A5082220927- Remote Sensing in Agriculture
- Remote-Sensing Image Classification
- Remote Sensing and Land Use
- Species Distribution and Climate Change
- Plant Water Relations and Carbon Dynamics
- Cryospheric studies and observations
- Atmospheric and Environmental Gas Dynamics
- Land Use and Ecosystem Services
- Smart Agriculture and AI
- Advanced Clustering Algorithms Research
- Remote Sensing and LiDAR Applications
- Solar Radiation and Photovoltaics
- Soil Geostatistics and Mapping
- Climate variability and models
- Data Management and Algorithms
- Hydrology and Watershed Management Studies
- Soil erosion and sediment transport
- Geochemistry and Geologic Mapping
- Spectroscopy and Chemometric Analyses
- Blind Source Separation Techniques
- Hydrology and Sediment Transport Processes
- Meteorological Phenomena and Simulations
- Face and Expression Recognition
- Data Mining Algorithms and Applications
- Atmospheric aerosols and clouds
BOKU University
2019-2025
Universitat de València
2008-2025
Geomatics (Norway)
2024
University of Twente
2015-2018
Parc Científic de la Universitat de València
2010-2014
Abstract The use of deep learning (DL) approaches for the analysis remote sensing (RS) data is rapidly increasing. DL techniques have provided excellent results in applications ranging from parameter estimation to image classification and anomaly detection. Although vast majority studies report precision indicators, there a lack dealing with interpretability predictions. This shortcoming hampers wider adoption by users community, as model’s decisions are not accountable. In that involve...
New Earth observation missions and technologies are delivering large amounts of data. Processing this data requires developing evaluating novel dimensionality reduction approaches to identify the most informative features for classification regression tasks. Here we present an exhaustive evaluation Guided Regularized Random Forest (GRRF), a feature selection method based on Forest. GRRF does not require fixing priori number be selected or setting threshold importance. Moreover, use...
Remote sensing optical sensors onboard operational satellites cannot have high spectral, spatial and temporal resolutions simultaneously. In addition, clouds aerosols can adversely affect the signal contaminating land surface observations. We present a HIghly Scalable Temporal Adaptive Reflectance Fusion Model (HISTARFM) algorithm to combine multispectral images of different reduce noise produce monthly gap free resolution (30 m) observations over land. Our approach uses from Landsat m 16...
In recent years, machine learning (ML) algorithms have gained substantial recognition for ecological modeling across various temporal and spatial scales. However, little evaluation has been conducted the prediction of soil organic carbon (SOC) on small data sets commonly inherent to long-term research. this context, performance ML SOC never tested against traditional process-based approaches. Here, we compare algorithms, calibrated uncalibrated models as well multiple ensembles their in...
Smallholder farmers cultivate more than 80% of the cropland area available in Africa. The intrinsic characteristics such farms include complex crop-planting patterns, and small fields that are vaguely delineated. These pose challenges to mapping crops from space. In this study, we evaluate use a cloud-based multi-temporal ensemble classifier map smallholder farming systems case study for southern Mali. combines selection spatial spectral features derived multi-spectral Worldview-2 images,...
The production of land cover maps through satellite image classification is a frequent task in remote sensing. Random Forest (RF) and Support Vector Machine (SVM) are the two most well-known recurrently used methods for this task. In paper, we evaluate pros cons using an RF-based kernel (RFK) SVM compared to conventional Radial Basis Function (RBF) standard RF classifier. A time series seven multispectral WorldView-2 images acquired over Sukumba (Mali) single hyperspectral AVIRIS Salinas...
Addressing the escalating climate crisis necessitates precise tools for evaluating nature-based solutions (NbCS). The BenchFlux project represents a significant advancement by developing scale-aware benchmarks carbon dioxide (CO₂) fluxes, leveraging flux tower measurements and Earth Observation (EO) data. Unlike existing scale-agnostic approaches, introduces methodology that explicitly accounts emergent, nonlinear behaviors inherent in dynamics across spatial temporal scales.The...
Soil organic carbon (SOC) is a fundamental contributor to soil functions and health.  Since SOC strong predictor of many important properties, it prominently featured in health assessments monitoring targets.  Yet, given its spatial heterogeneity, specific targets are highly debated. For example, the EU Directive for Monitoring Resilience proposed ratio between clay content (SOC:clay) as target, with 1/13 separating “degraded” from...
This letter introduces a simple method for including invariances in support-vector-machine (SVM) remote sensing image classification. We design explicit invariant SVMs to deal with the particular characteristics of images. The problem data can be viewed as encoding prior knowledge, which translates into incorporating informative support vectors (SVs) that better describe classification problem. proposed essentially generates new (synthetic) SVs from obtained by training standard SVM...
This paper presents a novel semisupervised kernel partial least squares (KPLS) algorithm for nonlinear feature extraction to tackle both land-cover classification and biophysical parameter retrieval problems. The proposed method finds projections of the original input data that align with target variable (labels) incorporates wealth unlabeled information deal low-sized or underrepresented sets. relies on combining two functions: standard radial-basis-function based labeled generative, i.e.,...
Monitoring vegetation changes over time is very important in dry areas such as Iran, given its pronounced drought-prone agricultural system. Vegetation indices derived from remotely sensed satellite imageries are successfully used to monitor at various scales. Atmospheric dust well airborne particles, particularly gases and clouds, significantly affect the reflection of energy surface, especially visible, short infrared wavelengths. This results with missing data (gaps) outliers while change...
Time series of phenological products provide information on the timings recurrent biological events and their temporal trends. This is key to studying impacts climate change our planet as well for managing natural resources agricultural production. Here we develop analyze new long term products: 1 km grids Extended Spring Indices (SI-x) over conterminous United States from 1980 2015. These (based Daymet daily temperature created by using cloud computing) allow analysis two primary variables...
The classification of the ever-increasing collections remotely sensed images is a key but challenging task. In this letter, we introduce use extremely randomized trees known as Extra-Trees (ET) to create similarity kernel [ET (ETK)] that subsequently used in support vector machine (SVM) novel classifier. performance classifier benchmarked against standard ET, an SVM with both conventional radial basis function (RBF) kernel, and recently introduced random forest-based (RFK). A time series...
Clustering is often used to explore patterns in georeferenced time series (GTS). Most clustering studies, however, only analyze GTS from one or two dimension(s) and are not capable of the simultaneous analysis data three dimensions: spatial, temporal, any third (e.g., attribute) dimension. Here we develop a novel algorithm called Bregman cuboid average triclustering with I-divergence (BCAT_I), which enables complete partitional 3D GTS. BCAT_I simultaneously groups along its dimensions form...
This brief addresses two main issues of the standard kernel entropy component analysis (KECA) algorithm: optimization decomposition and Gaussian parameter. KECA roughly reduces to a sorting importance eigenvectors by instead variance, as in principal components analysis. In this brief, we propose an extension method, named optimized (OKECA), that directly extracts optimal features retaining most data means compacting information very few (often just one or two). The proposed method produces...