D-InSAR Analysis of Sentinel-1 Data for Landslide Detection in Northern Morocco, Case Study: Chefchaouen

Land surface displacement caused by landslides is among the most damaging phenomena in northern Morocco. In this paper, we measure ground deformation in the Chefchaouen area which is a zone characterized by geological formations and structurally complex losses that promote instability (landslide, mudflow, block falls, etc.) leading to slow to extremely slow deformation phenomena, which require an interferometric study, using the DinSAR (differential interferometric synthetic Aperture Radar) technique with sentinel 1 images in bance C, which is a powerful tool for the detection and analysis of interferences and monitoring of ground deformations. We worked on four areas of the study area. Its points are provided by the direction of the roads, which generates Interferograms and then deformation maps with precision in mm/year.


Introduction
Landslides are among the geological hazards that are responsible for enormous human and natural losses (Lorenzo et al., 2020). The mountainous regions of Northern Morocco are characterized by high winter precipitation, responsible for a significant evolution of the slopes with the development of gravitational movements (GM), such as landslides, debris flows, collapses and intense gullies It has been widely exploited during the last three decades, allowing significant results in several fields, For instance, successful land deformation estimates have demonstrated the potential of this technique for a wide range of applications related to landslides (Colesanti et al., 2003;Hilley et al., 2004;Bovenga et al., 2012), land subsidence caused by ground wa-ter pumping (Tomás et al., 2005;Stramondo et al., 2007;Bell et al., 2008;Heleno et al., 2011) and mining (Colesanti et al.,2005;Jung et al., 2007), or urban monitoring Vallone et al., 2008;Cigna & Tapete, 2012). The aim of this document is to map the landslide points (PK) in Stehat accurately, Talambote, Derdara, Eljebha with accuracies in mm/year obtained by the Road Directorate of the Chefchaouen region and to establish deformation maps. The objective of this document is to validate the application of the DInSAR processing chain to the monitoring of the slip point (PK) in Stehat, Talambote, Derdara, Eljebha with accuracies in mm/year obtained by the Road Directorate of the Chefchaouen region, using open access data and free tools of the Copernicus program. In particular, a DInSAR processing chain has been proposed and deformation maps have been elaborated and will be discussed in detail.

Study Area
According to the historical table of landslides (Table 1), provided by the Road Directorate of Chefchaouen and the laboratory of Studies and Tests, the area that was started in this study is the province of Chefchaouen precisely Eljebha, Derdara, Bab Berred, Stehat and Talambote, These are points limiting the province of Chefchaouen (Figure 1(a)) which is a region located in the centre of the rifaine chain limited in the North by the Mediterranean Sea over a length of 120 km, in the South by the Provinces of Taounate and Sidi Kacem, the East by the Province of Al Hoceima and in the West by the Provinces of Tetouan and Larache with 600 m (Figure 1(b)) of altitude, it is characterized by a great diversity of the soil formed by limestone and siliceous layers.

Geological Setting of the Study Area
The Rifaine zone is formed by a geological structuring consisting of quartzite sandstones and dorsal limestones as well as flysch layers. This context is associated with a climate that is both Mediterranean and oceanic, as well as with the active tectonics of over-rection of the Rifaine domain ( Figure 2). Journal of Geoscience and Environment Protection

Landslide History of Chefchaouen Province
According to the "Roads Directorate, Provincial Equipment, Transportation and  The SAR data used in this study are acquired by the Sentinel1-A sensor and are downloaded free of charge from the Copernicus program. The choice of these data is made on several criteria (satellite mission, type of product, acquisition mode, orbit and polarization) as shown in the following table (Table 2).
We have downloaded the Sentinel-1 images from the Copernicus Open Access Hub in Single Aspect Complex (SLC) level 1 mode, with VV or VV-VH polarizations depending on availability at the various test sites, and Wide Interferometric (IW) acquisition mode. SLC products are based on focused SAR data, geo-referenced using orbit and attitude data from the satellite, and provided in oblique range geometry.
The pairs of Sentinel 1 images were chosen according to the landslides (Table   1) Images before and after the event.

Methodology
Synthetic Aperture Radar (SAR) is a system capable of obtaining complex, high-resolution images from large areas of terrain, usually located on board an orbital or aerial platform, but which can also be used in ground-based deployments (Yerro et al., 2014). D-InSAR, which uses the phase information of 2 SAR images at different times in the same area by satellite-loaded radar sensors and the DEM of this area to monitor surface deformation, is Differential Interferometry Synthetic Aperture Radar monitoring technology (Raucoules et al., 2007;Berardino et al., 2002).
It provides an image, called a differential interferogram, representing the ground motion between acquisitions with centimetric precision and decametric resolution.
It consists of exploiting the phase information contained in complex SAR imagery recorded by one or more sensors by varying an acquisition parameter (date, position and orientation of the radar, frequency, polarity of the transmitted/received wave) according to the desired application (Bamler & Hartl, 1998).
Is a reliable method for measuring small movements (1 cm) with high spatial resolution (10 m) over large areas (50 km 2 ) The interferometric phase between two SAR images includes the following contributions: In our study we collected the previous landslide points of the study area, we had a total of seven points since 2012. However, since the present study will be done with Sentinel 1 Radar imagery, we took the points since 2015 which makes the total of points to about 6 Sentinel 1 Radar images (Table 3), in our study we used Differential Radar Interferometry to generate differential interferograms as shown in (Figure 4) these interferograms were filtered using the GOLDSTEIN fitting and using the SNAP toolbox software, the phase unpacking was performed with the SNAPHU software, then the displacement was calculated from the unwrapped interferogram, finally the results were georeferenced using the Doppler terrain correction.

Coherence Image
After Co-registration the SLC images, coherence map is computed due to the complex correlation coefficient between two acquisitions. It gives the level similarity of pixels between the slave and master images in a scale from 0 to 1.

5(a)) the values close to 1 presented in white indicate a strong coherence that
means that there is no phase noise while the values close to 0 presented in black indicate a weak coherence that means that there is phase noise, The high elevation of the zone is focused on the eastern part of the zone, Eljebha ( Figure 5(b)), while that of the Bab berred zone (Figure 5(b)), is focused on the southern part of the zone.  (Figure 7(a)), low to medium in the Talambote zone, and high to very high in the Derdara zone (Figure 7(b)).

Interferogram Image
With regard to the generation of Interferograms, three components of three peers the interferometric phase carries a lot of information about the surface deformation (force and direction of movement) and the location of the surface break. Concerning the mapping of landslides from Sentinel-1 data, it has already been reported that the approach is based on the estimation of vertical displacement. The interferometric fringes represent a complete cycle of 2π. The fringes appear on an interferogram as arbitrary color cycles, each cycle representing half the wavelength of the sensor. The relative ground motion between two points can be calculated by counting the fringes and multiplying them by half the wavelength of the sensor (Meziane et al., 2018).
( Figure 8(a)) shows a differential interferogram generated from Sentinel-1 data, from the 8 June-26 July 2015 couple in Descending orbit, after subtracting all components, namely the atmospheric and residual components and the topography related component, thus the fringes presented show that the ground deformation extends up to 2.75 over the areas of interest (Figure 8(b)). Figure   9(a) shows the descending interferogram of the Stehat and Talambote zone with a time baseline of 51 days it is represented by a color cycle that shows a phase difference of −2.88 to 2.88 (Figure 9(b)) for the derdara talamobte zone, as shown by the interferogram in (Figure 10(a)), with a time baseline of 25 days which shows the coherence of the zone mean in this period from 29 July to 24 August (Figure 10(b)).

Discussion
The evaluation of the surface deformation caused by the landslide on the Chefchaouen area is carried out through the series of wrapped phase obtained in the fringes which are converted into an unwrapped phase thanks to the unwrapping which is carried out in the platform (SNAPHU), then the displacement maps to were calculated.
The deformation values vary from 0.04 to 0.1 m in the Bab Berred, Eljebha zone over the period from 8 June, 26 July 2015 (Figures 11(a)

Conclusion
The landslide-related ground displacement is analysed in this study by the D-InSAR technique, which has shown its ability to map landslide areas and to validate the landslide points obtained from the direction of roads in mountainous Figure 14. Displacement of the points of interest Eljebha, Bab Berred, Derdara, Talambote according to the years. Journal of Geoscience and Environment Protection areas with an accuracy of the order of mm/year, using the Sentinel-1 SNAP and SNAPHU toolbox for the unwrapping phase. The application of DInSAR in Northern Morocco is effective, because it allows to freely analyse and map the ground deformation in a remote area (Meziane et al., 2018). This process is useful and gives satisfactory results even if it is based on two SLC images to finally obtain a deformation map visualized in Google Earth; this result may have a positive influence on the adaptation of the D-InSAR method as an approach for monitoring landslides on roads.
In the future works, the PSInSAR technique will be used, and the processing chain of the chronological series of deformations of the same zone will be provided.