Earthen levees represent one of the most common structural measures used to reduce the severe effects of flood events. The hydraulic risk assessment on the flood-prone areas is typically achieved by assuming the levee system undamaged during floods, but the levees can fail due to different collapse mechanisms, among which overtopping and seepage/piping, due to infiltration process through the levee body and foundation, are the most frequent.
The monitoring of the levee status through techniques such as geophysical methods is fundamental to gather data for levee vulnerability analysis and to evaluate in real-time the possible happening of a breach formation.
In this context, this study describes a practical procedure for assessing levee vulnerability to seepage and an experimental monitoring system to be developed for an earthen levee along the Tatarena stream, located in the Umbria Region, central Italy. The selected area was damaged in 2020 by a levee failure due to the presence of animal burrows.
The monitoring system is based on combined geophysical methods: Electrical Resistivity Tomography (ERT), Ground Penetrating Radar (GPR) and Frequency Domain Electromagnetic Methods (FDEM). These methods collect different geophysical parameters (i.e. electrical conductivity and permittivity) that are correlated to the main hydraulic characteristics of the investigated subsoil, such as porosity, water content, permeability. Therefore, the geophysical methods could provide useful information on water infiltration process, making it possible to check the hydraulic status of the levees that is fundamental to identify possible critical conditions.
The work was carried out in two steps: 1) levee characterization; 2) monitoring system design and implementation.
First, the levee (approximately 570 m long) was analyzed using GPR, FDEM and ERT methods. The GPR (Dual-frequency antenna 300-800MhZ) was acquired by running profiles approximately 50 m long, defining a high-resolution image up to 2.0 m deep, while the electromagnetic method (Profiler EMP-400) through a continuous profile. The ERT is long 320m and a multichannel system with 72 electrodes (Syscal Pro instrument) and an electrode spacing of 1m was used. Therefore, an optimized roll – along protocol allowed to acquire a series of profiles (n.10) with 24 shift electrodes approach. All acquired data were elaborated and inverted (ZondRes2D software) to obtain the final electrical resistivity image. All the geophysical results highlighted an integrated interpretation allowing to define the characteristics of the levee and to optimally design the monitoring system.
Second, a monitoring system has been designed. It comprises a longitudinal ERT, consisting of 96 electrodes spaced 0.5 m (total length of 47.5 m). Cross-sectional ERTs are planned to be placed along two sections, located 19 m and 28.5 m from the beginning of the longitudinal ERT, consisting of 24 electrodes spaced 0.5 m (total length of 11.5 m).
In addition, a continuous self-potential measurement system is foreseen based on 40 electrodes located at a mean distance of 1.5 m (total system length 60 m).
The data recorded during the monitoring period will be used to assess the reliability of the estimate derived through the practical procedure.

How to cite: Dionigi, M., Bonaccorsi, B., Rizzo, E., Boldrin, P., Giampaolo, V., De Martino, G., Benigni, A., and Barbetta, S.: Assessing the earthen levees’ vulnerability through practical procedure and different monitoring techniques: the experimental site along the Tatarena stream, central Italy., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5884, https://doi.org/10.5194/egusphere-egu24-5884, 2024.

The characterization of discontinuities in rock slopes has traditionally been a time-consuming and expensive endeavor, the physical mapping itself involves extended stages as engineering geologists document key characteristics of the slope's discontinuities. Analyzing these discontinuities and determining their orientations are crucial for assessing the overall stability of a rock mass. However, recent advancements in remote sensing techniques such as UAV LiDAR, Terrestrial Laser Scanning (TLS) and iPhone 15 Pro LiDAR have revolutionized this process. Now, it is possible to remotely collect highly detailed point cloud data of a rock slope within a few hours and at significantly safe environment. These methods yield 3D point clouds and high-resolution images of rock outcrops, enabling the creation of three-dimensional reconstructions. The availability of such data has created new opportunities for collecting and assessing information about discontinuity characteristics, particularly geometric properties such as dip, dip direction, spacing, and persistence of joints in a rock slope. This paper presents a comprehensive review on capabilities of respective approach to characterize discontinuities on a rock slope, along with limitations and the advantages. In this study, several representative rock slopes underwent surveys using various techniques, including TLS, UAV LiDAR, and iPhone 15 Pro LiDAR. A comparative analysis of the results was conducted to determine the effectiveness of these new digital surveying and analysis approaches. The results of the study demonstrate better characterization of discontinuities by TLS with minor setbacks such occlusion and orientation bias during data collection. UAV LiDAR information provides comprehensive coverage of the rock outcrop with fewer point cloud density, while the LiDAR data acquisition from iPhone 15 Pro is limited to a specific distance from the rock slope. Indeed, the integration of these three approaches provides more inclusive analysis and characterization of rock discontinuities as each method compliments one another in terms of strength and limitation in characterizing rock slopes at various scales.

How to cite: Hellmy, A., Aripin, F., Jaapar, R., Mohamad, Z., Muhammad, R. F., and Che Omar, R.: Comparative Study of UAV LiDAR, Terrestrial Laser Scanner, and iPhone 15 Pro LiDAR in Rock Discontinuities Characterization for Rockfall Analysis, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7226, https://doi.org/10.5194/egusphere-egu24-7226, 2024.

In recent years, great efforts have been made by the research community to define and improve computational methods aimed at integrating DInSAR and GNSS data to estimate three-dimensional (3D) ground displacements maps. In particular, the SISTEM method (Guglielmino et al., 2011), which was the first that used the Strain Model (SM) to solve 3D deformations by combining GNSS and DInSAR data, has been both widely utilized and continuously improved since its proposal.

In this work, we propose the SISTEM PLUS method, an evolution of the original SISTEM method achieved through the adoption of a second order Strain Model.

The SISTEM PLUS method, like the original SISTEM one, is a linear in the parameters model solved by using the weighted least square (WLS). The usage of a second order Strain Model has yielded the following main advantages: a notable improvement in results accuracy and the provision of a mathematical framework for the effective integration of tilt measurements, in addition to DInSAR and GNSS data.

On the other hand, the usage of a second order Strain Model increases the complexity of this new method compared to the original SISTEM. Specifically, the SISTEM PLUS method requires the estimation of a greater number of unknown variables, which in turn, necessitates a larger set of available data to achieve robust estimations.

The proposed methodology was tested on both synthetic and experimental data, these last from GNSS and DInSAR measurements carried out on Campi Flegrei area during the 2016-2023 period. In order to appreciate the precision achieved on experimental test results, the estimated standard errors computed by Weight Last Square are provided. These new tests also allowed optimising the choice of specific parameters of the algorithm.

Currently the SISTEM PLUS method allows integration of DInSAR data taken from different geometries and different SAR sensors, (e.g. SENTINEL1, ALOS or CSK) and different kind of in situ data (GNSS, Levelling EDM and Tilt). We emphasize that, as the SISTEM PLUS method performs data integration through the resolution of a system of linear in the parameters strain model based equations, each equation being specific for a particular type of data, it has the built-in capability to integrate additional datasets, such as strain-meter and Distributed Fiber Optics to name a few. This is accomplished defining and adding a suitable set of additional strain model based equations, specifically tailored for the new datasets to be integrated, into the system of equations to be solved.

These potentialities will be fully exploited in future developments of the presented methods.

How to cite: Guglielmino, F., Puglisi, G., and Spata, A.: SISTEM PLUS: a second order Strain Model based method to integrate DInSAR and terrestrial geodetic data., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-9586, https://doi.org/10.5194/egusphere-egu24-9586, 2024.

Gravimetry and self-potential are passive geophysical techniques that can contribute to volcanic monitoring. The gravimetric method measures variations in the Earth's gravitational field due to density changes, which could be caused, for example, by magmatic intrusions and fluid accumulation. On the other hand, the spontaneous potential methods measure variations in the natural electric potential generated by the circulation of hydrothermal fluids related to the volcano dynamics. This technique is very useful to delineate the areas affected by hydrothermal activity. Combining both methods offers an integral perspective by detecting variations in mass distribution and natural electrical phenomena within the volcano. Therefore, their application allows us to see changes in the subsurface due to volcanic activity.

Applying these geophysical techniques in an active volcanic area could help better understand the temporal evolution of the volcanic activity. In this context, Tenerife represents the perfect scenario for applying these methods. This island consists of three ancient inactive volcanic systems (Anaga, Teno and Roques del Conde) located at the ends of the island, connected by three young rift zones to the center of the island where a giant volcanic caldera, known as Las Cañadas, is located. Inside the caldera and dominating the horizon is the Teide stratovolcano, which has a height of 3718 meters. The crater of Teide hosts intense hydrothermal activity with fumarole and relevant diffuse degassing.

For this purpose, since 2018, several gravimetric and spontaneous potential campaigns have been conducted each month on the island: from one side, 60 points spread over the three volcanic ridges, the caldera and the volcanic edifice of Teide were measured using a CG-6 Autograv™ Gravity Meter from SCINTREX. From the other side, 38 points distributed inside Teide’s crater were measured using a V-FullWaver datalogger from IRIS Instruments. The preliminary results show stable values for gravity, while for spontaneous potential, there are significant variations of the geoelectrical values due to variations in the hydrothermal activity.

We show how the continuous measurement of these parameters in time could contribute to the early identification of potential volcanic events in Tenerife, which is considered the island with the highest volcanic risk in the Canary Islands.

How to cite: Álvarez Hernández, A., Stoker, E., Neves, D. B., Reyes López, Á., Ortega Ramos, V., Jiménez-Mejías, M., Martínez van Dorth, D., García, R., D’Auria, L., and Pérez, N. M.: Gravity and spontaneous potential studies for volcano monitoring in Tenerife (Canary Islands), EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14878, https://doi.org/10.5194/egusphere-egu24-14878, 2024.

Coastal regions are vital for human settlements since they significantly contribute to the prosperity of numerous nations. However, their vulnerability to factors like erosion and pollution necessitates constant vigilance and proactive measures. Deltas, bays, and gulfs are particularly susceptible, facing risks from natural forces and human activities, demanding ongoing management strategies. In this context, the prediction of the wave dynamics interaction with the coastline and the seabed bathymetry plays a fundamental role. To obtain reliable forecasts about sea conditions, the numerical models for wave propagation necessitate accurate initial and boundary data from real-scale motions, as well as a fine-grain representation of the seabed bathymetry, which plays a major role in dispersion and refraction phenomena. However, these prediction models need to be calibrated with accurate information about the sea state in the area under test. Usually, such information is provided by in-situ sensors (e.g. wave buoys) and remote sensing devices (e.g. radars, video-monitoring systems) [1]. Remote sensing instruments are often preferred because they allow overcoming the main limitation of in-situ devices, that is the impossibility to provide spatial and temporal information about the sea state. Among the available remote sensing technologies, radar systems such as High Frequency and marine radar, have proven to be effective in measuring the wave spectra and retrieving the sea state information in coastal areas [2]. However, the major limitation of the aforementioned systems is related to their impossibility to retrieve spatial information about sea state very close to the coastline.

As a possible solution to such an issue, this communication aims at presenting the main activities and preliminary results achieved in the frame of the Italian PRIN-PNRR 2022 Project SEAWATCH - Short rangE K-bAnd Wave rAdar sysTem Close to tHe coast. Specifically, the SEAWATCH project, which started in December 2023, aims at expanding the capabilities of wave radar technology by developing an innovative short range K-band radar prototype that is suitable to perform sea state monitoring very close to the coastline. Thanks to its small size, weight, and low power supply, the proposed system is portable and allows performing on-demand surveys [3]. In this perspective, SEAWACTH addresses the safety of human life and structures in harbour and coastal zones. Finally, the developed system is expected to provide an improvement in the knowledge of the wave phenomena nearby the coast that, actually, are considered open issues in the scientific community.

Reference:

[1] P. Neill, M. Reza Hashemi, Chapter 7 - In Situ and Remote Methods for Resource Characterization, Editor(s): Simon P. Neill, M. Reza Hashemi, In E-Business Solutions, Fundamentals of Ocean Renewable Energy, Academic Press, 2018, Pages 157-191.

[2] Ludeno, M. Uttieri, Editorial for SI “Radar Technology for Coastal Areas and Open Sea Monitoring” JMSE, 2020, 8, 560.

[3] Gennarelli, et al. (2022). 24 GHz FMCW MIMO radar for marine target localization: A feasibility study. IEEE Access,10, 68240-68256.

Acknowledgment: This work was supported and funded by the EU—NextGenerationEU PNRR Missione 4— C2 Investimento 1.1, PRIN—SEAWATCH—Short-rangE K-bAnd Wave rAdar sysTem Close to tHe coast (B53D23023940001), and partially funded by the research project STRIVE (B53C22010110001).

How to cite: Ludeno, G., Antuono, M., Lugni, C., Contestabile, P., Vicinanza, D., Catapano, I., Esposito, G., Soldovieri, F., and Gennarelli, G.: Enhancing Coastal Monitoring: Short-Range K-Band Radar for Sea State Observation, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15923, https://doi.org/10.5194/egusphere-egu24-15923, 2024.

The underwater volcanic activity associated with deep-seated mantle processes represents a primary driver of the chemical and biogeochemical evolution of the global oceans. Hydrothermal activity is often a manifestation of submarine volcanism, where fluxes of heat and magmatic volatiles confer both potential hazard and opportunities of resource exploitation. Despite this, research on shallow submarine arc volcanoes is still in an early stage and only a few continuous seafloor observing infrastructures have been developed until now. 
The Kolumbo underwater volcano, located in the Aegean Sea, hosts one of the most active and dynamic hydrothermal vent fields, marking it - along with the proximity to the world-known Santorini island - a severe geohazard for a combination of reasons.
Within the framework of the SANTORY (SANTORini’s seafloor volcanic observatorY) project, funded by the Hellenic Foundation for Research and Innovation and with the financial support of the Municipality of Thira, between 2022 and 2023, three oceanographic cruises were performed on submarine Kolumbo volcano. The oceanographic surveys were mainly aimed at the deployment of integrated operating sensors of state-of-the-art technology, for in situ monitoring.

 A new-generation stand-alone multiparametric observatory has been developed at INGV Palermo and deployed at the bottom of the crater (500 meters depth) for the first time in December 2022. The battery powered module has been able to operate autonomously for a 10-month-long period, collecting a dense, heterogeneous dataset able to describe the activity of the hydrothermal reservoir, highlighting its intense dynamic along the time.

In June 2023, the observatory was recovered and re-deployed after brief maintenance operations including battery charging and data downloading. Finally, in October 2023, the observatory was definitely recovered.

Here we present for the first time a mid-term-long chemical-physical data series acquired (pH, temperature, hydrostatic pressure, turbidity, conductivity, dissolved methane) along with passive acoustic and the preliminary findings of the system evolution within the observing window.
A variety of local VT events likely sourced in the deeper portion of the plumbing system, together with several other minor seismic events related to fluid dynamics inside “fluid-filled” cracks and conduits has been revealed by passive acoustic data. Moreover the acoustic sensor recorded all the signals generated by the bubbles along the water column. The obtained results gave back an up to date picture of the ongoing Kolumbo degassing dynamics, hydrothermal and seismic activity. 

How to cite: Sciré Scappuzzo, S., Lazzaro, G., Longo, M., D'Alessandro, W., Grassa, F., Semprebello, A., Nomikou, P., Polymenakou, P., Rizzo, A. L., and Mallios, A.: Mid-term observation of the degassing dynamics of the Kolumbo submarine volcano (Aegean Sea) gained by new-generation stand-alone multidisciplinary seafloor observatory, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17579, https://doi.org/10.5194/egusphere-egu24-17579, 2024.

In Southern Italy, many natural disasters have induced significant human and material losses. The extensive damages caused by seismic events have highlighted the impelling need to improve the analyses of the risk assessment and the related planning of the economic resources for a proper environmental management. In this scenario, the most important scientific issue is the definition of a seismic hazard model using the most advanced data availability in order to include updated kinematic and thermo-rheological conditions.

This contribution describes the activities that will be developed during the Italian PRIN2022 PNRR project “funded by the European Union – Next Generation EU”. The project, entitled “Relation between 3D Thermo-Rheological model and seismic HAzard for the risk Mitigation in the urban areas of Southern Italy” (TRHAM), aims at proposing a new map of the seismic hazard of peninsular Southern Italy in order to provide a significant contribution for the seismic risk management. This newly devised map will be accomplished through the analysis and the integration of several pieces of information, including the 3D crustal, thermo-rheological, and kinematic models. In particular, the activities of the THRAM project will be organised in four Milestones (ML): the topic of ML1 is “Scientific Management and Dissemination”; the ML2 involves the definition of an integrated “3D crustal model from geophysical data”, while the retrieval of a “Thermo-Rheological model for Brittle/Ductile transition”, integrated with the kinematic one, is the main output of the ML3. Finally, the activities of the ML4 provide the “Probabilistic Seismic Hazard Analysis”, both at regional and local scales. Indeed, the computation of exceedance hazard curves for the urban zones of Abruzzo, Campania and Basilicata regions is an important output of the project. Three Research Units (RU) will collaborate for the activities of the THRAM project, whose multidisciplinary nature strongly emerges from the skills of the ten researchers constituting the RUs. The expertise of the team covers a wide spectrum of competencies within the geological and geophysical fields, including Thermo-Rheology, Seismology and Potential Fields, and Seismic Hazard and Crustal Kinematic model.

In conclusion, several intriguing and impacting features characterise the THRAM project. Specifically, the definition of a new seismic hazard map in Southern Italy will be relevant for exploring new paths in seismic risk assessment, and for increasing the awareness in the management of environmental resources. At the same time, the planned thermo-rheological studies are fundamental in the framework of the green energy exploitation of Southern Italy, as well as their integration with the kinematic data contributes to the increase in our comprehension of the Southern Apennines geodynamics, still poorly understood.

How to cite: Castaldo, R., De Landro, G., Carafa, M., Tizzani, P., Fedi, M., Zollo, A., Di Lorenzo, C., Di Naccio, D., Kastelic, V., and Taroni, M.: TRHAM PROJECT: Relation between 3D Thermo-Rheological model and seismic HAzard for the risk Mitigation in the urban areas of Southern Italy, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19411, https://doi.org/10.5194/egusphere-egu24-19411, 2024.

Tenerife is an active volcanic island belonging to the Canary Islands Archipelago. After its last eruption in 1909, the only visible volcanic manifestations consisted of weak fumaroles from the summit crater of Teide volcano, the most prominent geographic feature of the island. Tenerife has been monitored since 2016 by a seismic network consisting currently of 21 broadband stations.

Since October 2016, the seismic activity has shown an increase, corresponding to a relevant growth of the diffuse degassing from the summit of Teide. This has been interpreted as the effect of pressurization of the hydrothermal system triggered by repeated injections of fluids of magmatic origin.

Here, we present different analyses realized on the seismic catalogue of the island. First, we show the spatio-temporal variation in the hypocenter distribution, highlighting the temporal evolution of seismicity clusters. Then, we discuss the spatial and temporal variations of the Gutenberg-Richter b-value, which shows a remarkable increase and, in general, higher values around the central part of the island. We also highlight the presence of a significant long-period seismic activity, consisting of both isolated events and dense swarms.

Finally, we discuss the relationship between the seismicity and the internal structure of the island, inferred by recent seismic tomography studies, as well as the temporal correlation with other geophysical and geochemical parameters measured on the island.

How to cite: García Hernández, R., Fan, K., Morales Fernández, C., Ní Nualláin, K., Barrancos, J., Padilla, G. D., Martínez Van Dorth, D., Ortega Ramos, V., D'Auria, L., and Pérez, N. M.: Seismological investigation on the recent unrest of Tenerife (Canary Islands), EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18988, https://doi.org/10.5194/egusphere-egu24-18988, 2024.

Good knowledge on soil texture is a base for a sound land use management as one of the driving factors for soil fertility, and thus for a worldwide sustainable food production and safe drinking water supply using groundwater resources. Furthermore, soil texture is one of the driving factors of soil fertility. In particular, in countries of the Global South but also in other regions, high quality digital information on soil properties on regional level are rather scarce. While conventional soil inventories are time consuming, digital mapping of soil properties is a promising approach to close these gaps time and cost efficiently. For this purpose, a reliable method was developed within the project “ReCharBo” (Regional Characterisation of Soil Properties) to minimize field and laboratory work by combining gamma ray spectrometry with conventional soil survey at selected study areas characterized by different soil parent materials. Data acquisition was performed by using a portable gamma ray spectrometer and soil sampling at local scale as well as by helicopter-borne gamma ray spectrometry at regional level.

For the estimation of soil texture by gamma spectral data at different soil parent materials we developed classic multiple linear regression models based on  laboratory analyses of grain size and total Potassium and Thorium contents. To consider the soil parent materials, we calculated clay-, silt- and sand-Potassium-Ratios based on laboratory data and integrated them into the models. The statistical models were validated by dividing the data set randomly fifty-fifty into a training, and a validation data set. The results on the validation data set show that soil texture can be predicted with an error (RMSE) of 5.8% (Clay), 5.5% (Silt) and 4.6% (Sand) by gamma ray spectrometry. Based on these models, soil texture can reliably be estimated by gamma ray spectrometry accompanied by a scarce soil sampling in regions with poor data.

How to cite: Konen, L., Ibs-von Seht, M., Rückamp, D., Möller, A., Guggenberger, G., and Fries, E.: Estimation of soil texture in areas of different parent materials by using gamma ray spectrometry at local and regional scale, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-21274, https://doi.org/10.5194/egusphere-egu24-21274, 2024.

Cumbre Vieja volcano, at the southern part of La Palma Island, is the last stage on the geological evolution of the island, the fifth in extension (706 km²) and the second in elevation (2,423 m asl) of the Canarian archipelago. The recent volcanic activity in La Palma Island has taken place exclusively in Cumbre Vieja volcano in the last 123 ka, being the last volcanic activity the Tajogaite eruption, from 19 September to 13 December, 2021. We report herein the results of several intensive soil gas studies, focused on the non-reactive and highly mobile gas helium (He) in Cumbre Vieja. He has unique characteristics as a geochemical tracer: it is chemically inert and radioactively stable, non-biogenic, highly mobile and relatively insoluble in water. The geochemical monitoring of soil He emission at Cumbre Vieja started in 2002. Soil gas samples have been regularly collected at ~40 cm depth using a metallic probe at 600 sites for each survey and later the He content has been analysed by means of a quadrupole mass spectrometer. Spatial distribution maps have been constructed following the sequential Gaussian simulation (sGs) procedure to quantify the diffuse He emission from the studied area. The time series of diffuse He emission rate has shown significant increases before and during the occurrence of seismic swarms that took place in the period 2017-2021. During the eruptive period, a significant increase was also observed, with good temporal agreement with the increase of the volcanic tremor. Increases in diffuse He preceded peaks of diffuse CO₂ emission as expected by the characteristics of these gases. The absence of visible volcanic gas emissions (fumaroles, hot springs, etc.) at the surface environment of Cumbre Vieja, makes this type of studies in an essential tool for volcanic surveillance purposes.

How to cite: Pérez, N. M., Collins, H., Bonomini, I., Melián, G. V., Gironés, A., Asensio-Ramos, M., Padrón, E., Hernández, P. A., Rodríguez, F., and Padilla, G.: Diffuse He degassing monitoring of Cumbre Vieja volcano, La Palma, Canary Islands, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16451, https://doi.org/10.5194/egusphere-egu24-16451, 2024.

With increasing attention to urban temperature and outdoor thermal comfort, monitoring urban microenvironments at a lower cost is an effective method to supplement the spatiotemporal deficiencies of traditional monitoring networks. But widespread use of low-cost sensors has been hampered by uncertainty about their data quality. The calibration of low-cost sensors is key to promoting their large-scale application and increasing people's confidence in related research. The purpose of this study is to calibrate low-cost integrated environmental sensors and effectively improve their hourly data quality based on an IoT case study in Wuhan, China.

Based on the standards of 24 traditional weather stations in different locations of the meteorological regulatory authorities, this study applied a total of eight machine learning (ML) algorithms to calibrate low-cost sensors and compared their performance. The eight ML algorithms are: (a) Multiple Linear Regression (MLR); (b) Random Forest (RF); (c) K-Nearest Neighbors (KNN); (d) Gradient Boosting Regression Tree (GBRT); (e) Decision Tree (DT); (f) AdaBoost; (g) Bagging; (h) Extremely randomized Trees (Extra-Trees). Hourly raw data collected by 34 low-cost sensors deployed near traditional weather stations were calibrated, and the model was tested using ten-fold cross-validation. The two farthest locations are 121km apart in a straight line, and the maximum data collected from a single sensor is 12,406 hours. In addition, the model migration effects in different field scenarios were also considered, including six typical land surface types, namely built area, scrub, water, artificial surfaces, woodland, and cultivated land.

The results show that the random forest model shows better performance than other models on multiple low-cost sensors at different locations. By applying our method, it shows an average improvement with its R-squared value from 0.682 to 0.980, Root Mean Square Error (RMSE) from 5.989 to 1.355, and Mean Absolute Error (MAE) from 4.250 to 0.932. The random forest model has a better migration effect in similar scenarios. Using a model with a surface type that is more similar to the sensor to be calibrated, the average R-squared obtained by calibrating 34 sensors is 0.946, and the average MAE is 1.584. At the same time, the distance between the sensor to be calibrated and the best-performing migration model was also considered, with the farthest straight-line distance being 94km and the nearest being 7km.

This study introduces a calibration method for low-cost meteorological integrated sensors for long-term complex field environment monitoring. Moreover, we compared the migration effect of the random forest model in different typical scenes in the field. Similar surface types are more beneficial to model migration. Even in locations far apart, our model still has stable performance. The results show that this method can significantly improve data quality and increase user confidence in low-cost environmental sensor applications.

How to cite: Nan, F., Zeng, C., and Shen, H.: Calibration of integrated low-cost environmental sensors based on machine learning with multiple scenes, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7377, https://doi.org/10.5194/egusphere-egu24-7377, 2024.

The majority of ocean pollution stems from terrestrial sources, with more than 80% attributed to chemicals, industrial waste, toxic metals, plastics, sewage, and other land-based materials. Therefore, the management of ocean pollution must initiate from terrestrial interventions, necessitating the control of pollutants discharged from land sources and the crucial pursuit of identifying the roots of ocean pollution. Identifying the causes of ocean pollution enables the exploration of mitigation strategies for the entities responsible for pollution generation and facilitates the formulation of relevant policies. Furthermore, it can contribute to imposing costs for purification and raising awareness about the seriousness of ocean pollution. However, Rivers and streams that lead to the ocean are influenced by various changing factors as they pass through the land, making it a globally challenging task to pinpoint the sources of pollution.

Accordingly, in this study, we collected water quality observation data measured at five points in a specific estuary area and water quality data from five areas near the source of pollution, and conducted a study to calculate the contribution of pollution from the source. We performed statistical analysis and machine learning-based pollution source analysis, and developed an improved artificial intelligence model proposed in this study that complements the limitations of existing analysis methods. The variables used in the analysis were POC average concentration, POC δ13C, PN average concentration, PN δ15N, average concentration, and DOC δ13C. For basic data analysis, data distribution analysis, similarity/discrimination analysis, and clustering analysis of variables by branch were performed. Basic data analysis allowed for dividing the data's characteristics into four groups, but discrepancies in similarities emerged among items based on each analysis method, limiting the meaningfulness of the data analysis. For this, we analyzed which pollutants contaminated the five points in the river estuary using machine learning techniques such as XGBoost and a deep learning neural network, an artificial neural network model. The XGBoost analysis categorized pollution sources for each location into 1 to 2 categories, showing accuracies ranging from 51.04% to 99.92%. However, due to the intrinsic nature of machine learning, predicted values tend to maximize similarity to the most similar pollutant source, resulting in extreme values exceeding 99%. The artificial neural network analysis resulted in the classification of 2 or more pollution sources for each location, with accuracies ranging from 33.23% to 62.45%. This is considered a result of relatively lower accuracy due to the unique characteristics of each location.

To overcome the limitations of each model, this study created an integrated model that aggregates results from multiple models to determine the similarity. The analysis using the integrated model effectively identified pollution sources excellently without encountering extreme accuracy issues. To ensure the reliability of future pollution contribution assessment models, determining pollution contribution through isotopic fraction analysis will be necessary.

How to cite: Han, H.-G., Kim, T., Lee, C., and Park, Y.-G.: A Study Estimating the Contribution of Organic Contaminants to Ocean Pollution Using AI Techniques, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8296, https://doi.org/10.5194/egusphere-egu24-8296, 2024.