Geomatic Techniques for the Mitigation of Hydrogeological Risk: The Modeling of Three Watercourses in Southern Italy

Abstract

In recent decades, climate change has led to more frequent episodes of extreme rainfall, increasing the risk of river flooding. Streams and rivers characterized by short flow times are subject to rapid and impressive floods; for this reason, the modeling of their beds is of fundamental importance for the execution of hydraulic calculations capable of predicting the flow rates and identifying the points where floods may occur. In the context of studies conducted on three watercourses in Calabria (Italy), different survey and restitution techniques were used (aerial LiDAR, terrestrial laser scanner, GNSS, photogrammetry). By integrating these methodologies, multi-resolution models were generated, featuring a horizontal accuracy of ±16 cm and a vertical accuracy of ±15 cm. These models form the basis for the hydraulic calculations performed. The results demonstrate the feasibility of producing accurate models that are compatible with the memory and processing capabilities of modern computers. Furthermore, the technique set up and implemented for the refined representation of both the models and the effects predicted by hydraulic calculations in the event of exceptional rainfall (such as flow, speed, flooded areas, and critical points along riverbanks) serves as a valuable tool for improving hydrogeological planning, designing appropriate defense works, and preparing evacuation plans in case of emergency, all with the goal of mitigating hydrogeological risk.

1. Introduction

According to the definition provided by [1], hydrogeological risk is associated with the natural water cycle and the geological structure of the land. While its primary causes stem from natural factors—such as the country’s climate and geomorphological features—human activities also play a significant role. In recent decades, these anthropogenic influences have increasingly impacted and, in many cases, exacerbated natural processes.

A range of natural factors contribute to a territory’s susceptibility to hydrogeological risk, including its geology and lithology, topography, climate and rainfall patterns, soil characteristics, hydrographic features, and seismic activity. Among the physical factors that predispose the Italian territory (and in particular Calabria) to hydrogeological instability is its geological and geomorphological conformation, characterized by a complex orography and small hydrographic basins, which are therefore characterized by extremely rapid response times to precipitation. Localized and intense meteorological events combined with these characteristics of the territory can therefore give rise to violent phenomena characterized by very rapid kinematics (mudslides and flash floods) [2,3].

Hydrogeological risk is also strongly influenced by anthropogenic factors. Population density, ongoing urbanization, abandonment of mountain areas, illegal construction, widespread deforestation, environmentally harmful agricultural practices, and poor maintenance of slopes and waterways have all contributed to worsening land instability. These factors have further exposed the fragility of the Italian territory, increasing its vulnerability and the overall risk of such events [4].

According to the Higher Institute for Environmental Protection and Research (ISPRA), 7423 Italian municipalities—equivalent to 93.9% of the national total—are exposed to landslide and flood risk. Among these, 1701 municipalities contain areas classified as being of medium hydraulic hazard, while 4232 municipalities include zones characterized by both high landslide risk and medium flood risk [3].

Bearing in mind the general objective of mitigating hydrogeological risk, watercourse modeling is essential to perform hydraulic calculations that allow forecasting flow rates and identifying points where floods could occur. This information, in fact, can be used to design appropriate defense works and evacuation plans in the event of an alarm.

For several decades, geomatic techniques have been employed for the modeling of watercourses, often through integrated and complementary approaches.

Among these, aerial photogrammetry has long been the main technique [5]. Advances in digital camera resolution and computing power have enabled the use of techniques such as Structure from Motion (SfM) [6]. These methods enable the creation of accurate 3D models—even from imagery acquired with commercial-grade cameras—thus making them well-suited for both large-scale and very large-scale mapping applications. These techniques are also employed in laboratory studies on surface water motion [7].

A well-established technique is drone photogrammetry [8]. Thanks to the possibility of orienting the camera through the gimbal system, it is also possible to perform surveys of vertical surfaces, undercuts, and intrados of structures.

Significant progress has been made with the use of Light Detection and Ranging (LiDAR) technology, encompassing airborne, terrestrial (TLS), and Unmanned Aerial Vehicle (UAV)-based methods [9].

Airborne LiDAR (ALS) is widely employed for extensive mapping of riverine topography and shallow water. It offers rapid data acquisition over large areas, producing Digital Terrain Models (DTMs) that are valuable for floodplain delineation and geomorphological studies [10]. The possibility of taking acquisitions from above allows the surveying of large surfaces in a short time but at the same time constitutes a limitation of this technique, especially in correspondence with works of art (bridges). In such circumstances, elements that are fundamental for the execution of hydraulic calculations (like piers, bulkheads, pile caps, abutments) cannot be modeled. It is therefore necessary to carry out surveys of these elements using different techniques and to integrate them.

Terrestrial LiDAR (TLS), including mobile systems such as vehicle-mounted or vessel-mounted mobile systems, offers higher point densities and closer proximity to the riverbed, improving the accuracy of elevation models. This is particularly useful for detailed studies of riverbank morphology and sediment transport processes. Furthermore, the use of TLS allows the acquisition of engineering structures not visible from above [11]. If echosounders are installed on a vessel platform, the survey of bathymetry can be performed.

UAV-based LiDAR has emerged as a versatile tool for river modeling. Equipped with lightweight LiDAR sensors, UAVs can access challenging terrains and capture high-resolution data with very high point densities [12]. This capability is crucial for mapping intricate features like undercut banks and submerged areas, which are often missed by ALS.

For large areas, the satellite based Synthetic Aperture Radar (SAR) is increasingly used, which allows for the verification of any subsidence phenomena over time [13].

It is almost always necessary to integrate several techniques to obtain complete results thanks to their complementarity [14,15,16]. Each technology provides distinct contributions to channel modeling:

-

Photogrammetry offers visually detailed outputs and facilitates vegetation analysis;

-

ALS, due to its robustness and ability to penetrate vegetation, enables the generation of both Digital Surface Models (DSM) and DTM;

-

TLS delivers ultra-high precision, making it ideal for capturing fine-scale, localized features. It plays a crucial role in surveying elements that are not visible from aerial viewpoints.

-

The integration of these technologies allows for the creation of a comprehensive, high-resolution model of river systems.

It is also important to consider the use of multi-resolution models, which allow for lower point densities in areas with regular topography and higher densities where greater detail is required for computational accuracy (e.g., near artifacts located within the riverbed).

A problem to be addressed is the integration of different geomatics techniques for riverbed surveying, with the aim to combine datasets of different resolutions [17]. In fact, the techniques described above provide complementary data but are inherently heterogeneous in terms of accuracy, scale, and spatial density. Discrepancies in resolution often lead to issues such as data misalignment, interpolation artifacts, and difficulties in error propagation analysis. In recent years, the integration of multi-source geospatial data has become a central focus in geomatics, particularly in applications involving high-resolution terrain modeling, environmental monitoring, and hydrological analysis.

Among the most effective methods used to face these issues are data fusion techniques, hierarchical surface modeling, and hybrid interpolation schemes [18,19,20].

A multi-resolution model is obtained by combining multiple models created with different techniques; it is therefore necessary to address the issues related to registration [21,22]. For this purpose, common points are used, often detected with Total Stations (TS) or, better, with the use of Global Satellite Navigation Systems (GNSS) receivers that allow georeferencing. As regards referencing, particular attention must be paid to elevations: in fact, orthometric elevations (referred to the geoid) of good precision are necessary for hydraulic calculations to provide reliable results. It is therefore necessary to use the most accurate geoid models available for the area under study [23].

The final output of surveys carried out with geomatic techniques consists of DTM and DEM, which form the basis of hydraulic calculations carried out within a Geographic Information System (GIS) environment or through specialized software [24]. Hydraulic calculations for flood mapping use 1D and 2D modeling. While 2D models provide more accurate and detailed results, they require significantly greater computational resources. However, their use is becoming increasingly common due to advances in computing power [25].

The objective of this article is to present a multi-instrumental approach using geomatic techniques, which involves the development of heterogeneous models and their integration into a comprehensive multiresolution framework, with the aim of contributing to hydrogeological hazard assessment. The need for models with higher levels of detail in the presence of man-made structures, while allowing for lower point density in more regular areas to reduce memory and computational demands, led to the choice of the multi-instrument and multi-resolution approach adopted in this study.

The data were acquired as a part of a regional operational program [26], carried out through a collaborative effort among three Italian universities and a research center.

The article is structured as follows: It begins with a description of the surveyed watercourses, the instruments used and the techniques employed; this is followed by the presentation of the resulting products and the methods adopted for visualizing the outcomes of the hydraulic computations. The paper concludes with a discussion of the results and the final considerations.

2. Materials and Methods

2.1. Areas of Study and Their Peculiarities

The study areas are the riverbeds of three watercourses located in the Calabria region in southern Italy (Figure 1).

2.1.1. The Crati River

The Crati River basin covers an area of 2448.91 km². With its length of 90 km and the average annual flow of about 36 m³/s, it is the main watercourse of Calabria and the third in southern Italy (Figure 2a). The hydrographic network exhibits a hierarchical structure with eight Horton stream orders and consists of 21,080 channel segments (reaches), of which 1080 have lengths exceeding 1 km [27]. The river has a markedly torrential regime, alternating marked summer low water levels (minimum flow rate approximately 10 m³/s) with strong and sometimes disastrous winter floods (even over 3000 m³/s). On 24 November 1959, the river flooded, causing extensive damage to homes, shops, and department stores (Figure 2b). Characterized by steep slopes and a narrow bed in its first stretch, the river widens at the city of Cosenza, located at the confluence with the Busento River. In this stretch, there are several bridges that have been the subject of surveys.

Regarding flood prevention measures, the interventions carried out along the river include the replacement of multi-span bridges with single-span structures in the urban area it passes through, as well as the construction of earth embankments in the stretch of the river mouth downstream section, which is characterized by a high concentration of agricultural activities.

2.1.2. The Corace River

The Corace River basin has a surface area of 294.50 km². With a length of 48 km and an average annual flow rate of approximately 4.4 m³/s, the river flows into the Ionian Sea after a course of 48 km (Figure 3a). As for the hydrographic network, it shows a hierarchical structure with seven Horton stream orders and consists of 4033 channel segments, of which about 100 have lengths exceeding 1 km [28]. This river also has a markedly torrential regime, and several floods happened in the last decades (Figure 3b). In the section between 9 and 10 km from the river mouth, numerous road crossings are present, featuring structures built within the riverbed. As regards flood prevention measures, in recent years, the improvement of the embankments in the stretch of the river mouth has been carried out.

2.1.3. The Valanidi Fiumara

A fiumara is a short watercourse that remains completely dry for most of the year due to the scarcity of rainfall. Its bed is very wide and pebbly, with the presence of numerous boulders, evidence of the strong erosive and transport action developed during periods of flooding.

The Valanidi fiumara, located south of the city of Reggio Calabria, is the main watercourse in the municipality. It has a drainage basin covering 29.7 square kilometers and stretches for 19.91 km. About 1.5 km from its mouth, it splits into two branches, both characterized by a wide and dry bed, due to the presence of numerous check dams upstream (Figure 4a). The two branches cross the state road E90 that runs along the Ionian Sea. In 1953, after a series of heavy rains, its floods inundated the surrounding area, causing several victims and extensive damage (Figure 4b). Using data provided by the Regional Agency for Environmental Protection [29], the hydrological balance is based on an average precipitation value of about 1300 mm/year; surface runoff is estimated at 25% of the precipitation, and effective infiltration water is about 300 mm/year. The most recent risk mitigation interventions include the construction of weirs using metal gabions, reshaping through excavation, and building embankment walls reinforced with metal gabions.

2.2. Instruments Used for Data Acquisition

2.2.1. Aerial Photogrammetry

The aerial photogrammetric survey was performed by helicopter, which allowed the acquisition of shots with a flight height above the ground sufficiently constant even in areas with steep slopes. The average flight height was 500 m. The 40 MPixels camera DigiCAM-H/39, manufactured by IGI, Langenauer, Germany, allowed obtaining orthophotos with a Ground Sampling Distance (GSD) of 20 cm.

2.2.2. Aerial Laser Scanner

The ALS survey, realized simultaneously with the photogrammetric shots, was performed with a Riegl LMS-Q680i laser scanner (Riegl, Horn, Austria). The point cloud obtained has a density of 4 points/m². For georeferencing the point clouds, the helicopter was equipped with an Inertial Measurement Unit (IMU) and a GNSS receiver, both manufactured by Leica, Wetzlar, Germany; this equipment ensured an altitude precision of 15 cm.

2.2.3. Terrestrial Laser Scanner

To increase the point density near the artifacts present in the riverbed, a RIEGL LMS 420 i vehicle-mounted TLS was used. The instrument has a range of 800 m and a precision of 1 cm.

2.2.4. GNSS

Several Ground Control Points (GCPs) were surveyed through a Leica Viva GS15 GNSS receiver. The acquisition was performed in rapid-static mode.

2.2.5. Total Station

For the surveying of particular points (vertices on the pier caps, pile caps, etc.), a Leica 1200+ total station (Leica, Wetzlar, Germany) was used, with an angular precision of 1″, a distance precision of +/− 1 mm +/− 1 mm/km, and a range without prism of 1200 m.

2.3. Methodologies Adopted for the Integration of Data

To perform hydraulic calculations, a riverbed is represented with a 2.5D model, in which each pair of coordinates in the plan corresponds to a single height. In particular, DTM and DSM are used. The former are used to have the surface of the riverbed without vegetation, while the latter are useful especially in urban areas. The difference between DSM and DTM can provide information on the average height and density of vegetation present, which is useful for performing hydraulic calculations.

Several authors have proposed 3D modeling workflows that integrate multisensor and multiresolution data, using a combination of methods to produce point clouds and integrated 3D models [30]. In our case, we adopted the pipeline represented in Figure 5 for a 2.5D multisensor and multiresolution modeling, based on the integration of different techniques, aimed at generating riverbed models.

As for the laser scanner data, typical procedures for processing point clouds were used. After eliminating outliers and thinning out excessively dense areas, hole filling was performed. In the case of river surveys, this operation is particularly delicate: the holes in the meshes obtained, generally caused by undercuts or the presence of elements that hide the undetected areas, are often due to the presence of water. Filling the holes, in this case, would lead to the creation of incorrect models, which would distort the results of hydraulic calculations.

The coarse georeferencing of the aerial surveying was obtained through IMU and a GNSS receiver operating in kinematic mode. The acquisitions of the Continuously Operating Reference Stations (CORSs) present in the surroundings of the three surveyed areas have been used. Subsequently, GCPs surveyed on the ground by GNSS receivers or TS were used.

The terrain model obtained from the aerial laser scanner was then appropriately verified. The point clouds acquired in adjacent strips have overlapping areas. To evaluate the accuracy of their alignment, after selecting the common area, the deviation between the overlapping meshes is calculated. Based on the trend of the deviations, the registration can be refined.

Vertical registration errors can significantly affect the estimation of water surface elevation, while horizontal misalignments may distort flow direction and velocity, resulting in inaccurate representation of flood pathways. These errors have the potential to introduce systematic biases in flood inundation modeling. For instance, underestimation of flood extent may occur if artificial depressions are introduced into the terrain model, whereas overestimation may result from the incorrect placement or smoothing of elevated features such as levees. The impact of vertical inaccuracies is particularly pronounced in low-relief areas, where even minor elevation errors can lead to substantial deviations in predicted inundation extents.

The possibility of discriminating the first and the last pulse was exploited. This opportunity allowed obtaining both the DSM and the DTM of the surveyed areas (Figure 6). The DTM is used for hydraulic calculations in the riverbed area, but the DSM plays a fundamental role, especially in urbanized areas, to be able to predict the effects of a flood. The difference between DSM and DTM, i.e., the difference between the first and the last pulse, can provide, in the riverbed areas, the average height of the vegetation and its density, which allow an accurate evaluation of the roughness.

Georeferencing of the land survey was achieved using a GNSS receiver operating in rapid-static mode. As in previous cases, data from nearby CORS stations were utilized. Several tie points were selected, while the GCPs, specifically chosen to support the georeferencing of the acquired data, were surveyed. GCPs coincide with fixed elements (vertices of bridge railings, riverbank parapets, road signs, etc.) easily identifiable in the point clouds obtained from aerial and terrestrial shots (Figure 7). Some points were identified inside the riverbeds or at the natural riverbanks, essentially on boulders or tree trunks present (Figure 8); these were used exclusively for georeferencing the point clouds acquired by the TLS, since the ALS survey had been performed earlier on a different date.

As regards the heights, the Quasi-Geoid ITALGEO05 was used, which covers the Italian territory [23].

This model is derived from gravimetric measurements integrated with GPS/leveling (LEV) observations, which were performed to connect the IGM95 geodetic network to the national leveling network. It also incorporates vertical deviation measurements (astronomical) and takes into account the topographic conformation of the terrain using a Digital Terrain Model (DTM).

The registration of the point clouds obtained from ALS and TLS was obtained by exploiting the GCPs and common features. Common surfaces—specifically horizontal and sub-vertical ones—were utilized during the process. Vertical surfaces help refine horizontal alignment, whereas horizontal surfaces are used to improve vertical registration (Figure 9). This last operation is particularly important, given that the points’ elevations play a fundamental role in hydraulic calculations. The orthometric heights of the models obtained by TLS were chosen as reference.

Figure 9 shows the extrados of a bridge, which was used for the registration. The model derived from the points collected with TLS was used as the ground truth; the distances between the ALS points and the reference surface allow for an assessment of the vertical accuracy of the registration. In this case, the vertical accuracy is ±15 cm. A similar verification was carried out for the horizontal accuracy, using near-vertical surfaces (such as building walls or retaining walls, like those circled in red in Figure 10). The resulting horizontal accuracy is ±16 cm.

2.4. Integration of Datasets for Hydrological Modeling

From the point clouds acquired with the ALS for each river, a Triangulated Irregular Network (TIN) was obtained. To reduce computational load during hydrological simulations, a simplification step was applied to the TIN in relatively flat or homogeneous regions. A decimation algorithm, available in open-source tools [31], was used in these regular areas; this retained detail in complex terrain while significantly reducing the number of elements in the mesh. As for points acquired with TLS, the decimation is essential. While the points acquired with ALS are uniformly distributed (in our case, 4 points/m²), in the case of TLS, there is an excessive density in proximity to the instrument, which must be eliminated in order not to have an excessive calculation effort without having as a counterpart useful information for hydraulic calculations.

To construct a continuous and hydrologically accurate terrain model, we integrated the Triangulated Irregular Network (TIN) datasets acquired by ALS and TLS representing adjacent or overlapping regions of the study area. The integration aimed to eliminate discontinuities or mismatches that could adversely affect hydrological simulations such as flow routing or watershed delineation.

After preprocessing and registration, the procedure involved the following steps:

-

Shared edges and overlapping zones were identified. Duplicate or closely spaced vertices within these areas were removed or averaged, and edge snapping was applied to ensure seamless connectivity.

-

The merged point set was re-triangulated to form a new TIN structure using Delaunay triangulation. Special attention was given to preserving terrain features and ensuring topological correctness, particularly in areas critical for water flow such as ridges and depressions.

The integration process utilized QGIS (with GRASS and SAGA plugins) and MeshLab [31,32].

3. Results

The results obtained can be summarized as follows: (1) orthophoto, (2) meshing of ALS and TLS point clouds; (3) multiresolution TIN; (4) 2.5D model and views; (5) representation of results on the Regional Technical Map, and (6) representation on orthophotos with enhanced detail.

3.1. Orthophoto

Orthophoto is the first product, obtained through ALS acquisitions combined with aerial imagery. In our case, they were used not only as documentation but also to identify those structures designated as housing or production activities that insist on flood areas.

This information has proven valuable for assessing exposure and vulnerability and, consequently, for evaluating risk, i.e., the potential impact of a hazard, (Risk = Hazard × Exposure × Vulnerability).

In this context:

-

Hazard refers to the probability of a flood occurring with a return period of 200 or 500 years.

-

Exposure denotes what or who is at risk from the hazard—in this case, buildings, infrastructure, environmental heritage, economic activities, and human lives.

-

Vulnerability represents the degree to which those exposed are susceptible to harm or damage.

-

Risk components can be assessed either quantitatively or qualitatively. For the study area, the Hydrogeological Plan of the Calabria Region adopts a four-level evaluation scale—ranging from low to very high—for each component as well as for overall risk [33].

Figure 11 illustrates the residential and industrial settlements along the central section of the Valanidi fiumara, while Figure 12 outlines the designated risk zones in the middle and lower stretches of the watercourse.

3.2. Meshing of ALS and TLS Point Clouds

The union of point clouds acquired from ALS and from TLS offers the possibility of exploring the area of interest for hydraulic calculations. It is an aid in identifying singularities (artifacts, abrupt variations in slope of the ground bed, etc.) in proximity of which it is necessary to refine the TIN.

Figure 10 shows a central section of the Corace River. TLS-acquired points are shown in green, while ALS-acquired points are displayed in blue. In panel (Figure 10a), some retaining walls of which the surface was used for registration are highlighted with red ellipses. The bridge deck shown in Figure 9, used for vertical alignment during registration, is highlighted in yellow. In both panels of the figure, the different densities of the point clouds acquired by ALS and TLS are evident.

3.3. Multiresolution TIN

One of the most significant results obtained is the variable-resolution TIN, which was used for hydraulic calculations [24]. Figure 13 illustrates a section of the Corace River featuring a bridge (Figure 13a); the TIN generated from ALS data, with 3D models of the bridge plinths and piers overlaid, derived from TLS and TS data (Figure 13b); and the mesh refinement applied in the central area of the bridge (Figure 13c).

3.4. 2.5D Model and Views

From the TIN, a 2.5D model was derived and used for calculations and visualization. The views are particularly useful for visualizing and interpreting the results of hydraulic calculations—such as water depth, flow velocity, and flood extent—as they enhance the understanding of spatial patterns and flow behavior. In this study, the visualizations were generated using the software ER Mapper^® 7.2 [34]. Several thematic layers were represented, each illustrating different aspects of the hydraulic modeling results, including flow velocity, eta, and sediment deposition.

Figure 14a displays the 2.5D model of the fiumara Valanidi, illustrating water velocity during a 500-year return period flood event. Velocity values are represented using a color scale, increasing from violet to red. Several flooding points are visible along the river corridor. Panel (b) provides a zoomed-in view of the central area; on the hydraulic left, a critical point is highlighted where the riverbanks are insufficient to contain the flow, resulting in overtopping and local flooding.

3.5. Representation of Results on the Regional Technical Map

Various 2D maps were produced to visualize the modeling results over the base cartography of the study area. Figure 15 displays the Eta thematic layer—i.e., the water surface elevation—superimposed on a 1:5000 scale technical map.

3.6. Representation on Orthophotos with Enhanced Detail

The final output consists of representing the thematic data on an orthophoto. This was initially achieved using an image with a GSD matching the resolution of the hydraulic model’s calculation grid. Specifically, Figure 16a illustrates the deposition theme overlaid on an orthophoto with a pixel size of 5 × 5 m.

To enhance the detail of the representation, a higher-resolution orthophoto with a GSD of 1 m was used. Although the hydraulic calculations were performed on a 5 × 5 m grid, the finer 1 × 1 m cells—corresponding to image pixels—were introduced purely for visualization purposes. The deposition values within each 5 × 5 m cell were redistributed over the 25 corresponding 1 × 1 m image pixels using the following equation:

D_i = D + (h_m − h_i)

(1)

where

D is the deposition value in the 5 × 5 m cell

D_i is the deposition value in the ith 1 × 1 m cell

h_m is the orthometric height of the 5 × 5 m cell

h_i is the orthometric height of the 1 × 1 m cell

This method provides a smoother and more detailed visual representation without altering the resolution of the original hydraulic computations. Negative values of D_i are set to zero:

IF D_i < 0 then D_i = 0

(2)

A further refinement was then performed, to account for the residual deposition R, calculated as follows:

$$R=25 D-{\sum }_i{D}_i$$

(3)

This residual represents the difference between the total deposition assigned to the original 5 × 5 m cell and the sum of the interpolated values distributed among the 25 corresponding 1 × 1 m pixels. To preserve mass balance, the residual R was redistributed equally among the pixels where D_i ≥ 0. Let n be the number of such pixels. A second approximation of the deposition value, D_i,2, was then computed as follows:

IF D_i ≥ 0 then D_i,2 = Di + R/n

(4)

Due to the relatively small magnitude of R/n, no further iterative corrections were considered necessary.

The procedure described above enabled a more realistic representation, in which features such as building walls, retaining structures, and artificial embankments are clearly highlighted. The benefits of this refined visualization go beyond aesthetics: the increased level of detail improves the identification of man-made elements within potentially flooded areas. In some cases, the enhanced resolution also makes it possible to reassess results initially derived using coarser calculation cells, potentially leading to significantly different values and highlighting the need for more refined hydraulic calculation. This is clearly illustrated in Figure 16b, where not only are the artifacts more distinctly highlighted, but significant variations in the thematic values are also evident. In particular, on the right side of the figure, the deposition values are noticeably higher in the trench area of the E90 road, located between the two final branches of the fiumara.

4. Discussion

In the context of hydrogeological risk mitigation, the integration of ALS and TLS point clouds represents a key step toward developing accurate and actionable terrain models. The successful meshing of these datasets relies heavily on robust registration techniques. In this case, a multi-instrumental approach was adopted, which is reflected in the integrated use of various point cloud registration techniques. These included the use of ground control points (GCPs), alignment based on sub-vertical and horizontal surfaces, and subsequent refinement through the Iterative Closest Point (ICP) algorithm. The peculiarity of the presented application lies in both its spatial extent—comprising two segments of approximately 10 km and one of 40 km—and in the diverse nature of the watercourses analyzed, ranging from a typical river to fiumara-type torrents.

The multi-instrumental approach also supported the targeted identification of areas to be modeled using TINs with varying levels of resolution. The final multiresolution TIN offers several advantages. It supports the detection of small-scale features—such as incipient erosion, structural discontinuities, or sediment accumulation zones—that may be overlooked in uniform-resolution models. Furthermore, the enhanced detail in high-risk zones allows for more accurate hydraulic simulations, which are essential for assessing flood paths, sediment dynamics, and the impact of protective infrastructure.

Evidence supporting the above statements can be observed by comparing the 2.5D model outputs (Figure 14) with the risk map corresponding to the middle and lower sections of the Valanidi stream (Figure 12). The data refer to a flood event with a 200-year return period. Figure 14 clearly illustrates the stream overflowing as a result of a breach in the left-bank levee, a phenomenon also visible in the 2D representation (Figure 15). The inundation extends into areas classified as non-hazardous on the risk map, both outside and within the zone delineated by the stream’s two terminal branches.

The model’s high level of detail around anthropogenic structures enabled precise hydraulic simulations. These results can be highly valuable for the development of emergency response plans and the design of hydrogeological risk mitigation strategies by helping to prioritize intervention measures effectively.

A particular methodology has been introduced to refine the representation of the “deposit” theme, moving from the resolution of the hydraulic calculation to that of the orthophotos. The refined orthophoto serves as a valuable tool in the broader effort to mitigate hydrogeological risk. By integrating thematic data—such as deposition values—into a high-resolution, georeferenced image, it becomes possible to more effectively visualize not only the spatial distribution of key parameters but also the interaction between natural processes and human-made features. The ability to identify structures such as check dams, embankments, road trenches, and retaining walls directly from the orthophoto enhances the interpretation of hydraulic modeling results. Specifically, the enhanced resolution enables the precise delimitation of vulnerable zones and existing infrastructure. From a planning and decision-making perspective, such a product supports more informed risk assessments. It enables civil protection authorities and engineers to better locate intervention priorities, calibrate hydraulic models more accurately, and even reconsider the placement or effectiveness of structural mitigation measures. Therefore, the orthophoto is not merely a background image but an analytical tool that bridges hydraulic modeling and territorial planning, helping to translate technical results into practical actions aimed at reducing hydrogeological hazards.

For flood modeling, a comparison between the multi-instrument approach and alternative (typically single-method) methodologies can be made in terms of accuracy, reliability, and cost-effectiveness. The multi-instrument approach provides high topographic accuracy across all areas of the model, enabling detailed representation of channel geometry, floodplains, and infrastructure. In contrast, single-method approaches often result in limited accuracy or redundant data, particularly in complex terrains. In terms of cost-effectiveness, methodologies designed to generate multi-resolution TIN models incur higher costs due to the need for specialized equipment, skilled personnel, intensive data processing, and complex logistical operations. However, these challenges are often justified by the significantly superior results it delivers, making single-instrument approaches suitable only for limited areas or simpler terrains.

As part of the research activities on the Crati River, a model was developed using the traditional method based on surveys of cross-sections along the river course. Costabile and Macchione [35] conducted a comparison between the resulting models, which demonstrated the clear superiority of the methodology we proposed—enabling full 2D dynamic flood modeling—both in terms of hydraulic calculation accuracy and the quality of spatial representation.

5. Conclusions and Further Research

Several survey techniques (photogrammetry, ALS, TLS, GNSS) have been employed in the framework of activities carried out on three watercourses in Calabria (Italy).

A workflow integrating these survey techniques has been developed and applied to generate 2.5D multisensor, multiresolution riverbed models, achieving both horizontal and vertical accuracies of approximately ±15 cm.

The results confirm the feasibility of producing models that are both accurate and efficiently compatible with the memory and processing capabilities of modern computers.

Furthermore, the technique set up and implemented for the refined representation of both the models and the results of hydraulic calculations is a valuable tool for improving hydrogeological planning, designing appropriate defense works, and preparing evacuation plans in case of emergency, all with the goal of mitigating hydrogeological risk.

Further research will face the difficulties connected to riverbeds with dense vegetation along with the use of bathymetric data.