A deep learning approach for automatic identification of ancient agricultural water harvesting systems.

Despite the harsh climatic conditions in the Central Negev Desert, Israel, thousands of dry stonewalls were built across ephemeral streams between the fourth and seventh centuries CE to sustain productive agricultural activity. Since 640 CE, many of these ancient terraces have remained untouched but buried by sediments, covered by natural vegetation, and partially destroyed. The main goal of the current research is to develop a procedure for the automatic recognition of ancient water harvesting systems by incorporating two remote sensing datasets (a high-resolution color orthophoto and LiDAR-derived topographic variables) and two advanced processing methods (an object-based image analysis (OBIA) and a deep convolutional neural networks (DCNN) model). A confusion matrix of object-based classification revealed an overall accuracy of 86% and a Kappa coefficient of 0.79. The DCNN model achieved a Mean Intersection over Union (MIoU) value for testing datasets of 53. The individual IoU values of terraces and sidewalls were 33.2 and 30.1, respectively. The current study demonstrates how incorporating OBIA, aerial photographs, and LiDAR in the context of DCNN improves the identification and mapping of archaeological structures.

Evaluation of satellite Leaf Area Index in California vineyards for improving water use estimation.

Remote sensing estimation of evapotranspiration (ET) directly quantifies plant water consumption and provides essential information for irrigation scheduling, which is a pressing need for California vineyards as extreme droughts become more frequent. Many ET models take satellite-derived Leaf Area Index (LAI) as a major input, but how uncertainties of LAI estimations propagate to ET and the partitioning between evaporation and transpiration is poorly understood. Here we assessed six satellite-based LAI estimation approaches using Landsat and Sentinel-2 images against ground measurements from four vineyards in California and evaluated ET sensitivity to LAI in the thermal-based two-source energy balance (TSEB) model. We found that radiative transfer modeling-based approaches predicted low to medium LAI well, but they significantly underestimated high LAI in highly clumped vine canopies (RMSE ~ 0.97 to 1.27). Cubist regression models trained with ground LAI measurements from all vineyards achieved high accuracy (RMSE ~ 0.3 to 0.48), but these empirical models did not generalize well between sites. Red edge bands and the related vegetation index (VI) from the Sentinel-2 satellite contain complementary information of LAI to VIs based on near-infrared and red bands. TSEB ET was more sensitive to positive LAI biases than negative ones. Positive LAI errors of 50% resulted in up to 50% changes in ET, while negative biases of 50% in LAI caused less than 10% deviations in ET. However, even when ET changes were minimal, negative LAI errors of 50% led to up to a 40% reduction in modeled transpiration, as soil evaporation and plant transpiration responded to LAI change divergently. These findings call for careful consideration of satellite LAI uncertainties for ET modeling, especially for the partitioning of water loss between vine and soil or cover crop for effective vineyard irrigation management.

Estimating Daily PM2.5 and PM10 over Italy Using an Ensemble Model.

Spatiotemporally resolved particulate matter (PM) estimates are essential for reconstructing long and short-term exposures in epidemiological research. Improved estimates of PM2.5 and PM10 concentrations were produced over Italy for 2013-2015 using satellite remote-sensing data and an ensemble modeling approach. The following modeling stages were used: (1) missing values of the satellite-based aerosol optical depth (AOD) product were imputed using a spatiotemporal land-use random-forest (RF) model incorporating AOD data from atmospheric ensemble models; (2) daily PM estimations were produced using four modeling approaches: linear mixed effects, RF, extreme gradient boosting, and a chemical transport model, the flexible air quality regional model. The filled-in MAIAC AOD together with additional spatial and temporal predictors were used as inputs in the three first models; (3) a geographically weighted generalized additive model (GAM) ensemble model was used to fuse the estimations from the four models by allowing the weights of each model to vary over space and time. The GAM ensemble model outperformed the four separate models, decreasing the cross-validated root mean squared error by 1-42%, depending on the model. The spatiotemporally resolved PM estimations produced by the suggested model can be applied in future epidemiological studies across Italy.

The potential of the spectral 'water balance index' (WABI) for crop irrigation scheduling.

Hyperspectral sensing can detect slight changes in plant physiology, and may offer a faster and nondestructive alternative for water status monitoring. This premise was tested in the current study using a narrow-band 'water balance index' (WABI), which is based on independent changes in leaf water content (1500 nm) and the efficiency of the nonphotochemical quenching (NPQ) photo-protective mechanism (531 nm). The hydraulic, photo-protective and spectral behaviors of five important crops - grapevine, corn, tomato, pea and sunflower - were evaluated under water deficit conditions in order to associate the differences in stress physiology with WABI suitability. Rapid alterations in both leaf water content and NPQ were observed in grapevine, pea and sunflower, and were effectively captured by WABI. Apart from water status monitoring, the index was also successful in scheduling the irrigation of a vineyard, despite phenological and environmental variability. Conversely, corn and tomato displayed a relatively strict stomatal regime and/or mild NPQ responses and were, thus, unsuitable for WABI-based monitoring. WABI shows great potential for irrigation scheduling of various crops, and has a clear advantage over spectral models that focus on either of the abovementioned physiological mechanisms.

Multiscale mapping of species diversity under changed land use using imaging spectroscopy.

Land use changes are one of the most important factors causing environmental transformations and species diversity alterations. The aim of the current study was to develop a geoinformatics-based framework to quantify alpha and beta diversity indices in two sites in Israel with different land uses, i.e., an agricultural system of fruit orchards, an afforestation system of planted groves, and an unmanaged system of groves. The framework comprises four scaling steps: (1) classification of a tree species distribution (SD) map using imaging spectroscopy (IS) at a pixel size of 1 m; (2) estimation of local species richness by calculating the alpha diversity index for 30-m grid cells; (3) calculation of beta diversity for different land use categories and sub-categories at different sizes; and (4) calculation of the beta diversity difference between the two sites. The SD was classified based on a hyperspectral image with 448 bands within the 380-2500 nm spectral range and a spatial resolution of 1 m. Twenty-three tree species were classified with high overall accuracy values of 82.57% and 86.93% for the two sites. Significantly high values of the alpha index characterize the unmanaged land use, and the lowest values were calculated for the agricultural land use. In addition, high values of alpha indices were found at the borders between the polygons related to the "edge-effect" phenomenon, whereas low alpha indices were found in areas with high invasion species rates. The beta index value, calculated for 58 polygons, was significantly lower in the agricultural land use. The suggested framework of this study succeeded in quantifying land use effects on tree species distribution, evenness, and richness. IS and spatial statistics techniques offer an opportunity to study woody plant species variation with a multiscale approach that is useful for managing land use, especially under increasing environmental changes.

Derivation of land surface temperature for Landsat-8 TIRS using a split window algorithm.

Land surface temperature (LST) is one of the most important variables measured by satellite remote sensing. Public domain data are available from the newly operational Landsat-8 Thermal Infrared Sensor (TIRS). This paper presents an adjustment of the split window algorithm (SWA) for TIRS that uses atmospheric transmittance and land surface emissivity (LSE) as inputs. Various alternatives for estimating these SWA inputs are reviewed, and a sensitivity analysis of the SWA to misestimating the input parameters is performed. The accuracy of the current development was assessed using simulated Modtran data. The root mean square error (RMSE) of the simulated LST was calculated as 0.93 °C. This SWA development is leading to progress in the determination of LST by Landsat-8 TIRS.

An efficient approach for pixel decomposition to increase the spatial resolution of land surface temperature images from MODIS thermal infrared band data.

Land surface temperature (LST) images retrieved from the thermal infrared (TIR) band data of Moderate Resolution Imaging Spectroradiometer (MODIS) have much lower spatial resolution than the MODIS visible and near-infrared (VNIR) band data. The coarse pixel scale of MODIS LST images (1000 m under nadir) have limited their capability in applying to many studies required high spatial resolution in comparison of the MODIS VNIR band data with pixel scale of 250-500 m. In this paper we intend to develop an efficient approach for pixel decomposition to increase the spatial resolution of MODIS LST image using the VNIR band data as assistance. The unique feature of this approach is to maintain the thermal radiance of parent pixels in the MODIS LST image unchanged after they are decomposed into the sub-pixels in the resulted image. There are two important steps in the decomposition: initial temperature estimation and final temperature determination. Therefore the approach can be termed double-step pixel decomposition (DSPD). Both steps involve a series of procedures to achieve the final result of decomposed LST image, including classification of the surface patterns, establishment of LST change with normalized difference of vegetation index (NDVI) and building index (NDBI), reversion of LST into thermal radiance through Planck equation, and computation of weights for the sub-pixels of the resulted image. Since the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) with much higher spatial resolution than MODIS data was on-board the same platform (Terra) as MODIS for Earth observation, an experiment had been done in the study to validate the accuracy and efficiency of our approach for pixel decomposition. The ASTER LST image was used as the reference to compare with the decomposed LST image. The result showed that the spatial distribution of the decomposed LST image was very similar to that of the ASTER LST image with a root mean square error (RMSE) of 2.7 K for entire image. Comparison with the evaluation DisTrad (E-DisTrad) and re-sampling methods for pixel decomposition also indicate that our DSPD has the lowest RMSE in all cases, including urban region, water bodies, and natural terrain. The obvious increase in spatial resolution remarkably uplifts the capability of the coarse MODIS LST images in highlighting the details of LST variation. Therefore it can be concluded that, in spite of complicated procedures, the proposed DSPD approach provides an alternative to improve the spatial resolution of MODIS LST image hence expand its applicability to the real world.

A long-term spatiotemporal analysis of biocrusts across a diverse arid environment: The case of the Israeli-Egyptian sandfield.

Spatiotemporal data can be analyzed using spatial, time-series, and machine learning algorithms to extract regional biocrust trends. Analyzing the spatial trends of biocrusts through time, using satellite imagery, may improve the quantification and understanding of their change drivers. The current work strives to develop a unique framework for analyzing spatiotemporal trends of the spectral Crust Index (CI), thus identifying the drivers of the biocrusts' spatial and temporal patterns. To fulfill this goal, CI maps, derived from 31 annual Landsat images, were analyzed by applying advanced statistical and machine learning algorithms. A comprehensive overview of biocrusts' spatiotemporal patterns was achieved using an integrative approach, including a long-term analysis, using the Mann-Kendall (MK) statistical test, and a short-term analysis, using a rolling MK with a window size of five years. Additionally, temporal clustering, using the partition around medoids (PAM) algorithm, was applied to model the spatial multi-annual dynamics of the CI. A Granger Causality test was then applied to quantify the relations between CI dynamics and precipitation. The findings show that 88.7% of pixels experienced a significant negative change, and only 0.5% experienced a significant positive change. A strong association was found in temporal trends among all clusters (0.67 ≤ r ≤ 0.8), signifying a regional effect due to precipitation levels (p < 0.05 for most clusters). The biocrust dynamics were also locally affected by anthropogenic factors (0.58 > CI > 0.64 and 0.64 > CI > 0.71 for strongly and weakly affected regions, respectively). A spatiotemporal analysis of a series of spaceborne images may improve conservation management by evaluating biocrust development in drylands. The suggested framework may also by applied to various disciplines related to quantifying spatial and temporal trends.

A novel approach for long-term spectral monitoring of desert shrubs affected by an oil spill.

Crude oil pollution is a global environmental concern since it persists in the environment longer than most conventional carbon sources. In December 2014, the hyper-arid Evrona Nature Reserve, Israel, experienced large-scale contamination when crude oil spilled. The overarching goal of the study was to investigate the possible changes, caused by an accidental crude oil spill, in the leaf reflectance and biochemical composition of four natural habitat desert shrubs. The specific objectives were (1) to monitor the biochemical properties of dominant shrub species in the polluted and control areas; (2) to study the long-term consequences of the contamination; (3) to provide information that will assist in planning rehabilitation actions; and (4) to explore the feasibility of vegetation indices (VIs), along with the machine learning (ML) technique, for detecting stressed shrubs based on the full spectral range. Four measurement campaigns were conducted in 2018 and 2019. Along with the various stress indicators, field spectral measurements were performed in the range of 350-2500 nm. A regression analysis to examine the relation of leaf reflectance to biochemical contents was carried out, to reveal the relevant wavelengths in which polluted and control plants differ. Vegetation indices applied in previous studies were found to be less sensitive for indirect detection of long-term oil contamination. A novel spectral index, based on indicative spectral bands, named the "normalized blue-green stress index" (NBGSI), was established. The NBGSI distinguished significantly between shrubs located in the polluted and in the control areas. The NBGSI showed a strong linear correlation with pheophytin a. Machine learning classification algorithms obtained high overall prediction accuracy in distinguishing between shrubs located in the oil-polluted and the control sites, indicating internal component differences. The findings of this study demonstrate the efficacy of indirect and non-destructive spectral tools for detecting and monitoring oil pollution stress in shrubs.

Source-sink relations of sunflower plants as affected by a parasite modifies carbon allocations and leaf traits.

Sunflower broomrape (Orobanche cumana) is a root holoparasitic plant causing major damage to sunflower (Helianthus annuus L.). Parasite infection initiates source-sink relations between the parasite (sink) and the host (source), allocating carbohydrates, water and nutrients to the parasite. The primary aim of the current study was to explore responses of sunflower to broomrape parasitism, specifically to examine alternations in leaf area, leaf mass per area (LMA), mesophyll structure and root hydraulic conductivity. Leaf changes revealed modifications similar to described previously in shade adapted plants, causing larger and thinner leaves. These traits were accompanied with significantly higher root hydraulics. These changes were caused by carbohydrate depletion due to source-sink relationships between the host and parasite. An Imazapic herbicide (ALS inhibitor) was used for controlling broomrape attachments and by to investigate the plasticity of the traits found. Broomrape infected plants which were treated with Imazapic had leaves similar to non-infected plants, including mesophyll structure and carbon assimilation rates. These results demonstrated source-sink effects of broomrape which cause a low-light-like acclimation behavior which is reversible.

Recognition of Orobanche cumana Below-Ground Parasitism Through Physiological and Hyper Spectral Measurements in Sunflower (Helianthus annuus L.).

Broomrape (Orobanche and Phelipanche spp.) parasitism is a severe problem in many crops worldwide, including in the Mediterranean basin. Most of the damage occurs during the sub-soil developmental stage of the parasite, by the time the parasite emerges from the ground, damage to the crop has already been done. One feasible method for sensing early, below-ground parasitism is through physiological measurements, which provide preliminary indications of slight changes in plant vitality and productivity. However, a complete physiological field survey is slow, costly and requires skilled manpower. In recent decades, visible to-shortwave infrared (VIS-SWIR) hyperspectral tools have exhibited great potential for faster, cheaper, simpler and non-destructive tracking of physiological changes. The advantage of VIS-SWIR is even greater when narrow-band signatures are analyzed with an advanced statistical technique, like a partial least squares regression (PLS-R). The technique can pinpoint the most physiologically sensitive wavebands across an entire spectrum, even in the presence of high levels of noise and collinearity. The current study evaluated a method for early detection of Orobanche cumana parasitism in sunflower that combines plant physiology, hyperspectral readings and PLS-R. Seeds of susceptible and resistant O. cumana sunflower varieties were planted in infested (15 mg kg-1 seeds) and non-infested soil. The plants were examined weekly to detect any physiological or structural changes; the examinations were accompanied by hyperspectral readings. During the early stage of the parasitism, significant differences between infected and non-infected sunflower plants were found in the reflectance of near and shortwave infrared areas. Physiological measurements revealed no differences between treatments until O. cumana inflorescences emerged. However, levels of several macro- and microelements tended to decrease during the early stage of O. cumana parasitism. Analysis of leaf cross-sections revealed differences in range and in mesophyll structure as a result of different levels of nutrients in sunflower plants, manifesting the presence of O. cumana infections. The findings of an advanced PLS-R analysis emphasized the correlation between specific reflectance changes in the SWIR range and levels of various nutrients in sunflower plants. This work demonstrates potential for the early detection of O. cumana parasitism on sunflower roots using hyperspectral tools.

Hyperspectral Technologies for Assessing Seed Germination and Trifloxysulfuron-methyl Response in Amaranthus palmeri (Palmer Amaranth).

Weed infestations in agricultural systems constitute a serious challenge to agricultural sustainability and food security worldwide. Amaranthus palmeri S. Watson (Palmer amaranth) is one of the most noxious weeds causing significant yield reductions in various crops. The ability to estimate seed viability and herbicide susceptibility is a key factor in the development of a long-term management strategy, particularly since the misuse of herbicides is driving the evolution of herbicide response in various weed species. The limitations of most herbicide response studies are that they are conducted retrospectively and that they use in vitro destructive methods. Development of a non-destructive method for the prediction of herbicide response could vastly improve the efficacy of herbicide applications and potentially delay the evolution of herbicide resistance. Here, we propose a toolbox based on hyperspectral technologies and data analyses aimed to predict A. palmeri seed germination and response to the herbicide trifloxysulfuron-methyl. Complementary measurement of leaf physiological parameters, namely, photosynthetic rate, stomatal conductence and photosystem II efficiency, was performed to support the spectral analysis. Plant response to the herbicide was compared to image analysis estimates using mean gray value and area fraction variables. Hyperspectral reflectance profiles were used to determine seed germination and to classify herbicide response through examination of plant leaves. Using hyperspectral data, we have successfully distinguished between germinating and non-germinating seeds, hyperspectral classification of seeds showed accuracy of 81.9 and 76.4%, respectively. Sensitive and resistant plants were identified with high degrees of accuracy (88.5 and 90.9%, respectively) from leaf hyperspectral reflectance profiles acquired prior to herbicide application. A correlation between leaf physiological parameters and herbicide response (sensitivity/resistance) was also demonstrated. We demonstrated that hyperspectral reflectance analyses can provide reliable information about seed germination and levels of susceptibility in A. palmeri. The use of reflectance-based analyses can help to better understand the invasiveness of A. palmeri, and thus facilitate the development of targeted control methods. It also has enormous potential for impacting environmental management in that it can be used to prevent ineffective herbicide applications. It also has potential for use in mapping tempo-spatial population dynamics in agro-ecological landscapes.

Characterization of the biosolids composting process by hyperspectral analysis.

Composted biosolids are widely used as a soil supplement to improve soil quality. However, the application of immature or unstable compost can cause the opposite effect. To date, compost maturation determination is time consuming and cannot be done at the composting site. Hyperspectral spectroscopy was suggested as a simple tool for assessing compost maturity and quality. Nevertheless, there is still a gap in knowledge regarding several compost maturation characteristics, such as dissolved organic carbon, NO3, and NH4 contents. In addition, this approach has not yet been tested on a sample at its natural water content. Therefore, in the current study, hyperspectral analysis was employed in order to characterize the biosolids composting process as a function of composting time. This goal was achieved by correlating the reflectance spectra in the range of 400-2400nm, using the partial least squares-regression (PLS-R) model, with the chemical properties of wet and oven-dried biosolid samples. The results showed that the proposed method can be used as a reliable means to evaluate compost maturity and stability. Specifically, the PLS-R model was found to be an adequate tool to evaluate the biosolids' total carbon and dissolved organic carbon, total nitrogen and dissolved nitrogen, and nitrate content, as well as the absorbance ratio of 254/365nm (E2/E3) and C/N ratios in the dry and wet samples. It failed, however, to predict the ammonium content in the dry samples since the ammonium evaporated during the drying process. It was found that in contrast to what is commonly assumed, the spectral analysis of the wet samples can also be successfully used to build a model for predicting the biosolids' compost maturity.

Diurnal emissivity dynamics in bare versus biocrusted sand dunes.

Land surface emissivity (LSE) in the thermal infrared depends mainly on the ground cover and on changes in soil moisture. The LSE is a critical variable that affects the prediction accuracy of geophysical models requiring land surface temperature as an input, highlighting the need for an accurate derivation of LSE. The primary aim of this study was to test the hypothesis that diurnal changes in emissivity, as detected from space, are larger for areas mostly covered by biocrusts (composed mainly of cyanobacteria) than for bare sand areas. The LSE dynamics were monitored from geostationary orbit by the Spinning Enhanced Visible and Infrared Imager (SEVIRI) over a sand dune field in a coastal desert region extending across both sides of the Israel-Egypt political borderline. Different land-use practices by the two countries have resulted in exposed, active sand dunes on the Egyptian side (Sinai), and dunes stabilized by biocrusts on the Israeli side (Negev). Since biocrusts adsorb more moisture from the atmosphere than bare sand does, and LSE is affected by the soil moisture, diurnal fluctuations in LSE were larger for the crusted dunes in the 8.7 µm channel. This phenomenon is attributed to water vapor adsorption by the sand/biocrust particles. The results indicate that LSE is sensitive to minor changes in soil water content caused by water vapor adsorption and can, therefore, serve as a tool for quantifying this effect, which has a large spatial impact. As biocrusts cover vast regions in deserts worldwide, this discovery has repercussions for LSE estimations in deserts around the globe, and these LSE variations can potentially have considerable effects on geophysical models from local to regional scales.

The effect of differential growth rates across plants on spectral predictions of physiological parameters.

Leaves of various ages and positions in a plant's canopy can present distinct physiological, morphological and anatomical characteristics, leading to complexities in selecting a single leaf for spectral representation of an entire plant. A fortiori, as growth rates between canopies differ, spectral-based comparisons across multiple plants--often based on leaves' position but not age--becomes an even more challenging mission. This study explores the effect of differential growth rates on the reflectance variability between leaves of different canopies, and its implication on physiological predictions made by widely-used spectral indices. Two distinct irrigation treatments were applied for one month, in order to trigger the formation of different growth rates between two groups of grapevines. Throughout the experiment, the plants were physiologically and morphologically monitored, while leaves from every part of their canopies were spectrally and histologically sampled. As the control vines were constantly developing new leaves, the water deficit plants were experiencing growth inhibition, resulting in leaves of different age at similar nodal position across the treatments. This modification of the age-position correlation was characterized by a near infrared reflectance difference between younger and older leaves, which was found to be exponentially correlated (R(2) = 0.98) to the age-dependent area of intercellular air spaces within the spongy parenchyma. Overall, the foliage of the control plant became more spectrally variable, creating complications for intra- and inter-treatment leaf-based comparisons. Of the derived indices, the Structure-Insensitive Pigment Index (SIPI) was found indifferent to the age-position effect, allowing the treatments to be compared at any nodal position, while a Normalized Difference Vegetation Index (NDVI)-based stomatal conductance prediction was substantially affected by differential growth rates. As various biotic and abiotic factors may form distinctions in growth, future precision agriculture studies should consider its spectral effect on physiological predictions.

Comparison of UV-absorbing nets in pepper crops: spectral properties, effects on plants and pest control.

In horticultural crops, the use of screens to protect plants is the usual strategy in the Mediterranean area. Screen manufacturers offer a range of netting that vary in their UV-absorbing properties. We compared the photoeffects of seven different screens. Sweet pepper trials were conducted at the Gilat Research Center, Israel, where the spectral properties of the nets and their influence on pest infestation and crop development were evaluated. UV transmittance varied among the materials studied ranging from 40% to 70% of the incident radiation. BioNet white and P-Optinet, which absorbed and reflected the highest amount of UV radiation, performed the best protection against the main pepper pest (thrips, whiteflies and broad mites). Spectral measurements also showed that the photosynthetically active radiation differentially penetrated the nets, which together with the amount of UV absorbed by the screenings, resulted in a range of plant height and chlorophyll content. A global understanding of the UV-absorbing nets' effect on pepper crops and their pests was evaluated in this work because of the importance of these screens to integrated pest management and sustainable agriculture production.

Natural vegetation phenology assessment by ground spectral measurements in two semi-arid environments.

Natural vegetation in semi-arid regions is characterized by three ground features, in addition to bare surfaces - biological soil crusts, annuals, and perennials. These three elements have distinguishable phenological cycles that can be detected by spectral ground measurements and by calculating the weighted normalized difference vegetation index (NDVI). The latter is the product of the derived NDVI of each ground feature and its respective areal cover. Each phenological cycle has the same basic elements - oscillation from null (or low) to full photosynthetic status and back to a stage of senescence. However, they vary in phase. The biological soil crusts show the earliest and highest weighted NDVI peak during the rainy season, and their weighted NDVI signal lasts longer than that of the annuals. The annuals are dominant in late winter and early spring while the perennials predominate in late spring and during the summer.