Thermal Infrared Domain Research Articles

Learning Unsupervised Cross-Domain Model for TIR Target Tracking

The limited availability of thermal infrared (TIR) training samples leads to suboptimal target representation by convolutional feature extraction networks, which adversely impacts the accuracy of TIR target tracking methods. To address this issue, we propose an unsupervised cross-domain model (UCDT) for TIR tracking. Our approach leverages labeled training samples from the RGB domain (source domain) to train a general feature extraction network. We then employ a cross-domain model to adapt this network for effective target feature extraction in the TIR domain (target domain). This cross-domain strategy addresses the challenge of limited TIR training samples effectively. Additionally, we utilize an unsupervised learning technique to generate pseudo-labels for unlabeled training samples in the source domain, which helps overcome the limitations imposed by the scarcity of annotated training data. Extensive experiments demonstrate that our UCDT tracking method outperforms existing tracking approaches on the PTB-TIR and LSOTB-TIR benchmarks.

Analysis of the Spectral Contrast between Fossil-bearing and Fossil-depleted Limestone from the Kachchh Region, NW India Using Reflectance and Emittance Spectroscopy Data

ABSTRACT We analyzed the reflectance and emittance spectra of representative samples of mega and microfossil-bearing limestones and identified the mineralogical controls, which may have led to the preservation of mega and microfossils in the limestone within the spectral domain of 400-2500 nm and 8000-14000 nm. In this regard, we collected and analyzed the reflectance and emittance spectra of fossil-bearing and fossil-depleted limestone samples using respective reflectance and emittance spectra of their dominant constituent minerals. Reflectance spectra of mega and microfossil-bearing limestones have prominent absorption features diagnostic to iron and clay minerals, while emittance spectra of these rocks confirm the presence of silica in the fossil-bearing limestone samples. Spectral features of iron minerals are identified based on the presence of absorption minima at 480 nm and 900 nm, while clay minerals have features at 2200 nm. Presence of silica was identified with emissivity minima at 9000 nm. Laboratory spectral observations are substantiated by field observation and mineralogical studies. Fossil-depleted limestone samples, on the other hand, are characterized by the absence of iron oxide, clay and silica. Spectral contrast between fossil-bearing and the fossil-depleted limestone have allowed us to infer certain mineralogical controls essential for fossil preservation and the role of iron, clay minerals in preserving fossils is discussed in detail. Further, spectral analysis in visible, near infrared, short wave infrared and thermal infrared electromagnetic domain may provide rapid and non-destructive mineralogical assessment of fossil-bearing and fossil-depleted limestone.

Preparation and Application of Self-adaptive Porous Polyethylene for Mulch Film

Porous polyethylene has emerged as a consequential constituent within the realm of infrared thermal management, owing to its commendable transmittance in the thermal infrared domain and noteworthy high scattering capabilities within the visible spectrum. In this study, we have fabricated porous polyethylene and conducted a comprehensive investigation into its adaptable emissivity alterations within the infrared range, while concurrently upholding a pronounced scattering proficiency in the visible band. Furthermore, we have proffered the prospective application of the prepared adaptive porous polyethylene in the context of mulch film. Empirical has unequivocally demonstrated that the utilization of adaptive porous polyethylene can effectively engender adaptive cooling and insulation modulation in the realm of subterranean plant growth, by harnessing its efficient performance across both the visible and thermal infrared spectra. Simultaneously, the material continually scatters visible light, thereby providing supplemental illumination to plant leaves and fruits in a retrograde manner. This breakthrough affords an efficacious and cost-effective energy-conserving apparatus in the domain of agricultural production.

Quantifying Thermal Infra-Red directional anisotropy using Master and Landsat-8 simultaneous acquisitions

Satellite observations in the Thermal Infra-Red (TIR) domain provide valuable information on Land Surface Temperatures, Evapo-Transpiration and water use efficiency and are useful for monitoring vegetation health, agricultural practices and urban planning. By 2030, there will be 3 new high-resolution global coverage satellite TIR missions in space, all of them with fields of view larger than ± 30°. Directional anisotropy in TIR can affect the estimation of key application variables, such as temperature, and are typically studied by means of field campaigns or physical modelling. In this work, we have evaluated directional effects using simultaneous measurements from Landsat-8 and the ± 45°field of view MASTER airborne TIR sensor from NASA. Differences as high as 6 K are observed in the surface temperatures derived from these simultaneous observations. Those differences are attributed to directional effects, with the greatest differences associated with hotspot conditions, where the solar and satellite viewing directions align. Five well studied parametric directional models have then been fitted to the temperature differences, allowing the amplitude of the measured directional effects to be reduced below 1 K, with small variations between models. These results suggest that a simple correction for directional effects could be implemented as part of the ground segment processing for the upcoming missions.

A thermal radiation directionality correction method for the surface upward longwave radiation of geostationary satellite based on a time-evolving kernel-driven model

Thermal radiation directionality (TRD) characterizes the anisotropic signature of most surface targets in the thermal infrared domain. It causes significant uncertainties in estimating surface upward longwave radiation (SULR) from space observations. In this regard, kernel-driven models (KDMs) are suitable to remove TRD effects from remote sensing dataset as they are computationally efficient. However, KDMs requires simultaneous multi-angle observations as inputs to be well calibrated, which yields a difficulty with geostationary satellites as they can only provide a single-angle observation. To overcome this issue, we proposed a six-parameter time-evolving KDM that combines a four-parameter SULR diurnal variation model and a two-parameter TRD amplitude model to correct the TRD effect for single-angle estimated SULR dataset of geostationary satellites. The significant daytime TRD effect when solar zenith angle is within 60° can be effectively eliminated. The modeling accuracy of the time-evolving KDM is evaluated using a simulated SULR dataset generated by the 3D Discrete Anisotropic Radiative Transfer (DART) model; the TRD correction method based on the new time-evolving KDM is validated using a two-year single-angle estimated SULR dataset derived from data of the Advanced Baseline Imager (ABI) onboard Geostationary Operational Environmental Satellite-16 (GOES-16) against in situ measurements at 20 AmeriFlux sites. Results show that the proposed time-evolving KDM has a high accuracy with an R2 > 0.999 and a small RMSE = 1.5 W/m2; the TRD correction method based on the time-evolving KDM can greatly reduce the GOES-16 SULR uncertainty caused by the TRD effect with an RMSE decrease of 4.5 W/m2 (22.1%) and mean bias error decrease of 7.9 W/m2 (62.7%). Hence, the proposed TRD correction method is practically efficient for the operational TRD correction of SULR products generated from the geostationary satellites (e.g., GOES-16, FY-4A, Himawari-8, MSG).

Efficient thermal infrared tracking with cross-modal compress distillation

The key issue of thermal infrared tracking is to use neural networks to represent the target effectively and efficiently in the thermal infrared domain. The lack of thermal infrared trainable datasets makes it difficult to train a robust infrared object tracker from scratch, and the time-consuming convolution operations also make the tracking slow. To address the above problems, we proposed cross-modal compression distillation to represent thermal infrared objects for tracking, by leveraging an off-the-shelf RGB model with knowledge distillation. Specifically, cross-modal distillation is performed to effectively transfer knowledge from RGB modality to thermal infrared modality by inputting paired RGB and thermal infrared images into two branches of a Siamese network. Additionally, based on the teacher–student model architecture, the feature extractor is compressed into a lightweight model by model pruning and multi-level deep feature matching. Experimental results on LSOTB-TIR and PTB-TIR datasets show that the thermal infrared object tracking models distilled by our proposed method achieved faster tracking speed with better performance than the baseline RGB tracker by gaining an improvement of 1.5% Success Rate, 2.2% Precision, and 1.9% Normalized Precision, 58 frames per second (FPS) on LSOTB-TIR dataset, respectively.

Exploring Deep Learning Ear Recognition in Thermal Images

Ear biometrics is an appealing choice for human recognition in various daily activities, ranging from surveillance to commercial-related applications. It has multiple advantages in identifying non-cooperative individuals in visible unconstrained environments. Moreover, Infrared imagery is an alternative sensing modality for ear recognition systems in the dark or when there is no control over illumination. In this work, we investigate the performance of deep learning models for ear recognition in the visible and thermal infrared domains. We examine multiple convolutional neural network architectures and develop different strategies to enhance learning. We use a dual-band dataset for our experiments, recently collected for multi-pose (full frontal to full profile) face recognition applications. We also use other ear datasets for learning, the AWE visible Ear Dataset & the WVU MIWR Profile Face Dataset, for visible domain and thermal domain learning, respectively. Our experiments achieve a 98.76% Rank-1 identification accuracy, 0.7596% Equal Error Rate (EER) verification for the visible domain ear recognition, and 96.93% Rank-1 identification accuracy, 2.4541% EER verification for the thermal domain ear recognition.

Optimizing the Protocol of Near-Surface Remote Sensing Experiments Over Heterogeneous Canopy Using DART Simulated Images

Optical canopy models that connect land surface properties and satellite-observed radiance must be validated before being used. These models include the bidirectional reflectance distribution function (BRDF) models in the visible and near-infrared domains, and directional brightness temperature (DBT) models in the thermal infrared domain. Near-surface experiments have been extensively conducted to evaluate the modeling accuracy, including ground-, tower-, and aircraft-based measurements. Indeed, it should be noted that in situ measured BRDF/DBT results are sensitive to the experiment protocol, such as sensor moving orientation, flight height, and sampling frequency. A practical tool for optimizing the in situ measurement protocols is needed in the community of remote sensing modeling. For that, we devised a virtual experiment framework based on the discrete anisotropic radiative transfer (DART) 3-D radiative transfer model that is capable of simultaneously simulating both the BRDF/DBT pattern and the images acquired by in situ cameras. Here, as an optimization case, we use it to determine the optimal sensor flight orientation over heterogeneous vegetated canopies (a row-planted scene with three solar angles and a discrete scene with three solar angles) for measuring their DBT distribution. Results showed considerable errors (i.e., image-extracted DBT minus DART-simulated DBT) exist for sensor flight orientation along the canopy rows ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R^{2}$ </tex-math></inline-formula> = 0.24 and root mean square error (RMSE) = 4.32 K), and they become much smaller ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R^{2}$ </tex-math></inline-formula> = 0.94 ~ 0.98 and RMSE = 0.82 ~ 1.03 K) in other typical orientations (e.g., cross row plane, solar principal plane, and cross solar principal plane). The critical azimuth offset relative to the row direction that can ensure an acceptable RMSE < 1 K is quantified as atan(3*Unitwidth/Scenesize) based on a series of intensive simulations by this new tool. However, the RMSE of the discrete scene is not sensitive to the flight orientation. Such accuracy differences in various protocols were experimentally verified over row-planted maize using a 4-D tower in Huailai, Hebei, China. The result highlights the great potential of this newly designed DART-based virtual experiment to optimize near-surface experiment protocols.

Light Transport Induced Domain Adaptation for Semantic Segmentation in Thermal Infrared Urban Scenes

Semantic segmentation in urban scenes is widely used in applications of intelligent transportation systems (ITS). In urban scenes, thermal infrared (TIR) images can be captured in weak illumination conditions or in the presence of obscuration (e.g., light fog, smoke). Therefore, TIR images have great potential to endow automated intelligent vehicles or assist navigation systems. However, TIR imaging is blurry and low-contrast due to the absorption by atmospheric gases and heat transfer effect. Hence, TIR semantic segmentation in urban scenes has rarely been explored even though it has a wide range of scenarios in ITS. To overcome this limitation, we analyze the light transport of TIR light. Our analysis reveals that contours are the reliable features shared by TIR and Visible Spectrum (VS) light. Inspired by this, we attempt to transfer joint features from VS domain to TIR domain. Thus, we propose a curriculum domain adaptation method to guide the TIR urban scene semantic segmentation task from VS domain through contours. Moreover, to evaluate the proposed model, we build TIR-SS: an open-for-request dataset consisting of TIR images and pixel level annotations of 8 classes in urban scenes. Qualitative and quantitative experimental results on the dataset indicate that the proposed domain adaptation method outperforms related methods on this TIR semantic segmentation task.

Thermal remote sensing for plant ecology from leaf to globe

Abstract Surface temperatures are mechanistically linked to vegetation biophysical and physiological processes. Although remote sensing in the thermal infrared (TIR) domain can offer novel insights into the impacts of changing surface temperatures on vegetation, the transformative potential of remote sensing for plant ecology has not yet been realized. Remotely sensed surface temperatures can be used to derive stomatal behaviour and identify stressful environmental conditions in near‐real time. Plant species, traits and structural characteristics can be evaluated with high spectral resolution TIR emissivity. Beyond canopy scales, thermal remote sensing can enhance the inferences obtained from manipulative experiments and empirical evidence, providing unique insight into shifts in species ranges and phenology with changing climate conditions. Scaling leaf traits, canopy structure and regional patterns require an integrated understanding of both process and technology. Theory linking surface temperatures to vegetation dynamics is summarized from an energy balance perspective. We outline scaling considerations including the impacts of morphology on leaf energy balance, canopy structure influences on convective heat exchange and potential confounding impacts of non‐vegetated surfaces. Synthesis. We introduce a unifying framework to link leaf to globe through thermal remote sensing. Recent and emerging advances in sensors, data availability and analytics, together with synergies between TIR remote sensing and other data sources, present a timely opportunity for ecologists to advance our understanding of plant physiology, ecology and biogeography with thermal remote sensing.

GFSNet: Generalization-friendly siamese network for thermal infrared object tracking

The lack of large labeled training datasets hinders the usage of deep neural network for Thermal Infrared (TIR) tracking. Regular practice is to train a tracking network with large-scale RGB datasets and then retrain it to the TIR domain with limited TIR data. However, we observe that existing Siamese-based trackers can hardly generalize to TIR images though they achieve outstanding performance on RGB tracking. Therefore, the main challenge is the generalization problem: How to design a generalization-friendly Siamese tracking network and what affects the network generalization. To tackle this problem, we introduce the self-adaption structure into Siamese network and propose an effective TIR tracking model, GFSNet. GFSNet is successfully generalized to different TIR tracking tasks, including ground target, aircraft and high-diversity object tracking tasks. To estimate generalization ability, we present a notion of Growth Rate, the improvement of overall performance after retraining. Experimental results show that the Growth Rates of GFSNet exceed state-of-the-art SiamRPN++ by more than 7 times, which indicates the great power of GFSNet in generalization. In addition to experimental validations, we provide the theoretical analysis of network generalization from a novel perspective, model sensitivity. By performing some tests to analyze the sensitivity, we conclude that the self-adaption structure helps GFSNet converge to a more sensitive minimum with better generalization to new tasks. Furthermore, when compared with popular tracking methods, GFSNet maintains comparable accuracy while achieving real-time tracking with the speed of 112 FPS, 5 times faster than other TIR trackers.

Analysis of Multispectral Drought Indices in Central Tunisia

Surface water stress remote sensing indices can be very helpful to monitor the impact of drought on agro-ecosystems, and serve as early warning indicators to avoid further damages to the crop productivity. In this study, we compare indices from three different spectral domains: the plant water use derived from evapotranspiration retrieved using data from the thermal infrared domain, the root zone soil moisture at low resolution derived from the microwave domain using the Soil Water Index (SWI), and the active vegetation fraction cover deduced from the Normalized Difference Vegetation Index (NDVI) time series. The thermal stress index is computed from a dual-source model Soil Plant Atmosphere and Remote Evapotranspiration (SPARSE) that relies on meteorological variables and remote sensing data. In order to extend in time the available meteorological series, we compare the use of a statistical downscaling method applied to reanalysis data with the use of the unprocessed reanalysis data. Our study shows that thermal indices show comparable performance overall compared to the SWI at better resolution. However, thermal indices are more sensitive for a drought period and tend to react quickly to water stress.

The Monitoring Nitrous Oxide Sources (MIN2OS) satellite project

The Monitoring Nitrous Oxide Sources (MIN2OS) satellite project aims at monitoring global-scale nitrous oxide (N2O) sources by retrieving N2O surface fluxes from the inversion of space-borne N2O measurements that are sensitive to the lowermost atmospheric layers under favorable conditions. MIN2OS will provide emission estimates of N2O at a horizontal resolution of 1° × 1° on the global scale and 10 × 10 km2 on the regional scale on a weekly to monthly basis depending on the application (e.g., agriculture, national inventories, policy, scientific research). Our novel approach is based on the development of: 1) a space-borne instrument operating in the Thermal InfraRed domain providing, in clear sky conditions, N2O mixing ratio in the lowermost atmosphere (900 hPa) under favorable conditions (summer daytime) over land and under favorable and unfavorable (winter nighttime) conditions over the ocean and 2) an atmospheric inversion framework to estimate N2O surface fluxes from the atmospheric satellite observations. After studying three N2O spectral bands (B1 at 1240–1350 cm−1, B2 at 2150–2260 cm−1 and B3 at 2400–2600 cm−1), a new TIR instrument will be developed, centered at 1250–1330 cm−1, with a resolution of 0.125 cm−1, a Full Width at Half Maximum of 0.25 cm−1 and a swath of 300 km. To optimally constrain the retrieval of N2O vertical profiles, the instrument will be on-board a platform at ~830 km altitude in a sun-synchronous orbit crossing the Equator in descending node at 09:30 local time in synergy with two other platforms (Metop-SG and Sentinel-2 NG) expected to fly in 2031–32 aiming at detecting surface properties, agricultural information on the field scale and vertical profiles of atmospheric constituents and temperature. The lifetime of the MIN2OS project would be 4–5 years to study the interannual variability of N2O surface fluxes. The spectral noise can be decreased by at least a factor of 5 compared to the lowest noise accessible to date with the Infrared Atmospheric Sounding Interferometer-New Generation (IASI-NG) mission. The N2O total error is expected to be less than ~1% (~3 ppbv) along the vertical. The preliminary design of the MIN2OS project results in a small instrument (payload of 90 kg, volume of 1200 × 600 × 300 mm3) with, in addition to the spectrometer, a wide field and 1-km resolution imager for cloud detection. The instruments could be hosted on a small platform, the whole satellite being largely compatible with a dual launch on VEGA-C. The MIN2OS project has been submitted to the European Space Agency Earth Explorer 11 mission ideas.

Assessment of Five Thermal Infrared Kernel-Driven Models Using Limited Multiangle Observations

There are five widely used kernel-driven models in the thermal infrared domain designed for the angular correction of land surface temperature (LST), including three-parameter Roujean Lagouarde (RL), Vinnikov, RossThick-LiSparseR (Ross–Li), LiStrahlerFriedl-LiDenseR (LSF-Li), and four-parameter Vinnikov-RoujeanLagouarde (Vinnikov-RL). Their fitting accuracies with hundreds of observation angles (i.e., sufficient angle) were studied; however, the fitting ability of these five models with limited observation angles is unknown, which makes it difficult to choose the appropriate one in applications. To solve this problem, 30 600 groups of multiangle directional brightness temperature (DBT) datasets were simulated by the unified optical-thermal 4-stream model considering scattering by arbitrary inclined leaves (4SAIL) model considering ten different leaf area index values, three leaf inclination distribution functions, two hotspot factors, 17 different component temperatures, five solar zenith angles, and six solar azimuth angles. Each group contains DBT values in 21 960 viewing directions [i.e., 61 viewing zenith angle (VZA) <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\times $ </tex-math></inline-formula> 360 viewing azimuth angle (VAA)]. We assume that all limited observations are in the plane with VAA = 180°/0° and VZA changing from −60° to 60° with a step of 10°. There are 13 candidate angles to be selected. Five, seven, nine, and 11 angle sampling schemes include 225, 400, 225, and 36 limited multiangle combinations, respectively. Each combination was used to drive these five kernel-driven models to fit 21 960 DBTs for 30 600 groups of 4SAIL simulations. The root-mean-square error (RMSE) of each combination and mean RMSE of all 886 combinations were used to assess the overall fitting ability of five kernel-driven models. In addition, 1 k errors were added to the driven DBTs to evaluate the models’ robustness. Four groups of airborne measured DBTs were adopted to validate the assessment conclusions. Results show that the recommended order of these five models driven by 5–11 multiangle DBTs is Vinnikov-RL, LSF-Li, Vinnikov, Ross–Li, and RL when the driven DBTs do not contain errors; Vinnikov-RL, Vinnikov, LSF-Li, Ross–Li, and RL when the driven DBTs contain 1k errors; and Vinnikov-RL, LSF-Li, Ross–Li, RL, and Vinnikov for four groups of airborne measured datasets.

Editorial for the Special Issue “Remote Sensing Monitoring of Land Surface Temperature”

The combination of the state-of-the-art in the thermal infrared (TIR) domain [...]

Prediction of leaf traits of lianas and trees via the integration of wavelet spectra in the visible-near infrared and thermal infrared domains

Predicting leaf traits using models based on spectroscopic data can provide essential information to advance ecological research and future Earth system models. Most current models are based on Partial Least Squares Regression (PLSR) algorithms that attempt to predict a set of leaf traits of several plant groups using leaf spectra. However, PLSR models tend to be inconsistent in describing the importance of absorption features when used to predict leaf traits. Likewise, the effect of contrasting absorption features of different plant groups on the prediction and evaluation of PLSR models is not well understood. Hence, this study focuses on using wavelet spectra to overcome current PLSR's limitation and improve leaf trait predictions. Specifically, we explored the use of visible–near-infrared (0.45–1.0 μm) and mid- long-wave infrared spectra (2.55–11 μm) to predict three-leaf traits of lianas and trees: Leaf Mass Area (LMA), Water Content (WC), and Equivalent Water Thickness (EWT). We also compare the effect of life forms on the prediction of traits by using sun leaves collected from 14 liana species and 21 tree species (n = 700) from a Neotropical Dry Forest. On each leaf, reflectance measurements were performed for both selected spectral regions; then, leaf traits were estimated from a leaf segment. Leaf reflectance was first resampled and then processed using continuous wavelet transformation (CWT) to derive the wavelet spectra. PLSR models linking the leaf traits and the reflectance or wavelet spectra were compared. Our results reveal that PLSR models based on wavelet spectra require fewer components to predict traits (13–16) than those based on reflectance (25–29). In addition, PLSR models' performance (e.g., R2) of testing datasets tend to be higher for models based on wavelet spectra (LMA = 0.83; WC = 0.77; EWT = 0.68) than reflectance (LMA = 0.78; WC = 0.76; EWT = 0.49). Wavelet spectra models also seem to better characterize absorption features that drive the variability of leaf traits than models based on reflectance. However, life forms play an essential role in model performance, where the prediction of lianas' traits presenting lower R2 (R2 = 0.61 ± 0.25) than trees' traits (R2 = 0.69 ± 0.15) regardless of the type of spectra or leaf trait. Our findings highlight the use of wavelet spectra to overcome limitations of the PLSR models for predicting leaf traits and the need to explore potential bias associated with plant groups on the model evaluations.

Evaluation of Multiple Methods for the Production of Continuous Evapotranspiration Estimates from TIR Remote Sensing

Continuous daily estimates of evapotranspiration (ET) spatially distributed at plot scale are required to monitor the water loss and manage crop irrigation needs. Remote sensing approaches in the thermal infrared (TIR) domain are relevant to assess actual ET and soil moisture status but due to lengthy return intervals and cloud cover, data acquisition is not continuous over time. This study aims to assess the performances of 6 commonly used as well as two new reference quantities including rainfall as an index of soil moisture availability to reconstruct seasonal ET from sparse estimates and as a function of the revisit frequency. In a first step, instantaneous in situ eddy-covariance flux tower data collected over multiple ecosystems and climatic areas were used as a proxy for perfect retrievals on satellite overpass dates. In a second step, instantaneous estimations at the time of satellite overpass were produced using the Soil Plant Atmosphere and Remote Sensing Evapotranspiration (SPARSE) energy balance model in order to evaluate the errors concurrent to the use of an energy balance model simulating the instantaneous IRT products from the local surface temperature. Significant variability in the performances from site to site was observed particularly for long revisit frequencies over 8 days, suggesting that the revisit frequency necessary to achieve accurate estimates of ET via temporal upscaling needs to be fewer than 8 days whatever the reference quantity used. For shorter return interval, small differences among the interpolation techniques and reference quantities were found. At the seasonal scale, very simple methods using reference quantities such as the global radiation or clear sky radiation appeared relevant and robust against long revisit frequencies. For infra-seasonal studies targeting stress detection and irrigation management, taking the amount of precipitation into account seemed necessary, especially to avoid the underestimation of ET over cloudy days during a long period without data acquisitions.

Alfalfa (Medicago sativa L.) crop vigor and yield characterization using high-resolution aerial multispectral and thermal infrared imaging technique

Alfalfa (Medicago sativa L.) is an important forage crop grown worldwide for animal feed, green manure, and as a land cover. However, very few approaches exist for timely field scale mapping of crop status, yield and quality attributes for management of inputs, harvest and storage resources, budgeting, crop insurance, etc. This study aims to apply high-resolution aerial multispectral and thermal infrared remote sensing (7 cm/pixel) to characterize above crop attributers. Imaged were two crop cutting cycles in 2018 season. Eight crop vigor index (VI) and a Crop Water Stress Index (CWSI) features were derived from collected imagery data. Modified Non-Linear Index (MNLI), Modified Simple Ratio (MSR) and CWSI reliably evaluated the spatial variations in crop vigor and stress traits (Coefficient of variation [CV] in the ranges of 24–69%). Yield was then predicted with indices as predictor variables through nine simple linear regression (LRs, variable: one image feature per model), seven multiple linear regression (MLRs, variables: one VI and CWSI per model), a stepwise linear regression (SLR), a partial least square regression (PLSR) and a least absolute shrinkage and selection operator (LASSO) models. The SLR, PLSR and LASSO initially used all image features for model training. Amongst simple models, MLR-4 (Variables: MNLI and CWSI) performed the best (Root mean square error [RMSE] = 0.45 kg, R2 = 0.64) and LR-5 (Variable: MNLI) was the second-best model (RMSE = 0.51 kg, R2 = 0.54). The complex SLR, PLSR and LASSO models predicted yield with similar accuracy as MLR-4 (RMSE in the ranges of 0.45–0.46 kg, R2 in the ranges of 0.63–0.64). MNLI (canopy vigor) and CWSI (stress) were significant and sufficient for effective alfalfa crop status and yield prediction for their non-saturation and non-linearity features. Overall, high-resolution aerial remote sensing in the visible-NIR and thermal infrared domain showed potential for site-specific crop monitoring.

Estimation of Soil Salinity Under Various Soil Moisture Conditions Using Laboratory Based Thermal Infrared Spectra

Soil salinization is a world-wide phenomenon that threatens ecological environment and agricultural production. Modeling soil salt content (SSC) is a big challenge because of its huge spatiotemporal variation and the interference of soil water content (SWC) and soil salt types. Prior studies showed more interest in the use of hyperspectral reflectance, while few studies focused on thermal infrared band domain. In this study, we arranged samples with three salt types and several levels of SWC and measured the soil emissivity for each sample at each level of SWC. We employed both original and derivate emissivity to figure out the relationship between SSC and soil thermal infrared spectra, then used partial least squares regression to estimate SSC. Finally, the optimal model was determined with the evaluation criteria, RPD (ratio of performance to deviation) for predictive ability and AICc (corrected Akaike Information Criterion) for simplicity. The models were applied to estimate SSC and coefficient of determination (R2) of 0.67, 0.71, 0.69 and 0.7, and root mean relative error of 4.03, 3.78, 3.92, 3.86 (g/100 g) was obtained, respectively, for NaCl, Na2SO4, Na2CO3 and all salt types. The study provided a comparison result of three salt types for soil salinity estimation and a criterion for modeling effectively and succinctly and should have potential applications in the future.

A general framework of kernel-driven modeling in the thermal infrared domain

Radiometric measurements in the Thermal Infrared (TIR) domain exhibit an angular variation over most surface types, known as the Thermal Radiation Directionality (TRD) phenomenon. A primary objective of the ongoing development of TRD physical models is to perform a correction of the angular effects to obtain comparable land surface temperature products. In practice, it is advised to handle only the models having a limited number of input parameters for the purpose of operational applications. The use of semi-empirical kernel-driven models (KDMs) appears to be a good tradeoff between physical accuracy and computational efficiency as it was already demonstrated through a broad usage in the optical domain. It remains that the existing state-of-the-art 3-parameter TIR KDMs (RossThick-LiSparseR, LiStrahlerFriedl-LiDenseR, Vinnikov, and RoujeanLagouarde) underestimate the hotspot phenomenon, especially for continuous canopies marked by a narrow peak. In this study, a new general framework of TIR kernel-driven modeling is proposed to overcome such issue. It is a linear combination of three kernels (including a base shape kernel, a hotspot kernel with adjustable width and an isotropic kernel) with the ability to simulate the bowl, dome and bell shapes in the solar principal plane. Four specific 4-parameter models (Vinnikov-RoujeanLagouarde, LiStrahlerFriedl-RoujeanLagouarde, Vinnikov-Chen, and LiStrahlerFriedl-Chen, named “base shape kernel - hotspot kernel”) within the new framework were studied to assess their abilities to mimic the patterns of the directional brightness temperature for both continuous and discrete vegetation canopies. These four 4-parameter KDMs and four 3-parameter KDMs were comprehensively evaluated with 306 groups of simulated multi-angle datasets generated by a modernized analytical 4-stream radiative transfer model based on the Scattering by Arbitrarily Inclined Leaves (4SAIL), and a Discrete Anisotropic Radiative Transfer (DART) model considering different solar zenith angles (SZA), canopy architectures and component temperatures, and 2 groups of airborne measured multi-angle datasets over continuous maize and discrete pine forest. Results show that the four 4-parameter KDMs behave better than the four existing 3-parameter KDMs over continuous canopies (e.g. R2 increases from 0.661~0.970 to 0.940~0.997 and RMSE decreases from 0.17~0.71 to 0.07~0.16 when SZA = 30°) and discrete canopies (e.g. R2 increases from 0.791~0.989 to 0.976~0.996 and RMSE decreases from 0.10~0.84 to 0.08~0.21 when SZA = 30°). The new general framework with four parameters (three kernel coefficients and an adjustable hotspot width) improves the fitting ability significantly, compared to the four existing three-parameter KDMs, given the addition of one more degree of freedom. Results show that the coefficients of the base shape kernel, hotspot kernel and isotropic kernel are related to the temperature difference between leaf and background, temperature difference between sunlit component and shaded component, and the nadir brightness temperature, respectively. However, the estimated hotspot width depends on vegetation structure. The new kernel-driven modeling framework has the potential to be a tool for angular correction of multi-angle satellite observations and angular optimization of future multi-angle TIR sensors.