Temporal Changes of Leaf Spectral Properties and Rapid Chlorophyll-A Fluorescence under Natural Cold Stress in Rice Seedlings

Nowadays, hyperspectral remote sensing data are widely used in nutrient management, crop yield forecasting and stress monitoring. These data can be acquired with satellites, drones and handheld spectrometers. In this research, handheld spectrometer data were validated by chlorophyll-a fluorescence measurements under natural cold stress. The performance of 16 rice cultivars with different origins and tolerances was monitored in the seedling stage. The studies were carried out under field conditions across two seasons to simulate different temperature regimes. Twenty-four spectral indices and eleven rapid chlorophyll-a fluorescence parameters were compared with albino plants. We identified which wavelengths are affected by low temperatures. Furthermore, the differences between genotypes were characterized by certain well-known and two newly developed (AAR and RAR) indices based on the spectral difference between the genotype and albino plant. The absorbance, reflectance and transmittance differences from the control are suitable for the discrimination of tolerant-sensitive varieties, especially based on their shape, peak and shifting distance. The following wavelengths are capable of determining the tolerant varieties, namely 548-553 nm, 667-670 nm, 687-688 nm and 800-950 nm in case of absorbance; above 700 nm for reflectance; and the whole spectrum (400-1100 nm) for transmittance.

Potential application of spectral indices for olive water status assessment in (semi-)arid regions: A case study in Khuzestan Province, Iran

Spectral indices can be used as fast and non-destructive indicators of plant water status or stress. It is the objective of the present study to evaluate the feasibility of using several spectral indices including water index (WI) and normalized spectral water indices 1-5 (NWI 1-5) to estimate water status in olive trees in arid regions in Iran. The experimental treatments involved two olive cultivars (Koroneiki and T2) and four irrigation regimes (irrigated with 100%, 85%, 70%, and 55% estimated crop evapotranspiration [ETc]). The results obtained showed that olive trees subjected to the different irrigation regimes of 85%, 70%, and 55% ETc experienced soil water content (SWC) deficits by 4.5%, 12%, and 20.5% that of the control, respectively. Significant differences were observed among the treatments with respect to measured relative water content (RWC), SWC, and the spectral indices of WI and NWI 1-5. The normalized spectral indices combining NIR and NIR wavelengths were found more effective in tracking changes in RWC and SWC than those that combine NIR and VIS or VIS and VIS wavelengths, respectively. Spectral indices were closely and significantly associated with RWC (.63**<R2<.77**) and SWC (.51**<R2<.67**). Among all the spectral indices investigated, NWI-2 showed the least consistent associations with RWC (ranging from 4-15% lower than the other indices examined) and SWC (ranging from 1-23% lower than the others). Based on the pooled data on spectral indices, RWC, and SWC collected during the study period, WI, NWI-1, NWI-4, and NWI-5 showed stronger correlations with RWC and SWC than did NWI-3 and NWI-2. In conclusion, the spectral indices of WI and NWI 1-5 measured at the leaf level are found useful as fast and non-destructive estimators of plant water stress in arid regions.

Based on machine learning algorithms for estimating leaf phosphorus concentration of rice using optimized spectral indices and continuous wavelet transform

Remotely estimating leaf phosphorus concentration (LPC) is crucial for fertilization management, crop growth monitoring, and the development of precision agricultural strategy. This study aimed to explore the best prediction model for the LPC of rice (Oryza sativa L.) using machine learning algorithms fed with full-band (OR), spectral indices (SIs), and wavelet features. To obtain the LPC and leaf spectra reflectance, the pot experiments with four phosphorus (P) treatments and two rice cultivars were carried out in a greenhouse in 2020-2021. The results indicated that P deficiency increased leaf reflectance in the visible region (350-750 nm) and decreased the reflectance in the near-infrared (NIR, 750-1350 nm) regions compared to the P-sufficient treatment. Difference spectral index (DSI) composed of 1080 nm and 1070 nm showed the best performance for LPC estimation in calibration (R² = 0.54) and validation (R² = 0.55). To filter and denoise spectral data effectively, continuous wavelet transform (CWT) of the original spectrum was used to improve the accuracy of prediction. The model based on Mexican Hat (Mexh) wavelet function (1680 nm, Scale 6) demonstrated the best performance with the calibration R² of 0.58, validation R² of 0.56 and RMSE of 0.61 mg g^-1. In machine learning, random forest (RF) had the best model accuracy in OR, SIs, CWT, and SIs + CWT compared with other four algorithms. The SIs and CWT coupling with the RF algorithm had the best results of model validation, the R² was 0.73 and the RMSE was 0.50 mg g^-1, followed by CWT (R² = 0.71, RMSE = 0.51 mg g^-1), OR (R² = 0.66, RMSE = 0.60 mg g^-1), and SIs (R² = 0.57, RMSE = 0.64 mg g^-1). Compared with the best performing SIs based on the linear regression models, the RF algorithm combining SIs and CWT improved the prediction of LPC with R² increased by 32%. Our results provide a valuable reference for spectral monitoring of rice LPC under different soil P-supplying levels in a large scale.

Monitoring Spatiotemporal Vegetation Response to Drought Using Remote Sensing Data

Environmental factors such as drought significantly influence vegetation growth, coverage, and ecosystem functions. Hence, monitoring the spatiotemporal vegetation responses to drought in a high temporal and adequate spatial resolution is essential, mainly at the local scale. This study was conducted to investigate the aspatial and spatial relationships between vegetation growth status and drought in the southeastern South Dakota, USA. For this purpose, Landsat 8 OLI images from the months of April through September for the years 2016-2021, with cloud cover of less than 10%, were acquired. After that, radiometric calibration and atmospheric correction were performed on all of the images. Some spectral indices were calculated using the Band Math toolbox in ENVI 5.3 (Environment for Visualizing Images v. 5.3). In the present study, the extracted spectral indices from Landsat 8 OLI images were the Normalized Difference Vegetation Index (NDVI) and the Normalized Multiband Drought Index (NMDI). The results showed that the NDVI values for the month of July in different years were at maximum value at mostly pixels. Based on the statistical criteria, the best regression models for explaining the relationship between NDVI and NMDI_Soil were polynomial order 2 for 2016 to 2019 and linear for 2021. The developed regression models accounted for 96.7, 95.7, 96.2, 88.4, and 32.2% of vegetation changes for 2016, 2017, 2018, 2019, and 2021, respectively. However, there was no defined trend between NDVI and NMDI_Soil observed in 2020. In addition, pixel-by-pixel analyses showed that drought significantly impacted vegetation coverage, and 69.6% of the pixels were negatively correlated with the NDVI. It was concluded that the Landsat satellite images have potential information for studying the relationships between vegetation growth status and drought, which is the primary step in site-specific management.

Fusion of Different Image Sources for Improved Monitoring of Agricultural Plots

In the Valencian Community, the applications of precision agriculture in multiannual woody crops with high added value (fruit trees, olive trees, almond trees, vineyards, etc.) are of priority interest. In these plots, canopies do not fully cover the soil and the planting frames are incompatible with the Resolution of Sentinel 2. The present work proposes a procedure for the fusion of images with different temporal and spatial resolutions and with different degrees of spectral quality. It uses images from the Sentinel 2 mission (low resolution, high spectral quality, high temporal resolution), orthophotos (high resolution, low temporal resolution) and images obtained with drones (very high spatial resolution, low temporal resolution). The procedure is applied to generate time series of synthetic RGI images (red, green, infrared) with the same high resolution of orthophotos and drone images, in which gray levels are reassigned from the combination of their own RGI bands and the values of the B3, B4 and B8 bands of Sentinel 2. Two practical examples of application are also described. The first shows the NDVI images that can be generated after the process of merging two RGI Sentinel 2 images obtained on two specific dates. It is observed how, after the merging, different NDVI values can be assigned to the soil and vegetation, which allows them to be distinguished (contrary to the original Sentinel 2 images). The second example shows how graphs can be generated to describe the evolution throughout the vegetative cycle of the estimated values of three spectral indices (NDVI, GNDVI, GCI) on a point in the image corresponding to soil and on another assigned to vegetation. The robustness of the proposed algorithm has been validated by using image similarity metrics.

Multi-dimensional variables and feature parameter selection for aboveground biomass estimation of potato based on UAV multispectral imagery

Aboveground biomass (AGB) is an essential assessment of plant development and guiding agricultural production management in the field. Therefore, efficient and accurate access to crop AGB information can provide a timely and precise yield estimation, which is strong evidence for securing food supply and trade. In this study, the spectral, texture, geometric, and frequency-domain variables were extracted through multispectral imagery of drones, and each variable importance for different dimensional parameter combinations was computed by three feature parameter selection methods. The selected variables from the different combinations were used to perform potato AGB estimation. The results showed that compared with no feature parameter selection, the accuracy and robustness of the AGB prediction models were significantly improved after parameter selection. The random forest based on out-of-bag (RF-OOB) method was proved to be the most effective feature selection method, and in combination with RF regression, the coefficient of determination (R²) of the AGB validation model could reach 0.90, with root mean square error (RMSE), mean absolute error (MAE), and normalized RMSE (nRMSE) of 71.68 g/m², 51.27 g/m², and 11.56%, respectively. Meanwhile, the regression models of the RF-OOB method provided a good solution to the problem that high AGB values were underestimated with the variables of four dimensions. Moreover, the precision of AGB estimates was improved as the dimensionality of parameters increased. This present work can contribute to a rapid, efficient, and non-destructive means of obtaining AGB information for crops as well as provide technical support for high-throughput plant phenotypes screening.

Crop Nitrogen Retrieval Methods for Simulated Sentinel-2 Data Using In-Field Spectrometer Data

Nitrogen (N) is one of the key nutrients supplied in agricultural production worldwide. Over-fertilization can have negative influences on the field and the regional level (e.g., agro-ecosystems). Remote sensing of the plant N of field crops presents a valuable tool for the monitoring of N flows in agro-ecosystems. Available data for validation of satellite-based remote sensing of N is scarce. Therefore, in this study, field spectrometer measurements were used to simulate data of the Sentinel-2 (S2) satellites developed for vegetation monitoring by the ESA. The prediction performance of normalized ratio indices (NRIs), random forest regression (RFR) and Gaussian processes regression (GPR) for plant-N-related traits was assessed on a diverse real-world dataset including multiple crops, field sites and years. The plant N traits included the mass-based N measure, N concentration in the biomass (Nconc), and an area-based N measure approximating the plant N uptake (NUP). Spectral indices such as normalized ratio indices (NRIs) performed well, but the RFR and GPR methods outperformed the NRIs. Key spectral bands for each trait were identified using the RFR variable importance measure and the Gaussian processes regression band analysis tool (GPR-BAT), highlighting the importance of the short-wave infrared (SWIR) region for estimation of plant Nconc-and to a lesser extent the NUP. The red edge (RE) region was also important. The GPR-BAT showed that five bands were sufficient for plant N trait and leaf area index (LAI) estimation and that a surplus of bands effectively reduced prediction performance. A global sensitivity analysis (GSA) was performed on all traits simultaneously, showing the dominance of the LAI in the mixed remote sensing signal. To delineate the plant-N-related traits from this signal, regional and/or national data collection campaigns producing large crop spectral libraries (CSL) are needed. An improved database will likely enable the mapping of N at the agro-ecosystem level or for use in precision farming by farmers in the future.

Integration of Radiometric Ground-Based Data and High-Resolution QuickBird Imagery with Multivariate Modeling to Estimate Maize Traits in the Nile Delta of Egypt

In site-specific management, rapid and accurate identification of crop stress at a large scale is critical. Radiometric ground-based data and satellite imaging with advanced spatial and spectral resolution allow for a deeper understanding of crop stress and the level of stress in a given area. This research aimed to assess the potential of radiometric ground-based data and high-resolution QuickBird satellite imagery to determine the leaf area index (LAI), biomass fresh weight (BFW) and chlorophyll meter (Chlm) of maize across well-irrigated, water stress and salinity stress areas in the Nile Delta of Egypt. Partial least squares regression (PLSR) and multiple linear regression (MLR) were evaluated to estimate the three measured traits based on vegetation spectral indices (vegetation-SRIs) derived from these methods and their combination. Maize field visits were conducted during the summer seasons from 28 to 30 July 2007 to collect ground reference data concurrent with the acquisition of radiometric ground-based measurements and QuickBird satellite imagery. The results showed that the majority of vegetation-SRIs extracted from radiometric ground-based data and high-resolution satellite images were more effective in estimating LAI, BFW, and Chlm. In general, the vegetation-SRIs of radiometric ground-based data showed higher R² with measured traits compared to the vegetation-SRIs extracted from high-resolution satellite imagery. The coefficient of determination (R²) of the significant relationships between vegetation-SRIs of both methods and three measured traits varied from 0.64 to 0.89. For example, with QuickBird high-resolution satellite images, the relationships of the green normalized difference vegetation index (GNDVI) with LAI and BFW showed the highest R² of 0.80 and 0.84, respectively. Overall, the ground-based vegetation-SRIs and the satellite-based indices were found to be in good agreement to assess the measured traits of maize. Both the calibration (Cal.) and validation (Val.) models of PLSR and MLR showed the highest performance in predicting the three measured traits based on the combination of vegetation-SRIs from radiometric ground-based data and high-resolution QuickBird satellite imagery. For example, validation (Val.) models of PLSR and MLR showed the highest performance in predicting the measured traits based on the combination of vegetation-SRIs from radiometric ground-based data and high-resolution QuickBird satellite imagery with R² (0.91) of both methods for LAI, R² (0.91–0.93) for BFW respectively, and R² (0.82) of both methods for Chlm. The models of PLSR and MLR showed approximately the same performance in predicting the three measured traits and no clear difference was found between them and their combinations. In conclusion, the results obtained from this study showed that radiometric ground-based measurements and high spectral resolution remote-sensing imagery have the potential to offer necessary crop monitoring information across well-irrigated, water stress and salinity stress in regions suffering lack of freshwater resources.

Sensing Architecture for Terrestrial Crop Monitoring: Harvesting Data as an Asset

Very often, the root of problems found to produce food sustainably, as well as the origin of many environmental issues, derive from making decisions with unreliable or inexistent data. Data-driven agriculture has emerged as a way to palliate the lack of meaningful information when taking critical steps in the field. However, many decisive parameters still require manual measurements and proximity to the target, which results in the typical undersampling that impedes statistical significance and the application of AI techniques that rely on massive data. To invert this trend, and simultaneously combine crop proximity with massive sampling, a sensing architecture for automating crop scouting from ground vehicles is proposed. At present, there are no clear guidelines of how monitoring vehicles must be configured for optimally tracking crop parameters at high resolution. This paper structures the architecture for such vehicles in four subsystems, examines the most common components for each subsystem, and delves into their interactions for an efficient delivery of high-density field data from initial acquisition to final recommendation. Its main advantages rest on the real time generation of crop maps that blend the global positioning of canopy location, some of their agronomical traits, and the precise monitoring of the ambient conditions surrounding such canopies. As a use case, the envisioned architecture was embodied in an autonomous robot to automatically sort two harvesting zones of a commercial vineyard to produce two wines of dissimilar characteristics. The information contained in the maps delivered by the robot may help growers systematically apply differential harvesting, evidencing the suitability of the proposed architecture for massive monitoring and subsequent data-driven actuation. While many crop parameters still cannot be measured non-invasively, the availability of novel sensors is continually growing; to benefit from them, an efficient and trustable sensing architecture becomes indispensable.

Screening of Croatian Native Grapevine Varieties for Susceptibility to Plasmopara viticola Using Leaf Disc Bioassay, Chlorophyll Fluorescence, and Multispectral Imaging

In the era of sustainable grapevine production, there is a growing demand to define differences between Vitis vinifera varieties in susceptibility to downy mildew. Croatia, as a country with a long tradition of grapevine cultivation, preserves a large number of native grapevine varieties. A leaf disc bioassay has been conducted on 25 of them to define their response to downy mildew, according to the International Organisation of Vine and Wine (OIV) descriptor 452-1, together with the stress response of the leaf discs using chlorophyll fluorescence and multispectral imaging with 11 parameters included. Time points of measurement were as follows: before treatment (T₀), one day post-inoculation (dpi) (T₁), two dpi (T₂), three dpi (T₃), four dpi (T₄), six dpi (T₅), and eight dpi (T₆). Visible changes in form of developed Plasmopara viticola (P. viticola) sporulation were evaluated on the seventh day upon inoculation. Results show that methods applied here distinguish varieties of different responses to downy mildew. Based on the results obtained, a phenotyping model in the absence of the pathogen is proposed, which is required to confirm by conducting more extensive research.

Sensor-based evaluation of maize (Zea mays) and weed response to post-emergence herbicide applications of Isoxaflutole and Cyprosulfamide applied as crop seed treatment or herbicide mixing partner

BACKGROUND

Some maize post-emergence herbicides obtain their crop/weed selectivity only through the use of chemical crop safeners. Safeners improve the tolerance of maize to herbicidal active ingredients. In order to investigate the crop response to safener (cyprosulfamide) spray application and seed treatment, greenhouse and field trials were conducted on three maize development stages (2-, 4-, and 6-leaf stage). Visual estimations on crop vitality were compared to ground-based and airborne hyperspectral and multispectral sensors.

RESULTS

The reduction of cyprosulfamide by 88% when applied as seed treatment did not significantly reduce maize biomass yields at the field. The crop deterioration in both trials was stronger in the cyprosulfamide seed treatments compared to the spray applications but was found to be transient in the field trial. The hyperspectral sensor and multispectral camera data correlated with R² = 0.84 (CropSpec Vegetation Index) and R² = 0.64 (Green Normalized Difference Vegetation Index).

CONCLUSION

The sensor-based collection of crop responses to treatments enables early, quantifiable and auditor-independent assessments. In particular, the airborne multispectral imagery assessment of field experiments provides more detailed and comprehensive information than visually collected data. © 2019 The Authors. Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Development and Evaluation of a New Spectral Disease Index to Detect Wheat Fusarium Head Blight Using Hyperspectral Imaging

Fusarium head blight (FHB) is a major disease threatening worldwide wheat production. FHB is a short cycle disease and is highly destructive under conducive environments. To provide technical support for the rapid detection of the FHB disease, we proposed to develop a new Fusarium disease index (FDI) based on the spectral data of 374–1050 nm. This study was conducted through the analysis of reflectance spectral data of healthy and diseased wheat ears at the flowering and filling stages by hyperspectral imaging technology and the random forest method. The characteristic wavelengths selected were 570 nm and 678 nm for the late flowering stage, 565 nm and 661 nm for the early filling stage, 560 nm and 663 nm for the combined stage (combining both flowering and filling stages) by random forest. FDI at each stage was derived from the wavebands of each corresponding stage. Compared with other 16 existing spectral indices, FDI demonstrated a stronger ability to determine the severity of the FHB disease. Its determination coefficients (R²) values exceeded 0.90 and the RMSEs were less than 0.08 in the models for each stage. Furthermore, the model for the combined stage performed better when used at single growth stage, but its effect was weaker than that of the models for the two individual growth stages. Therefore, using FDI can provide a new tool to detect the FHB disease at different growth stages in wheat.

An Assessment of Surface Water Detection Methods for Water Resource Management in the Nigerien Sahel

Water is a scarce, but essential resource in the Sahel. Rainfed ephemeral ponds and lakes that dot the landscape are necessary to the livelihoods of smallholder farmers and pastoralists who rely on these resources to irrigate crops and hydrate cattle. The remote location and dispersed nature of these water bodies limits typical methods of monitoring, such as with gauges; fortunately, remote sensing offers a quick and cost-effective means of regularly measuring surface water extent in these isolated regions. Dozens of operational methods exist to use remote sensing to identify waterbodies, however, their performance when identifying surface water in the semi-arid Sahel has not been well-documented and the limitations of these methods for the region are not well understood. Here, we evaluate two global dynamic surface water datasets, fifteen spectral indices developed to classify surface water extent, and three simple decision tree methods created specifically to identify surface water in semi-arid environments. We find that the existing global surface water datasets effectively minimize false positives, but greatly underestimate the presence and extent of smaller, more turbid water bodies that are essential to local livelihoods, an important limitation in their use for monitoring water availability. Three of fifteen spectral indices exhibited both high accuracy and threshold stability when evaluated over different areas and seasons. The three simple decision tree methods had mixed performance, with only one having an overall accuracy that compared to the best performing spectral indices. We find that while global surface water datasets may be appropriate for analysis at the global scale, other methods calibrated to the local environment may provide improved performance for more localized water monitoring needs.

The GA-BPNN-Based Evaluation of Cultivated Land Quality in the PSR Framework Using Gaofen-1 Satellite Data

Rapid and efficient assessment of cultivated land quality (CLQ) using remote sensing technology is of great significance for protecting cultivated land. However, it is difficult to obtain accurate CLQ estimates using the current satellite-driven approaches in the pressure-state-response (PSR) framework, owing to the limitations of linear models and CLQ spectral indices. In order to improve the estimation accuracy of CLQ, this study used four evaluation models (the traditional linear model; partial least squares regression, PLSR; back propagation neural network, BPNN; and BPNN with genetic algorithm optimization, GA-BPNN) to evaluate CLQ for determining the accurate evaluation model. In addition, the optimal satellite-derived indicator in the land state index was selected among five vegetation indices (the normalized vegetation index, NDVI; enhanced vegetation index, EVI; modified soil-adjusted vegetation index, MSAVI; perpendicular vegetation index, PVI; and soil-adjusted vegetation index, SAVI) to improve the prediction accuracy of CLQ. This study was conducted in Conghua District of Guangzhou, Guangdong Province, China, based on Gaofen-1 (GF-1) data. The prediction accuracies from the traditional linear model, PLSR, BPNN, and GA-BPNN were compared using observations. The results demonstrated that (1) compared with other models (the traditional linear model: R² = 0.14 and RMSE = 91.53; PLSR: R² = 0.33 and RMSE = 74.58; BPNN: R² = 0.50 and RMSE = 61.75), the GA-BPNN model based on EVI in the land state index provided the most accurate estimates of CLQ, with the R² of 0.59 and root mean square error (RMSE) of 56.87, indicating a nonlinear relationship between CLQ and the prediction indicator; and (2) the GA-BPNN-based evaluation approach of CLQ in the PSR framework was driven to map CLQ of the study area using the GF-1 data, leading to an RMSE of 61.44 at the regional scale, implying that the GA-BPNN-based evaluation approach has the potential to map CLQ over large areas. This study provides an important reference for the high-accuracy prediction of CLQ based on remote sensing technology.

A New Vegetation Index Based on Multitemporal Sentinel-2 Images for Discriminating Heavy Metal Stress Levels in Rice

Heavy metal stress in crops is a worldwide problem that requires accurate and timely monitoring. This study aimed to improve the accuracy of monitoring heavy metal stress levels in rice by using multiple Sentinel-2 images. The selected study areas are in Zhuzhou City, Hunan Province, China. Six Sentinel-2 images were acquired in 2017, and heavy metal concentrations in soil were measured. A novel vegetation index called heavy metal stress sensitive index (HMSSI) was proposed. HMSSI is the ratio between two red-edge spectral indices, namely the red-edge chlorophyll index (CI_red-edge) and the plant senescence reflectance index (PSRI). To demonstrate the capability of HMSSI, the performances of CI_red-edge and PSRI in discriminating heavy metal stress levels were compared with that of HMSSI at different growth stages. Random forest (RF) was used to establish a multitemporal monitoring model to detect heavy metal stress levels in rice based on HMSSI at different growth stages. Results show that HMSSI is more sensitive to heavy metal stress than CI_red-edge and PSRI at different growth stages. The performance of a multitemporal monitoring model combining the whole growth stage images was better than any other single growth stage in distinguishing heavy metal stress levels. Therefore, HMSSI can be regarded as an indicator for monitoring heavy metal stress levels with a multitemporal monitoring model.

Evaluation of high yielding soybean germplasm under water limitation

Limited information is available for soybean root traits and their plasticity under drought stress. To date, no studies have focused on examining diverse soybean germplasm for regulation of shoot and root response under water limited conditions across varying soil types. In this study, 17 genetically diverse soybean germplasm lines were selected to study root response to water limited conditions in clay (trial 1) and sandy soil (trial 2) in two target environments. Physiological data on shoot traits was measured at multiple crop stages ranging from early vegetative to pod filling. The phenotypic root traits, and biomass accumulation data are collected at pod filling stage. In trial 1, the number of lateral roots and forks were positively correlated with plot yield under water limitation and in trial 2, lateral root thickness was positively correlated with the hill plot yield. Plant Introduction (PI) 578477A and 088444 were found to have higher later root number and forks in clay soil with higher yield under water limitation. In sandy soil, PI458020 was found to have a thicker lateral root system and higher yield under water limitation. The genotypes identified in this study could be used to enhance drought tolerance of elite soybean cultivars through improved root traits specific to target environments.

Estimation of soil moisture content from the spectral reflectance of bare soils in the 0.4-2.5 µm domain

This work aims to compare the performance of new methods to estimate the Soil Moisture Content (SMC) of bare soils from their spectral signatures in the reflective domain (0.4-2.5 µm) in comparison with widely used spectral indices like Normalized Soil Moisture Index (NSMI) and Water Index SOIL (WISOIL). Indeed, these reference spectral indices use wavelengths located in the water vapour absorption bands and their performance are thus very sensitive to the quality of the atmospheric compensation. To reduce these limitations, two new spectral indices are proposed which wavelengths are defined using the determination matrix tool by taking into account the atmospheric transmission: Normalized Index of Nswir domain for Smc estimatiOn from Linear correlation (NINSOL) and Normalized Index of Nswir domain for Smc estimatiOn from Non linear correlation (NINSON). These spectral indices are completed by two new methods based on the global shape of the soil spectral signatures. These methods are the Inverse Soil semi-Empirical Reflectance model (ISER), using the inversion of an existing empirical soil model simulating the soil spectral reflectance according to soil moisture content for a given soil class, and the convex envelope model, linking the area between the envelope and the spectral signature to the SMC. All these methods are compared using a reference database built with 32 soil samples and composed of 190 spectral signatures with five or six soil moisture contents. Half of the database is used for the calibration stage and the remaining to evaluate the performance of the SMC estimation methods. The results show that the four new methods lead to similar or better performance than the one obtained by the reference indices. The RMSE is ranging from 3.8% to 6.2% and the coefficient of determination R2 varies between 0.74 and 0.91 with the best performance obtained with the ISER model. In a second step, simulated spectral radiances at the sensor level are used to analyse the sensitivity of these methods to the sensor spectral resolution and the water vapour content knowledge. The spectral signatures of the database are then used to simulate the signal at the top of atmosphere with a radiative transfer model and to compute the integrated incident signal representing the spectral radiance measurements of the HYMAP airborne hyperspectral instrument. The sensor radiances are then corrected from the atmosphere by an atmospheric compensation tool to retrieve the surface reflectances. The SMC estimation methods are then applied on the retrieve spectral reflectances. The adaptation of the spectral index wavelengths to the HyMap sensor spectral bands and the application of the convex envelope and ISER models to boarder spectral bands lead to an error on the SMC estimation. The best performance is then obtained with the ISER model (RMSE of 2.9% and R2 of 0.96) while the four other methods lead to quite similar RMSE (from 6.4% to 7.8%) and R² (between 0.79 and 0.83) values. In the atmosphere compensation processing, an error on the water vapour content is introduced. The most robust methods to water vapour content variations are WISOIL, NINSON, NINSOL and ISER model. The convex envelope model and NSMI index require an accurate estimation of the water vapour content in the atmosphere.