriminate benthic covers that have similar spectral patterns (e.g., coral and algae), which cannot be achieved by multispectral sensors with a limited number of wavebands. An important process in such applications is separation of the bottom reflection from the absorption and scattering caused by the water column. Recently, inversion of a semi-analytical radiative transfer model using airborne hyperspectral data has become popular for simultaneous estimation of benthic cover, bathymetry, and water optical properties. However, it requires non-trivial settings and ancillary information, and generally is not user friendly. A cost-effective water-column correction method based on two-band combination has also been popular for multi-band remote sensing; however, it is not suited to hyperspectral data, because of the combinatorial proliferation of many hyperspectral wave bands. To fill the gap between the two approaches, in this paper, a new cost-effective method for water-column correction optimal for hyperspectral data is proposed. The method, called the multi-band bottom index (MBI), incorporates empirical radiative transfer theory and a traditional spectral enhancement technique (i.e., logarithmic residual method). A case study was conducted for the coral reef habitats around Ishigaki Island, using free hyperspectral satellite imagery (Hyperion). The result shows that our MBI approach has more robust performance in the estimation of the water attenuation coefficient than the existing water-column correction approach. The derived MBI map successfully mitigates water-column effects and enhances the spectral pattern of benthic materials. Comparison with in-situ spectra in the literature revealed that the overall spectral pattern of the MBI is reasonable for the benthic categories of coral, seagrass, and sandy bottom. The MBI approach is also promising for application to recently launched hyperspectral satellite sensors, such as HISUI, DESIS, and PRISMA.