Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 15
Encontrada(s) 1 página(s)

Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Modelling the impact land use change on flood risk: Umia (Spain) and Voglajna (Slovenia) case studies

  • Acuña Alonso, Carolina
  • Álvarez Bermúdez, Xana
  • Bezak, Nejc
  • Zupanc, Vesna
Accurate quantification and assessment flood hazards is critical for mitigation and prevention. This study proposes a methodological framework for analysing this hazard and its relationship to agricultural and forestry land use. The objective is to obtain an integrative methodology based on the HEC-HMS model and to apply different simulations, with land use changes as indicator in two different study areas, taking into account the different characteristics of each basin and the different regulations of each area. This was done in two river basins, the Umia Basin (north-western Spain) and the Voglajna Basin (eastern Slovenia). The hydrological models obtained showed a very high performance in the calibration and validation periods. In the Umia River Basin agricultural use has priority over forestry, with food security taking precedence over water security. Reforestation only upstream has been shown to be almost as beneficial as reforestation throughout the basin. However, the use of abandoned land currently limits the reduction of peak discharge. Nevertheless, for an event designed with a 100-years return period, a reduction of about 12% was achieved. The increase in agricultural use promoted by the administration could increase this peak by about 6%, thus increasing the flood hazard, but it has been shown that this use on less permeable soils and upstream would not significantly increase this peak (<1% for the event studied). On the other hand, the Voglajna Basin has a smaller catchment area and fragmented land use, and a mosaic landscape. For the 100-years return period the reduction in peak discharge is only few percent compared to the baseline scenario even if 30% of the agricultural land (about 7% of the total area) is changed to forest land use. The information provided by the simulations is a useful indicator that can be incorporated into management plans to ensure appropriate decision making by the administration. Not only for the application of nature-based solutions (NBS) and providing evidence-base for the NBS and the reduction of flood risk, but such information is key when the use changes from forestry to agriculture and vice versa, as it provides tools to ensure food and water security. In both case studies it was demonstrated how stakeholders need to undertake optimal and strategic planning and management in order to reduce the risk of flooding. In turn, the use of this modelling, as well as the calculation of scenarios from a perspective that not only evaluates different land use changes, but also incorporates different regulations, is presented as an innovative and realistic analysis., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. GPC-ED431B 2022/12, Xunta de Galicia | Ref. ED481B-2023-042, Universidade de Vigo/CISUG, Slovenian Research Agency | Ref. V2-2137, Slovenian Research Agency | Ref. J6-4628, Slovenian Research Agency | Ref. N2-0313, Slovenian Research Agency | Ref. P2-0180, Slovenian Research Agency | Ref. P-0048
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Application of remote sensing to understand the role of Galician feral horses in the biomass reduction of a shrub-grassland-dominated landscape

  • Janeiro Otero, Andrea
  • Álvarez Bermúdez, Xana
  • Dormann, Carsten F.
Galician forests in northwestern Spain are subject to frequent wildfires with high environmental and economic costs. In addition, due to the consequences of climate change, these fires are becoming more virulent, occurring throughout the year, and taking place in populated areas, in some cases involving the loss of human life. Therefore, forest fire prevention is even more relevant than mitigating its consequences. Given the costs involved in forestry work, alternative measures to reduce fuel load and create vegetation gaps are needed. One involves grazing by an endemic species of feral horses (Equus ferus atlanticus) that feed on thicket-forming gorse (Ulex europaeus). In a 100-ha forest fenced study area stocked with 11 horses, four 50 m2 enclosed plots prevented the access of these wild animals to the vegetation, with the aim of manipulating their impact on the reduction of forest biomass. The measurement of biomass volumes is an important method that can describe the assessment of wildfire risks, unfortunately, high-resolution data collection at the regional scale is very time-consuming. The best result can be using drones (unmanned aerial vehicles - UAVs) as a method of collecting remotely sensed data at low cost. From September 2018 to November 2020, we collected information about aboveground biomass from these four enclosed plots and their surrounding areas available for horses to forage, via UAV. These data, together with environmental variables from the study site, were used as input for a fire model to assess the differences in the surface rate of spread (SROS) among grazed and ungrazed areas. Our results indicated a consistent but small reduction in the SROS between 0.55 and 3.10 m/min in the ungrazed enclosured plots in comparison to their grazed surrounding areas (which have an SROS between 15 and 25 m/min). The research showed that radar remote sensing (UAV) can be used to map forest aboveground biomass, and emphasized the importance and role of feral horses in Galicia as a prevention tool against wildfires in gorse-dominated landscapes., Agencia Estatal de Investigación | Ref. TED2021-130241 A-I00
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Compensating local communities for mining transport: a game theoretical analysis

  • Valencia Toledo, Alfredo
  • Vidal Puga, Juan José
We propose a freight transportation problem to distribute the benefit provided by a corridor to compensate the communities on the route. Such distribution aims to avoid conflicts generated by the maintenance of the corridor or the ore transportation that causes external costs to communities that can block the route, forcing more expensive alternatives. We propose a cooperative game theory solution based on stability and fairness principles. In particular, we present some reasonable properties to characterize a family of core assignment rules for determining compensation to the local communities in the corridor. We illustrate the method with a case study using data from Peru., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Agencia Estatal de Investigación | Ref. PID2020-113440GB-I00, Fondo Nacional de Desarrollo Científico y Tecnológico (Chile), Xunta de Galicia | Ref. ED431B 2022/03, Universidade de Vigo/CISUG
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Development of a convolutional neural network to accurately detect land use and land cover

  • Acuña Alonso, Carolina
  • García Ontiyuelo, Mario
  • Barba Barragáns, Diego
  • Álvarez Bermúdez, Xana
The detection and modeling of Land Use and Land Cover (LULC) play pivotal roles in natural resource management, environmental modeling and assessment, and ecological connectivity management. However, addressing LULCC detection and modeling constitutes a complex data-driven process. In the present study, a Convolutional Neural Network (CNN) is employed due to its great potential in image classification. The development of these tools applies the deep learning method. A methodology has been developed that classifies the set of land uses in a natural area of special protection. This study area covers the Sierra del Cando (Galicia, northwest Spain), considered by the European Union as a Site of Community Interest and integrated in the Natura 2000 Network. The results of the CNN model developed show an accuracy of 91 % on training dataset and 88 % on test dataset. In addition, the model was tested on images of the study area, both from Sentinel-2 and PNOA. Despite some confusion especially in the residential class due to the characteristics in this area, CNNs prove to be a powerful classification tool, Xunta de Galicia | Ref. ED481B-2023-042, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Geospatial mapping of carbon estimates for forested areas using the InVEST model and Sentinel-2: A case study in Galicia (NW Spain)

  • García Ontiyuelo, Mario
  • Acuña Alonso, Carolina
  • Valero Gutiérrez del Olmo, Enrique Maria
  • Álvarez Bermúdez, Xana
CO2 emissions have increased exponentially in recent years, so measuring and quantifying carbon sequestration is a step towards sustainable forest management and combating climate change. The overall goal of this study is to develop an accurate model for estimating carbon storage and sequestration for forest areas of the Atlantic Biogeographic Region. Specifically, the modelling and field sampling are carried out in the municipality of Baiona (Galicia, NW Spain), which was selected as a representative biome of this region. The methodology consists of carrying out two object-based image analysis (OBIA) classifications in spring and autumn to observe possible stocks of seasonal differences. Two carbon storage and sequestration models are built up (model 1 and model 2): model 1 for forest areas only and model 2 including all other land cover in the study area. Sentinel-2 geospatial data for 2021, Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) tools and geographic information systems (GIS) are used. A Kappa index of 0.92 is obtained for both classifications, thus ruling out any notable seasonal differences in the images used. The results from both models indicate that it is land covers associated with forest uses which store the most carbon in the study area, accounting for >50 % more than the other land covers. It is concluded that the methodology and data used are very useful for quantifying ecosystem services, which will help the governance of the region by implementing measures to mitigate some of the effects of climate change and help to create silvicultural models for the sustainable management of the Atlantic Biogeographic Region., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. GPC-ED431B 2022/12, Xunta de Galicia | Ref. ED481B-2023-042, Universidade de Vigo/CISUG
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Stable and weakly additive cost sharing in shortest path problems

  • Bahel, Eric
  • Gómez Rúa, María
  • Vidal Puga, Juan José
In a shortest path problem, agents seek to ship their respective demands; and the cost on a given arc is linear in the flow. Previous works have proposed cost allocations falling in the core of the associated cooperative game. The present work combines core selection with weak versions of the additivity axiom, which allows to characterize a new family of rules. The demander rule charges each demander the cost of their shortest path, and the supplier rule charges the cost of the second-cheapest path while splitting the excess payment equally between access suppliers. With three or more agents, the demander rule is characterized by core selection and a specific version of cost additivity. Convex combinations of the demander rule and the supplier rule are axiomatized using core selection, a second version of cost additivity, and two additional axioms that ensure the fair compensation of intermediaries. With three or more agents, the demander rule is characterized by core selection and a specific version of cost additivity. Finally, convex combinations of the demander rule and the supplier rule are axiomatized using core selection, a second version of cost additivity, and two additional fairness properties., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Agencia Estatal de Investigación | Ref. PID2020-113440GB-I00, Xunta de Galicia | Ref. GPC-ED431B 2022/03
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Past and present in the ecological connectivity of protected areas through land cover and graph-based metrics

  • Vidal Llamas, Antonio
  • Acuña Alonso, Carolina
  • Álvarez Bermúdez, Xana
Habitat reduction is significantly threatening biodiversity, making ecological connectivity which facilitates species movement across habitat patches, essential for human impacts mitigation, promoting genetic exchange, and enabling colonization of new areas. Ecological connectivity in the River Lérez Basin (Galicia, NW Spain), including three Natura 2000 sites: River Lérez, Serra do Cando, and Serra do Candán, was assessed. Land cover maps for the years 2013 and 2023 were created using Landsat 8-9 images and the random forest machine learning method. Hardwood forest habitat patches and ecological corridors were identified. Betweenness Centrality (BC) metric, along with other global structural connectivity indices such as Integral Connectivity Index (IIC), Connectivity Probability (CP), and Equivalent Connectivity (EC), were calculated. iNaturalist records were then downloaded to observe how the distribution of records resembled that of habitat and corridors for 2023. Some indices, such as the splitting patch index (SPLIT), effective meshsize (EM), edge length (EL), and edge density (ED), were also calculated at the patch level. Results showed that the area of hardwoods increased by 2.16% across the basin. The spatial patch patterns were similar in both years. Additionally, hardwood structural connectivity appeared to have improved over the study period, with IIC rising in 20.00%, PC increase by 16.67–18.92%, and EC 8.04–8.68%. However, some patches from 2013 had higher BC values due to the loss of certain connections. iNaturalist record distribution showed similarities with habitat patch and path distribution, with a lower average distance (406.06 m) compared to random points (854.12 m) for 2023. The patch indices indicated a reduction in fragmentation, with SPLIT decreasing by 53.95%, EM and EL increasing by 173.66% and 14.21%, respectively, while ED decreased by 9.29%. The integration of satellite imagery has been proven effective for generating land cover data for connectivity analysis. It also demonstrates that indices and graph-based approaches offer a data-efficient alternative to traditional models. Furthermore, the incorporation of citizen science platforms, despite some biases, complements technical methods by providing real-world insights into species distribution. This combination is considered a promising approach for future research, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Agencia Estatal de Investigación | Ref. PID2022-138374OA-I00, Universidade de Vigo/CISUG
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Influence of protected areas in the carbon cycle: a machine learning approach for future forecasting

  • Lago González, Pedro
  • Acuña Alonso, Carolina
  • Álvarez Bermúdez, Xana
Protected Areas (PAs) are destined to the conservation of natural resources, whose quantity and quality are menaced by Climate Change and anthropogenic pressure. The objective of this research was to quantify and forecast the Carbon Absorption of PAs, attending these factors. We used the Net Primary Productivity (NPP) and implemented machine learning algorithms, as Random Forests (RF), Gradient Boosting Trees (GBT), and Multilayer Perceptrons (MLPNN), to forecast it in four differentiated PAs of Galicia (NW Spain): the Central Massif, the Sil Canyons, Fragas do Eume, and the Tambre River. Two testing stages were carried, one in the areas where the model was trained, and another in the whole territory. Finally, we set several scenarios based on projections SSP2-4.5 and SSP 5-8.5 in and land use changes. GBT was the most accurate algorithm, with a Root Mean Squared Error (RMSE) of 0.05 kgC/m 2 (5.7% of the average NPP) and correlation of 0.9. RF obtained an error of 0.07 and correlation of 80%, and MLPNN 0.06 and 86%, respectively. The GBT obtained a RMSE of 0.04 kgC/m 2 and R 2 of 0.95 in the first test. These results were worsened in the second test, with an RMSE of 0.09 kgC/m 2 and 71% of correlation. For the SSP2-4.5 scenarios, a decrease around 7% can be expected, barely influenced by the land use. On the other hand, the SSP5-8.5 is expected to record a decrease of 5% in the NPP, with no significant differences between the land uses, but with a significant trend. All the pilot sites stick to this trend, except for the Tambre River, which was forecasted to record a slight increase of the NPP (~ 3%). The prediction and analysis of future scenarios can help the management of the territory, focusing on the mitigation of the effects of Climate Change in PAs. Graphical Abstract, Universidade de Vigo/CISUG, Agencia Estatal de Investigación | Ref. PID2022-138374OA-I00, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. ED481B-2023-042
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Contribución de congreso (ConferenceOutput)

Detección de cambios en el uso del suelo en el Corredor Ecológico Río Tea (NO España), Detection of Land Use Change in the River Tea Ecological Corridor (NW Spain)

  • García Ontiyuelo, Mario
  • Acuña Alonso, Carolina
  • Valero Gutiérrez del Olmo, Enrique Maria
  • Álvarez Bermúdez, Xana
Los cambios de uso del suelo y de la cobertura terrestre (LULCC) tienen una fuerte interrelación entre ellos, llegando a alterar los servicios ecosistémicos y contribuir negativamente al cambio climático. El objetivo general de este estudio es analizar diferentes fuentes de información para cuantificar el LULCC en el LIC Río Tea (NO de España) en el periodo 2015-2023. El área de estudio presenta múltiples coberturas con muy baja variabilidad espectral entre ellas y con una gran fragmentación del territorio, lo que dificulta la obtención de altos niveles de precisión. La clasificación de las coberturas del suelo se realizó mediante metodologías de Clasificación por Análisis de Imágenes Basadas en Objetos (OBIA) y Redes Neuronales Artificiales (ANN) a partir de imágenes de las plataformas de satélites multiespectrales Sentinel-2 y Planet Labs (RapidEye y PlanetScope). La precisión en la clasificación de las coberturas terrestres obtenida para los datos de Planet Labs y la metodología OBIA fue de un 80%, mientras que Sentinel-2 proporcionó una precisión de un 70%, empleando la metodología de ANN con los datos de Planet Labs, la precisión se sitúa en torno al 55%. Así pues, los datos utilizados y la metodología seguida influyen en los niveles de precisión obtenidos en las clasificaciones. Por último, se concluye que los datos geoespaciales disponibles son muy útiles para detectar y cuantificar los cambios en la ocupación del suelo. No se detectan grandes cambios en las coberturas terrestres, el principal cambio detectado es una transición de las coberturas vinculadas a los cultivos, las cuales en períodos de barbecho se detectan como suelo desnudo. Se confirma que la metodología utilizada contribuye a la planificación territorial y a la gestión forestal sostenible, facilitando futuras decisiones y planes de acción en la gobernanza de la región, Land use and land cover change (LULCC) is strongly interrelated, altering ecosystem services and contributing negatively to climate change. The overall objective of this study is to analyse different sources of information to quantify LULCC in the Río Tea SCI (NW Spain) in the period 2015-2023. The study area has multiple land covers with very low spectral variability between them and with a high fragmentation of the territory, which makes it difficult to obtain high levels of accuracy. Land cover classification was carried out using Object Based Image Analysis Classification (OBIA) and Artificial Neural Network (ANN) methods based on images from the Sentinel-2 and Planet Labs multispectral satellite platforms (RapidEye and PlanetScope). The accuracy of land cover classification obtained using Planet Labs data and the OBIA method was 80%, while Sentinel-2 gave an accuracy of 70% and the ANN method with Planet Labs data gave an accuracy of around 55%. Thus, the data used and the methodology applied influence the accuracy of the classifications. Finally, it is concluded that the available geospatial data are very useful for detecting and quantifying land cover changes. No major changes in land cover are detected; the main change detected is a transition of land cover associated with crops, which in fallow periods are detected as bare soil. It is confirmed that the methodology used contributes to spatial planning and sustainable forest management, facilitating future decisions and action plans in the governance of the region, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. GPC-ED431B 2022/12
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Assessing ecological connectivity in the Serra do Cando and Serra do Candán area of Galicia: A multitemporal classification and least-cost path modelling approach

  • Peco Costas, Carlos
  • Acuña Alonso, Carolina
  • García Ontiyuelo, Mario
  • Álvarez Bermúdez, Xana
Ecological connectivity is essential for mitigating the anthropogenic impact caused by urbanization, infrastructure, and the production of goods on natural habitats and their fragmentation. This study assesses the state of ecological connectivity between hardwoods habitats in different years for a Natura 2000 area in Galicia, in northwestern Spain, the Serra do Cando and Cand´ an. A supervised land cover classification was performed using two different machine learning algorithms, an Artificial Neural Network (ANN) and Random Forest (RF), and Sentinel-2 images from 2015 and 2022. A possible future land use scenario for the year 2029 was generated with Modules for Land Use Change Evaluation (MOLUSCE) plugin for QGIS from a Multilayer Perceptron ANN. Land use information was used to construct resistance surfaces on which ecological corridors were modelled as leastcost paths between habitat patches. The equivalent connected area (ECA) was calculated to quantify the level of connectivity and compare different time periods. Classifications achieved an accuracy of 91 % in RF and 88 % in ANN for the year 2015, and 92 % and 91 % respectively in 2022. The results for the year 2029 show a decrease in areas under crops and grassland according to RF and conifers in the case of ANN. The highest ECA values were reached in 2022 with 864 ha according to the RF-based methodology and 757 ha according to ANN. The area of hardwoods patches was the fundamental parameter that affects ECA. Combining remote sensing techniques with the least-cost paths method, including a simulation of future land use changes, has made it possible to compare the degree of ecological connectivity in different scenarios. This methodology shows the effects of land-cover changes and provides a tool to support decision making in land use planning, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Ministerio de Educación | Ref. FPU23/00964
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Merge-proofness and cost solidarity in shortest path games

  • Bahel, Eric
  • Gómez Rúa, María
  • Vidal Puga, Juan José
We study cost-sharing rules in network problems where agents seek to ship quantities of some good to their respective locations, and the cost on each arc is linear in the flow crossing it. In this context, Core Selection requires that each subgroup of agents pay a joint cost share that is not higher than its stand-alone cost. We prove that the demander rule, under which each agent pays the cost of her shortest path for each unit she demands, is the unique cost-sharing rule satisfying both Core Selection and Merge Proofness. The Merge Proofness axiom prevents distinct nodes from reducing their joint cost share by merging into a single node. An alternative characterization of the demander rule is obtained by combining Core Selection and Cost Solidarity. The Cost Solidarity axiom says that each agent’s cost share should be weakly increasing in the cost matrix., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Comision Europea, Agencia Estatal de Investigación | Ref. PID2020-113440GB-I00, Xunta de Galicia | Ref. ED431B 2022/03, Universidade de Vigo/CISUG
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Predictive modelling of cyanobacterial blooms: evaluating climate scenarios and management strategies for drinking water reservoirs

  • Acuña Alonso, Carolina
  • Seoane Martínez, Elena
  • García Ontiyuelo, Mario
  • Álvarez Bermúdez, Xana
Eutrophication has become a global environmental and ecological problem affecting the aquatic ecosystem and drinking water supply. This process is often manifested by an increase in Cyanobacterial Harmful Algal Blooms (Cyano-HABs), which have increased considerably in recent years due to climate change and increased anthropogenic pressure. To address these problems, explanatory models have been developed by identifying and quantifying the main factors affecting these phenomena using the Partial Least Square Path Modelling (PLS-PM) methodology. These models were applied to two reservoirs in Galicia (NW Spain) that have experienced recurrent Cyano-HABs, aiming to quantify the influence of climatic conditions, water physicochemical parameters, trophic state, flow regime, and nutrient concentrations (phosphorus and nitrogen) on the eutrophication process. Additionally, scenario analysis was performed to evaluate the impact of both management measures and future climate projections (RCP 4.5 and 8.5) for the years 2050 and 2075. The model developed for the A Baxe reservoir emerged as the most robust, explaining 64.1 % of the variance in eutrophication. Within this model, “Temperature” (W = 0.793) and “Chlorophyll-a” (W = 0.929) were identified as the most influential variables. Regarding the scenarios developed, it has been observed that the application of measures combined with environmental monitoring techniques are fundamental tools for mitigating the effects derived from eutrophication by reducing the cyanobacteria concentration by 10 % in a favourable scenario. The effective application of these models would represent a significant advance in reservoir management and decision-making, improving environmental governance and favouring water security management., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Universidade de Vigo/CISUG
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Artículo científico (JournalArticle)

Flooding study of the Loira River (Galicia, Spain): Importance of pre-evaluation in land management

  • Solla, David
  • Acuña Alonso, Carolina
  • Peco Costas, Carlos
  • Álvarez Bermúdez, Xana
Floods pose recurrent threats to numerous countries, carrying substantial social, economic and at times, catastrophic implications, especially in extreme scenarios. In addition, coastal areas are strongly threatened by extreme weather events and climate change impacts. Given these pressing challenges, it becomes imperative to conduct thorough assessments of future flood impacts, particularly in areas of significant environmental and socio-economic importance. This study addresses this need by focusing on the potential consequences of flooding in the Loira River, situated in the municipality of Marín, Pontevedra, Northwestern Spain. For this purpose, the free software IBER has been used, these is a free software that solves the 2D depth-averaged shallow water equations (SWEs) by using a finite volume scheme, with the domain being discretized with both structured and unstructured triangular or quadrilateral elements. The hydrological model has been established from the maximum flow for different return periods (T10, T100 and T500). Field observations, including channel widths, bridge configurations, and various parameters, further inform our analysis. Our findings reveal critical insights, including a maximum height differential of 0.48 m, particularly significant at the upstream bridge, exacerbating flooding risks in adjacent plains. Moreover, velocities along riverbanks reach hazardous levels, notably during the 500-year return period, necessitating urgent protective measures for local residents. The hydrological model also indicates that small-scale forestation has no significant effect on flood prevention. This research underscores the utility of advanced hydrological modelling in informing risk management strategies and underscores the necessity of pre-emptive hydrological analyses to mitigate future flood risks effectively., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. ED481B-2023-042, Xunta de Galicia | Ref. ED431B 2022/12
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Contribución de congreso (ConferenceOutput)

Integración de datos geoespaciales y machine learning para la evaluación del ecosistema acuático: un estudio en la Demarcación Hidrográfica Galicia-Costa (NO de España), Integration of geospatial data and machine learning for assessment of freshwater ecosystem: a study on the Hydrographical Demarcation Galicia-Costa (NW of Spain)

  • Lago González, Pedro
  • Acuña Alonso, Carolina
  • Valero Gutiérrez del Olmo, Enrique Maria
  • Álvarez Bermúdez, Xana
El estado de la biodiversidad en los ecosistemas de agua dulce es una métrica relevante con relación al nivel de contaminación de estos. En este aspecto, resulta fundamental ser capaces de medir esta biodiversidad, identificar y cuantificar las variables ambientales que influyen en mayor medida al estado del ecosistema. En este estudio se pretende, por un lado, analizar el estado de los ecosistemas de agua dulce de la Demarcación Hidrográfica de Galicia-Costa, evaluando su estado actual y su susceptibilidad a según qué variables; por otro lado, entrenar algoritmos basados en técnicas de machine learning, capaces de predecir el estado de la biodiversidad. Se llevó a cabo tomando el índice biológico METI como variable de calidad del ecosistema, parámetros fisicoquímicos de muestras de agua in situ, métricas paisajísticas para los usos Agrícola, Forestal, Artificial y Agua, parámetros físicos de las cuencas hidrográficas y densidad de población, obtenidos a partir del SIOSE y del Modelo Digital del Terreno. Los factores ambientales más relevantes para el estado de la biodiversidad son mayoritariamente variables de usos del suelo, lo que permite el modelado del METI con poco respaldo de variables fisicoquímicas, lo cual supone una potencial optimización en la detección de cambios en el estado de los ecosistemas acuáticos. Los algoritmos Random Forest (RMSE = 0.55, R2 = 0.82) y Deep learning (RMSE = 0.56, R2 = 0.83) mostraron predicciones aceptables en el área y período de estudio. El desarrollo de herramientas basadas el conjunto de datos geoespaciales y técnicas de machine learning es esencial para la mejora de la planificación integral sostenible de las cuencas hidrográficas, The state of biodiversity in the freshwater ecosystems is a relevant metric regarding their pollution conditions. On this issue, it is fundamental to be able to measure this biodiversity, identify and quantify the environmental variables which influence in the ecosystem state to a greater extent. Thus, this study aims, on one hand, to analyse the state of the freshwater ecosystems of the Hydrographical Demarcation of Galicia-Costa, evaluating their current situation and their susceptibility to other variables. In the other hand, it aims to train machine-learning-based algorithms capable of predicting this parameter. This was carried by taking the biological index METI as the ecosystem quality variable; physicochemical parameters of in-situ water samples; and land-use variables, physical parameters of the basins and population density, provided by the SIOSE and the Digital Terrain Model. The most relevant factors for the state of biodiversity were mainly land-use related variables, which allows modelling the METRI with little support of physicochemical variables. This implies a potential optimisation on the detection of changes in the state of the freshwater ecosystems. The Random Forest (RMSE = 0.55, R2 = 0.82) and Deep Learning (RMSE = 0.56, R2 = 0.83) algorithms showed acceptable predictions within the area and period of study. The development of tools based in the combination of geospatial data and machine-learning-based techniques is essential to improve the integral sustainable planification of the basins, Xunta de Galicia | Ref. ED481B-2023-042, Agencia Estatal de Investigación | Ref. PID2022-138374OA-I00, Agencia Estatal de Investigación | Ref. TED2021-130241A-I00
Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación Contribución de congreso (ConferenceOutput)

Evaluación de la conectividad en la cuenca del Río Lérez usando imágenes satelitales y machine learning

  • Vidal Llamas, Antonio
  • Acuña Alonso, Carolina
  • Álvarez Bermúdez, Xana
La reducción y fragmentación de hábitats es una de las principales causas de pérdida de biodiversidad. Por ello, asegurar la conectividad ecológica, entendida como el movimiento de las especies entre diferentes parches de hábitat, es vital para paliar los efectos negativos de la actividad humana, asegurando el intercambio genético y la colonización de nuevos espacios. Este trabajo tiene por objetivo evaluar la conectividad ecológica en la cuenca del río Lérez (Pontevedra, Galicia, NO España), donde se encuentran tres áreas incluidas en la Red Natura 2000 por la Directiva Hábitats. Se utilizaron imágenes Landsat 8-9 para crear un mapa de cobertura de suelo para los años 2013 y 2023 mediante el método de machine learning Random Forest. Se identificaron parches de hábitat y se generaron mapas de corredores ecológicos y de nodos de conexión entre los diferentes parches. Posteriormente, se calculó la métrica de centralidad de intermediación (BC, por su acrónimo en inglés) junto con otros índices globales de conectividad estructural como el índice Integral de Conectividad (IIC), la Probabilidad de Conectividad (PC) y Conectividad equivalente (EC). Se utilizaron mustélidos como grupo de referencia, los cuales tienen una preferencia de hábitat por el bosque de frondosas. Los resultados mostraron como la superficie del bosque de frondosas se habría visto incrementada. Los patrones espaciales resultaron ser parecidos en ambos casos, sin embargo, los índices PC Y EC mostraron una aparente mejora en 2023 con respecto a 2013 en cuanto a la conectividad del bosque de frondosas. El uso de índices y grafos requiere de una baja cantidad de información previa respecto a modelos de poblaciones y metapoblaciones por lo que constituyen una herramienta potente para la evaluación de la conectividad cuando no se cuenta con información precisa, medios para obtenerla o no se requiere un nivel de detalle tan alto., Agencia Estatal de Investigación | Ref. TED2021-130241A-I00, Xunta de Galicia | Ref. ED481B2023-042, Universidade de Vigo/CISUG