**3. Results**

#### *3.1. Remote Sensing*

The analyzed satellite images and aerial pictures give a consistent idea of the changes in the land use of the area and the visibility of the paleohydrographic traces over the last 60 years. In particular, recent high-resolution true-color composite images allow mapping these features with submetric accuracy. The traces mainly consist of cropmarks with vegetation suffering from hydraulic stress during the growing season, and dead patches in July and August. Significant changes can be observed over very short periods (compare, e.g., Figure 4d,f) during the hot season, while sensitivity is low in spring time (e.g., Figure 4e) and is lost in winter, when the fields are covered by scarce vegetation and are bare and plowed, respectively. By comparing satellite images mostly taken during the growing season (i.e., those that best show the traces) differences in vegetation colors allow the identification of buried morphologies of two distinct reaches, hereafter named Reach A and Reach B, and a crevasse

splay. Reach A consists of two major bends, called Bend 1 and Bend 2 (Figure 4b). The NDVI and NDMI images derived from Sentinel-2 summer images (Figure 5c,d) clearly show the di fferential behavior of the paleomeanders in comparison to the external floodplain and the inner portion surrounded by Reach A; the vegetation growing on more permeable sandy soil is less healthy than the one living on finer sediments. The paleohydrographic traces are much more evidenced by NDVI and NDMI during the summer season and they clearly help in identifying the general pattern of the abandoned channels. However, the rather low resolution of the Sentinel-2 images does not contribute significantly to discriminating in detail the specific morphological and sedimentological features composing the paleohydrographic traces.

**Figure 4.** Remote-sensing results: (**a**) aerial photogram of the study area enhancing the poorly visible fluvial pattern during winter time; (**b**) meander belt reconstruction and fluvial morphology identification in the study area—white dashed lines highlight evidence of scroll-bar morphologies; and (**<sup>c</sup>**–**f**) satellite images (2016, 2013, 2012) providing information about fluvial morphology on the basis of seasons and soil use.

**Figure 5.** Processed satellite images: (**a**) RGB combination on Sentinel-2 image to highlight petrographic differences in the study area; (**b**) band 8, of Sentinel-2 image of the study area, showing paleochannel morphologies; and (**<sup>c</sup>**,**d**) NDVI and NDMI indexes calculated on Sentinel 2 images (2018), respectively, enhancing different plant health and hydraulic stress.

Bend 1 is an open bend, with an SSW–NNE bend axis, and is characterized by a sinuosity of about 2.2 and a radius of curvature of ca. 140 m. The scroll-bar pattern is particularly clear from cropmarks, in the northcentral portion of the bend (Figure 4c,d), showing a different signal compared to the residual channel fill (i.e., light cropmark when the others are dark, and vice-versa), testifying the progressive growth of the meander bend. The channel fill displays a width of about 15–20 m and can be better defined where bounded by opposite-trending scroll-bar patterns, like in the upstream side of the bend. The riffle-to-riffle distance on the channel fill is about 260 m.

Bend 2 is a low sinuosity bend (i.e., 1.12), with an NW–SE bend axis, characterized by an estimated radius of curvature and ri ffle-to-ri ffle distance of 135 and 230 m, respectively. Bend 2 is sited upstream from Bend 1, and shows a scroll-bar pattern that testifies a progressive expansional growth style (sensu [77]) of the bend.

Reach B forms a bend occurring south of Reach A, but is less visible from satellite images and its sinuosity cannot be defined. The radius of curvature is ca. 350 m and the axis of the bend trends ca. SW–NE, although satellite images do not show a clear bar-scroll pattern and the position of the relevant channel fill. Reach B cuts over Reach A suggesting that it developed after a chute channel that cut o ff Reach A [78], which was later abandoned. Additionally, along the eastern side of Reach B, (Figure 4d,f) a divergent pattern of minor channels point to a local development of a crevasse splay sourced from the downstream side of the bend. Several straight dark stripes, with a width of about 1 m, located on Reach B, can be interpreted as traces of abandoned ditches that were associated with a drainage system dating back to Renaissance times and dismissed later on (Figure 4d).

#### *3.2. 3D Electrical Conductivity Model from EMI*

The inversion of the EMI data produced a 3D volume of electrical conductivity values. The results are shown in Figure 6, where we elected to show the volume sliced horizontally at eight depth levels down to a depth of nearly 8 m below ground, corresponding each to a layer selected in the inversion approach. Note that the inversion was conducted with an Occam approach, but using a limited number of layers compatible with the information contained in the six di fferent acquisition configurations obtainable with the CMD Explorer instrument.

An arcuate sedimentary body having low electrical conductivity (i.e., a resistive body) is clearly visible at a depth between 1 and 6 m below ground. The internal boundary of this arcuate body (see Slice 5 in Figure 6b) is fully visible in the maps, and shows a radius of curvature and a sinuosity of ca. 60 and 2.3, respectively. The external boundary (see Slice 5 in Figure 6b) slightly debouches from the maps, but its radius of curvature and sinuosity can be estimated to be ca. 135 and 2.2, respectively (Figure 6b), as also confirmed by remote-sensing results. Orientation of the outer boundary of this body fits with the orientation of meander Bend 1 of Reach A, as depicted by remote-sensing analyses, and is also consistent with the associated scroll pattern (Figure 4b), suggesting that these low-resistivity deposits represent the point-bar body associated with meander Bend 1 of Reach A. Of course, the main contribution of the EMI data is to provide continuous and extensive depth information that is not available from remote sensing. In the shallower layers (Slices 1–5), the arcuate point-bar body presents low conductivity values with σ < 20 mS/m, and its conductivity is still close to 40 mS/m at about 5–6 m below ground (Slices 6 and 7; Figure 6b). Note that the width of the most resistive part of the bar is clearly shrinking with depth, thus showing the 3D shape of the sand body. At larger depths (Slice 8—below 6.1 m) conductivity increases up to 180 mS/m, delimiting the base of the bar body. It must be noted, however, that the CMD Explorer provides, as a rule of thumb, reliable information only down to 6 m below ground and thus Slice 8 is e ffectively an extrapolation due to the need to have an infinite semispace at the bottom of the electrical conductivity model, and thus should be considered with care. Although the point-bar body shows a fairly homogeneous electrical resistivity, a subtle increase in resistivity values defines a 20 m narrow, NNE–SSW trending belt in the SE corner of Slices 2 to 6. The location of this belt fits with that of the abandoned channel forming the meander Bend 1 as apparent in satellite images (Figure 4b), and suggests that the higher resistivity values are linked to the coarser material of the deposits filling the abandoned channel. Deposits surrounding the low-resistivity point-bar body show conductivity values spanning from 80 to 250 mS/m, with values close to 100 mS/m down to 3 m below ground, increasing to 250 mS/m below 3 m. Comparison between geophysical data and geomorphic evidences sugges<sup>t</sup> that these electrically conductive sediments represent floodplain deposits in which the Bend 1 meander was cut, thus developing the related point-bar sedimentary body.

**Figure 6.** Geophysical results: (**a**) 3D view of the eight 2D conductivity maps; (**b**) the eight 2D maps showing difference in conductivity values and highlighting point-bar, channel-fill and floodplain morphologies.
