Quantitative Estimation of Altimetric Variations Using Residual DEM from Topographical Maps between Two Epochs 1893 and 1985: Case Study: The Endorheic Basin of Enfidha-Tunisia

This article endeavours to analyse the recent deformation in the Enfidha region. This analysis has been carried out using the Residual Digital Elevation Model (DEM). It is the altimetric difference between two DEM generated from the contour lines of two topographical maps over a period of 100 years. This deformation has been studied by some authors who report the presence of recent ground movements by comparing contour lines with a downward trend in elevation from 1893 to 1985. In 2006, this study area was marked by the presence of two earthquakes that occurred in several coastal cities located in the northeastern part of Tunisia. Our study involves a quantitative estimation of altimetric variations under a Geographic Information System (GIS) environment. Our proposed methodology aims at the mapping of residual (DEM) and the extraction of parameters that have a morphological and morphostructural signature. The extraction of quantitative morphostructural parameters requires the integration of multi-source and multi-scale data. This can only be done if the problem of heterogeneity at the level of scale and coordinate system is solved through the use of GIS tools and the obtainment of the vectorial shapefile format. Then, in order to compare the DEM generation errors with reference to recent and old data, they must be projected in the same projection system and on the same scale. The available data are two topographic maps of Enfidha which represent two different epochs. The first one is an old topographic map of 1893 (type 1922) at a scale of 1:50,000 and the second one is a recent topographic map of 1985 at a scale of 1:25,000. These topographical maps have the Lambert (IGN) projection system. This How to cite this paper: Rebai, N. and Moussi, A. (2019) Quantitative Estimation of Altimetric Variations Using Residual DEM from Topographical Maps between Two Epochs 1893 and 1985: Case Study: The Endorheic Basin of Enfidha-Tunisia. Journal of Geographic Information System, 11, 429-456. https://doi.org/10.4236/jgis.2019.114027 Received: May 9, 2019 Accepted: August 9, 2019 Published: August 12, 2019 Copyright © 2019 by author(s) and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/


Introduction
Since the early orogenic phases, the Tunisian Sahel region and its foreland have recorded distinct cycles of uplift, erosion and displacement estimated displacement estimated in some regions at a few hundred metres [1] [2]. The repeated folds generally result from maximum compressive stresses oscillating between an NW-SE pole [3] [4] [5] and an NS pole [6] [7]. This causes the individualization of a complex system of folds and evolves integrally in the molded faults on the faults solicited by the same compressions. The generated structures are always altered by an intense erosion that gives these folded structures their geomorphological imprints. These imprints can be observed and analyzed using the different hydrographic networks that cross them. These networks, once plotted on a topographic map, allow to qualify and quantify the tectonic movements deployed over a given period of time. In our case, this period is about 100 years. This work has four main objectives: 1) to measure, using morphometric indices, the amplitude of the movements responsible for the collapse of the endorheic basin of the Enfidha region 2) to develop a morphological pattern in line with that of the tectonics of the region 3) to present an evolutionary scenario of this region compatible with the general geodynamic framework of the Atlasique Tunisie 4) and to relate this evolution with that observed in the western Mediterranean during the quaternary period [8] [9]. To achieve these objectives, we have collected data from the analysis of a Digital Elevation Model (DEM) of two topographic maps with two different periods from 1893 (type 1922) at the 1/50,000 scale and from 1985 at the 1/25,000 scale with the Lambert (IGN) projection system. The validation of recent deformation is done by direct observation in the field. This is only possible in the GIS environment and the use of GIS tools that enable the shapefile vector format to solve the problem of heterogeneity, superposition and application of mathematical operators. The various studies carried out using morphometric index and structural analyses [10]- [21] indicate that the DEM, the hydrographic network can significantly determine the existence of N. Rebai, A. Moussi Journal of Geographic Information System tectonics or erosion. In the present work, we are interested in quantifying and highlighting the recent deformation by applying a comparative approach to the DEM of two topographic maps from 1985 to 1893. The assessment of the results is carried out through field observations in two sites. Essentially, a methodological approach has been developed in this context. It is based on the residual (DEM), which corresponds to an arithmetical difference of two DEM.

The Geomorphological Context
Enfidha region is situated between Zaghouan mountain range and the sea. It is located in the middle of a plain in the Tunisian Sahel.

The Geologic Context
The Oligocene facies distribution throughout the sandstone pole and the carbonate pole in the Sahel and in offshore domains [22] reveals some latitudinal and longitudinal lineaments. In the Tunisian Sahel eastern margin, along with the Gulf of Hammamet and Cape Bon, the Miocene and Pliocene sedimentary series are most often buried below the surface. In the northeastern Sahel, the Miocene and Pliocene basins are also characterized by synclinal gutter formation associated with folded structures along the EW tectonic corridors of El Haouaria, Tazoghrane, Hammamet-Maamoura and the N-S of Jriba-Enfidha [23].
Until today, The study area was also marked by the predominance of Quaternary deposits with recent soils and alluvium occupying the Enfidha plain and the mountain foothills, as well as the oligo-Mio-Pliocene sandstone deposits ( Figure   2).
Significantly, the study area includes the Enfidha block [7] [24], which corresponds to a folded and faulted zone. The geologic structures of Enfidha are closely associated with the N-S axis [25]. Thus, this axis limits our sector from the west. It corresponds to a deep suture due to the heterogeneity of the bedrock [25]. This axis separates a stable area in the East from another unstable area in the West [25]. It is due to the reduction in the atlasic direction folds (NE-SW) which become NS as they approach the Haffouz-Kairouan area in the south [25].
Towards the east of this N-S axis, oriental plains are developing, dominated by Mio-plioquaternary outcrops [23] [26] [27] [28]. In this way, the processing and interpretation of petroleum seismic profiles, seismic profiles, gravity data and N. Rebai, A. Moussi    in the eastern platform which is characterized, during the Mesozoic period, by a slower and more active subsidence during the Cenozoic period. The style of its deformation is essentially marked by the play of horsts and grabens, associated with folds of a large curvature radius [38]. The tectonic deformations recognized by seismic data in depth [23] [39] affect only narrow, elongated areas oriented in three major directions: N45, N100-120 and N160-180. These zones, which are active at several geologic and tectonically complex times, delimit large areas with few or no deformations [38]. The Miocene deformation is much less intense than in the other parts of Tunisia. Indeed, the eastern platform is only affected by the N45 direction folds, often accompanied by reverse faults and associated with the N90-110 dextral and N160-180 senestral faults [39]. The Plio-Quaternary deformation proofs are recognized outcrops in small areas, most often in the extension of the Atlas structures and as a result of accidents detected by seismic reflection [39]. This deformation is represented by a succession of dissymmetrical anticlines, with straightened NW flanks, and by N90 dextral and N160 senestral strike-slip, all probably due to pre-existing accidents [39]. According to [30], in the further east of Zaghouan accident and precisely 40 kilometres away, is located the new overlapping front of the Atlas Mountains of Tunisia. This is a NNE-SSW to North-South accident, connecting Chrichira in the south, to Enfidha in the north along more than 100 km [30].

The Background and Purpose of the Study
The context of this study is essentially based on [40] previous work, develops a the comparative examination of the isohypses of the old Enfidha topographic map (n˚43, 1/50,000, lifted in 1892), the new Enfidha NE topographic sheets (n˚43-NE, 1/25,000, published in 1990) and the topographic surveys of the railway level control at entrance of the city. According to a conventional approach and with recourse to multisource data, this comparison permits us to remark that in the Enfidha plain, the contours are clearly shifted upstream. This displacement expresses a downward revision of the equivalent altitudes of a subsidence trend. This phenomenon is apparent the case of the Enfidha plain, around Sebkhet el Kelbia and Bled Saadia (Jbibina sheet, n˚48) [40].
This study focuses on the implementation of a methodological approach

Methodological Approach
Our methodological approach presented by the flowchart (Figure 3), is based mainly on the generation of residual (DEM) in a GIS environment that makes possible the processing of spatial information and the estimation of the altimetric variation in the Enfidha lowland.

Scanning Topographic and Geologic Maps
The resolution of the scan affects the quality of the digitization afterwards, espe-N. Rebai, A. Moussi Journal of Geographic Information System cially when using documents on a large scale; where the boundaries of buildings, roads, etc. can be "incorrectly" reproduced. However, the use of a very high resolution is not always necessary since it is impossible to exceed the accuracy of the initial document, and moreover the size of the scanned document becomes more difficult to handle with a computer later on. In general, a resolution between 250 dot per inch and 300 dot per inch is sufficient.
Paper topographic and geologic information, supplied by the (Office of Topography and Cartography and National Mining Office), were scanned directly to RGB color TIFF files using a high-resolution, large format drum scanner at an optical resolution of 300 dpi.

The Georeferencing Process
The first step in georeferencing, or warping, the TIFF was to generate a neatline coverage with precise corner coordinates. A preexisting neatline or index grid could also be used as a reference. Georeferencing process allows the image to be displayed in a precise coordinate system and cartographic coordinates to be used to assign a spatial location to map features. Finally, all the elements of a map layer have a specific geographical position and extent that allow them to be located on or near the earth's surface.
The digitized topographic of both 1893 and 1985 and geologic maps were geo-referenced in the ArcMap environment using the kilometric grid of the Lambert North Tunisia coordinate system. An error measurement (residual error) is returned when the general formula (polynomial transformation) is calculated and applied to the control point. This error is the difference between the final position of the origin point and the actual position specified (destination point). The total error is calculated using the root mean square errors (RMSE) of all residues. This value describes the homogeneity of the transformation between the different control points (links).

The Digitization Process
A blank shapefile was created for every topographic and geologic GIS layer before digitizing began. The next operation concerns the digitization of contours lines from the three maps of Enfidha, one of 1893 at 1/50,000 and two SW/SE maps of 1985 at 1/25,000 under the ArcGis software. Similarly, the geologic map is digitized to present the lithological formations and site locations of field observations used to validate the obtained results with respect to the altitude variation between two periods of 1893 and 1985 by the residual (DEM). ESRI ArcGIS version 10.2.2 software was used, as well as various vector data formats (ArcInfo covers, shape files and geospatial databases) and raster images (ArcGrid, TIFF) to perform the conversion. During the completion of the project, geodatabase topological tools were used as (must not self intersect) for contour lines and (must not have gaps and must not overlap) for geologic formation contacts.

The Homogenization of the Scale Map
This operation concerns the homogenization of the scale by adjusting the two N. Rebai, A. Moussi Journal of Geographic Information System topographic maps (SW, SE at 1/25,000 scale) to 1/50,000 scale. As is known, the 1/25,000 map contains more detail than the 1/50,000 map so it is more reasonable to make the transition from 1/25,000 to 1/50,000 scale without loss of information that is not possible otherwise, where you cannot add and provide non-existent details to the 1/25,000 scale map. In fact, the applied treatment is based on the transition from the 5 m equidistance, of the topographic map at a scale of 1/25,000, to the 10 m of the topographic map at a scale of 1/50,000. This treatment is based on the spatial request by the selection of 10 m countour line interval and not 5 m of the 1/25,000 map. This process provides the same contour lines configuration of the 1/50,000.

The Overlay and Comparison of the Contours of the Two Topographic Maps
After

The Generation and Validation of DEM
Based on cartographic modelling that involves an ordered sequence of elementary GIS operations on maps operating on raw data as well as on derived data, the aim is to stimulate a spatial modelling process [41].
In fact, Digital Elevation Models provide true terrain configuration information in the form of morphometric and topographic parameters [13]

The Performance of Error Statistics and the Evaluation of the
Accuracy of the DEM Thus, the data obtained by digitizing the contour lines on the various topographic maps can be transformed into a (unique) database.
The study area topographic data that is available in a vector format for the study area is derived from the digitization of the topographic map of the Enfidha  Table 1).
The statistical results analysis ( Figure 5) shows relatively symmetrical distributions (Sk > 0) associated with a leptokurtic orography [52], uniformly distributed orographic volumes and ridge lines developed with very pronounced topographic maxima. A positive kurtosis (Ks > 3) indicates that the tails allow more observations (Figure 5(a)). Asymmetric distributions ( Figure 5(b)) on the left (Sk > 0) and (Ks < 3) are associated with a platykurtic orography [52].  The external validation method (with control data), described by [45], assesses the quality of a parameter noted z, corresponding to any local parameter that can be extracted from the DEM of n control points, knowing that the first parameter generally evaluated is the elevation. The assessment in point i is noted ẑi . The value of the control data at this point is noted zi.
with an average of errors defined as follows (3): According to [57], for the internal validation (without control data), the DEM's compliance with landforms can be controlled by visualizing the parameters derived from the surface because they are more sensitive to the accuracy of the altitude values and artifacts of the interpolation method than at the altimetric level. [58] also proposes to visualize the result of the DEMs filtering (for example a Laplacian filter to detect the objects contours). The quality of a DEM depends essentially on a series of choices made during the construction of this DEM. These choices concern both the basic data used to obtain all the reference  [59]. Indeed, it can be analyzed with regard to genealogy (source data, acquisition and construction methods), the morphological context (terrain characteristics) and also the timeliness of the data [45] [60] [61]. It is necessary to work on a common grid rather than a specific grid for each survey. It is also important to use the same interpolators (e.g. Ordinary Kriging) applied to different topographic data sets. DEM is built at 10 m intervals (73 rows × 100 columns) by an "exact interpolator", such as the ordinary kriging. It is a local stochastic (or geostatistical) method developed by [62] [63], with one-time databases of two date maps. The one must have almost the same value of the root of the Mean Square Error (RMSE) ( Table 1 and Figure 6), a low distribution in the plain and a high distribution in the hills (Figure 7).
The map of the average interpolation errors generated by a calculation on both rasters by the following equation:  Table 2).

Residual DEM
The main idea behind this article is to perform a simple raster subtraction between two DEM with different dates and perfectly identical footprint. The obtained result is the arithmetic difference between the DEM of 1893 and that of 1985. As shown in the flowchart (Figure 3), the generated DEM is in "Raster" format, with image oriented data presented as a group of juxtaposed square pixels, which effectively mark the arithmetic pixel by pixel difference (GRID).
This GRID difference reflects our use of two "GRID" DEM with identical alignment and cell size used to calculate the two models generated in 1893 and 1985. However, the difference in DEM in vector format presents relatively exaggerated errors with a very rough estimate by setting the same coordinates for each point on the map where the arithmetic difference remains dependent on having the Z for the same points from one companion to another.    We have calculated the altitude differences on the same grid and applied an interpolator to generate the DEM. The comparison of the residual between the DEM enables the visualization and quantification of the altitude variation over the entire sector. These variations involve slope or local subsidence variations [64].

DEM and Derived Models
As previously mentioned in the introduction, this work is based on the analysis of recent deformations by the residual digital elevation model (DEM) that reveals morphostructural responses.
Through simple and complex spatial analyses, our methodology consists in extracting several morphometric parameters from the DEM in order to study the morphostructural behavior and the recent deformation of the investigated area.
In this study, two spatial analysis levels were performed. The first is a complex spatial analysis that allows us to identify the second and third order morphometric parameters: topographic profile, slope profile and topographic residual ( Figures 9-11). All these parameters permit us to determine the morphostructural behavior and establish a geomorphological framework for the studied massifs.
The second is a simple spatial analysis that makes it possible to identify first-order morphometric parameters (slope map, relief map, three-dimensional views) (Figures 10-12).

Residual Topography
The residual topography is also a grid surface of a high analytical importance [15] [66] [68] [70]. It is produced by the arithmetic subtraction of the recent surface (1985) from the old surface of 1893 (type 1922) (Figure 9).
From the DEM, an automatic extraction is carried out according to a topographic section chosen to indicate the variation in altitude over the entire field and projected on the map.

N. Rebai, A. Moussi Journal of Geographic Information System
An examination of the residual map reveals very large topographic residual intervals between 1 and 45 m and between −7 and −35 m that appear in the hillside mountain (Figure 9). The residual map (Figure 9) shows an estimated absolute subsidence of 3 to 7 meters in the Enfidha plain.   Consequently, while considering a 100 m shift between the line contours of 1893 (type 1922) and those of 1985, the one can either descend or rise abruptly of almost 20 m of altitude as shown in the topographic profile (AB) (Figure 11).
Thus, considering a 100 m offset between the contour lines of 1893 (type 1922) and those of 1985, one can suddenly descend or rise by nearly 20 m in altitude as also indicated in the topographic profile (AB) (Figure 11).
In the plain, the margin of error risk is almost insignificant and a displacement of a few meters will not have a significant effect on the residue as indicated in the topographic profile (CD) (Figure 11). The altitude varies between 1 and 2 m for a 100 m displacement.
It is undeniable that taking into consideration the margin of error in georeferencing, digitization and modelling of the digital terrain model (interpolation) (Figure 11  calculated by the first derivatives [76] (slope, orientation) or secondary derivatives (e.g. curvatures), the obtained accuracy will be different. In order to better characterize the investigated morphostructures, we generate a grid DEM with a 10 m resolution. From these, we generated slope maps, relief maps, slope profiles and topographic profiles.
Subsequently, a digital contrast is applied to the produced slope map in order to highlight the slope gradients and their trend in the geologic terrain. The final products maps illustrated in figures (Figure 10, Figure 11) are a 2.5 dimension with a distribution of the slope and an elevation of the value.
Topographic and slope profiles.
The topographic and slope profiles are automatically generated from two DEM.
The idea is to extract altimetric variations or slope gradients along a topographic trace through the DEM. The comparison between the (AB) profile (1893) and the (AB) profile (1985) illustrates the subsidence of the Enfidha plain, which is about 2 to 3 meters in height ( Figure 11).
The topographic maps and profiles show two geographical compartments with a very clear morphological contrast. There are mountain ranges on the one side, and very cut-out plateaus and plains on the other (Figures 10-12).

The Effect of Lithology on the Behavior of the Hydrographic Network
In the [77] work, it is revealed that the hydrographic network configuration results from the simultaneous interactions of many and complex factors. If water is incompressible by nature and flows by the shortest path from the highest point (summit, ridge …) to the lowest point (sea, lake, river …), obstacles that modify its path. The causes are geologic, tectonic, climatic (paleoclimate), vegetation and human action. They have an important role in the evolution of the rivers flow paths [77]. It appears that the geomorphological context of a given area correlates with the morphology of hydrographic networks [78] [79], which makes it possible to understand the path of these networks and therefore can provide information in relation to the geomorphology of the associated areas.
The analysis of ( Figure 13) uncovers the fact that on the same lithological facies (Quaternary), hydrographic networks of different forms are established. For this reason, the causes of the hydrographic network layout modification are not of a pure geologic (lithological) order, but are also related to active tectonics.

Geographical Classifications of the Hydrographic Network
The geometry of the hydrographic network highlights two categories of faults.
• Evidence of E-W and NW-SE faults: The study area is marked by a main hydrographic network with a NNW-SSE direction. We also note that these rivers flow towards the endorheic zone of Enfidha. This proves a flow from NNW to the SSE. The traces of these rivers still show concavity bends to the West and convexity to the East (Figure 13).
In the entire study area, the elbow geometry is aligned in an E-W direction.
This argues in favor of the existence of an E-W direction fault with dextral strike-slip.
• Evidence of the E-W and N-S faults: The hydrographic network of the Sahel sebkhas watersheds follows the synclinal gutters and the East-West and North-South faults in a well-regulated manner ( Figure 13). Right-angled stream deviations occur at the interference corners of the two accident networks [28].

Field Validation
The obtained result from the analysis of residual DEM and hydrographic network approves the existence of subsidence in the Enfidha plain varying between 3 and 7 m (±50 cm). In the course of this analysis, we have carried out a field work in the Enfidha region to observe evidence of recent deformation. This is illustrated by the presence of recent relay tectonics, marked in particular by the morphology of the contour lines ( Figure 4). We have chosen two very interesting observation sites located between the relief and the quaternary deposit of the Enfidha plain ( Figure 14).
The Jgarneya region is located in the north, 6 km away from the town of Enfidha ( Figure 14).  In the first observation site, an analysis of the Mio-Pliocene outcrop formed by conglomerates, sandstones and clays have been analysed. The analysis detects the presence of two combined fault networks in both directions: NE-SW and NW-SE ( Figure 15 and Figure 16) as well as the set of a N120 fault showing the dextral offset with a normal set. This offset can be estimated at 25 cm. The subsidence of SE behavior in olive trees is controlled by another fault in the N45 (NE-SW) direction ( Figure 15). The fracture intensity clearly shows the distensive NE-SW and NW-SE direction regime ( Figure 16).
In the second observation site, tectonic analysis shows the presence of a global fault network (Figure 17), with a sinistral normal direction of N120 and N135. The global fault network tilt a movable deposit and are filled with stony alluvium. The outcrop of Lower Miocene Lumachelle Sandstone indicates the existence of a normal fault with a direction of N45 and another fault with a direction of N145 affecting the Roman Ruins ( Figure 17 and Figure 18).
The fracture intensity clearly shows the distensive NE-SW and NW-SE direction regime (Figure 18).
The previous work [30] indicates the existence of a subsurface fault that extends straight through the study area. On the other hand, following our observations and according to the tectonic index found, the existence of relay faults affect Miocene deposits and even Quaternary deposits, so the relative date of these faults is Miocene-Quaternary, marking an active tectonic. In the North of Enfidha, there is Henchir Jgarneya. It is the ancient Uppenna, where the ruins of an important fortress and a church are located, there is a mosaic bearing epitaphs of various bishops and martyrs is found. The Roman ruins are affected by dextral faults (N142-155), showing a lateral displacement of a few centimeters in the Hammam and the mosaics (Figure 19(d)). The calculated tensor for minor faults illustrates a strike-slip regime characterized by a σ1 = N017 and σ3 = N085 [80].

Conclusions
Our proposed methodology aims to quantify the variation in altitude by restoring two DEM from two topographic maps, produced at two different epochs, in 1893 and 1985. A hydrographic network, which is a complementary morphological study, is also used to reveal morphostructural responses considered as indicators of recent deformation in the case of the Enfidha region.     We conclude that the analysis of recent deformations by GIS is significant and promising. However, the results can be improved by setting up a GPS surveillance network that allows direct measurements of recent deformation.