modeling water stress as an indicator of red palm

MODELING WATER STRESS AS AN INDICATOR OF RED PALM WEEVIL INFESTATION USING FIELD SAMPLING, WORLDVIEW-3 REFLECTANCE, AND LABORATORY ANALYSIS A. Bannari 1, T.A. Almansoori 2, A.M.A. Mohamed 3, A. El-Battay 1, and N.A. Hameid 1 1 Department

of Geoinformatics, College of Graduate Studies, Arabian Gulf University, Manama, P.O. Box: 26671, Kingdom of Bahrain, Tel: (973) 1723-9545; Fax: (973) 1723-9552 E-mail: [email protected] 2 Department of Biology, College of Science, University of Bahrain, P.O. Box 32038, Sakhir, Kingdom of Bahrain 3 Department of Life Sciences, College of Graduate Studies, Arabian Gulf University, Manama, P.O. Box: 26671, Kingdom of Bahrain,

Abstract This study focuses for the first time on the water content modeling as an indicator for red palm weevil (RPW) stress-attacks using field sampling, Worldview-3 (WV-3) reflectance measurements, water content estimation at the laboratory (WC-Lab), several water stress indices (WSI) and statistical analysis. Based on field identification and sampling, 100 date palm trees were considered and divided into four classes of RPW stress-attacks: dead, severely attacked, moderately attacked, and healthy trees (young and mature trees). Spectral measurements were acquired over each sample using Analytical Spectral Devices (ASD). Then, they were resampled and convolved using WV-3 spectral response profiles and the Canadian radiative transfer code (CAM5S). For model calibration, only 80 samples were considered, while 20 samples were used for validation purposes. Obtained results indicate that the proposed models based on PTWSI-4 and SRWI offer an important alternative to discriminate among different levels of RPW stress-attack. They discriminate significantly among the considered stress classes (R2 ≥ 0.85) using second order regression (p < 0.05). The validation of PTWSI-4 model revealed a significant correlation with the WC-Lab values (R2 of 0.95), and an acceptable RSME of 8.5 %. Index Terms: Modeling, date palm trees, red palm weevils, water content, water stress, spectral indices, ASD, Worldview-3.

1. INTRODUCTION Around the world, several factors have affected the natural development of date palm trees [1], especially the invasive alien insect pests such as the red palm weevil (RPW), Rhynchophorus ferrugineus, Olivier [2]. It was first reported on the Indian coconut palm trees (Cocos nucifera) in 1906 [3]. By mid-1980s, RWP was discovered in the Arabian Gulf countries and has become one of the most destructive pests of date palms trees in the region [4,5]. Since then, the weevil has rapidly expanded its geographical distribution reaching several countries in Asia, North Africa, Europe, Oceania, Central America, and the Caribbean [6]. The rapid spread is probably due to international trade in infested palm trees coupled with the absence of reliable techniques to detect the early signs of RPW infestation [2]. Infested palm tree does not reveal any visible symptoms until they are in advance stages of infestation [3]. Likewise, the internal structure of the palm tree attacked severely and destroyed, without any distinctive visual signs of external physical deterioration. In the literature, several detection methods have been developed to identify the RPW infestation in the field [7,8]. Unfortunately, the application of each of the developed methods has suffered logistical and implementation issues or used for only a limited number of trees. However, at the early stages of infestation, careful observation of palm trees may reveal several symptoms, which can be related to bio-physiological issues developed in

978-1-5386-7150-4/18/$31.00 ©2018 IEEE

response to the weevil attack [9,10]. Indeed, the ability of the trees to transport water and nutrients to the crown and distribute the photoassimilates via the vascular system is severely affected by the infestation. In response to the RPW larvae aggressive feeding, which drastically damages the vascular bundles, the rate of the net photosynthesis decreases significanly [11]; whereas the transpiration rate increases drastically [12] causing a significant reduction in the water use efficency of the infested trees [11]. Based on aerial thermal images, water stress in RPW infested palm trees was evident 25 days post-infestation and three weeks before any visual symptoms were observed [13]. The change of water conductivity due to slow water transport through the damaged vascular tissues causes water stress at the tree crown [14]. At such stage, the symptoms of severe injury become physically visible in form of drying the outer leaves and fruit bunches as well as toppling of the trunk, in case of very severe and extensive tissue damage [3,7]. Without a doubt, remote sensing science can be used as an alternative for rapid and early detection of RPW stress-attack over relatively large areas because the bio-physiological variations, including the water status of the crown, can be used as indicators in such studies. The measured reflectance of the palm tree crown is dominated by liquid water absorption in the 900-2500 nm region [15,16]. In laboratory studies, Carter [17] reported a significant correlation between NIR (720 - 1000 nm) and SWIR (1000 - 2500 nm) reflectance and crop canopy water content. Removing background signals related to the structure of the canopy, spectral derivation showed that the estimation of plant water content is improved at wavelengths of intermediate water absorption (1450 nm), while wavelengths of near-total water absorption (1940 and 2500 nm) showed little to no sensitivity to variations in liquid water content [18,19]. Tucker [20] suggested that the spectral interval between 1550 and 1750 nm was the best-suited region for monitoring plant-canopy water status from space. Based on the relative depth of the 970 nm absorption feature, water stress spectral indices were correlated with ground measures of plant water content at both leaf and canopy scales [19,21]. For the first time, this paper reports the progress made to improve the understanding of the bio-physiological processes governing the link between the water content of date palm tree crown and the RPW stress-attacks. In this perspective, a model developed especially for WV-3 satellite data is proposed in this study. To achieve these objectives, a fieldwork was organized, and samples were collected from 100 date palm trees with various degrees of infestation (i.e., stress-attacks). Then, in the BRDFGoniometric-Laboratory, the bidirectional reflectance was acquired above each samples using an ASD (Analytical Spectral Devices) spectroradiometer [22]. All of the measured spectral signatures were resampled and convolved in the solar-reflective spectral bands of the WV-3 sensor [23]. The water content information collected under field conditions was quantified at the laboratory (WC-Lab) and was used for model calibration and validation. Statistical analysis was conducted between WC-Lab and WC-Predicted using several regression analysis functions at the significance level p < 0.05.

9056

IGARSS 2018

2.

MATERIALS AND METHODS

2.1. Study site The Kingdom of Bahrain (26° 00’ N, 50° 33’ E) is an Archipelago located in the Arabian Gulf, east of Saudi Arabia and west of Qatar. It has a semi-arid to extremely arid environment. The main Island is characterized by high summer temperatures (45°C) and a moderate winter temperature with an average of 17°C. The rainy season extends from November to April, with an annual average precipitation of 72 mm. Mean annual relative humidity is over 70% due to the surrounding Arabian Gulf water, with an annual average potential evapotranspiration rate of 2099 mm [24]. Historically, date palm trees have been an integral part of the rich heritage of Bahrain and have played a fundamental role in the country’s economy and society. Unfortunately, the revolutionary changes that have accompanied the discovery of oil have had a massive impact on the cultivation of date palms. In adition, climatic change impacts, limitation of freshwater, soil salinization, pest and diseases, especially RPW infestation, have caused significant decline in the total number of the trees. Indeed, over the last decades in Bahrain, the number of date palm trees has decreased from 892,000 to 534,600 covering approximately 2,330 ha [25].

The average of forty spectra measured with the ASD over each palm tree sample was resampled and convolved in the solarreflective spectral bands of WV-3 sensor characteristics using the Canadian Modified Herman transfer radiative code (CAM5S) adapted by Teillet and Santer [23]. CAM5S simulates the signal received at the top of the atmosphere from a surface reflecting solar and sky irradiance at sea level considering the relative spectral response profiles characterizing the filters of each WV-3 band. 2.5. Water content quantification After spectroradiometric measurements, water content was quantified at the laboratory (WC-Lab) according to Ceccato et al. [28] method considering the 100 palm samples. For each sample, ten leaflets were selected randomly from individual fronds. Then, subsamples were collected from the tip, mid and basal regions of each of the ten leaflets. Each subsample was weighed to obtain the fresh weight. They were then oven-dried at 80oC for 48 hours (until no change in the weight was observed by further drying) to find the dry weight. Subsequently, the percentage of WC-Lab was calculated using the following equation [19,28]:

(1)

2.2. Field sampling For statistical analysis and validation of different levels of RPW infestation, more than 100 date palm trees in five different fields were inspected by experts in remote sensing and entomology, along with a specialist in the RPW National Management Program from the Agriculture Affairs, Ministry of Works, Municipalities Affairs and Urban Planning, the Kingdom of Bahrain. Based on the levels of crown and trunk infestation, the trees were grouped into four classes: dead, severely attacked, moderately attacked, and healthy trees (young and mature trees). Representative leaflet samples were collected from the top of the crown of each palm tree of the considered classes. They were immediately placed in plastic-bags and stored in a cooler for transportation from the fields to the laboratory and, subsequently, used for spectroradiometric measurements and water content quantification. 2.3. Spectroradiometric measurements Spectroradiometric measurements were acquired in the BRDFGoniometric laboratory above each selected sample using an Analytical Spectral Devices (ASD) spectroradiometer [22]. This instrument is equipped with two detectors operating in the VNIR and SWIR, between 350 and 2500 nm. It acquires a continuous spectrum with a 1.4 nm sampling interval from 350 to 1000 nm and a 2 nm from 1000 to 2500 nm. The ASD resamples the measurements in 1-nm intervals, which allows the acquisition of 2151 contiguous bands per spectrum. The sensor is characterized by the programming capacity of the integration time, which allows an increase of the signal-to-noise ratio as well as stability. The data were acquired at nadir with a FOV of 25° and a solar zenith angle of approximately 5° by averaging forty measurements. The ASD was installed on a BRDF-Goniometric-System with a height of approximately 65 cm over the target, which makes it possible to observe a surface of ≈ 830 cm2. A laser beam was used to locate the center of the ASD-FOV. The reflectance factor of each sample was calculated by rationing target radiance to the radiance obtained from a calibrated “Spectralon panel” in accordance with the method described in [26]. Corrections were made for the wavelength dependence and non-lambertien behavior of the panel. 2.4. Worldview-3 spectral bands simulations Launched on August 13 2014, WV-3 is the first commercial satellite operated by DigitalGlobe with very high spatial resolution; multipayload and 16 bands super spectral resolution. This new technology provides a compromise between spatial and spectral information. It captures the electromagnetic spectrum in eight bands at visible and near infrared (VNIR) as well as in eight other bands at short-wave infrared (SWIR) wavelengths [27].

Where FW and DW are, respectively, the fresh and the dry weight of the subsamples. 2.5. Modeling As mentioned earlier, plant liquid water at the canopy scale absorbs solar radiation strongly in a series of absorption features in the NIR and SWIR. Based on these features, Gao [29] proposed the Normalized Difference Water Index (NDWI) for vegetation liquid water estimation exploiting the NIR (860 nm) and SWIR (1240 nm) bands. In fact, absorption by vegetation liquid water near 860 nm is negligible, and the weak liquid absorption at 1240 nm is present. The multiple scattering at the crown then enhances the water absorption and, consequently, the NDWI becomes more sensitive to the changes in liquid water content of vegetation canopies. Moreover, the used bands (860 and 1240 nm) are both located in atmospheric windows, where water vapour absorption is very small and atmospheric aerosol scattering effects are insignificant [16]. Furthermore, the Normalized Multi-band Drought index (NMDI), was proposed by Wang and Qu [30] for remote sensing of both soil and vegetation water content mapping using three bands in the NIR and SWIR (860, 1640 and 2130 nm). The two shortwave infrared water stress indices (SIWSI-1 and SIWSI-2) were proposed for monitoring plant leaf water content in a semi-arid environment [31]. The configurations of these indices were derived from the NIR (860 nm) and the SWIR (1230-1250 nm or 1628-1652 nm), respectively. The Normalized Difference Infrared Index (NDII) was proposed for the leaf of “Partina-alterniflora” canopies water content estimation using the NIR region (819 and 900 nm), and the SWIR region (1240 and 1640 nm) [32]. Based on a simulation study, Zarco-Tejada and Ustin [33] showed the dependency of the Simple Ratio Water Index (SRWI) on leaf-level variables such as leaf structure and dry matter content. Based on the water content agitation caused by RPW stressattack in the eight WV-3 SWIR spectral bands, Bannari et al. [10] proposed five Palm Tree Water Stress Indices (PTWSI). They showed that the indices NDWI, SRWI, SIWSI-1, SIWSI-2, and NDII are sensitive to palm trees water agitation caused by RPW stress-attacks (R2 ≈ 95%). Nevertheless, they revealed only 10% to 55% water content dynamics range. Contrariwise, the PTWSI differentiated amongst different levels of RPW water stress-attack classes with R2 of 90%, and enhanced significantly water content dynamic range (WCDR) for a maximum of 90% or 100%. In this study, we integrated these indices in the process of water stress modeling as an indicator for RPW stress-attack. The following are the equations of these indices using WV-3 spectral bands: NDWI = (nir- swir-1) / nir + swir-1) NMDI = (nir- swir-3 + swir-5) / (nir+ swir-3 - swir-5)

9057

(2) (3)

SIWSI-1= (swir-1 - nir) / (swir-1 + nir) SIWSI-2= (swir-3 - nir) / (swir-3 + nir) NDII = (nir- swir-3) / nir + swir-3) SRWI = nir/ swir-1 PTWSI-1 = (swir-1- swir-4) / swir-8) PTWSI-2 = (swir-1- swir-5) / swir-4) PTWSI-3 = (swir-1- swir-2) / swir-4) PTWSI-4 = (swir-1- swir-6) / (swir-1+ swir-6) PTWSI-5 = (nir-1- swir-2) / (nir-1+ swir-2)

to the WCDR. These R2 values indicate that all the models used in this investigation reveal significant correlations (R2 ≥ 0.85); except for the model based on NMDI, which showed a low R2 (0.42). Nevertheless, the models based on the NDWI, SIWSI-1, and PTWSI-5 expressed the WCDR in a limited dynamic range between 40 and 45%. Unluckily, this saturation and limitation of WCDR cannot probably discriminate clearly among different RPW stressattack classes. The models based on PTWSI-2 and PTWSI-3 indices show similar results in term of differentiation among RPW stressattack with a significant correlation (R2 of 88%), enhancing substantially the WCDR (130 to 160%), and overestimating the amount of water in the crown of the moderate attacked trees.

(4) (5) (6) (7) (8) (9) (10) (11) (12)

2.6. Statistical processing All the resampled and convolved WV-3 SWIR spectral bands were individually fitted through the WC-Lab. Then, among the 100 tree samples, only 80 were used for modeling considering the above eleven WSI equations and different statistical functions. The second order polynom was the most sensitive, providing the most significant correlation values (R2) at p < 0.05. For validation purposes, the strength of the relationship between observed (WCLab) and predicted values (WC-Predicted) and the RMSE were evaluated. 3. RESULTS AND DISCUSSIONS The spectral signatures of the four palm tree classes are illustrated in Fig. 1. Significant spectral variability is observed among these considered classes: dead, severely attacked, moderately attacked, and healthy trees. The visible bands show the effects of pigmentation constituents such as chlorophyll and carotenoids; whereas the red-edge and NIR bands express the biomass status, while the eight SWIR bands illustrate the water content change progressively from the healthy to the dead palm trees. Table 1 summarizes the R2 between these WV-3 SWIR spectral bands and WC-Lab. Aside from the SWIR-1 band, the other seven bands correlated significantly (R2 ≥ 0.78) with the quantified WC-Lab.

Table 2: Model equations and R2 between the considered WSI and WC-Lab (p ˂ 0.05), and water content dynamic range (WCDR). WCDR Index Model equations R2 (%) PTWSI-2 y = -14,38x2 + 67,64x - 10,58 0.88 160 PTWSI-3 y = -27.22x2 + 81.00x + 1.37 0.88 130 PTWSI-1 y = -8,56x2 + 43,38x + 0,50 0.88 70 PTWSI-4 y = 28,09x2 + 97,70x - 6,35 0.89 70 SRWI y = -18.19x2 + 124.26x – 69.20 0.85 75 NDII y = 60,34x2 + 81,70x + 20,13 0.87 70 SIWSI-2 y = 60,34x2 - 81,70x + 20,13 0.87 75 PTWSI-5 y = -40,13x2 + 165,97x + 2,23 0.88 45 NDWI y = 191,75x2 + 185,54x + 35,79 0.85 40 SIWSI-1 y = 191,75x2 - 185,54x + 35,79 0.86 40 NMDI y = -586,51x2 + 853,02x - 251,61 0.42 30

Fig. 2. Relationship between WC-Lab and PTWSI-4 considering different RPW stress-attack classes.

Fig. 1. Example of spectral signatures of the four palm tree classes. Table 1: R2 values of the regression analysis between the WV-3 SWIR spectral bands and WC-Lab of the four considered palm tree classes (p ˂ 0.05). Band SWIR-1 SWIR-2 SWIR-3 SWIR-4 SWIR-5 SWIR-6 SWIR-7 SWIR-8 R2 0.15 0.78 0.73 0.73 0.83 0.83 0.81 0.82

Model fit statistics were calculated between each WSI presented above and WC-Lab considering the 80 sample data set and using several regression functions (p ˂ 0.05). Although the exponential function has given acceptable results, a second-order polynomial function allowed the most significant results. Table 2 summarizes the obtained equation models and the R2 values between the considered WSI and WC-Lab, as well as the sensitivity of each model

The other models based on PTWSI-1, PTWSI-4, SRWI, NDII and SIWSI-2 differentiate the considered palm tree classes with significant correlation coefficients (R2 ≥ 0.85). Although they showed a limited variability in water content range (70 to 75%), they illustrate a good linearity between the WC-Lab values (from 6 to 60%) and the derived indices values. For instance, Fig. 2 shows the behaviour of the model based on PTWSI-4 as a function of WC-Lab for all RPW classes [full set of results not shown due to space limit]. Although this model relatively underestimates the severely attacked class, in general, its values increase progressively from dead to severely attacked, moderate attacked, and finally healthy trees, with a significant fit (Fig. 2 and Table 2). Moreover, it is not affected by saturation and linearity problems. Furthermore, the models based on NDII and SIWSI-2 indices show similar trend results fits (R2 of 87%, WCDR from 70 to 75%), underestimating the amount of water in the crown, and they show a low aptitude to discriminate between the water content of the young and mature trees. In contrast, the PTWSI-1 and SRWI models moderately overestimate the water content prediction. Although the SRWI model display an offset bias around 50%, it still sensitive to the water content change in the crown of the considered palm tree

9058

classes, even between young and mature trees. The statistical validation of PTWSI-4 model using the other 20 independent samples that were not involved in the modeling process are illustrated in Fig. 3. This statistical analysis was conducted using linear polynomial regression (p < 0.05) between the observed (WCLab) and the predicted water content (WC-Predicted) using the model based on PTWSI-4. It was evident from the results obtained that this model discriminate significantly among the different RPW stressattacks classes with an excellent R2 value (0.95) and an acceptable RSME of 8.5%.

[9]

[10] [11]

[12] [13] [14]

[15] [16]

Fig. 3. Linear regression between WC-Lab and WC-Predicted using the model based on PTWSI-4 considering the four RPW stress-attack classes.

[17] [18]

4. CONCLUSIONS These preliminary results indicate that the proposed models for water content prediction at the crown of date palm trees based on PTWSI-4 and SRWI using WV-3 high spectral resolution simulated data offer a potentially viable and important alternative to discriminate among different levels of RPW stress-attack. However, these positive results require further analyses and validation considering other fields, using WV-3 imagery data, during further cycles of RPW activity and over different seasons.

[19]

5. ACKNOWLEDGEMENTS

[23]

The authors would like to thank the Arabian Gulf University for the financial support (Grunt number: TS-GIF.004_2015-2017).

6. REFERENCES [1] [2] [3] [4]

[5] [6] [7] [8]

[20] [21] [22]

[24] [25]

FAOSTAT, “Crop Production in 2008”, Statistics Division, Food and Agriculture Organization of the United Nations, Italy, 2009. Murphy, S.T. and Briscoe, B.R. “The red palm weevil as an alien invasive: Biology and prospects for biological control as a component of IPM”. BioControl, 20, 35-45, 1999. Al-Dosary, N.M.N. et al. “Review on the management of red palm weevil Rhynchophorus ferrugineus Olivier in date palm Phoenix dactylifera L.” Emir. J. Food Agric. 28(1), 34-44, 2016. Dembilio, O. and Jaques, J.A. “Biology and Management of Red Palm Weevil”. Chapter 2 (pages 13-35) in W. Wakil, J.R. Faleiro and T.A. Miller (eds.), Sustainable Pest Management in Date Palm: Current Status and Emerging Challenges. Springer, Dordrecht, Netherlands, 429 pages, 2015. Faleiro, J.R. et al. “Integrated Pest Management of Palm Pests” (p. 439–497). In Abrol, D.P. (ed), Integrated Pest Management in the Tropics. New India Publishing Agency, New Delhi, India, 2016. CABI/EPPO “Rhynchophorus ferrugineus”. Distribution Maps of Plant Pests. Wallingford, UK: CABI, Map 258 (4th revision), 2016. Soroker, V. and Colazza, S. (Editors) “Handbook of Major Palm Pests: Biology and Management”. Wiley-Blackwell, 344 pages, 2017. Massa, R. et al. “Experimental and numerical evaluations on palm microwave heating for Red Palm Weevil pest control”. Scientific Reports, 8 p. https://www.nature.com/articles/srep45299.epdf, 2017

[26] [27] [28] [29] [30] [31] [32] [33]

9059

Bannari, A. et al. “Bio-Physiological Parameters Retrieval and Statistical Analysis for Red Palm Weevils Stress-Attack Prediction Using WorldView-3 data”. IGARSS-2016, 10-15 July, Beijing, China, pp. 7129-7132, 2016. Bannari, A. et al. “Water Stress Detection in Palm Threes Canopies as an Indicator for Red Palm Weevil Stress-Attack using WorldView-3 data”. IGARSS-2017, 23-28 July, Texas, USA, pp. 4000-4003, 2017. Al-Jabr, A.M. et al. “Effects of Red Palm Weevil Rynchophorus ferrugineus (Olivier) Infestation on Gas Exchange Capacity of Two Date Palm Phoenix dactylifera L. Cultivars”. J. of Biological Sc., 7, 1270-1273, 2007. Bokhari, U. G. and Abuzuhira, R. “Diagnostic tests for red palm weevil, Rhynchophorus ferrugineus infested date palm trees”. Arab Gulf J. of Scientific Research, 10(3), 93-104, 1992. Soroker, V. et al. “Early detection and monitoring of red palm weevil: approaches and challenges”. Proceedings AFPP - Palm Pest Mediterranean Conference, Nice (France), 2013. Cohen, Y. “Morphology and Physiology of Palm Trees as Related to Rhynchophorus ferrugineus and Paysandisia archon Infestation and Management.” In Soroker, V. and Colazza, S. (Eds.) Handbook of Major Palm Pests: Biology and Management. Wiley-Blackwell, pp. 3953, 2017. Moran, M.S. et al. “Opportunities and limitations for image-based remote sensing in precision crop management”. Remote Sens. of Envir., 61, 319–-346, 1997. Gao, B.-C. and Goetz, A.F.H. “Extraction of dry leaf spectral features from reflectance spectra of green vegetation”. Remote Sens. of Environment, 47, pp. 369-374, 1994. Carter, G.A. “Primary and secondary effects of water content on the spectral reflectance of leaves”. American Journal of Botany, 18(7), 916– 924, 1991. Rollin, E.M. and Milton, E.J. “Processing of high spectral resolution reflectance data for the retrieval of canopy water content information”. Remote Sens.of Envir., 65, 86–92, 1998. Champagne, C. et al. “Validation of a Hyperspectral Curve-Fitting Model for the Estimation of Plant Water Content of Agricultural Canopies”. Remote Sens. of Envir., 87, pp.148-160, 2003. Tucker, C.J. “Remote sensing of leaf water content in the nearinfrared”. Remote Sens. of Envir., 10, pp. 23-32, 1980. Gamon, J.A. et al. “Water expressions from hyperspectral reflectance: Implications for ecosystem flux modeling”. Proc. of the Eight Annual AVIRIS Workshop. Pasadena, CA: JPL Publication, CD-ROM, 1999. Analytical Spectral Devices, ASD Inc. (1999) http://www.asdi.com/products-spectroradiometers.asp. Teillet, P.M. and Santer, R.P. “Terrain Elevation and Sensor Altitude Dependence in Semi-Analytical Atmospheric Code”. Canadian J. of Remote Sensing, 17(1), 36-44, 1991. FAO “Bahrain: Geography, Climate and Population”, 2016. http://fao.org/nr/water/aquastat/countries_regions/bhr/index.stm Thuraya, A.A. et al. “Date Palm Status and Perspective in Bahrain”. In: Al-Khayri, J.M., et al. (Eds.), Date Palm Genetic Resources and Utilization, vol. 2: Asia and Europe, pp 353-386, 2015. Jackson, R.D. et al. “Hand-held Radiometry”. U.S. Department of Agriculture Science and Education Administration, Agricultural Reviews and Manuals, ARM-W-19, Phoenix, Arizona, U.S.A., 1980. DigitalGlobe “Core Imagery Product Guide v2.0”, 67 p., 2014. Ceccato, P. et al. “Detecting vegetation leaf water content using reflectance in the optical domain”. Remote Sens. of Envir., 77(1), 2233, 2001. Gao, B.C. “NDWI-a normalized difference water index for remote sensing of vegetation liquid water from space”. Remote Sens. of Envir., 58, 257-266, 1996. Wang, L. and Qu, J.J. “NMDI: A normalized multi-band drought index for monitoring soil and vegetation moisture with satellite remote sensing”. Geophysical Research Letters, 34(2), 1-5, 2007. Fensholt R. and Sandholt, I. “Derivation of a shortwave infrared water stress index from MODIS near- and shortwave infrared data in a semiarid environment”. Remote Sens. of Envir., 87, 111–121, 2003. Hardisky, M.A. et al. “The Influence of Soil Salinity, Growth Form, and Leaf Moisture on-the Spectral Radiance of partina alterniflora Canopies”. Photog. Eng. and Remote Sens., 49, 77–83, 1983. Zarco-Tejada, P.J. and Ustin, S.L. “Modeling canopy water content for carbon estimates from MODIS data at land EOS validation sites”. IGARSS-2001, 1, pp. 342–344, 2001.