Research axis 1 (R1) : social and biophysical organizations
Actions in detail
Action I: Aerial photogrammetry
Mohamed MAAOUI , Denis FEURER (IRD/LISAH)
The use of the SfM (structure from motion) technique in aerial photogrammetry has evolved considerably in recent years in several areas. SfM algorithms produce a three-dimensional model of the identifiable objects on the input images. A crucial aspect during this photogrammetry process is to assess the quality of the DSM produced. To do this, the necessary number of ground control points as well as the doming effect (James and Robson, 2014) are the two main problems to be solved.
The main idea of this work is to test a new approach which consists of adding images taken at an altitude twice as high as that of the main site. On an 8-hectare site located on the fortuna watershed in Cap Bon, 41 images were taken at an altitude of 90 m and 27 images at an altitude of 190 m. Three scenarios were tested. L5 uses 5 GCPs and images taken at an altitude of 90 m. L10 equals L5 plus 5 other GCPs. LH5 equals L5 plus images taken at 190 m. the results obtained, by analyzing the altitude estimation errors on 63 validation points, show that L5 is the most affected by doming and that it has an RMSE of 0.19 m. L10 is less affected by doming with an RMSE of 0.10 m. While LH5 has the best results, no doming with an RMSE of 0.06 m. In conclusion, we can say that the use of the multi-altitude strategy during photogrammetry processes reduces the number of ground control points necessary for the calculation and also makes it possible to avoid the effect of doming.
Action II: Geophysical mapping on the agricultural plot in the Public Irrigated Perimeter of Merguellil
M2 internship: “Mapping and tomography of electrical resistivity and spontaneous potential in the study of the spatial variability of soil properties in an agricultural plot of the Irrigated Perimeter of Merguellil, Kairouan”
Wafa CHARNI, (Soil and Environmental Sciences)
Co-supervisors : Hajer AZAIEZ, (CERTE), Gaghik HOVHANNISSIAN (UMR 242 iEES Paris)
M2 internship : “Application of electromagnetic mapping in the frequency domain in the study of the spatial variability of soil properties and their water functioning in a plot irrigated with saline water from the Irrigated Perimeter of Merguellil (El Haouareb, Region of Kairouan)”. Wael JOUINI, (Soil and Environmental Sciences)
Co-supervision: Mohamed HACHICHA, (INRGREF), Gaghik HOVHANNISSIAN, (UMR 242 iEES Paris)
Hydro geophysical prospecting (1D SEV electrical and 1D TDEM electromagnetic surveys) in the El Haouareb dam area
Objective: “Characterization of surface formations by different geophysical methods for understanding the hydrodynamic functioning of the downstream part of the El Haouareb dam”.
A one-week field campaign was carried out at the end of June – beginning of July 2018. About 20 electromagnetic and electrical (SEV-1D) soundings (TDEM-1D) were carried out. Another campaign is planned for early December 2018.
Action III: Scénarios de variables hydro-météorologiques
M2 internship : Nesrine FARHANI,
Co-supervisors: Julie CARREAU, (IRD/HSM) Gilles BOULET (IRD/CESBIO), Rim ZITOUNA-CHEBBI (INRGREF), Zeineb KASSOUK (INAT)
| Goals |
| Create climate forcing scenarios in present and past climate, i.e. value fields on grids that reproduce the spatio-temporal variability and the intensities of hydro-meteorological variables. The main issues identified are the strong spatio-temporal variability of these variables on the southern shore of the Mediterranean, the “no-Gaussianity” of the distributions, i.e. the fact that the Normal law is not appropriate in several cases and the relatively poorly instrumented context sites studied. The first step is to set up a virtual station, i.e. a station that takes advantage of the information contained in three geographically and meteorologically very close stations in order to fill in the gaps. |
| Approach/methodology |
| A stochastic approach based on generalized linear regression models makes it possible to model the conditional distribution of a hydro-meteorological variable at a given site by taking into account different effects (seasonal and diurnal cycles, geographical, persistence and inter-variable dependence). Four distribution families are considered: Normal, Heteroscedastic Normal (Pa, Tair, HuR, U, V), Gamma (Rg, Precip) and Bernoulli (occurrence of Precipitation). Once the regression models are fitted to the data, they allow simulation in conditional mode, i.e. by preserving the observed values and generating values according to the conditional distribution where there are gaps |
| Principle results |
| Adaptation and validation of a stochastic approach that models the conditional distribution of hydro-meteorological variables. Different effects are accounted for by introducing covariates into generalized linear regression models. This approach makes it possible to fill in the gaps (imputation in the statistical sense) by exploiting the inter-variable and inter-site links and by respecting the spatio-temporal variability and the distribution of the intensities of the variables. |
Action IV : Low-cost spectroradiometer for proxy-detection.
B. Mougenot (IRD CESBIO), P. Fanise, (IRD CESBIO). V Dantec, (IRD CESBIO), S. Gascoin,(IRD CESBIO) Z. Kassouk, (INAT), G. Nougaret, (IRD CESBIO)G. Boulet, (IRD CESBIO)
| Goals |
| Realization of a low-cost device, making it possible to perform spectral measurements at an infra-hour time step in the near solar domain, for process studies by proxy-detection (spectral measurements at short distance). |
| Approach/methodology |
| The spectral signature reflects the composition of the surface states: vegetation, soil, water, snow, etc. and allows early monitoring of the effect of processes such as water stress on vegetation. It also involves validating or simulating in the field measurements made by operational or future satellites. |
| Principle results |
| The micro-spectrometer developed here was tested on a melon field in the Merguellil plain in comparison with radiometers simulating the bands of the “Sentinel-2” satellite. The first results show a consistency of the measurements, the expected sensitivity to temperature, the inter calibration of the incident and reflected sensors must be refined |
Action V: Identification of determinants of infiltrability by satellite data
Project involving: UMR LISAH (Montpellier), UMR TETIS (Montpellier), DG/ACTA Sol (Tunis)
Work supported by the TOSCA A-MUSE project, 2018-2019. “Multi-temporal analysis of SENTINEL 2 and 1 data for the monitoring of observable characteristics of the ground surface, related to infiltrability”.
Soil infiltrability is the key property of continental surfaces that controls a wide variety of ecosystem services on water risks and resources, including regulating services (e.g., aquifer recharge, useful soil reserve recharge, runoff attenuation and floods). And the infiltrability of cultivated Mediterranean or semi-arid soils can be explained and reasonably predicted from observable characteristics of the soil surface and their modalities: structural characteristics (“roughness” or size of clods, presence/absence of crusts) and soil cover characteristics (mineral elements -pebbles-; or organic elements -litter and vegetation cover-).
The traditional methods of characterization of soil infiltrability (e.g., measurement by double rings) and characterization of the soil surface in connection with infiltrability (e.g., measurement of roughness by rugosimeter) are difficult to envisage at the parcel scale. (relevant scale for users such as basin managers or agro environmental design offices) and are often cumbersome and expensive to set up to be maintained throughout a crop cycle. The availability of Sentinel-2 multi-spectral data with high temporal repeatability offers a promising and complementary field of data for mapping these observable first-order features for soil infiltrability. This work aims to analyze the potential of Sentinel-2 optical data for the spatial monitoring of observable characteristics of the soil surface, in connection with the infiltrability of agricultural soils.