La productividad del agua de los cultivos (PA) es un indicador importante del uso del agua agrícola; uno de los grandes retos del sector agrícola es producir una mayor cantidad de alimentos con la menor cantidad de agua aumentando la PA. Los valores de la PA dependen del rendimiento y la evapotranspiración del cultivo, donde la evapotranspiración actual de los cultivos varía entre las diferentes zonas agroclimáticas y es una función de los tipos de híbridos, la fecha de madurez relativa, las prácticas agronómicas y las condiciones ambientales. El presente estudio tuvo como finalidad estimar la evapotranspiración actual (ETa) y la productividad del agua (PA) en el cultivo de maíz forrajero, mediante la integración de sensores remotos, datos climáticos de reanálisis y el modelo de simulación AquaCrop. La ETa observada se determinó usando el modelo AquaCrop, calibrado con datos de campo y parámetros de la cobertura vegetal; mientras que la ETa estimada se obtuvo a partir del modelo SEBAL, implementado en Google Earth Engine (GEESEBAL), utilizando imágenes Landsat 8 y 9 y datos climáticos de reanálisis. Los resultados mostraron que los sensores remotos sobreestimaron la ETa observada, con una sobreestimación promedio del 13 %. Por otro lado, la productividad del agua (PA) estimada con sensores remotos fue ligeramente menor (2.79 y 2.62 kg/m³) que la observada (3.19 y 2.95 kg/m³), con una diferencia promedio del 12 %. Estos resultados indican que los sensores remotos, junto con datos climáticos de reanálisis, pueden adaptarse como una alternativa viable para el monitoreo espacial de la productividad del agua en zonas agrícolas extensas. No obstante, es necesario trabajar en la reducción del error, por ejemplo, incorporando modelos de estimación espacial del rendimiento o utilizando más de un modelo de balance de energía, como METRIC, SSEBop, entre otros.
En este estudio se evaluaron distintos esquemas de adquisición de sistemas de radar de apertura sintética (SAR), considerando variaciones en la longitud de onda, la polarización y el ángulo de incidencia para el monitoreo hidrológico de los ecosistemas de humedal del interior de las islas del delta del Río Paraná. Con este propósito, se analizaron las variaciones de la señal retrodispersada del centro de las islas dominados por cortadera (Scirpus giganteus) a partir de tres series temporales de imágenes de las misiones satelitales Sentinel-1 (C-VH/VV-32,2°) y ALOS-2 PALSAR-2 (L-HH/HV-26,1° y L-HH/HV-37,5°), en relación con las fluctuaciones ordinarias de las condiciones de anegamiento. Dado que la señal retrodispersada en ambientes de humedal está influenciada por múltiples factores como las variaciones en el nivel de agua y saturación del suelo, la rugosidad superficial, y el estado fenológico de la vegetación; se aplicaron modelos aditivos generalizados (GAM) para analizar la influencia de estos factores en la variabilidad de la señal retrodispersada en cada configuración de adquisición satelital. Los resultados de esta investigación indican que la configuración más adecuada para el monitoreo de la dinámica hidrológica de los humedales del delta del Río Paraná corresponde a la señal en banda L, con polarización HH y ángulo de incidencia empinado (<28°). Esto se debe a que la combinación de la banda L con ángulos de incidencia verticales permite una mayor penetración de la señal a través de la cobertura vegetal de las cortaderas, aportando información más sensible del estado de inundación del humedal.
Hyperspectral target detection (HTD) faces significant challenges, including the dimensionality and substantial spectral variability. The objective of this work is to develop robust and computationally efficient matched filter-based detectors that overcome these limitations through ensemble learning. Although conventional matched filters (MF) are widely used as a primary and straightforward method in this field, they suffer from covariance matrix instability in high-dimensional spaces and spectral confusion in complex target-background environments. Therefore, this paper presents three improved ensemble-based variants of MF: the random subspace matched filter (RS-MF), its enhanced version with an adaptive target signature (Adaptive RS-MF), and the selective subspace matched filter (Selective RS-MF). Experiments conducted on three benchmark hyperspectral target detection datasets demonstrate that the proposed variants not only improve detection performance over the conventional MF but also outperform other existing HTD methods. The final results show that Selective RS-MF achieves an area under the receiver operating characteristic curve (AUC-ROC) of approximately 0.99, providing robust and effective solutions for HTD in diverse environments, while maintaining computational efficiency.
La hidrovía Paraguay-Paraná es un corredor estratégico para Sudamérica ya que conecta con las rutas oceánicas y posibilita el comercio exterior de sus países. El Puerto San Pedro (Buenos Aires, Argentina) se localiza en el km 274 de la misma, sobre la margen derecha del río Paraná, en una zona con una hidrodinámica compleja debido a las interacciones entre el cauce principal con sistemas secundarios. En mayo de 2025, se registraron precipitaciones extremas que afectaron la operatividad del puerto, debido a la ocurrencia de depósitos de sedimentos que impidieron la navegabilidad en su canal de acceso. En este estudio se determinó la dinámica hídrica y sedimentológica en las inmediaciones del Puerto San Pedro durante el evento de precipitaciones extremas utilizando imágenes satelitales y datos in situ. Con imágenes de la misión satelital SWOT (Surface Water and Ocean Topography) se estimaron las pendientes de la superficie del agua en el sistema, las cuales constituyeron un dato clave para la estimación del caudal y del transporte de sedimentos. Complementariamente, con imágenes ópticas de las misiones Sentinel 2 y Planet se analizó la evolución hidro-sedimentológica del sistema. Asimismo, se comparó la granulometría del sedimento del lecho, antes y después de ocurrido el evento. Los resultados demuestran la dinámica particular del escurrimiento generada durante el evento, con la ocurrencia de una inversión del flujo que favoreció un transporte de sedimentos significativo hacia el canal de acceso al puerto, afectando la operatividad del mismo. Este trabajo resalta la importancia de comprender la dinámica hidro-sedimentológica vinculada a eventos hidrometeorológicos extremos en sistemas fluviales complejos, para prevenir y mitigar los posibles impactos socio-económicos asociados (e.g. en vías navegables y puertos). Asimismo, se resalta en este estudio, el aporte clave de la información remotamente sensada proporcionada por la misión SWOT para la estimación de diferentes parámetros hidráulicos.
The increasing availability of high-resolution satellite imagery, georeferenced field measurements, and big data management tools facilitates the development of predictive models that improve agricultural planning. This type of support is particularly valuable in regions like the central-western pampas (Argentine) where the adoption of cover crops such as Vicia villosa Roth is limited due to their high yield variability. This study evaluates the individual capacity of different vegetation indices derived from Sentinel-2 imagery to serve as input variables in models for predicting vetch seed yield. Four groups of indices, calculated from different spectral bands and across multiple dates between vegetation growth and seed maturation were considered. For each index, temporal observations were stacked to form a multiband image and associated with yield measurements recorded by harvesters. Using this information, supervised classification models were trained and validated using the Random Forest algorithm. The study was conducted with available data from seven fields in the region. Results showed that indices derived from red-edge bands exhibited the highest discriminative capacity, even when the input stack was reduced to observations from the final stages of the crop phenological cycle. In addition, certain indices incorporating corrections related to soil background effects or photosynthetically active area also achieved high classification performance. This analysis contributes to establishing a solid base for developing a predictive tool applicable to vetch seed production in central-western Pampas.
During 2024, wildfires in the Pantanal, Brazil, posed a significant threat to the world’s largest wetland. This study presents a scene classification approach using Sentinel-2 imagery from fire-affected and non-affected areas, based on deep learning techniques. First, a model was experimentally trained using the EuroSAT dataset to classify grasslands and herbaceous vegetation. Subsequently, a transfer learning technique was applied to images of the fire-affected Pantanal. Finally, the VGG-19 architecture was trained from scratch, without using pretrained parameters. Considering these three experiments, training with the EuroSAT dataset achieved a loss of 0.202 after 50 epochs. The model with transfer learning, using the weights learned during the experimental training, obtained a validation loss of 0.15 and an accuracy of 95%, whereas training from scratch reached a validation loss of 0.07 and an accuracy of 97%. The results indicate similar performance between the two strategies, demonstrating that in-domain training can be considered an effective method for training CNNs applied to satellite imagery, without compromising classification accuracy, even in the presence of significant differences in the contextual information of the images. This approach is presented as a methodological alternative for environmental monitoring in complex tropical regions, characterized by high spatial and spectral heterogeneity.
Deforestation, caused by human activities, leads to the loss of vegetation and natural resources that are essential to ethnic communities. This study addressed the lack of information on the anthropogenic processes of deforestation in ethnic territories by analyzing its causes. The deforestation rate and the geospatial distribution of socioeconomic activities were evaluated between 2016 and 2023. A generalized linear model (GLM) with gamma distribution and logarithmic link function was implemented to explain the causes of deforestation. Between 2016 and 2018, the highest rate of deforestation was in the indigenous community. In this community, agriculture (40 degrees%o) and forestry (33 degrees%o) were the main economic activities, while in the afro-descendant community, mining (29 degrees%o) and forestry (26 degrees%o) were the most important. The GLM showed that deforestation increases with distance from existing forests, indicating that more remote areas are targeted for new economic activities, such as agricultural expansion or timber extraction. This geospatial analysis highlights the relationship between socioeconomic activities and the loss of vegetation cover in ethnic territories.
The transient yet abrupt reduction of the ozone layer known as a "mini-ozone hole" can locally increase surface ultraviolet (UV) radiation, with potential implications for human health and ecosystems. In South America these phenomena have been sparsely documented, underscoring the need for detailed studies. Here we present the first characterization of a mini-ozone hole detected in March 2025 at the Observatorio Atmosf & eacute;rico de la Patagonia Austral (OAPA), R & iacute;o Gallegos, Argentina (51.55 degrees S, 69.23 degrees W) and assess its surface radiative impact. We employed satellite observations Tropospheric Monitoring Instrument (TROPOMI/Sentinel-5 Precursor) and Ozone Monitoring Instrument (OMI/NASA EOS-Aura), ground-based measurements Syst & egrave;me d'Analyse par Observation Z & eacute;nithale spectrophotometer (SAOZ) and YES UVB-1 solar radiometer and a reanalysis product from Multi Sensor Reanalysis version 2 (MSR-2). Long-term climatologies for the study area were computed to identify anomalous total ozone column (TOC) events and to quantify the response of the UV index (UVI) and the daily erythemal dose. Event detection used a climatological threshold (mu-2 sigma) and was complemented by UVI radiative simulations generated with a parametric model. Results show a simultaneous decrease in TOC recorded by SAOZ (-16%) and TROPOMI (-19%) relative to the climatological value on 24 March 2025, reaching percentiles below 1%. This reduction corresponded to a theoretical increase in the erythemal dose of +30% under clear-sky conditions compared with the climatology for that day; observed values showed a more moderate rise (+9%), attributable to attenuation by clouds, given that the troposphere above the site has a very low aerosol content. Satellite maps confirmed the presence and spatial evolution of the mini-ozone hole over southern South America. This study demonstrates the value of integrated satellite-ground monitoring for detecting extreme ozone events at subpolar latitudes. The proposed methodology is transferable to other regions and contributes to improved understanding of the risks associated with acute UV
The water quality of the high-Andean lakes in Ecuador has been scarcely studied using remote sensing due to multiple factors. Among these, the difficult access stands out, which limits the acquisition of field data necessary for model calibration and validation. Additionally, the high cloud cover and the high sunglint risk of the lakes located close to the equatorial line, aggravated by the daily mountain breeze, further complicate the acquisition of useful images. In this study, the concentration of chlorophyll-a was assessed in two distinct water bodies: the oligotrophic Lake Atillo and the hypereutrophic Yambo Lake. For this purpose, in situ data were compared with automatic products derived from Sentinel-2 images processed using four atmospheric correction methods: ACOLITE and the three variants of C2RCC (C2RCC, C2X, and C2X-COMPLEX). The results indicate that, in Atillo, the best performance was achieved with the standard version of C2RCC, followed by the chl_oc2 model implemented in ACOLITE. Conversely, in Yambo, the best statistical results were obtained with C2X-COMPLEX, followed by the chl_re_bramich model in ACOLITE. Although ACOLITE showed slightly inferior statistical results in both lakes, its ability to effectively correct sunglint makes it, along with C2RCC, a valuable tool for monitoring eutrophication in these systems. It is worth noting that the use of free software (ACOLITE and C2RCC) and openly accessible images (Sentinel-2) facilitates the implementation of temporal monitoring and spatial analysis programs for chlorophyll-a in these high-Andean lakes, offering a viable alternative for assessing their trophic status.
La reducción transitoria pero abrupta de la capa de ozono conocida como “mini-agujero de ozono”, puede incrementar de forma puntual la radiación ultravioleta en superficie, con efectos potenciales sobre la salud y los ecosistemas. En Sudamérica estos fenómenos han sido escasamente documentados, lo que resalta la necesidad de estudios detallados. En este trabajo se presenta la primera caracterización de un mini-agujero detectado en marzo de 2025 sobre el Observatorio Atmosférico de la Patagonia Austral (OAPA), Río Gallegos, Argentina (51,55° S, 69,23° O) y se evalúa su impacto radiativo en superficie. Se utilizaron observaciones satelitales del Tropospheric Monitoring Instrument (TROPOMI/Sentinel-5 Precursor) y Ozone Monitoring Instrument (OMI/NASA EOS-Aura), mediciones de superficie del espectrofotómetro Système d’Analyse par Observation Zénithale (SAOZ) y del radiómetro UVB-1 YES, y el producto de reanálisis Multi Sensor Reanalysis versión 2 (MSR-2). Se calcularon climatologías de largo plazo en el área de estudio con el fin de identificar eventos anómalos de la columna total de ozono (CTO) y cuantificar la respuesta del índice UV (IUV) y de la dosis eritémica diaria. La detección del evento se basó en un umbral climatológico (μ−2σ) y se complementó con simulaciones radiativas del IUV modeladas con un modelo paramétrico. Los resultados muestran una disminución simultánea de la CTO registrada por SAOZ (−16%) y TROPOMI (−19%) respecto del valor climatológico del día 24 de marzo 2025, alcanzando percentiles <1%. Esta reducción se tradujo en un incremento teórico de la dosis eritémica de +30% bajo condiciones despejadas respecto a la climatología para ese día, mientras que las observaciones reales indicaron un aumento más moderado (+9%), atribuible a la atenuación por nubosidad, dado que la tropósfera sobre el sitio tiene típicamente muy escaso contenido de aerosoles. Los mapas satelitales confirmaron la presencia y evolución espacial del mini-agujero sobre el sur de Sudamérica. Este estudio evidencia la utilidad del monitoreo integrado satélite–superficie para la detección de eventos extremos de ozono en latitudes subpolares. La metodología propuesta es transferible a otras regiones y contribuye al entendimiento de los riesgos asociados a episodios agudos de exposición a radiación UV.
Tesis doctoral Autor: Bryan Alemán-Montes Directores: Dr. Pere Serra Ruiz y Dra. Alaitz Zabala Torres Lugar: Universitat Autònoma de Barcelona Fecha: 19/09/2025 Calificación: Sobresaliente Cum Laude Disponible: http://hdl.handle.net/10803/695313
La deforestación, causada por actividades antrópicas, provoca la pérdida de vegetación y recursos naturales esenciales para las comunidades étnicas. Este estudio abordó la falta de información sobre los procesos antrópicos de la deforestación en territorios étnicos, analizando sus causas. Se evaluó la tasa de deforestación y la distribución geoespacial de actividades socioeconómicas, entre 2016 hasta 2023. Se implementó un modelo lineal generalizado (GLM) con distribución gamma y función de enlace logarítmico para explicar las causas de la deforestación. Entre 2016 y 2018, la mayor tasa de deforestación fue en la comunidad indígena. En esta comunidad, la agricultura (40%) y el aprovechamiento forestal (33%) fueron las actividades económicas principales, mientras que, en la comunidad afrodescendiente, la minería (29%) y el aprovechamiento forestal (26%) eran las más importantes. El GLM mostró que la deforestación aumenta a mayor distancia de los bosques existentes, indicando que áreas más alejadas son objetivo de nuevas actividades económicas, como la expansión agrícola o la extracción de recursos maderables. Este análisis geoespacial resalta la relación entre las actividades socioeconómicas y la pérdida de cobertura vegetal en territorios étnicos.
La calidad de las aguas de las lagunas altoandinas de Ecuador ha sido escasamente estudiada mediante teledetección debido a múltiples factores. Entre estos se destaca la dificultad de acceso, que limita la obtención de datos de campo necesarios para la calibración y validación de modelos. Adicionalmente, la alta nubosidad y el elevado riesgo de sunglint debido a la proximidad de estas lagunas a la línea ecuatorial, agravado por el viento frecuente, complican aún más la adquisición y uso de imágenes útiles. En este trabajo, se evaluó la concentración de clorofila-a en dos cuerpos de agua: la laguna oligotrófica de Atillo y la hipereutrófica de Yambo. Para ello, se compararon datos in situ con productos automáticos derivados de imágenes Sentinel-2, procesadas mediante cuatro métodos de corrección atmosférica: ACOLITE y las tres variantes de C2RCC (C2RCC, C2X y C2X-COMPLEX). Los resultados indican que, en Atillo, el mejor rendimiento se obtuvo con la versión estándar de C2RCC, seguida del modelo chl_oc2 implementado en ACOLITE. Por el contrario, en Yambo, los mejores estadísticos correspondieron a C2X-COMPLEX, seguido del modelo chl_re_bramich de ACOLITE. Si bien ACOLITE mostró resultados ligeramente inferiores en términos estadísticos en ambas lagunas, su capacidad para corregir eficazmente el sunglint lo convierte —junto con C2RCC— en una herramienta valiosa para el monitoreo de la eutrofización en estos sistemas. Cabe destacar que el uso de software libre (ACOLITE y C2RCC) e imágenes de acceso gratuito (Sentinel-2) facilita la implementación de programas de seguimiento temporal y análisis espacial de clorofila-a en estos lagos altoandinos, ofreciendo una alternativa viable para el análisis de su estado trófico.
Do & ntilde;ana National Park is located in the southwest of the Iberian Peninsula. This protected area faces significant environmental challenges due to more frequent and extreme drought events and plays a key role as a biodiversity hotspot. Understanding carbon dynamics is essential to advance our knowledge of climate change effects on natural land covers and the ecosystem services they provided. In this study, a Light Use Efficiency (LUE) model has been applied to estimate Gross Primary Production (GPP) for two types of featured ecosystems of Do & ntilde;ana: xeric shrubland, characterized by its summer drought resistance, and the seasonal marshes, which flood and dry up every year, and it is the habitat of grasslands and aquatic plants dependent on hydroperiod. Model validation was carried out using in-situ data collected by Energy and Carbon Flux Towers (Eddy covariance, EC) installed in both ecosystems. LUE model inputs were based on remote sensing and reanalysis data: i) Sentinel-2 Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), ii) solar radiation reanalysis data (ERA5-Land), and iii) Sentinel-2 Land Surface Water Index (LSWI). Our methodology has proven to be accurate, with root mean square error (RMSE) lower than 0.50 g C / m2 and coefficients of determination (R2) of 0.82 for marshes and 0.67 for xeric shrublands. Once validated, this remote sensing and LUE model-based approach enabled the long-term monitoring of carbon dynamics for two different ecosystems of Do & ntilde;ana, contributing to a better understanding of the responses of these natural systems to climate change during the study period.
Object-based image analysis (OBIA) is increasingly employed to enhance land cover classification accuracy from satellite imagery. This gap is addressed by systematically evaluating three leading boosting models - Gradient Boosting (GB), Extreme Gradient Boosting (XGBoost), and Light Gradient Boosting (LightGBM) for mapping six land cover classes in a complex mountainous landscape in Yen Bai province, Vietnam. We implement an OBIA using the Simple Non-Iterative Clustering (SNIC) algorithm for image segmentation on Sentinel-2 data. Spectral bands and derived indices (NDVI, NDBI, MBI) were used as predictive features. The findings demonstrate that all three models achieve high accuracy, with XGBoost emerging as the superior model, yielding an Overall Accuracy (OA) of 0.903 and a Kappa coefficient of 0.884. A key finding from the variable importance analysis is the differing feature reliance among algorithms: while GB and XGBoost prioritized the raw spectral information from the Red band (Band 4), LightGBM favored the derived Normalized Difference Built-up Index (NDBI). This research provides critical insights into the comparative strengths of boosting algorithms in an object-based framework, offering a valuable reference for selecting optimal models for high-resolution land cover monitoring initiatives.
Maize (Zea mays L.) is a fundamental cereal in global food security, but its vulnerability to water stress compromises its productivity and threatens food availability. This study analyzed the relationship between the crop water stress index (CWSI), obtained from thermal images captured by the Zenmuse H20T camera, and various vegetation indices derived from the MicaSense RedEdge-MX Dual. The analysis included machine learning (ML) models such as random forest (RF), k-nearest neighbors (KNN), and gradient boosting regression (GBR). The results showed that RF was the most accurate model for predicting CWSI in maize, with a coefficient of determination (R²) of 0.80, a root mean square error (RMSE) of 0.13, and a mean absolute error (MAE) of 0.09. KNN achieved an R² of 0.78, an RMSE of 0.13, and an MAE of 0.09, while GBR reached an R² of 0.79, an RMSE of 0.14, and an MAE of 0.10. The red band (668 nm) played a crucial role in RF (70.69%) and GBR (50.92%), whereas in KNN, the simple ratio (SR) index showed the highest importance (36.40%). These findings confirm the superiority of ML models over traditional regression approaches for estimating CWSI in maize. Despite the satisfactory results, the algorithms underestimated CWSI values derived from thermal images, which highlights the need to refine these models to improve their accuracy in future agricultural applications.
El Parque Nacional de Doñana, situado en el suroeste de la Península Ibérica, se enfrenta a unos desafíos ambientales sustanciales, con eventos extremos de sequía cada vez más frecuentes y un papel clave como hotspotde biodiversidad. La comprensión de la dinámica del carbono es esencial para avanzar en el conocimiento sobre los efectos del cambio climático en las cubiertas naturales y en los servicios ecosistémicos que proveen. En este estudio, se ha aplicado un modelo de eficiencia en el uso de la luz (LUE) para estimar la producción primaria bruta (PPB) en dos tipos de ecosistemas representativos de Doñana: el matorral xerófilo, caracterizado precisamente por su resistencia a la sequía estival, y la marisma estacional, que sufre ciclos de inundación y desecación y alberga pastizales y vegetación acuática dependientes del hidroperiodo. La validación del modelo se realizó con datos in situ medidos a través de torres de flujos de energía y carbono (Eddy Covariance, EC) instaladas en ambos ecosistemas. En la aplicación del modelo LUE se han usado únicamente datos procedentes de teledetección y de reanálisis: i) Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) de Sentinel-2, ii) datos de reanálisis de radiación solar (ERA5-Land) y iii) el índice multiespectral Land Surface Water Index (LSWI) para las imágenes de Sentinel-2. Esta metodología ha demostrado ser muy adecuada, con valores de error cuadrático medio (RMSE) inferiores a 0,50 g C/m2 y coeficientes de determinación (R2) de hasta 0,82 en la marisma y 0,67 en el matorral xerófilo. Una vez validado, este método basado en el modelo LUE y con datos de teledetección y reanálisis permitió efectuar el seguimiento de la dinámica de carbono en Doñana, contribuyendo así a una mejor comprensión de la respuesta de estos sistemas naturales frente a los eventos extremos en el periodo analizado.
In the hyperspectral remote sensing images processing, band selection is an essential task for many specific applications, including supervised classification. The objective of this work is to compare the performance of the classical strategy, which involves variable selection as a preliminary step to classification, with new proposals of penalized algorithms that perform classification and variable selection simultaneously. For the comparison, an extract of a hyperspectral image EO-1 Hyperion, covering an area in the province of C & oacute;rdoba, Argentina, was used. Additionally, a simulation study was conducted. The obtained results show that penalized algorithms are more effective in selecting relevant bands while providing good predictive properties, mainly in the context of high dimensionality, that is, when the size of the training sample is small relative to the number of variables.
Shrub communities of Cistus ladanifer L. (gum rockrose) are one of the most characteristic, extensive, and prone to wildfire of Mediterranean ecosystems. In addition, these shrublands have a remarkable potential for the extraction of subproducts, which are highly valuable in the pharmaceutical, food and cosmetic industries. Therefore, estimating their biomass is essential to manage and prioritize their use, to calculate their carbon content and CO2 capture as well as predicting their fire behaviour. In this study, we aim to estimate the fuel load of gum rockrose shrublands in southern Spain based on airborne LiDAR data from National Aerial Orthophotography Plan (PNOA), considering fuel load as the total accumulated biomass per unit area in this vegetation type. For this purpose, non-destructive field inventories were carried out with measurements of mean height and shrub cover in 143 circular plots in Andalusia region. These two fuel variables were used as inputs in an existing specific equation to estimate the fuel load for C. ladanifer. Two different approaches were compared to estimate the fuel load of gum rockrose shrublands by means of linear regression analysis: (i) direct estimation (DE), consisting of the adjustment that directly relates fuel load to LiDAR data; and (ii) indirect estimation in two steps (IE), based on the adjustment of equations to estimate the input variables (shrub height and cover) from LiDAR data. Better goodness-of-fit statistics were obtained in the DE equation than in the IE approach, explaining 72 degrees%o and 70 degrees%o of the observed variability, respectively. These results can be valuable for the development of gum rockrose biomass mapping for use in fire prevention and suppression as well as in harvesting planning for the extraction of their products.
Spain is one of the largest wine producers in the world, therefore, viticulture is key to its economy. The Spanish wine industry has incorporated remote sensing techniques in the different stages of production, mostly aimed at vegetation mapping, pest detection and disease control, however, there are few studies related to the determination of production and yield in vineyards. For this reason, based on various vegetation spectral indices NDVI, NDRE, LAI, MSAVI2, TCARI, OSAVI, among others, and values of Leaf Area Index, LAI, different non-parametric models were generated, using principal component analysis and neural networks, which have been widely studied and implemented in various fields. The products obtained showed an estimation error RMSE of 16.19 t and 5.53 t/ha, in relation to productivity and yield respectively, from the analysis of principal components, and, 10.32 t and 4.23 t/ha, respectively, in the case of neural networks, showing an improvement when using this last technique. This study was carried out in the vineyards of Vi & ntilde;a Arnaiz, located in the municipality of Haza (Burgos).
