This PDF file contains the front matter associated with SPIE Proceedings Volume 11528, including the Title Page, Copyright information, and Table of Contents.

The event continues to be the leading international conference for researchers, engineers and scientists involved in the latest developments in all aspects of earth observation, next generation satellites, atmospheric propagation, and imaging analytics.

Spectral reflectance and backscattering’s of remotely-sensed data presents significant details of different objects on the Earth's surface in numerous applications of space technology. In this study, satellite datasets such as Sentinel-1 i.e. synthetic aperture radar (SAR) and Sentinel-2 satellite datasets i.e. multispectral data is used in order to effectively classify crop patterns of rainfed regions of India. The study areas have been selected in Jhansi district of Uttar Pradesh, Buldhana district of Maharashtra and Ballari district of Karnataka. These three (3) selected districts contains rainfed and irrigated agricultural lands. In the initial stage baseline map of cropland has been prepared by the following concepts; a baseline map is a map including the total cropland area i.e. Rabi and Kharif and non-cropped area’s i.e. man-made (infrastructures + urban), permanent water, permanent vegetation (forest), and other land cover’s depending on the area (for instance wetlands, etc.). This is generated by using annual temporal descriptors (TD). In the next, Sentinel-2 MSI high-resolution satellite data has been further used for crop pattern map derived by applying the supervised classification following the machine learning algorithms (MLA) techniques. The accuracy level of crop pattern map of Ballari district is 84.04%, Jhansi district is 89.3%, and of Buldhana district is 83.09%. The total crop land area in Ballari district is 0.582 million ha, with the total crop types covered area is cotton 32%, groundnut 23%, maize 29%, pearl millet 10% and paddy 6%. The total cropland area in Jhansi district is 0.507 million ha, with the total crop types covered area is black gram 65%, paddy 10%, groundnut 17%, and peppermint 8%. The total cropland area in Buldhana district is 0.686 million ha, with the total crop types covered area is cotton 24%, maize 7%, pigeon pea 16%, sorghum 25% and soybean 28%, respectively.

The unprecedented loss of habitats and biodiversity in the UK has been particularly evident during the past few decades and strongly emphasises the need for monitoring habitat states and dynamics on a more regular and consistent basis. Timely availability of free, high resolution Sentinel optical and radar data offers an opportunity to develop capacity for monitoring and informing natural resource management and interventions. This study demonstrates a workflow to utilise the full potential of Sentinel-2 datasets through a systematic, scalable approach for routinely mapping dominant vegetation genus/species in the natural and semi-natural habitats across the landscape. The present study is part of the Living Wales project which is developing a landcover monitoring framework for Wales with operational capabilities. The project utilizes the Food and Agriculture Organisation Land Cover Classification System (LCCS) as the base taxonomy. The LCCS provides capability to integrate Essential Environmental Descriptors (EEDs) of the environment retrieved from Earth observation (EO) data to generate the land cover maps. A precursor classification step first involved identification and separation of natural/semi-natural vegetated and cultivated areas across Wales into woody (trees and shrubs) and herbaceous lifeforms. Within the natural/semi-natural lifeform category, spectrally distinct genus or species were separated first, followed by further segregation of groups with similar spectral response. A machine learning algorithm was optimized by identifying appropriate spectral variables and time windows for the stepwise separations. This study generates most recent dominant species/genus map for Wales with validation and pixelwise probability statistics. The map represents a composite period of four years (2015-19), with capacity for routinely update (e.g., for 2016-20), thereby ensuring temporal consistency. The method demonstrates a systematic EO-based approach for genus/species mapping by efficiently utilising dense time-series information from ~10 m resolution Sentinel-2 data.

Principal Component Analysis (PCA) is applied to Sentinel 2 Multi-Spectral Instrument (MSI) imagery to investigate its ability to detect deforestation through direct analysis of resulting Singular Values and Principal Component loading matrices. Initial work aims to compare deforestation detection in small areas across North-East Rondônia, Brazil with previous deforestation studies. Subsequently, a deforestation analysis of the Sentinel 2 MSI using PCA in Sub-Saharan Africa is presented. Standardised PCA is applied through the Singular Value Decomposition (SVD) of channel standardised, resampled input imagery. First, cropped sub-areas of input imagery are considered, termed local PCA. Local PCA is applied to images separately (separate rotation) and to two-year image composites (merged rotation). Both approaches were found to detect forest cover changes, with separate rotation allowing for the generation of time-series data. The change detection resolution of both approaches is relatively low; being able to detect only if change has occurred in the general area and not the exact location of greatest change. In order to improve change detection resolution and identify the sub-areas of greatest change, separate rotation PCA is applied on a pixel scale. A simple statistical threshold is used to implement bi-temporal change detection, demonstrating the ability of this approach to detect forest cover changes at a higher resolution. Lessons learned from the separate rotation local PCA technique is used to analyse forest cover changes in an area in Sub-Saharan Africa.

The paper considers the application of artificial neural networks for drought monitoring using Sentinel-1 and Sentinel-2 satellites’ data in the South of Ukraine. From the data collected from several weather stations equipped with suction pressure measuring Watermark sensors in 2017-2018, we formed 5 datasets containing suction pressure; NDVI and NDWI values calculated using Sentinel-2 images; digital number values of co and cross-polarized radar data obtained from the Sentinel-1 images. Four datasets were used to train a neural network, and the fifth one - for accuracy checking. We used a multilayer perceptron neural network to detect the dependencies between suction pressure and normalized values of spectral indices combined with radar data. The best accuracy within the training datasets was obtained for the neural network with one neuron in one hidden layer and one neuron in the input layer with sigmoidal transfer function (maximal relative error of 28.0% or 17 kPa in absolute values). Estimates for the testing dataset described the actual data with an average relative error of 32.0%. The neural network better estimated lower levels of moisture content that is essential while predicting soil droughts. We also compared the results of neural network assessment with several spectral drought indices. As a proof of adequacy, for three images acquired in 2018, the estimates obtained by the neural network correlated with ln(TVX) (temperature vegetation index) with R=0.44-0.62. The neural network approach with the use of Sentinel imagery, however, allows evaluating soil moisture content with greater time resolution.

Precision Agriculture (PA) has a fundamental role in the sustainability of agricultural systems, supporting decisionmaking of agricultural crops, increasing yield and quality in production. In the present research a PA approach for viticulture was made combining remote sensing data and robotic monitoring. With this approach it was intended to perform a spatial-temporal analysis of the grapevine phenology, according the 3 periods of the grape’s biological cycle – reproductive cycle, peak of the season and vegetative dormancy - corresponding to the years of 2017/18, for a specific area of the Green Wine Region, from Celorico de Basto (Portugal). The proposed methodology is based in the automation of spatial analyses through Geographical Information Systems (GIS), Google Earth Engine (GEE) and Python programming language. GEE was used for image acquisition and processing data of several indices, as Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI) and Visible Atmospherically Resistant Index (VARI). Regarding the geoprocessing of environmental factors, it was considered the following parameters: precipitation, temperature and soil moisture. Afterwards, NDVI was selected for a space-time analysis of the vineyard phenology, once this index represents a close dynamic to the vineyard biological cycle. From the relation between environmental factors and NDVI it was possible to interpret the space-time dynamics of the vineyard phenology. Finally, a spatial interpolation of yield and NDVI was made to understand the influence of NDVI in the yield. It can be assumed that the NDVI does not have a statistically significant influence on vineyard yield.

Farmland segmentation is crucial and getting increasingly important role in the field of agricultural insurance and digital agriculture. To accurately achieve insured crop area and disaster loss assessment, precision smallholders’ farmland segmentation and mapping are necessary. Deep learning technology has demonstrated its strength and has out-performed state-of-the-art alternatives in many fields. In this study, we aim to explore the effectiveness of five classic deep semantic segmentation models on the smallholder-wise farmland segmentation from remote sensing images. Five FCN-based segmentation models were selected including U-Net, PSPNet, DeepLabV3+, DANet, and CCNet, which were originally proposed for natural or medical image segmentation. The study area locates in Jiaxiang County in the north China. We used GF-1 at 2m spatial resolution with 4-band multispectral images (red, green, blue, and near-infrared) acquired based on image fusion. Comparative experiment results showed that DeepLabv3+ achieved the best mIoU of 89.82%. PSPNet, DANet, CCNet and U-Net obtained lower but similar mIoU with: 89.63%, 89.66%, 89.59%, and 89.15%. The OA metric of the all the five models were above 94.8%. The results indicate that the FCN-based deep semantic segmentation networks are effective for larger-scale smallholder-wise farmland segmentation with high spatial resolution multispectral satellite images.

Satellite imagery contains valuable large-scale information for precision farming. However, the low-resolution of satellite images can make it challenging to extract crop status information due to mixed pixels, in particular within multi-species crop stands like grass-clover for silage. In contrast, proximal high-resolution images with centimeter to sub-millimeter scale contain cures for single crop species pixels. However, these are often sparsely sampled due to computational limitations. In this paper, we present a preliminary attempt to enrich multispectral satellite images with crop stand population intelligence extracted from sparsely proximal RGB samples. The system attempts to reinforce satellite imagery based on proximal indicators lowering the risk of faulty interpretation knowledge base for future farm management information system (FMIS). A semantic segmentation algorithm is utilized to find the ratio of grass, clover, and soil across proximal images. Sentinel-2 as satellite imagery is employed as the 10-meter ground sampling distance input of the system and the grass, clover, and soil ratios are the output gained simultaneously. The system includes 1) a method where the proximal images and satellite imagery are preprocessed and then aligned with each other; and 2) a non-linear Multi-Layer Perceptron (MLP) extracting grass, clover, and soil ratio. Estimation results present promising correlation between clover, grass, soil, and Sentinel-2. Although, more data with higher diversity of clover-grass mixture is required to confirm the distinction of clover and grass.

Quantifying wheat’s production is essential to support food security management. It can be achieved with empirical models developed with the information provided by vegetation indices (VI). This work evaluated the performance of different time series of VI for the predictive modelling of wheat production and yield in Spain comparing two sources of cropland masks: wheat mask using Common Agricultural Policy declarations (CAP), and arable land from Corine Land Cover (CLC). Both sources were used to analyse the improvement derived from considering specific wheat masks. The wheat production and yield were modelled using time series of MODIS NDVI and EVI2 (2006 to 2016) from weekly surface reflectance products (MOD09Q1 v6) at 250 meters. The sum of VI values of one month after the maximum was used as this period is related with yield and production. VI indicators were filtered and aggregated to NUTS-3 level. The cropland masks were obtained either by combining the parcel boundaries with the CAP wheat reports, or from the CLC arable land category of 2006 and 2012 maps. Production (t) and yield (t ha-1) estimates were obtained from official statistics. Subsequently, different regression analyses were carried to build an overall model and single models for some NUTS2. Models using CAP wheat masks outperformed those of CLC, predicting more accurately production than yield. The best performance for production models using CAP was that of EVI2 in Castille and Leon (R²=96% and Normalized Relative Error (NRE)=14.72%) and the best for CLC that of EVI2 in Spain (R²=55% and NRE=58.01%). Regarding yield modelling, CAP with EVI2 in Aragon was the best (R²=81% and NRE=10.57%) as well as CLC with EVI2 in Spain overall model (R²=50% and NRE=22.34%). The findings of this work suggest that the use of specific crop masks is of paramount importance for the predictive modeling of crop production.

The EU-funded DIONE project (grant agreement No. 870378) offers an innovative close-to-market (TRL7) solution seeking to improve the traditional methods of agricultural monitoring. The project introduces a cloud-based Software as a Service (SaaS) system architecture, building on a fusion of novel technologies that will support the forthcoming needs of the modernized Common Agriculture Policy (CAP) and the “Greening” perspectives, with an automated area-based monitoring system. In particular, an interoperable and harmonized system is designed, connecting large volumes of Earth Observation data (Satellite, UAV, and in-situ) and user-generated highly precise geolocated data (geo-tagged photos, soil measurements, etc.). DIONE’s system architecture encompasses customized and third-party frameworks, where heterogeneous and multi-source data are stored, processed and managed using Artificial Intelligence (AI) algorithms. These harmonized, curated and open accessed data are then provided as Open Geospatial Consortium (OGC)-compliant, web-service layers (WMS, WFS, and WCS). Furthermore, the proposed solution formulates a scalable, flexible, interoperable, and semantically enriched environment, taking advantage of a Spatial Data Infrastructure (SDI) framework capabilities, whilst allowing an interactive connection among different tools and components through RESTful APIs. Our approach establishes a novel, cloud-based, accurate and inexpensive agriculture monitoring solution, enabling the real-time provision of multi-source data to relevant stakeholders such as Paying Agencies, Policy Officers and Control and Certification Bodies, and other domain experts. The system architecture was formulated exploiting a codesign methodology, aiming to ensure a long-term and sustainable solution. Two large scale demonstration will take place in Lithuania and Cyprus, evaluating the system capabilities in real-life and operational conditions.

Flash flood disasters frequently occur in Dir Lower, causing heavy losses and diluting the capability of sustainable development of the local economy. Flash floods are one of the natural disasters which harm the life and infrastructure. These types of floods come due to high rainfall. In the present study flash flood were assessed in District Dir lower watersheds, Khyber Pakhtunkhwa Pakistan. Assessment was conducted using remote sensing and GIS techniques. Two approaches were used to assess the flooding risk for five watersheds of Dir lower: The morphometric ranking approaches and The El-shamy’s approaches. Both approaches utilize the morphometric parameters for the flash flood susceptibility. The results achieved based on two approaches which identify the sub watersheds with High flash flood susceptibility, Moderate flash susceptibility, and low flash flood susceptibility. Both approaches have the same result in the study area which indicates that in sub watershed B1, B2, and B3 have High flash flood susceptibility which is 60% of total watersheds, B5 have moderate, and B4 have low flash flood susceptibility. The adaption of methodology can apply to estimate the flash flood susceptibility in Dir Lower.

Spire Global operates the world’s largest and rapidly growing constellation of CubeSats performing GNSS based science and Earth observation. The Spire constellation, performs a variety of GNSS science, including radio occultation (GNSS-RO), ionosphere and space weather measurements, and precise orbit determination. In December 2019, Spire launched two new satellites to perform GNSS reflectometry (GNSS-R), with plans for two more GNSS-R satellites to be launched in mid-2020. Due to its agility and rapid launch cycle, averaging launches of four to eight satellites every six weeks, Spire has the unique ability to improve performance and add capabilities on-orbit that are impossible with traditional, risk-averse satellite missions. GNSS-R is a relatively new technique based on a passive bistatic radar system. The potential of space-borne GNSS-R observations for ocean and land applications has been demonstrated by other GNSS-R missions, including the NASA Cyclone Global Navigation Satellite System (CYGNSS) and the UK’s Technology Demonstration Satellite, TechDemoSat (TDS-1). We present initial results from these new Spire GNSS-R satellites that are primarily focused on retrieving soil moisture but also estimate other Earth surface properties such as ocean wind speeds and flood inundation/wetland mapping. Prior to the launch of Spire’s GNSS-R satellites and in preparation for Level-2 data production, we developed algorithms and processing chains for land applications. We will present Spire's soil moisture retrieval method using CYGNSS observations. We evaluated the implemented soil moisture change detection algorithm by comparing the Spire’s daily soil moisture product with NASA’s Soil Moisture Active Passive (SMAP) observations and in-situ soil moisture measurements. The results of study indicate remarkable retrieval skills of the GNSS-R technique for soil moisture monitoring at a medium spatial resolution. Spire’s GNSS-R satellites are tuned for land applications with a series of hardware and software optimizations for better signal calibration and acquiring many more data per satellite compared to CYGNSS. A more robust GNSS-R soil moisture retrieval at finer spatial resolution will be possible in the near future after having more Spire satellites in orbit. Spire’s current and future GNSS-R satellites will provide unprecedented sub-daily global coverage and fine spatial resolution. Such intensive data acquisition is of great importance for many land and ocean applications.

Estimating actual crop evapotranspiration (ET) is a critical component in tracking crop water availability and managing irrigation. Various methods presently exist for modeling crop ET utilizing remotely sensed data and imagery as inputs. While more traditional remote sensing platforms like satellite and manned aircraft provide useful information, limitations exist due to lack of spatiotemporal resolution and high cost. Unmanned aerial systems (UAS) now offer greater opportunities for collecting remotely sensed data with enhanced spatiotemporal resolution that may lead to increased accuracy in estimating crop (ET), making near real-time irrigation management more feasible. The two-source energy balance (TSEB) model is one of the methods of estimating crop ET from surface energy balance fluxes. In this research, multispectral and thermal infrared imagery collected with a UAS over maize and soybean fields throughout the growing season at different vegetative stages were used in the TSEB model to estimate spatially distributed surface energy balance fluxes and daily crop ET. The estimated spatially distributed surface fluxes and ET were weighted and aggregated using a two-dimensional flux footprint and validated against measured fluxes from Eddy Covariance systems located within the fields.

This study investigates the use of Sentinel-2A (S2A) and Landsat-8 (L8) OLI for monitoring of turbidity in reservoir waters. Using observed in situ data from 18 sampling stations for Chebara Reservoir in Kenya, the study developed an empirical multivariate regression model for turbidity estimation from atmospherically corrected, band adjusted and spectral resolution standardized S2A and L8 bands. Best results for turbidity estimation were obtained from the regression of in situ data with B2 (blue) and B3 (green) bands as [Rrs(B2/B3)^2+Rrs((B2/B3)] for S2A and [Rrs((B3/B2)] for L8. Both S2A and L8 retrieved turbidity with high and nearly equal accuracy of R^2 < 0.75 from the visible and NIR bands, with nearly similar RMSE of 0.5 NTU and NMAE% being higher for S2A by more than 30% as compared to L8’s average NMAE% of 15%. The study shows that for both S2A and L8 sensors, and the proposed empirical regression algorithm suffices in the rapid and cost-effective quantification of turbidity inland reservoir waters. Using spatial interpolation for the visualization of the correlation between the predicted and observed turbidity, the L8 results were found to be more significant than the turbidity estimations using S2A bands.

Vegetation phenology, the study of the timing of biological cycles of plants and their relation with environmental factors, is considered an important ecological indicator of climate change. Different ecological processes, such as water and carbon cycles, energy fluxes or species interactions, can be altered by changes in the phenological patterns of plants. Furthermore, these changes could also have important consequences on economy (e. g. agriculture, forestry). Iberian Peninsula is one of the regions with the greatest diversity of ecosystems in European continent. It is therefore an excellent natural laboratory for monitoring vegetation dynamics. The goal of this study was to monitor the vegetation phenology dynamics across ecoregions of the Iberian Peninsula for the period between 2001 and 2017. 782 8-day composites of NDVI (Normalised Difference Vegetation Index) images were produced from the surface reflectance product MOD09Q1 at a spatial resolution of 250 meters. A Savitzky-Golay filter was applied to smooth the NDVI time series and a threshold-based method was used to extract three phenometrics: the start of the growing season (SOS), the end of growing season (EOS) and the length of the growing season (LOS). Results of this research showed how both SOS and EOS are significantly different between the northern and southern ecoregions. Cantabrian mixed forests and Pyrenees conifer and mixed forests presented the latest SOS and EOS. Iberian conifer forests, Northwest Iberian montane forests and Northeastern Spain and Southern France Mediterranean forests showed the highest internal variability in the phenological dates, which may be related with the behaviour of different land covers (e. g., phenology of natural vegetation vs. crop phenology) and the altitude effects on climatic conditions (e. g. increases in precipitation, decrease of the temperature). Phenological patterns of the Iberian ecoregions could contribute to improve the understanding of the potential environmental effects of climate change.

Vegetation stress is a major widespread problem in many countries because of climate change and anthropogenic activities. Deforestation and forest degradation phenomena may be caused for several reasons such as infrastructure development, agriculture, collection of wood energy, forest exploitation. Over the last decade, a severe decline in vegetation was observed in the Congo Basin rainforest, the second-largest tropical forest in the world, behind the Amazon. Therefore, actions are required to monitor and detect vegetation stresses to mitigate their negative impacts on human life, wildlife, and plant communities. Vegetation stress can be estimated using three different methods: field measurements, meteorological data, and remote sensing. The present study is mainly focused on satellite remote sensing. The main objective is to develop and test new indices of vegetation-soil dryness based on the surface emissivity. Until now, the problem has been attacked through indices such as the normalized differential vegetation index (or NDVI). The problem of NDVI is that it is a greenness index and is not capable to distinguish bare soil from senescent vegetation, whereas this distinction is important especially when forest degradation followed by eventual regeneration occurs and when dealing with semi-arid regions, where we could have desert sand. We propose to follow the strategy of using surface emissivity (ε), which is more closely related to surface type and coverage. By properly using surface emissivity in the infrared we can define a set of channels that are particularly sensitive to bare soil, green, and senescent vegetation. From these emissivity channels, we can derive a suitable emissivity contrast index or ECI, which is sensitive to green vegetation, senescent vegetation, and bare soil, therefore overcoming the NDVI limitation concerning its capability to distinguish bares soil from senescent vegetation. The analysis is performed with CAMEL (Combined ASTER and MODIS Emissivity for Land) database from 2000 to 2016.

Phosphorus (P) is an important parameter participated in the process of metabolism, photosynthesis, and energy exchange of crops. A growing number of studies have focused on effects of arbuscular mycorrhizal fungi (AMF) inoculation on crop P uptake. In this context, efficient and nondestructive monitoring of the changes of leaf P content (LPC) in inoculated crops is of vital. In this study, hyperspectral remote sensing was explored in an attempt to diagnose P deficiency in the inoculated and non-inoculated soybean plants. Greenhouse pot experiment was conducted under drought stresses, and measurements of leaf spectral reflectance and LPC were carried out at the 30th, 45th and 64th days after inoculation. We transformed the raw spectral reflectance (R) into the first derivation (FD), second derivation (SD), reciprocal (1/R), reciprocal logarithm (log(1/R)) and first derivation of log(1/R) (log’(1/R)). Results indicated that the AMF-inoculated plant had significantly higher LPC than the counterparts under different drought stresses. Analysis of the correlation between LPC and the raw and five transformed reflectance in the 350-2500 nm spectral range indicated that the green bands center around 545 nm and 567 nm, as well as NIR band center around 832 nm were the most sensitive (r>0.73). The kernel ridge regression (KRR) of LPC with the sensitive bands selected from the raw/ transformed reflectance was performed, showing that FD, 1/R and log(1/R) produced excellent results in LPC assessment, with determination of coefficient (R²) all larger than 0.70. Validation with independent samples revealed that the log(1/R)-KRR model achieved the strongest and superior prediction accuracy, with R2 of 0.93, RMSE of 0.23 g/kg and RRMSE of 7.8%, respectively. Our results indicate that the log(1/R)-KRR derived from hypersectral remote sensing data can provide the most suitable estimation model for describing the dynamic changes of LPC in the AMF-inoculated soybean.

Fires are a widespread phenomenon in Indian forests. Limited fore-knowledge of the impending ignition potential is one of the main causes for not-so-effective forest-fire management in the country. The present study is an approach to link the forest fire trends and the Keetch–Byram Drought Index (KBDI) for assessing forest fire forewarning. Fire trends were analysed based on forest fire incidents detected by MODIS remotely-sensed data in the Shivalik Forest tracts of Uttarakhand State of India over a period of 16 years (2001-2016) to identify vulnerable window of most-prone weeks. The study has also used KBDI to calculate the drought build-up in an area based on which severity of impending forest fires can be predicted. The index was tested in the study area for the years 2014-2016. This is one-of the first attempts for inclusion of KBDI in understanding forest fire regime in Indian conditions. Based on the weekly-trends obtained from forest fire points during 2001-2016 affirm the most vulnerable time period for occurrence of forest fire in the study region is from week 14 to week 22. Further, the performance of KBDI proved to be a valid parameter in forecasting severity of forest fire as its pattern over the 3-years of study period (2014-2016) showed the same dynamics. The minimum value of KBDI i.e. ‘dip’ value is significant for its use as a predictor to severity of the fire season in the ensuing summer a full 10 weeks in advance! The findings of the present study pave the path for identifying the high risk areas and severity of possible fires in the following fire season by correlating remote sensing and GIS techniques with Fire trends, peak period and Keetch Byram Drought Index. This study has a potential for adoption in countries like India for modern forest fire preparedness and management in a cost-effective manner.

Meteorological modeling takes data captured from multiple sources that is then processed by data mining techniques to predict environmental changes. The most commonly used machine learning techniques for processing meteorological data are decision trees, rule-based methods, neural networks, naive Bayes, Bayesian belief networks, and support vector machines. These techniques require accurate data for effective models to be simulated. Meteorological datasets can contain outliers and errors that can significantly skew the accuracy of the generated models that are relied upon for many sectors of society including agriculture, natural disasters, and meteorological forecasting. This paper proposes a method to eliminate outliers from meteorological data to enhance the accuracy of models by applying a blind thresholding algorithm to the principal components (PCs) obtained from L₁ and L₂ norm Principal Component Analysis to identify and discard outliers in the dataset.

Natura 2000 is the largest network of habitat and biodiversity protection areas. However, it remains unclear how strongly or to what extent the human activities impact the areas or what kind of activities are conducted within. Our objective is to analyze the human impact in all Natura 2000 sites, based on satellite data. We have developed an algorithm in Google Earth Engine using JavaScript, for satellite data temporal analysis of protected areas, the analysis using two datasets. First, we used Landsat archive to derive the biophysical composition index (BCI), allowing to evaluate both impervious surface (positive and relatively high values) and vegetation (negative and low values) abundance. BCI was derived for each year from Landsat scenes acquired in August, from 2000 to 2019. Mean BCI value in each site was plotted to calculate the slope of linear trend line, highlighting both the magnitude and sign of human activities changes per year. Secondly, we calculated the mean value within each site from the Human Modification dataset (gHM) that provides a cumulative measure of human modification of terrestrial lands globally at 1 km2 resolution, in 2016. The results showed an average increase in BCI values of 0.011 per year for the 24262 sites, suggesting a slight increase of impervious surfaces and thus human activities. Only Malta, Belgium, Hungary and Cyprus out of 28 countries recorded a decrease of human activities within sites in the last 20 years. Nine countries recorded a very slight increase (<0.01 per year) and 13 countries recorded increase in BCI value between 0.01 and 0.02 per year. Czech Republic and Estonia registered the highest increase in human activities within sites, with an average of 0.022 and 0.039 per year respectively. The gHM dataset analysis showed an average value of 0.42 for all Natura 2000 sites, suggesting a moderate to high modified land in 2016. The lowest values were recorded in Finland (0.24), Latvia (0.25) and Estonia (0.26), being among the nine countries that recorded moderate modified land within sites (0.1-0.4). The rest of the 19 countries recorded high modified land in Natura 2000 areas with values between 0.4 and 0.7, the highest in Malta (0.68), Luxembourg (0.63), Netherlands (0.56) and Czech Republic (0.55). These results suggest that within Natura 2000 sites, the regulations regarding human activities are too loose or not fulfilled and should be enforced in order to enhance biodiversity protection.

Europe is covered by distinct climatic zones which include semiarid, the Mediterranean, humid subtropical, marine, humid continental, subarctic, and highland climates. Land use and land cover change have been well documented in the past 200 years across Europe where land cover grassland and cropland together make up 39%. In recent years, the agricultural sector has been affected by abnormal weather events. Climate change will continue to change weather patterns, for some regions, this will have a positive impact. For other regions, it will be negative. Hail, heavy rainfall, floods, and droughts may lead to a reduction in agricultural land in some regions with an additional impact on biodiversity. For that reason, major climatic and hydrological variables such as temperature, precipitation, vegetation index, and soil moisture must be regularly monitored and the trends of their deviations from the normal must be observed. This research investigates the quantity of potato production in the future in two stages. In the first stage, Autoregressive Integrated Moving Average (ARIMA) models are used for yield and soil moisture forecasting. The models are trained on a dataset acquired in time series from the Soil Moisture and Ocean Salinity (SMOS) satellite from 2010 to 2019 and are compared by statistics on potato production from EUROSTAT at the national scale. In the second stage, the approach uses Support Vector Machine (SVM) architecture of Sentinel 2 Vegetation Index satellite imagery together with weather and Sentinel 1 surface soil moisture information for the EU-28 member countries by season 2018 for single potato fields. Agricultural statistical data such as crop calendar, map, production area, yield, and acreage are collected from Eurostat while vector data collection from Land Use and Land Cover Survey (LUCAS) and Land Cover from national to small scale agriculture (CORINE). The ARIMA output performances measure Mean Squared Error (MSE) and Root Mean Squared Error (RMSE). RMSE for soil moisture forecasting is 0.3792, 0.2816, 0.2679, 0.2110, and 0.1791 for East, West, South, North, and Centre EU respectively. MSE for potato yield forecasting is 0.0954, 0.081, 0.0589, 0.0447, and 0.0199 for North, Centre, East, South, and West EU respectively. The number of known crop field areas and the distance between them are different for each country and this alters the dimension of the classification input. Two countries of EU-28 do not contain information on potato location while the input dimensions of another two are too big to perform. Therefore, the sensitivity analysis of SVM has been implemented for a total of 24 states. The classification accuracies for kernel Radial Basis function are 98.53% in Austria, 99.39% in Centre of UK, 99.64% in Italia, 99.12% in Bulgaria and Greece, 99.43% in Sweden for Centre, West, South, East, and North EU, respectively, with the Sentinel Vegetation Index. The classification accuracies change with the amount of ground-truth data. The results show that some geographical and socio-economic biases like the chemical composition of the soil or the degree of automation in the agricultural sector are influencing in the statistical analysis.

Fast and reliable tests for the new coronavirus are urgently needed. Current Polymerase Chain Reaction based virus detection approaches are typically time-consuming and expensive. Technologies capable of providing a fast, real-time and non-contact detection of virus contamination and real-time virus classification are not yet available. Here, we demonstrate the potential of a fluorescence detection technique along with machine-learning based classification for virus detection. The ultraviolet (UV) light irradiated virus emits a fluorescent signal with a characteristic spectrum, which is regarded as a fingerprint for the virus. We analyzed eight virus samples including a heat-inactivated SARS-CoV-2 (virus causing COVID-19) and collected a number of emission spectra. Machine learning techniques are applied to discriminate among the candidate viruses via classifying a number of spectra data collected. First, Principle Component Analysis (PCA) was applied to reduce spectra data dimensionality. Then support vector machine (SVM) with various kernel functions (kernelSVM), k-nearest-neighbor (k-NN) and Artificial Neural Networks (ANN) methods were used to classify these viruses with dimension-reduced data from PCA. We found that dimension-reduced data in 3 principal components (PCs) space performs better than that in 2 PCs space in the machine learning algorithms mentioned above. Variance ratio analysis is able to explain nearly 95% of variance which allows nearly 100% accuracy of predictions for 25% data test set randomly chosen from the whole dataset. Finally, cross validation (CV) analysis is applied to kernel-SVM and k-NN methods.

The features of location and functioning of wind–powered drive of agricultural airplane special equipment. A mathematical model of a wind propeller in non–uniform wind stream is developed. Software package to comprehensively assess wind–powered drive conditions and performance indicators is generated and tested taking into account experimental data, as well as airplane performance and flight modes. The main regularities, qualitative and quantitative features of the wind drive application in aviation operations and proposals for its future improvement and use, as well as for ensuring the quality of aviation spraying of agricultural aircraft with a wind drive are formulated.

This article presents a novel methodology for monitoring and assessment of dynamics of phenological development of crop vegetation, based on orthogonal transformation of multispectral satellite images from Sentinel-2. In this work two indices using the Greenness and Wetness components of the Tasselled Cap Transformation (TCT) model are applied. The essence of the TCT model is the unitary matrix for orthogonal transformation, which is fixed for each individual sensor. The presented methodology implements a model for automated processing of Sentinel-2 images, aimed at obtaining Greenness and Wetness Tasselled Cap indices. The applied indices are defined as Normalized Differential Greenness Index (NDGI) and Normalized Differential Wetness Index (NDWNI). These indices are strongly sensitive to small changes and reflect quantitatively the vegetation dynamics in defined temporal periods. NDGI and NDWNI are especially suitable for determination of timing of different growth stages and for monitoring of the health status of the studied crops during these stages. The manifestation of the indices and their dynamics is closely related to variations of chlorophyll and moisture content, which are observed during the transition from one stage to another.

In countries with insufficient infrastructure related to water, it is difficult to identify hydro-meteorological characteristics due to insufficient measurement data. Rainfall data derived from satellites can be an alternative for insufficient hydro-meteorological measurement data. The purpose of this study is to develop a program that estimates the approximate inundation range using rainfall data derived from satellites and global topographic data

The Mekong River is the 12th longest river in the world and is an important river called the lifeline of Southeast Asia. The Mekong River and its tributaries usually rain heavily in August. This causes the water level to rise sharply and floods begin to fill the vast floodplain. The purpose of this study is to analyze the amount of flooding in the Mekong River basin using precipitation data calculated by satellites, and to estimate the flood area.

Precipitation plays an important role in understanding the mechanisms and interactions of Earth's water and energy balance. It is the main input data for hydro-meteorological models and climate research. In this study, a study was conducted to evaluate the accuracy of satellite rainfall by comparing GPM (IMERG) satellite rainfall data with ASOS ground rainfall data of the Korea Meteorological Administration. The study area was conducted by dividing the four major river basins in Korea.

Remote sensing of the Earth (RSE) is one of the main objective sources of information about the earth's surface. With the development of unmanned aerial vehicles (UAVs), it became possible to take aerial photos with high spatial resolution, which can more accurately identify objects. But due to the fact that the mass use of UAVs for remote sensing of the Earth has become relatively recent, there are no ready-made solutions for automated processing of UAV images. The purpose of the study is to increase the reliability of interpretation of UAV images by developing a method of automated processing based on conceptual modeling. Analysis of methods for thematic interpretation of UAV images showed that none of them provides sufficient segmentation quality without additional adjustment to the subject area. It was found that a combination of methods will improve the result of interpretation. When developing the method of automated processing of UAV images and its software implementation, the method of conceptual modeling of subject problems was used, which ensures the adequacy of syntactic representations (including various images), allows you to control the logic of solving problems and reduces the number of errors at the stage of its software implementation. Using the error matrix and the formula for calculating the Kappa Cohen index, the reliability of thematic interpretation of images of forest areas was assessed. 59 (52.2%) of the 113 trees shown in the picture were correctly identified by the standard watershed method, and 80 (70.8%) - by the developed method. Thus, the developed method made it possible to improve the identification of forest objects in UAV images by 18.6%. In the future, the development of this method can be carried out to determine the characteristics of the identified trees: age, species, height, timber stock.

Along with the growth of economic activity in the northern regions of Russia, there is an increasing need to assess their impact on the ecosystems of the northern territories. Due to the severe climatic conditions and low recovery capacity of northern ecosystems, it is important to quickly identify and respond to emerging negative processes. When monitoring terrestrial ecosystems, a key role is played by the availability of up-to-date, constantly updated information about the state of the northern ecosystem. One of the main sources of operational information is ultra-high spatial resolution images from unmanned aerial vehicles. Intelligent technologies and neural networks, in particular, are used to solve various point problems of thematic interpretation of spatial data on images, but are not used in combination because of the lack of elaboration of the mechanism for linking them together, and are also practically not focused on the problems of territory management. The article describes a formalized approach to combining a pool of intelligent technologies at two levels for the regular receipt of up-to-date information about ecosystem objects that enters the dynamic monitoring and management system of the northern territories. Within this approach, when obtaining information about the state of natural resources for monitoring northern ecosystems, the conceptual apparatus of different subject areas are taken into account: the area of territorial management tasks; natural resources of the ecosystem; remote sensing of the Earth. It is proposed to develop a mechanism for linking concepts from subject areas, their characteristics and dependencies, and building a semantic model, its formal description. The implementation of the process of obtaining information based on combining a sequence of neural networks of such modern architectures as U-Net and Res-Net is considered in this paper. The proposed approach provides a unified interaction of information manipulation processes based on various intelligent technologies for the regular receipt of up-to-date information about the state of natural resources in the framework of a dynamic monitoring and management system for the northern territories.

The main objective of GrasSat project is a fully operational system in form of desktop and mobile application, which provides a complementary tool for managing grassland production, mainly for medium and large farms in Poland and Norway. Combining the effectiveness of the application with the support of external advisors is the key to improve grass production management. Experience of the team of remote sensing and grassland specialists will be the firm foundation of the tools to be prepared within the project.

Wetlands systems are at the risk to changes in hydrological regime under climatic variation. Wetland habitat is sensitive and responses quick to climate changes. Wetlands cover 6% of the world’s land surface and contain about 12% of the global carbon pool, play an important role in the global carbon cycle (International Panel on Climate Change (IPCC); In a time of global climate change, wetlands conditions are not known well enough. That’s why the research that we have been carried out since couple of years which are based on satellite data and in-situ data are important for depicting the changes in hydrological and vegetation parameters. Biebrza Wetlands are our main area of research, because they are: • one of the largest area in Europe covered with marshes, swamps, and wet meadows • 60 000 ha of flat river valley covered with hydrogenic soils such as peat in various stages of mouldering • habitat of 271 bird species • protected as a National Park, Natura 2000 and RAMSAR sites The results are also connected to serve as an indicator of UN Sustainable Development Goal 6.6.1 - Change in the extent of water-related ecosystems over time.