This PDF file contains the front matter associated with SPIE Proceedings Volume 12727, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Insect populations are declining globally, making systematic monitoring essential for conservation. Most classical methods involve death traps and counter insect conservation. This paper presents a multisensor approach that uses AI-based data fusion for insect classification. The system is designed as low-cost setup and consists of a camera module and an optical wing beat sensor as well as environmental sensors to measure temperature, irradiance or daytime as prior information. The system has been tested in the laboratory and in the field. First tests on a small very unbalanced data set with 7 species show promising results for species classification. The multisensor system will support biodiversity and agriculture studies.

The temperature rise due to climate change is expected to increase the frequency of droughts. These droughts not only pose a significant threat to agriculture and food security, but also increase the risk of additional disasters such as wildfires. Therefore, accurate drought monitoring and prediction are crucial. Predicted drought information can help strengthen policies related to agriculture and water management and prepare for drought response. Generally, the surface drought condition is monitored through satellite-based drought indices. Among various drought indices, the vegetation health index (VHI) which comprehensively combines temperature and vegetation status, is mainly used. In this study, we propose a model for predicting VHI time series data using Convolutional Long Short-Term Memory (ConvLSTM). ConvLSTM is a model that combines Convolutional Neural Network (CNN) and LSTM and can learn both temporal and spatial characteristics of time series data while preserving its spatial features. Therefore, it is being used in various fields such as image and video processing, and weather forecasting, where local features need to be considered. The study area is South Korea, and long-term weekly VHI data provided by NOAA were used for short-term prediction. The proposed model can be useful for drought prediction considering local features.

Remarkably fast enhancement of machine learning for estimating spatiotemporal dataset suggest new ways to remote sensing and atmospheric fields in using data-driven modeling. Especially, many kinds of optimizable model structure for spatio-temporal database such as LSTM (Long Short Term Memory) or token-mixing based methodology have been suggested and replaced ‘state of the art’ models in representing numerical phenomenon over the earth. In addition to this, vast amount of observation database has been reanalyzed and assimilated as large amount of atmospheric variables such as EAR5 or MERRA-2 which can cover densely along spatial and temporal aspect. In this research, we compare LSTM based model (PredRNN) with token-mixing based model(AFNO) in forecasting precipitation as quantitative guideline for application of newly suggested machine learning model. 8 different atmospheric variables (wind components, temperature, relative humidity, water vapor and geopotential, total precipitation) over 4 pressure level (500, 750, 850, 1000) from ERA5 and RAR(Radar-AWS Rainrates) database from KMA (Korea Meteorological Administration) is adopted as input and output data to train and evaluate models.

Crop type information is crucial for several agricultural applications and even useful for different agencies to take important decision. Therefore, accurate crop mapping is required which is a simple yet critical issue in agriculture. Remote sensing has played an important role in acquiring necessary crop data. The free available Planet Scope dataset with a spatial resolution of 3m has generated new opportunities for mapping small land holding. The objective of the study is to map a Mustard field in Haridwar district, as it is considered as one of the important catch crop in the study area. Random Forest (RF) and Classification and Regression Tree (CART) algorithms of machine learning have been in this study. Further, spectral indices images of NDVI, EVI2, NDRE1 and BBI were generated from the original data set. On the temporal PlanetScope dataset, separability analysis is first carried out using the transformed divergence approach. This gives us the optimal three band combination and best time stamp for mapping mustard, which is then used in the study together with spectral indices for mapping mustard crop. Hyperparameter tuning is done to achieve high accuracy, and utilizing the optimized value of the parameters, the classification is carried out using the aforementioned powerful algorithms. The classification findings demonstrate that RF (85.78%) offers a more accurate result than CART (77.75%) in terms of total accuracy. However, both classifiers offer approximately same result in the field of agricultural. For example, RF classifies mustard with an accuracy of 93.33% while CART achieves 90.69%, while for mapping other crops, RF achieves 91% accuracy and CART achieves 84.67%. However, RF provides more precise mapping for Mustard than CART does. According to the results of the study, Random Forest produces the best outcomes when original data and spectral indices are combined.

In the year 2012 the Ministry of environment and waters (MOEW) in Bulgaria initiate the development of “Methodology for defining the adjoining lands and flooding rivers stripes in Bulgaria” based on Water Directive 2000/60/ЕС requirements and its transposing into national law. The methodology should be applied in cases when governmental bodies issue permits for use and water use of a water body upon art. 46 of the Water Act in Bulgaria. The methodology can be used also in the development of Flooding Risk Management Plans. The research investigates the possibilities of remote sensing as a tool for the verification of initial data for generating DEM and verification of data for the bedding surface of the studied watershed to the place of use or water use from the river section. The results show that remote sensing data from Sentinel-2 contribute for the timeliness of the data that have been used in different stages of the Methodology. The correlation among components water, vegetation and soil have been examined by estimating the indices NDVI, NDWI, SAVI, MSAVI 2. The interpretation of the analysis is useful when calculate the maximum runoff formed in the studied watershed. The probability of an increase in the water levels of the studied river section and flooding of the lands belonging to it can be predicted by applying the methodology.

The study aims to detect ground standing water in cropland during the spring/early summer season in eastern South Dakota, USA. The goal is to develop a reliable and accurate method that can distinguish ground surface open water from vegetation, which is often mistakenly identified as water. To achieve this, the study utilized Sentinel-1 synthetic aperture radar (SAR) data due to its high reliability, short revisit time, and free availability. A total of 159 sites were selected and surveyed, including 78 water sites and 81 non-water sites, located between Brookings, SD and Sioux Falls, SD, USA. The SAR data were preprocessed at both VV and VH polarizations for both water and non-water sites. In previous work, we used maximum likelihood estimation (MLE) of the density functions with a shifted Rayleigh distribution. In this paper, a Neyman-Pearson test for SAR data classification is developed using the Rayleigh priors at the dual-polarization. The developed method demonstrates good performance in distinguishing between water and non-water sites, providing an alternative approach to ground water detection that is important for precision agriculture, hydrologic and environmental studies.

The country-wide monitoring service generates CBK-Floods product, which provides the current surface water cover over Poland. The automatic detection algorithm has been developed. It uses Copernicus Sentinel-1 radar satellite images as well as proposed improved model of valleys derived from fusion of various data sources (e.g., Copernicus Riparian Zones, LIDAR, flood hazard zones). The overall accuracy of the algorithm is around 86%. The map is updated after each pass of the satellite and shows different stages of inundation: new water extent, areas with long-lasting water and those from which water has receded in the last days. Two kinds of information are generated: (1) flood water extent; and (2) hydroperiod regime. Information about flood water extent is of critical importance for rescue and crisis management activities. Availability of recent water cover maps can support rapid situational assessment and influence decision processes taken in regional and local crisis management centers during flood. Information about hydroperiod regime allows the proper management of water needed for agriculture and can be an indicator of the state of ecosystems present in the valley. In 2022 service worked in pre-operational mode and produced a series of surface water maps for the entire Poland. In 2023 service will go into operational mode. The water extent maps will be available to visualize in the Sat4Envi Crisis Management Portal and downloadable from its repository. In this paper, we aim to present the data processing chain applied in the flood monitoring system, including the surface water detection method and the way of visualizing the final product. We present the limitations of the service based on satellite radar data and give examples of the use of flood products.

Estimating the accurate position of the shoreline over time can be challenging. We propose a new approach to delimit the shoreline from remotely sensed images simply and accurately. This approach treats pixels as simultaneous radiometric measurements, and by examining the distance between iso-radiometry lines, it assumes the shoreline in the position where the radiometric behaviour changes more suddenly. The analyses carried out applying this coastline extraction approach underscore its significant flexibility. Specifically, the approach yielded promising outcomes even if applied to a variety of different and dissimilar types of remotely sensed products. Indeed, the intrinsic straightforwardness and low computational load of the approach qualify it as a promising tool for time-series production.

Forest ecosystems are well known to provide microrefugia for many species in the context of climate change, due to their buffered and more stabilized microclimatic conditions compared with macroclimatic fluctuations outside forests. Forest structure is a major driver of microclimate resulting in spatial heterogeneity in managed forests. Spatializing the buffering effect of forest canopies on understory temperatures requires spatially contiguous predictors, which can be derived from remote sensing. ALS and satellite optical data have already proven their interest to map forest microclimates. However, no study has yet been tested the potential of SAR and especially Sentinel 1 data that allow intra- and inter-year spatiotemporal analyses of forest structure. The objective of this study is therefore to test the potential of SAR, combined with ALS and optical data, to map forest buffering capacity. We derived metrics describing both the vertical and horizontal heterogeneity of the canopy over the year 2021 using Sentinel 1 and 2 time series and ALS data that cover the French state forest of Blois. These metrics were then related to understory microclimate data collected across 52 plots during the leaf-on period in 2021. Model selection was performed to identify the best metrics to predict microclimate. Finally, the best model was used to map temperatures across the whole forest. Results showed that SAR-derived metrics performed well combined with ALS and optical data, with models explaining up to 90% of variance in microclimate effect. This study highlights the potential of SAR data and its complementarity, especially with optical data, to map forest thermal environment at fine spatiotemporal resolutions.

Operational deforestation detection for forest early warning systems (EWS) is a hot topic in Earth observation today. Due to the persistent cloud cover in tropical regions, active microwave is regarded as one promising technology for EWS. Despite significant progress in the last decade, a reliable, genuine deforestation EWS is still lacking, because the development of powerful algorithms demand a near perfect understanding of the forest backscatter nature. Building upon 9-years of ALOS-2 long-term pantropical forest observations with various breakthrough findings, we introduce the next-generation algorithm for L-band SAR deforestation detection which realize the first real EWS in the tropics with unmatched accuracy and speed.

The western Himalayan region is highly vulnerable to climate change due to its fragile ecosystem, complex topography, and high dependence on natural resources, which is expected to have significant impacts on its vegetation. In this study, we investigate the vulnerability of Western Himalayan vegetation to climate change using machine learning algorithms. We analyzed remote sensing data of the region to estimate temperature, precipitation, and other variables relevant to vegetation growth. We then used GIS-based open-source software and machine learning algorithms to study the variables significant for predicting vegetation vulnerability to climate change. The study results indicate that the Western Himalayan ecosystem is highly vulnerable to climate change, and the region is likely to experience significant changes in ecosystem vulnerability and resilience in the future. The study also highlights the importance of incorporating machine learning algorithms and GIS software in assessing the vulnerability of ecosystems to climate change.

We have to employ remote sensing data, such as Normalized Difference Vegetation Index (NDVI), from different satellite systems/programs for our multi-decade monitoring programs on ground. Calibration is essential to integrate different sourced NDVI data. NDVI is an essential layer for generating the Australian national Forest Productivity Index (FPI), which is a key component of the Australian Full Carbon Accounting Model (FullCAM). FullCAM is used to calculate Australia’s greenhouse gas emissions from the land sector. This study/project developed a correlation algorithm by using statistical regression analysis of long-term average monthly NDVI derived from a random points network over Australia, which aimed to calibrate 1km monthly MODIS/NDVI grids into AVHRR/NDVI-orientated NDVI layers to produce consistent Australian national Forest Productivity Index (FPI) products without rewriting the complex FPI model. The calibrated NDVI layers have been used to (1) accurately reproduce national FPI from 2001 to 2019. The national 1km monthly and annual FPI grids calculated from calibrated NDVI well replicated the national spatial pattern and conserved spatial details locally compared to those produced using AVHRR/NDVI; (2) generate consistent FPI products since 2020. This study provides a practically useful reference to other similar NDVI applications.

The former battlefields of World War I (WWI) provide an interesting framework for studying the long-term impacts of ancient anthropogenic disturbances on current ecosystem functioning. The 47 map sheets of devastated regions at 1:50,000 scale edited in 1920 by the geographic service of French army locate the areas heavily damaged by trenches and bombing, the destructed cities, roads and forests inventoried at the end of WWI. As they stand, these scanned maps are not usable under a geographic information system (GIS). A protocol was implemented on 5 sheets to compare the effect of two transformation models (thin plate spline and polynomial order 3) and the number of ground control points on the quality of georeferencing. A second protocol based on morphological operators, color space transformation and K-means clustering classification was tested on 12 different map sheets to extract areas heavily damaged and figures of punctual destructions. Neither significant effects of transformation model or number of ground control points were confirmed. The local thin plate spline method exacerbates non-natural local distortions linked to the research of ground control points and the simplification of physical objects on the maps. With polynomial order 3 transformation and 50 ground control points, residuals vary from 35 to 70 m depending on the map. The second protocol extracted accurately data of interest, with an accuracy varying between 0.31 and 100% depending on data to extract and their presence or absence on the map sheets. The resulting shapefiles are now available and workable in GIS.

Protein-rich soybean crops have a strategic importance for food production worldwide. Initiatives to increase their yield without compromising the environment include the use of remote sensing technologies to monitor their cultivation using spectral data. The red edge position (REP) is among the most used spectrally-derived information in this area. It is strongly correlated to the plants' chlorophyll contents and it can provide a reliable indication of changes in their nutrient status. Besides the availability of nutrients, the plants' photosynthetic capacity is also affected by other abiotic factors, notably light exposure. Variations in the red to far-red (R/FR) ratios of light impinging on soybean leaves are believed to trigger shade-avoidance responses that contribute to their photosynthetic efficiency. To date, the extent of possible connections between variations in the REP and R/FR ratios of soybean leaves remains unclear. In this paper, we address this open question using available measured spectral reflectance and transmittance data obtained for two groups of soybean specimens characterized by distinct chlorophyll contents. More specifically, we examine the impact that their distinct pigmentation levels have on their respective REP and R/FR ratios. The potential ramifications of our findings include not only the enhancement of the procedures employed in the monitoring and management of soybean crops through the combined use of these indices, but also the strengthening of the current knowledge about the intertwined physiological processes responsible for these plants' highly adaptive photosynthetic apparatus.

In Thailand, rice yield is normally derived from Crop Cutting Experiments (CCEs) and found lately for agricultural planning policy. The potential of remote sensing is widely adopted for crop monitoring and yield estimation; however, there are few research using satellite data and rice biophysical parameters in Thailand. Thus, the objective of study is investigating the potential of Optical (Sentinel-2) and Synthetic Aperture Radar (SAR- Sentinel-1) for estimating rice biophysical and rice yield by developing a linear regression model in the three representative’s provinces located in the Chao Phraya River delta, Thailand.

Crop phenological phases have traditionally been observed from the ground, which is a labor-intensive and time-consuming activity that also lacks spatial variability due to the sparse and limited network of ground data, if any are available. In view of this, remote sensing can provide a low-cost avenue to systematically monitor and detect phenological phases from space. The most common approach for retrieving vegetation phenology from remotely sensed time series is the dynamic threshold method. However, only a few number of studies have attempted to calibrate and optimize these derived phenological metrics to relate them to actual crop growth stages. Accordingly, this study attempted to develop a framework to optimize the derivation of phenological phases for three major crops (winter wheat, corn and sugar beet) in Germany by investigating the optimal thresholds and comparing the performances with ground-truth observation data. To this end, the Normalized Difference Vegetation Index (NDVI) time series covering Germany and for two cropping seasons 2019 and 2020 were obtained and derived from a 10 x 10 km tiling grid of Sentinel-2 analysis ready data using a specific decentralized cloud platform that combines both a set of satellite imagery (petabytes of data) with huge analysis capabilities on a very large scale. Since cloud contamination is typically the major drawback for estimating phenology with optical satellite data, the study suggests first a new smoothing and gap filling method (UE-Whittaker) that is based on both envelope detection and the Whittaker filter and that, in the end, constructs high-quality NDVI time series that are suitable for phenological analysis. Based on these generated time series, the estimation of various phenological phases of crops as well as threshold optimisation and calibration were carried out as a second step. In which we traverse the thresholds from 0 to 1 with an increment of 0.01 for each specific phase and finds the optimal threshold when the lowest error value is obtained between the satellite-derived DOY and the observed DOY in ground data from the year 2019. Later on, these optimum thresholds were used to derive phenological phases from the next year, and the results of the calculation of the root-mean-square error (RMSE) and the mean absolute error (MAE) between the ground reported in-situ phenology observations and those derived from satellite data reveal that they ranged typically between 3 days and 2 weeks for nearly all the phenological phases. The findings demonstrate how calibrating and optimising the derivation of different phenological phases of crops using only optical data could produce a timely and accurate information on crop growth and its condition for a large area which can be used in agricultural management, crop yield estimation, and several other related applications.

Rice biophysical is relevant with rice yield and able to analyses the correlation with satellite data. Thus, the study aims to investigate which parameters significant with rice yield. The study conducts in three provinces located in the Chao Phraya River delta in Thailand, which is main rice cultivated area of country. The primary data use based on the field experiments in the wet season rice of 2017 and separated different 5 growth stages (e.g., seeding, tillering, panicle, flowering, and harvesting). Several of rice biophysical are collected such as agricultural practices, stem density, water depth, height, Leaf Area Index (LAI), chlorophyll contents, wet-dry biomass, and rice yield. Then, the average of rice biophysical are demonstrated with the different rice variety and irrigation. The dynamics of rice biophysical are distinguish from others; then, the study analyses with the Pearson correlation at P-value 0.05 and two-tailed significance. The result suggests the appropriate rice biophysical based on rice yield.

Rice paddy fields in semiarid northeastern Thailand are a unique feature of traditional agriculture in the region, presenting unique agricultural practices and challenges, particularly in water management in semiarid conditions. This study aims to improve rice production efficiency and sustainability by understanding how these methods arose and how they could be enhanced to have an impact on agricultural practices in other regions with similar problems. This study developing a phenology-based method using the unsupervised classification of Sentinel-1 and 2 time-series data to identify rice paddy fields. Vegetation Index (VI) time series were used for identifying variations in the canopy of rice fields during growth stages. Monthly Sentinel-1 and 2 time series data between January 2020 and December 2022 were classified using k-means clustering to identify regions with similar phenological patterns. This methodology produced maps with a 10-meter resolution of rice field extent, intensity, and crop calendar. Validation was performed using the MS-700 Spectroradiometer and time-series image observed from the top of an automatic weather station. The results indicate that the proposed method based on phenology is cost-effective and capable of accurately mapping rice fields and growth stages across large areas. This study also highlights the importance of rice paddy fields in semiarid northeastern Thailand and how the developed methodology can help improve holistic water resources management and sustainability in the region and other similar areas.

Food production is one of the significant challenges for the world's population. Countries like Brazil, with a vast territorial dimension and good availability of resources, stand out in the production of grains, especially soy. Soy cultivation requires care and management to ensure phytosanitary production and reduce the risk of diseases such as Asian Soybean Rust (ASR) caused by the fungus Phakopsora pachyrhizi. In Brazil, soy cultivation occurs in the spring/summer (September/March), with greater solar energy and rainfall in the country. Brazil has established a fallow period to reduce the risk of ASR, which prohibits planting outside the agricultural calendar. However, there is the possibility of authorizing planting in the floodplains of the tropical plains of the Formoso River basin, Tocantins, Brazil. The government of the State of Tocantins created the State Program for the Control of ASR, authorizing the planting of soybeans during the dry season (April to September) through registration and monitoring of areas. However, other plantings, such as beans, with a shorter cycle and less water demand, also occur. This study aims to monitor the soybean crop development phases considering data collected in the field by the Agricultural Defense Agency (ADAPEC) and digital processing using deep-learning techniques of Sentinel-1 image time series. The phenological differences of cultivation farms enabled agricultural mapping and the fight against ASR. The digital processing steps of the Sentinel-1 time series dataset (10 m resolution) consisted of image preprocessing using Sentinel Application Platform (SNAP); time series filtering using Savitzky-Golay; evaluation of deep learning methods (Long Short-Term Memory - LSTM, Bidirectional LSTM - Bi-LSTM, Gated Recurrent Unit - GRU, and Bidirectional GRU - Bi-GRU); and accuracy analysis. However, the classification has some erroneous portions that can be improved by increasing the number of classes and samples in future works.

In this study, we compare the performance of texture descriptors and spectral vegetation indices for the classification of a hemiparasitic plant that grows on host trees, known as mistletoe. For this purpose, we computed 180 image features, including GLCM, Gabor, and LBPs, as well as spectral vegetation indices, from multispectral aerial image sets. Our image feature database is then classified using Support Vector Machines, with optimized hyperparameters, and accuracy metrics are reported in order to evaluate the contribution of specific feature sets for our application. In addition, we make use of feature selection algorithms in order to determine which combination of descriptors improves the classification process. The study has important implications for the remote sensing community, as it can provide insights into the use of texture and spectral descriptors for classification of the mistletoe species known as Struthanthus Interruptus. The results of the study can be used to develop more effective tools to monitor the spread of the pest in urban parks, which can help to preserve trees and ensure their long-term health. Overall, the study contributes to the growing body of research on the use of remote sensing technologies, in conjunction with artificial intelligence techniques, to monitor urban environments.

The TUBIN spacecraft launched in June 2021 is tasked with the detection of high-temperature events. To this end, it employs two microbolometer focal plane arrays and a complementary sensor sensitive in the visible range. Using simulated data, a fire detection algorithm was developed. Within the first 18 months of mission operations, several hundred scenes containing wildfires and other high-temperature events were captured. The detected high-temperature events can be categorized into volcanic features, vegetation fires and artificial events such as gas flares. Fire maps were manually created for all scenes meeting the criteria of nadir or close to nadir imaging and the availability of recent calibration data. These maps were validated using secondary space-borne instruments such as MODIS, VIIRS. The individual validation of the data was performed with satellite data with close temporal proximity to the time of image capture. The available fire products feature various ground sampling distances (GSD) mostly lower than the approximately 150 m GSD of the TUBIN sensor suite. The fire detection algorithm is based on the brightness temperature values of the pixels. Through a series of steps, cloud and background pixels are isolated from candidate fire pixels that are further evaluated based on their relative response. This paper evaluates the results from the TUBIN fire detection algorithm on the gathered data and determines the accuracy within the data set.

Mediterranean-type ecosystems are among the most affected by global climate change due to an increase in droughts and fires. Sentinel-2 satellites are currently among the best alternative for operational vegetation properties monitoring because of their temporal revisit and global coverage. The increasing availability of spaceborne imaging spectrometer (e.g. DESIS, PRISMA, EnMAP) and the preparation of missions ensuring global accessibility (e.g. CHIME, BIODIVERSITY) will enable the estimation of vegetation traits with better accuracies. The SENTHYMED project aims to study the complementarity between multi- and hyperspectral images to evaluate Mediterranean forest functional traits. The objective is to estimate canopy pigment, leaf water and dry matter contents from physical model inversion using DART radiative transfer model. A preliminary step is to study the influence of DART optical properties parametrization on remote sensing image simulation in order to simulate scenes as accurately as possible. Two forests in the South of France, mainly composed of evergreen oaks and pubescent oaks, with heterogeneous canopy structure, were studied. UAV LiDAR data were first acquired and converted into voxel matrices of plant area density values with AMAPVox. Pytools4dart was then used to build the mock-ups, handle DART parameterization and generate images in spectral reflectance unit at canopy level. Several simulations were implemented, assigning different optical properties to the underground and to the canopy. These images were compared to airborne AVIRIS-Next generation acquisitions, acquired close to the field campaign that took place in June 2021 and where in-situ measurements were collected for calibration and validation of DART simulations.

Sugarcane and maize (corn) crops are extensively used in the production of biofuel worldwide. These plants share important physiological and morphological traits. Both belong to the group of C4 species characterized by the presence of unifacial leaves. Moreover, their stress adaptation mechanisms make them less susceptible to adverse conditions elicited by climate changes. Given these aspects and their economical value, one would expect that there is no shortage of data for these plants, particularly with respect to their foliar spectral responses. After all, such data is essential for the efficacy of precision farming and remote sensing strategies devised to obtain an ecologically sustainable increase in the yield of these crops. However, this is not the case, with the data scarcity situation being markedly more serious for sugarcane. Because of that, and considering their physiological and morphological similarities, investigations on the spectral responses of C4 plants are usually conducted using data obtained from maize specimens, with the resulting findings often being implicitly extended to sugarcane. This raises the question of whether the level of comparability between the foliar spectral responses of these two species is sufficient to support such an approach. In this paper, we aim to contribute to the elucidation of this question. Using measured reflectance data obtained for maize and sugarcane leaves, we compute selected spectral features associated with these specimens, and assess possible discrepancy trends. We then discuss data availability issues in this area, and identify relevant topics for future research that will likely require comprehensive measured spectral datasets for these plants.

Accurate quantification of precipitation partitioning into evapotranspiration and runoff is important for global water balance estimation and water resources managements. The Budyko framework is a simple yet robust solution to parameterize precipitation partitioning and has been widely applied for studying catchment-level water and energy fluxes. However, substantial variations between the observed and Budyko-predicted evaporative indices have been observed. Many studies have attributed the scatter around the Budyko curve to catchment characteristics (e.g., vegetation and soil property), which are not directly accounted for in the Budyko framework. However, modified Budyko-type equations that consider catchment characteristics are not transferable between regions and the interannual catchment behaviours still fail to follow the adjusted Budyko trajectories. To explore if the pronounced Budyko scatter in humid catchments has a systematic pattern caused by measurable catchment properties, this study comprehensively investigated the relationship between Budyko scatter and multiple catchment biophysical features from both spatial and temporal perspectives. Results reveal that for humid catchments, topography and seasonal cumulative moisture surplus can explain the spatial distributions of Budyko scatter with r higher than 0.65, whereas soil properties and vegetation indices explained little of the variance (r≤0.30). Temporally, the interannual variability of Budyko scatter was negatively correlated with annual average vegetation indices, particularly for catchments with relatively low vegetation cover. Overall, this study provides valuable insights to the interpretation of Budyko framework and offers possible solutions to improve its performance to predict the spatio-temporal variability of water balances.

The Atmosphere-Land Exchange Inverse (ALEXI) a two-source energy balance model was developed to estimate ET. The Visible Infrared Imaging Radiometer Suite (VIIRS) a polar satellite used in this research to provide 375-m resolution compared to other geostationary satellite data which have more than 1 km resolution. VIIRS acquires images of the Globe on daily basis; day/night images. The ALEXI model takes advantage of day/night thermal infrared imaging to produce daily regional ET estimates using a LST differential to retrieve energy balance components between midmorning after sunrise and before noon local time. Daily Evapotranspiration maps were produced with 15^o X 15^o grid size (Tile). We ran ALEXI for Tile 153, over Brazil for the years 2013-2018. We created a website called Global Daily Evapo-Transpiration (GloDET) where we publish these maps at (https://glodet.nebraska.edu). The ALEXI estimated ET values were compared with ground data from eddy covariance flux towers. ALEXI ET results were extracted at the towers locations for 2013-2016, to serve as comparison for each tower with energy balance closure. The linear correlation was excellent for all sites with R2 between 0.78 - 0.88, for different types of vegetation.

Traditional pixel-based algorithms, considering only spectral information and ignore spatial information, have limitations to provide better accuracy of surface heat fluxes from high-resolution images. Based on the high-resolution satellite images, this paper systematically analyzes the feasibility of combining the object-based approach with traditional physical model to estimate surface heat fluxes.Sentinel-3 surface temperature and Sentinel-2 multi-spectral data were input to the energy balance Two-Source Energy Balance (TSEB) model to estimate the surface heat fluxes. Pixel-based TSEB model was firstly employed at 10m. An multi-layer experiments framework was constructed to explore the applicability of the object-based method. The object-based approach is introduced into TSEB model and the multi-scale segmentation algorithm of eCognition is used to segment the images and extract the surface objects. Two object -based strategies, estimating heat fluxes before or after aggregating objects properties, were used to analyze the influence of the different strategies on the results. Object-based method and different inversion strategies are compared with pixel -based results. The results show that, comparing with the pixel-based method, the object-based method is beneficial to map surface heat fluxes and can improve the estimation accuracy of the TSEB model, which is mainly influences by the vegetation-related parameters.

Daily evapotranspiration (ET) estimation using remote sensing procedures is important for proper irrigation management based on crop type and at field scales. This research aimed to estimate daily ET for grassland, bare soil, corn, soybeans, cottonwood, and red cedar through the hybrid modelling approach named Spatial EvapoTranspiration Modeling Interface (SETMI) applied in southwest Nebraska, USA, based on two source model using satellite image inputs, and weather datasets. Multispectral and thermal infrared imagery from satellite sensors joined with climate, crop cover classification image, and weather datasets were used to estimate ET for the period 2008-2013. SETMI was applied using multispectral and thermal infrared imagery from Landsat 7 and 8. SETMI model considers the ET obtained from the two-source energy balance model at satellite overpass time and was validated using latent heat fluxes measured with an Eddy covariance system (EC) on grassland. Crop coefficient (Kc) was estimated using modeled ET and daily reference ET over the season. Modeled ET showed a strong correlation to the ground data from EC, with ET presenting R² equal to 0.96. Overall, the maximum average Kc over the period was satisfactory for all land uses ranging from 0.45 for bare soil to 1.2 for corn. On average, bare soil showed the lowest Kc, while corn and soybeans had the highest values. The SETMI model produced adequate estimated daily Kc values over the years through the TSEB model, confirming the applicability of the model in estimating ET.

Estimating actual evapotranspiration (ETa) and crop coefficients (Kc) is crucial for efficient water management in sugarcane crops in Colombia. Because the implementation of water use policies in agriculture has made water management strategies increasingly demanding, the use of surface energy modeling tools such as the Surface Energy Balance Algorithm for Land (SEBAL) and the Simple Algorithm for Evapotranspiration Retrieving (SAFER) has gained importance. These tools not only estimate the amount of water based on evapotranspiration for each satellite-image pixel, but also determine Kc values along with information gathered by weather stations. This study aimed to estimate the ETa and Kc of sugarcane using the SEBAL and SAFER algorithms. Thirteen central pivots located in the sugarcane-growing region of the Cauca river valley were studied for three consecutive crop cycles between 2018 and 2021. Analyses were performed on Landsat 8 satellite images for both modeling algorithms. Reference evapotranspiration (ETo) data from 37 weather stations of the Colombian sugar sector were also interpolated to determine Kc values. SEBAL-based analysis results for ETa showed maximum values of 4.2 mm.day^-1 and minimum values of 1.02 mm.day^-1, while SAFER yielded maximum values of 3.8 mm.day^-1 and minimum values of 1.02mm.day^-1. SEBAL-based analysis results for Kc indicated maximum values of 0.9 and minimum values of 0.2, while those of SAFER indicated maximum values of 0.7 and minimum values of 0.1. Although the ETa and Kc values estimated using these models were lower compared with those of FAO 56 and should be calibrated using a direct method, these models showed potential to be applied in water management programs for Colombia’s Cauca river valley.

The arrangement of plant roots and their overall structure, known as root system architecture (RSA), plays an important role in acquiring water and nutrients essential for plant growth and development. Moreover, the RSA demonstrates remarkable adaptability to environmental stresses, making it a central factor in plant adaptation. Root traits, including root length, root diameter, root length density (RLD), and the presence of root hairs, play a crucial role in optimizing resource utilization within the soil and enhancing productivity. In particular, root hairs play a crucial role in the overall health and functioning of plants. These microscopic, hair-like structures extend from the surface of root cells and greatly increase the root’s surface area, which accounts for approximately 70% of the total root area. The characteristics of root hairs, such as their length and density, significantly enhance soil nutrients and water uptake. Considering these advantages, it is difficult to observe root hairs in a scene with low resolution. Therefore, we proposed a study using deep learning-based image super-resolution methods as a pre-processing step that helps to reconstruct finer details and structures within the root hairs, leading to a more accurate representation of their morphology, to understand the improvement in the response of root hairs under different environmental conditions and their impact on nutrient and water uptake, models need to be evolved.

Models using remote sensing to estimate evapotranspiration (ET) and water requirements of agricultural crops have been used for monitoring and managing water resources. The objective of this study was to use the SETMI hybrid model through multispectral and thermal imaging to estimate the ET and water requirement of agricultural crops using center pivot irrigation systems through the variable rate irrigation (VRI) for the State of North Dakota in the USA for the years 2021 and 2022. The SETMI model adequately estimated the ET and water requirements in VRI systems, improving irrigation management and the use of water resources.

In this paper, an approach for joint estimation of grassland LAI and Cab from unmanned aerial vehicle (UAV) hyperspectral data was proposed. Firstly, based on PROSAIL (PROSPECT+SAIL) model, 15 typical hyperspectral VIs were simulated and analyzed to identify optimal VIs for LAI and Cab estimation. Secondly, three different combinations of VIs were tested in the construction of LAI and Cab inversion matrices. Thirdly, 3-layer 2-dimension VIs matrix was generated to determine the relationship between VIs and LAI values and Cab values. Finally, LAI and Cab were joint retrieved according to the cells of VIs matrix.

This paper presents a field trial aimed at evaluating the performance of a micrometeorological weather station developed by IOSB. Which aims to sample a larger area through point measurements. Goal of the trial was to sample meteorological data of a larger scale by extensive point measurements. The weather station is equipped with a variety of sensors, including GPS, temperature, humidity, air pressure, wind speed and direction, as well as particle and aerosol measurements (PM1, PM2.5, and PM10). Additionally, radiation variables such as UV and IR measurements (IR: response maximum 820nm, UV: response range 280 – 430 nm), as well as aerosol optical thickness, can be measured. The weather station is also equipped with a camera for wider optical applications such as turbulence determination, cloud tracking, and positioning. During several intense field trials, including in Jordan, the performance and accuracy of the micrometeorological weather station in measuring weather parameters were evaluated and a variety of errors were corrected. The data collected by the weather station are compared with measurements from other meteorological instruments to evaluate the reliability and accuracy of the weather station. The micrometeorological weather station has the potential to be a cost-effective and reliable solution for measuring weather parameters in various applications such as precision agriculture, environmental monitoring, and climate research.

In Italy, the utilised agricultural area (UAA) is equal to approximately 41.8% of the surface area of the entire state: optimising agricultural production using new technologies therefore makes it possible to improve the performance of the soil and, consequently, its wellbeing, both necessary conditions for both the environmental protection of ecosystems and the conscious management of resources. Many fruit and vegetable varieties produce ethylene during ripening (among them, climacteric ones even after they have been harvested). Being able to monitor the concentration of ethylene in an agricultural field or greenhouse (or, in the case of climacteric fruits and vegetables in harvest warehouses) makes it possible to optimise their harvest, manage their packaging and sale, and reduce waste and wastage. The DIAL (Differential Absorption Lidar) technique is able to measure the concentration profile of the species in the atmosphere. In this work the possibility to estimate the ethylene concentration using a DIAL is evaluated by numerical simulations. The interference of other chemicals, such as water vapour, is assessed and the use of multiwavelength approaches is analysed to improve the accuracy of the measurements, and different hardware configurations are proposed.

Precision farming and remote sensing have seen an unprecedented development in the last decade. The growing interest in this domain has led to the development of robust and accurate processing pipelines to evaluate nutrient management and irrigation practices, among others. Problems such as crop classification have gained significant attention in Southern Italy due to unique challenges such as water scarcity and the spread of cultivar-specific diseases (i.e., Xylella fastidiosa). Here, we present a technological platform hosted by the ReCaS HTC/HPC cluster based in Bari, Italy, for the automated segmentation of common crops in Southern Italy, specifically the Apulia region, in very high-resolution aerial (VHR) RGB images. In particular, we discuss the adoption of a Deep Convolutional Neural Network (DCNN) which uses a lightweight EfficientNet-B0 architecture, for patch-wise land cover classification and compare its performance with a standard machine learning algorithm (Random Forest) fed with Haralick features. The DCNN, pre-trained on ImageNet-1000 and fine-tuned on a 4-class problem, including vineyard, olive groves, arable land, and “no-crop”, had the highest performance with an overall accuracy of 77±5 % when performing a repeated spatial cross-validation. The experimental results demonstrate the effectiveness of the proposed approach in achieving high accuracy in land cover classification, although a misclassification between arable land and “no-crop” was observed, as they share similar vegetation textural patterns. The lightweight EfficientNet-B0 architecture provides a good balance between accuracy and computational efficiency, making it a suitable choice for processing very high-resolution aerial images. The processing pipeline has been successfully implemented and deployed on the high-performance computing (HPC) platform, leveraging Apache Mesos as the underlying framework. To ensure efficient execution of tasks, the Chronos job scheduler has been employed to submit the execution of Docker containers. By utilizing specialized hardware, including Nvidia V100 and A100 GPUs, the pipeline can effectively handle and process substantial volumes of data within tight timeframes. The proposed approach is highly versatile and can be easily adapted to various precision farming applications. The use of Docker technology facilitates easy deployment and portability across different environments. Additionally, the adoption of a lightweight DCNN architecture allows to efficiently exploit parallel computing resources enabling seamless scalability and, therefore, handling of massive computational tasks across broader regions of interest.

Environmental Impact Assessment (EIA) processes have well-established requirements for baseline data describing the status and trends of rapidly changing environments. Examination of these requirements demonstrates a strong capacity for Earth Observation (EO) science to support EIA processes. This capacity has not been matched with EO uptake indicating substantial persistent barriers, including: (1)-EO science awareness; (2)-data availability and usability; and (3)-technological solutions and analytics capacity. The Canada Centre for Mapping and Earth Observation is at the mid-point of a ten-year Earth Observation for Cumulative Effects (EO4CE) research program to integrate EO within Canada’s EIA processes. The EO4CE program has employed a wide variety of data systems (optical, microwave, gravity, etc.) and methods (machine learning, big data systems, etc.) to build biosphere, hydrosphere, and cryosphere data products with national scale coverage and regional scale detail. Next steps for the program include preparing for inclusion of next-gen sensors, improving data production frameworks, and addressing awareness and uptake issues through focused communication and demonstration. This presentation will provide an overview of EO4CE implementation, results, and lessons learned which would be applicable to similar initiatives.

The Satellite Application Facility on Land Surface Analysis (LSA SAF) produces and provides access to remotely sensed variables for the characterization of terrestrial ecosystems, such as land surface fluxes and biophysical parameters, taking full advantage of the EUMETSAT satellites and sensors. In this work, a procedure for the joint estimation of LSA SAF vegetation parameters is proposed. The approach includes the use of multi-task learning with gaussian processes (MTGP). The MTGP learns a shared covariance function on input features and a covariance matrix over tasks. Unlike the single output approaches, the proposed multi-task captures the inter-task dependencies among outputs. Two comparison exercises were undertaken to assess the estimation power of the MTGP as compared to single output algorithms such as standard gaussian processes regression (GPR), neural networks (NN), and random forest (RF). First, we evaluate the performance of MTGP in the context of deriving CO2 fluxes such as the gross primary production (GPP), net ecosystem exchange (NEE), and total ecosystem respiration (TER) blending SEVIRI/MSG and eddy covariance (EC) data. In addition, the MTGP prediction power was also assessed for the joint estimation of LAI, FAPAR, and FVC in a hybrid approach using radiative transfer model simulations and AVHRR/MetOp observations. The results show that MTGP outperforms the single output approaches in terms of accuracy. The MTGP multi-task optimization links outputs in such a way that the relationships among the biophysical parameters are better described obtaining a more robust model and therefore improving the accuracy of the estimates. The findings pave the way for future multi-task implementations in order to derive consistent outputs and accurate estimates of vegetation properties from remote sensing.

Lychee is a tropical evergreen fruit which is mainly produced in South-East Asian countries. The north-eastern state of Assam in India is one of the largest exporters of lychee which mainly comes from areas around Tezpur town in the district. Health monitoring is critical in the region which is capable to trigger alarm in case of abnormalities during the growing season itself such that necessary steps can be taken to limit the damage. The proposed method tracks the photosynthetic activity through indices as an indicator of the health. This supervised approach considers the usual growth pattern as training set to define the normal growth using satellite image derived indices. The behavior pattern on the test time series sequence which is not conforming to the usual growth pattern is tagged as anomalous incorporating the temporal variation of the indicators across seasons and within class variability of the different types of the lychee trees present in the area of study.

According to Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora (Habitats Directive) each member state should maintain or restore at favorable conservation status natural habitats and species of wild fauna and flora. The study examines the possibilities for application of remote sensing in projects for improving the conservation status of habitats in NATURA 2000 network in Bulgaria. The research focuses on exploring the actual condition of certain habitats before and after implementing restoration measures like removing invasive or ruderal flora types. Sentinel-2 images and have been used to detect changes in plant species distribution and drone to verify that the invasive plant species have been removed. GIS database has been developed which can be used in future monitoring on surveyed area. The results can be used as a verification tool to measure the effectiveness of the restoration measures implemented and in long-term aspect to track the sustainability of similar projects in Bulgaria.

Infrared imaging is a well-known non-invasive technology that in recent years has gained great interest in precision agriculture field. Plants are subjected to a wide range of biotic stresses caused by pathogenic bacteria, fungi, nematodes, and viruses that reduces productivity. In this work wild rocket (Diplotaxis tenuifolia) plants inoculated with the soil-borne pathogens Rhizoctonia solani Kühn, Sclerotinia sclerotiorum (Lib.) and Fusarium oxysporum f. sp. raphani were monitored daily in laboratory by means of the infrared imaging. Plant monitoring was performed with both active and passive approaches. The results obtained showed that the infrared imaging methods tested are promising for early diagnosis of soil-borne diseases by allowing their detection a few days before they are detectable through a visual analysis. These findings open up the possibility of developing new imaging systems for both proximal and remote sensing.

Information on crop harvest events has become valuable input for models related to food security and agricultural management and optimization. Precise large scale harvest detection depends on temporal resolution and satellite images availability. Synthetic Aperture Radar (SAR) data are more suitable than optical, since the images are not affected by clouds. This study compares two methods for harvest detection of soybean in Vojvodina province (Serbia), using the C-band of Sentinel-1. The first method represents a maximum difference of ascending VH polarization backscatter (σ_VH) between consecutive dates of observation. The second method uses a Radar Vegetation Index (RVI) threshold value of 0.39, optimized to minimize Mean Absolute Error (MAE). The training data consisted of 50 m point buffers’ mean value with ground-truth harvest dates (n=100) from the 2018 and 2019 growing seasons. The first method showed better performance with Pearson correlation coefficient r=0.85 and MAE=5 days, whereas the calculated metrics for the RVI threshold method were r=0.69 and MAE=8 days. Therefore, validation was performed only for the method of maximum VH backscatter difference where mean values of parcels with ground-truth harvest dates for 2020 had generated the validation dataset (n=67). Performance metrics (r=0.83 and MAE=3 days) confirmed the suitability for accurate harvest detection. Ultimately, a soybean harvest map was generated on a parcel level for Vojvodina province.

Snow cover is among the most important features of the Earth's surface and a crucial element of the cryosphere that affects the global energy balance, water, and carbon cycles. Accurate monitoring of this land surface component is of particular significance as snowmelt provides between 50%–80% of the annual runoff in the temperate (boreal) regions and significantly impacts the hydrological balance during the warm season. Limited reserves of soil moisture during the winter period can lead to all types of droughts, including green-water drought, which is expressed by reduced water storage in soil and vegetation. Green-water drought causes a variable effect across landscape components, on the functions and ecosystem services (ES) they provide. The present study aims to track the snow cover dynamics in the transitional seasons of the year when the snow cover is most unstable and to differentiate its territorial distribution depending on elevation and slope exposure. The study area covers the mountainous territories of Bulgaria and the seasons from 2016 to 2022. To achieve the aim of the study, we used Sentinel-2 images and calculated the Snow Water Index (SWI). SWI uses spectral characteristics of the visible, shortwave infrared (SWIR), and near-infrared (NIR) bands to distinguish snow and ice pixels from other pixels, including water bodies which is crucial for the accurate monitoring of snow cover dynamics. The obtained results were validated using VHR images for pre-selected test areas.

The present study aims to monitor the Snow Cover Extent (SCE) of the mountainous region of Bulgaria (13905 km²), located 1000m above sea level, for eight years. Information is important for calculation of Snow Water Equivalent (SWE), hydrological runoff modeling, forecasting, and assessing flood events. Global Warming and Climate Change and their impacts, such as a constant increase in recorded high-temperature levels, frequent droughts, water scarcity in the summers, and less-snow winters, have a significant effect on agriculture, hydrology, forests, and ecology in Bulgaria. The present research uses the available cloudless optical data of Sentinel-2-MSI for snow cover monitoring concerning the decrease in snow distribution during the last decade. Sentinel-2 satellite imagery, from October to May, for the period between 2016 and 2023, was generated and exported from Google Earth Engine (GEE). Normalized Differential Snow Index (NDSI) and Snow Water Index (SWI) were calculated, and the resulting output indices rasters were post-processed and inspected additionally to obtain thresholding classifications, masking out the areas covered by shadows (topographic), water bodies, forests, etc., and snow cover area distribution. The results obtained in the study can be used and integrated for climate change observations and research at the local and regional levels.

Intelligent agriculture increasingly relies on modern technologies for reliable monitoring of crops and timely detection of areas in which special intervention is needed to ensure the planned yields. This paper represents the results of a study of the possibilities of fusion and processing of multi-spectral and thermal data from satellite systems and such from remotely controlled platforms (drones) for the decision making support in intelligent agriculture. The study examined images with different resolutions, such as from the Sentinel-2, Landsat and Planet Labs satellite systems, as well as multi-spectral images from the commercial drones DJI Phantom 4 Multispectral and thermal images (thermograms) from the DJI Mavic 2 Enterprise Advanced.

Nowadays, modern technologies are more and more accessible, cheaper and easier to use. Part of them are the multisensor remotely controlled aerial platforms or the so-called “drones”. This paper presents the results of a research for the possibilities of mixing and processing multi-spectral and thermal images from aerial drone sensors for the intelligent agriculture support. Time series of multi-spectral images from the DJI Phantom 4 Multispectral and thermal images from the DJI Mavic 2 Enterprise Advanced over same area are processed. An approach of a new methodology for data fusion of drone multispectral and thermal (LWIR) images data processing for intelligent agriculture support is proposed.

Precision agriculture has been at the cutting edge of research during the recent decade, aiming to reduce water consumption and ensure sustainability in agriculture. The present study aims to estimate the actual water requirements of crop fields based on the Crop Water Stress Index, combining multiple and multiscale data, such as infrared canopy temperature, air temperature, air relative humidity, near-infrared and thermal infrared image data, taken above the crop field using an innovative aerial micrometeorological station (AMMS), and two more compatible and advanced cameras, a multispectral and a thermal mounted in an Unmanned Aerial Vehicle (UAV), along with satellite-derived thermal data. Moreover, ground micrometeorological stations (GMMS) were installed in each crop. The study area was situated in Trifilia (Peloponnese, Greece) and the experimentation was conducted on two different crops, potato, and watermelon, which are representative cultivations of the area. The analysis of the results showed, in the case of the potato field, that the amount of irrigation water supplied in the rhizosphere far exceeds the maximum crop needs reaching values of about 394% more water than the maximum required amount needed by the crop. Finally, the correlation of the different remote and proximal sensors proved to be sufficiently high while the correlation with the satellite data was moderate. The overall conclusion of this research is that proper irrigation water management is extremely necessary and the only solution for agricultural sustainability in the future. The increasing demand for freshwater, mainly for irrigation purposes, will inevitably lead to groundwater overexploitation and deterioration of the area's already affected and semi-brackish coastal aquifers.

With recent advancements in water-quality analytical technology and the increasing popularity of the Internet of Things (IoT), the market demand for compact and durable automated water-quality monitoring devices has grown substantially. However, the current existing online monitoring devices mostly featuring a single-light source to monitor turbidity and chemical oxygen demand (COD), two critical indicators of natural water bodies, which tend to be influenced by interfering substances that limits their ability to measure more complex water-quality parameters. To address these issues, a new modularized water-quality monitoring device equipped with multi-light sources (UV/VIS/NIR) has been designed and implemented. This device can measure the photo-intensity of scattering, transmission, as well as reference light simultaneously, and coupled with different water-quality prediction models to provide accurate estimates for tap water (Turbidity<2 NTU, relative error < 17.8%), environmental sample (Turbidity<400 NTU, relative error < 2.3%) and industrial wastewater (COD<300 mg/L, relative error < 17.6%). The study results suggest the new designed optical module can effectively monitor turbidity and COD in different water samples and provide alerts for water treatment in high concentration, thereby enabling automated water quality monitoring in future.

The aim of the study is to present results derived from monitoring the flood event on river Stryama, located in Karlovo municipality, Bulgaria occurred on 02.09.2022 due to extreme intensive rainfalls. During the flood event the rainfall had increased up to 250 l/m², and the water level of the Stryama River in some regions had reached up to 3 meters in 8 hours. Stryama river is situated in central Bulgaria, Plovdiv district, it springs up from Stara planina mountain and its length is 110 km. The applied methodology in the following survey includes use of Sentinel-2 MSI optical data and Tasseled Cap Transformation (TCT) of selected satellite imagery for change detection and estimating the territorial extent of areas affected by the flood waters. Satellite imagery of different temporal points were chosen before and after the flood event in order to track the water dynamics around the riverbed. The calculation of the spatial and temporal characteristics of the river waters were accomplished by segmenting of Sentinel-2 multispectral imagery. The application of the matrix for Tasseled Cap Transformation segments the optical images in 3 components: TCT-brightness, TCT-wetness, TCT-greenness. On the basis of TCT-wetness component and its values the dynamics and territorial distribution of river waters were monitored for the chosen temporal period. On the basis of TCT-greenness component and Normalized Differential Greenness Index (NDGI) an assessment of the impact of flood waters on the vegetated areas was made.

Remote sensing drone utilization technology is rapidly developing and is being used in a variety of ways, from seed sowing to disease management and maintenance. Crop growth environment analysis and crop prediction vary depending on the climate, soil environment, topography, and applied technology of the target area. Crop growth is a complex trait determined by various factors such as genotype, growing environment, and interactions. To accurately predict growth conditions and growth, it is necessary to fundamentally understand the functional relationship between these interaction factors through data analysis. Interpretation of growth-related relationships requires both a comprehensive dataset and powerful algorithms in the model. This study aimed to build a model using drone imaging and AI technology to develop a model for the cultivation status and growth prediction of various crops grown in the field. The development model included the entire process of drone image acquisition, image processing, AI algorithm application, farmland information, crop status, and growth information production. This paper presents the overall configuration for the construction of the growth prediction model and the results of the AI-based cultivation area extraction model conducted in the first stage. Classifying cultivated crops by field is important for identifying the cultivated area and predicting yield. The development of drone remote sensing (RS) and AI technology has made it possible to precisely analyze the characteristics of field crops with images. The purpose of this study was to create and evaluate an AI-based cultivated crop classification model using the reflectance and texture characteristics of drone RGB images. The major crops cultivated during the crop classification survey period were kimchi cabbage, soybean, and rice. The texture applied in this development model is the texture characteristic of Haralick using GLCM (Gray Level Co-occurrence Matrix). A total of 8 factors were used to create the model: mean, variance, contrast, homogeneity, correlation, ASM (Angular Second Moment), homogeneity, and dissimilarity. Two AI models, SVC and RFC, were built in this study. For the SVC-based classification model, the hyperparameters C and gamma were set to 1.5 and 0.01, respectively, and a radial basis function (RBF) kernel was used. The cross-validation accuracy was 0.88 and the test set accuracy was 0.91. The maximum depth of the RFC-based classification model was set to 8 and the number of trees was set to 500. The cross-validation accuracy of the RFC-based model was 0.95, and the test set accuracy was 0.89. The learning time of the two models was 90 seconds for the SVC model and 7,200 seconds for the RFC model. The SVC-based classification model was evaluated as advantageous when considering classification accuracy and learning time. The findings of this study are expected to improve the precision of crop cultivation area identification using AI technology and be used as a useful tool for agricultural production management and forecasting by farmers and the government.