4.1. Quantitative Description
Lake waters were conveyed into a collector channel, which ended at the
incile where the waters were introduced into the emissary. On the lakeside of the
incile there was the elliptic tank, the first tank of the outlet. In the deepest part, this corresponded to an elliptic basin, which, in the upper part, widened into a triangular-shaped volume. The south wall was curved towards the inside of the basin, giving the tank a peculiar shape (
Figure 2). The structure had, in the center of the bottom part, an opening with a second gate adjustable by a maneuver room placed immediately above the collector tunnel. This tunnel ran below all the
incile’s structures to the last tank: the trapezoidal tank. The water coming from the lake, after passing the second gate, fell into a third tank, smaller than the previous ones, but deeper: the trapezoidal tank, from the actual geometric shape it has in Brisse and De Rotrou’s plan and in all the other representations. On the longest side of the trapezium, there was a gate used to regulate or to block the water flow coming from the lake before it went inside the tunnel of the emissary.
Just after the elliptical tank, a short passage (narrow 3 m, long 8 m, and deep 10.5 m) connected the first tank with a presumed large basin. The shape of the latter is not exactly known, since Brisse and De Rotrou did not consider it necessary to excavate its entire perimeter. The other iconographic sources, such as Fabretti, Piranesi, and Melchiorri, who describe this intermediary basin assuming that it had a hexagonal configuration, could only speculate about its shape since it was probably covered with soil.
As regards to the functionalities of the incile's tanks, they probably had to serve as the basin of expansion and, slowing down the running water, for decanting, collecting, and removing the various debris that had managed to pass there. The same could be for the presumed hexagonal tank.
The incile illustrated in Torlonia’s Atlas, also described in other iconographic sources, was the incile in its final configuration. Most likely, this work has changed over time during the drying phases of the lake and the tanks were built in different phases, adapting the incile’s structure to the level of the lake’s waters gradually receding. The final configuration may have taken decades to be achieved. The alleged hexagonal tank may have existed in a specific phase of the Roman drainage work and, when no longer necessary, it was destroyed and then other parts of the incile were built.
In the virtual reconstruction of the incile some hypotheses have been made, especially about the area behind the larger side of the trapezium constituting the tank. Its internal articulation is, in fact, only partially illustrated in Torlonia’s Atlas and some fundamental data necessary for its reconstruction are missing. The first problem, regarding this tank, is the access to the maneuver chamber. According to the description expressed by Brisse and De Rotrou, there were two different tunnels. The north one drove directly to the artificial emissary, while the south tunnel led to the area with the emissary’s gate valve. However, Torlonia’s engineers did not care about fully describing these areas, so it is not clear how they were originally built. In the first edition of the Atlas, written in French and English, Brisse and De Rotrou describe the north entrance as a descending tunnel, while the south one is addressed as a short tunnel, with a staircase directly connected to the floodgates room. The filling and sanding of a large part of the incile, especially the central basin, make the reconstruction more difficult, multiplying the possible hypotheses.
The back wall of the trapezoidal basin, leant against the slope of Mount Salviano, was the head of the emissary tunnel, where two distinct parts were distinguished. At the bottom, there was the entrance to the emissary tunnel, whose floor was at the same level as the bottom of the trapezoidal tank. A third gate, adjustable by an overhead operating room, possibly served the entrance to the tunnel. A special service staircase probably connected the tunnel for possible inspections and maintenance interventions, once the water flow was blocked by means of the third gate.
In the upper part there were service rooms spread over two floors: In the lower floor, there was the maneuvering room for the third gate. The two floors were presented with a three-arched loggia-like structure. In the virtual reconstruction proposed here, we assumed that the three arches were built with two rows of bricks, using the typical measures of the Roman bricks for them.
On the back of this upper wall, behind the three arches, it is possible to see in the detailed Torlonia section, a door or an old passage bricked up (
Figure 3a). There is no clear evidence that can define and explain its possible role. Nevertheless, in Piranesi’s print (
Figure 5) a pathway is represented with several steps. It can be assumed there was an access route with a ladder walled opening.
The philological reconstruction of this now lost structure would be very useful also for studying the
incile functionality and understanding the most innovative knowledge of Romans, at that time, about the technologies used for lake drainage and reclamation works. For this purpose, the data concerning the lake waters levels are fundamental. Some information about the levels and depth variations of the Fucino waters (performed during the period in which regular measurements were made), comes from the middle of the 18th century, up to the final draining of Fucino Lake in the 19th century. There are also studies that speculate on the variations of the waters of the lake even in previous centuries [
27]. The variations of the water levels, from 1783 to 1875, are shown in the Brisse and De Rotrou Atlas.
4.2. The Relocation of the Roman Incile
Today there are no ruins that can provide evidence as to where exactly the ancient
incile was located. Furthermore, the area has changed a lot during the last centuries and deep alterations occurred in the territory, especially due to the 19th century works and modern human activities, such as agriculture or urbanization. Analyzing the historical sources and the current structures, a hypothesis has been formulated here. Torlonia’s drainage works envisaged, in an intermediate phase, the reuse of the
incile with some modifications of the ancient structures. Some parts of Torlonia’s modified
incile correspond to the well n°32 of Torlonia’s final emissary. In
Figure 8, the ancient emissary’s drawing is found, modified by the addition of two wells (evidenced with a blue elliptic) and the new structures are shown on the right, according to Brisse and De Rotrou’s representation. Today the well n°32 is still present in the territory and can be observed in the countryside near the place where the
incile was situated (Borgo Incile di Avezzano, AQ) (
Figure 9). In order to place the
incile in the territory, an emissary’s survey has been performed with traditional technology (
Figure 8), while the surrounding area was detected with UAV photogrammetry.
Since the incile is part of a significant hydraulic work, the quality of geographical data is very important especially with regards to the elevations. For this reason, accurate data are necessary, particularly considering that the aim of the incile’s reconstruction is to investigate the hydraulic opera’s functionalities. Consequently, the territory data, provided by traditional surveys, are integrated by aero photogrammetric data.
The territory has been acquired by DJI Mavic Pro camera, whose technical specifications are reported in
Table 1.
The UAV is equipped with a GNSS sensor (GPS + GLONASS), whose accuracy is equal to a few meters. The entire area, which starts from the well n°32 to the Cunicolo Maggiore, was detected (
Figure 10). This is an area 300 m large and 1000 m long. The flight altitude was 70 m with respect to the ground of the starting point (Cunicolo Maggiore). A lower value could not be used due to the presence of orographic obstacles near the Cunicolo Maggiore. Anyway, a ground simple distance (GSD) of about 4 cm/pixel was obtained for a large part of the map, which includes the area of major interest around the well n°32. The obtained accuracy of cloud point is generally greater than the GSD value and this value of repeatability can be considered adequate for the application in this work. Nevertheless, to improve the quality of the model, 5 ground control points (GCPs) were used as reference and two of them, in particular, were assigned to external solid elements of the emissary. Those points, distributed over the whole area of interest, allow to georeference the model and permit to scale exactly it. They were chosen between the fiducial points provided by the territory’s Agency of L’Aquila province (
https://www1.agenziaentrate.gov.it/servizi/Monografie/ricerca.php). A key-feature of those points is that they are clearly distinguishable and consequently the use of markers for detecting them from the photos was not required. Six control points (CPs), moreover, were identified, whose altitude is known. Some of them were marked with the use of 30 × 30 cm squares.
All the previous information was taken into account for defining the flight plan, which has been created using the web-based application DroneDeploy. The map of the flight plan is evidenced in
Figure 10. In that image, the yellow line represents the drone’s path, while the green points identify where photos have been captured. These photos were taken with an overlap of 75% for the front and of 65% for the side. In order to minimize the distortion effect, the viewing angle
of the camera was fixed to
deg for all the photo acquisitions. Consequently, some details of the vertical walls of the surrounding houses were not captured during the survey, but they were considered as irrelevant for our final goal.
Table 2 reports a summary of the parameters of the UAV survey. In
Figure 10, the square points correspond to the GCPs and the triangular points identify the CPs.
At the end of the mission, about 390 photos were acquired, which have been processed in the software Pix4D (developed by Pix4D SA, Route de Renens 24, 1008 Prilly, Switzerland), one of the most used photogrammetry software. During the initial processing, all the information concerning the camera and the photos were extracted, such as orientation of the camera, focal length, geolocation info, etc. Then tie points were identified by the use of a bundle adjustment approach. At the end a densified point cloud was obtained, which was used to create the 3D textured mesh. The settings parameters are quoted in
Table 3.
After about 4 h of elaboration by using an Intel Xeon CPU E5-2680 (64 Gb RAM) work-station, the point cloud was obtained (
Figure 11).
The scaling and rototranslation of the point cloud were carried out by using as reference three points, two of them located respectively on the well n°32 (point D of
Figure 10) and the Cunicolo Maggiore (point A of
Figure 10). The other point (point C of
Figure 10) is located outside the tunnel alignment. The well n°32 and the Cunicolo Maggiore are reference elements of the Emissary and they were used by ancient Romans to determine the direction and the depth of the Emissary tunnel. It was for this reason that the point cloud was transformed using these elements instead of other geo-referenced points. The transformed point cloud was verified by checking the elevations of the 5 CPs which are referenced in altitude. In
Table 4, the errors for elevation (in meters) are reported.
Then the 3D mesh, with the real texture of the territory applied, was generated. On this mesh, the virtually reconstructed
incile has been relocated, so that the ancient
incile is inserted in the territory as it appears today. Its position error should be limited to a few meters. The final renderings of how the reconstructed
incile could appear nowadays are reported in
Figure 12. In the construction of this picture, the point of view of the representation of Piranesi (
Figure 5) and the same level of the water have been used.
As regards to materials and textures for the incile’s reconstruction, they have been chosen taking into account what can be deduced by the available iconographic references and in compliance with the typical building technologies used in Roman times for similar works. In particular, the texture applied to the 3D models comes from a reworking of pictures of the Cunicolo Maggiore (the monumental tunnel connected to the Roman emissary) and from the modern Torlonia incile, where the same type of materials can be found. The textures from the Cunicolo Maggiore represent the uncertain Roman stone masonry with appeals in rows of bricks partially used in the ancient incile (opus mixtum). The carved stones of the walls of the modern incile with indentations are a reference for the limestone material used for reconstructing the entrance of the tunnel in the ancient incile.
In some cases, the textures have been constructed by composing single elements; this is the case with the arches with two orders of bricks (
Figure 13) or with the stone coverings made in
opus reticulatum.
Figure 14 shows the longitudinal section of the Roman
incile in a current setting performed by the 3D virtual reconstruction.
All renderings have been generated by using Cinema 4d R.20 (developed by Maxon Computer GmbH, Max-Planck-Str. 20, 61381 Friedrichsdorf Germany).