Calibration and Inversion of Large-scale Electromagnetic Induction Measurements of Agricultural Fields and Comparison with Leaf Area Index and Sun-induced Fluorescence Measurements

Recently, paleochannel structures were detected using electromagnetic induction (EMI) measurements and satellite based leaf area index (LAI) patterns on agricultural fields near Selhausen (Germany). This paleochannel system influences the amount of water that is available for crops. For two fields, the LAI patterns were correlated to the EMI measurements. In this thesis, additional fields were investigated with EMI for increased depths and high-resolution airborne sun-induced fluorescence (SIF) and LAI measurements were performed to estimate plant activity and available leaf area, respectively. The crop patterns present in the SIF and LAI were compared with the EMI apparent electrical conductivity (ECa) patterns. The EMI measurements were performed with two multi-receiver sensors, allowing measurements over nine depths of investigations (DOI). Three fields were measured, one field (F12), located in the Lower Terrace, showed relatively high ECa values. The second field (F13), located in the Upper Terrace, showed generally low ECa values. The third field (F10) was for two-third located in the Lower Terrace and one-third in the Upper Terrace and showed an abrupt change in ECa values. The filtered ECa values were interpolated to a 1-by-1 m grid. The different measurement directions were still visible after interpolation of F13, indicating that different ECa values were measured in perpendicular directions. The differences could only be observed for the northern part of the field, that was recently ploughed by the farmer. Further investigations showed that the ECa values were higher in the ploughing direction, with differences up to 2 mS/m. The SIF and LAI were measured using the HyPlant imaging spectrometer. The high spectral resolution of HyPlant allows the retrieval of fluorescence using the Fraunhofer Line Depth Principle. A simple ratio of the fluorescence, retrieved at 680 and 760 nm, clearly showed the within field variability and was compared with ECa values using a linear regression analysis. High SIF values were observed above the Lower Terrace, whereas lower values were observed above the Upper Terrace. Small-scale patterns of higher SIF values in the Upper Terrace originate from the paleochannel system. Regression analysis of the interpolated ECa maps of F13 resulted in coefficients of determination (R2) ranging from 0.32 to 0.64 for SIF and 0.25 to 0.55 for LAI. Both regressions showed increasing correlation with larger DOI, indicating that the deeper part of the soil plays an important role in the crop performance. To obtain quantitative ECa values and being able to perform a large-scale EMI inversion, a calibration was performed based on predicted ECa values from ERT and VES measurements and inversions. A regression analysis between measured and predicted ECa values resulted in good, but different regressions. The corresponding R2 values were not decisive for finding the optimal calibration parameters. The ERT-based calibration resulted in high scale factors for the CMD-SpecialEdition. The resulting calibrated ECa maps contained negative values for the low conductive part of the field. Comparison of a two, three, and four-layer VES inversion, showed that a three-layer VES inversion could be used to calibrate the ECa values. The calibrated ECa maps are inverted for a multi-layer earth using a parallelized version of the shuffled complex evolution (SCE) approach. The inverted EMI data for a three-layer model contained a negative middle layer, indicating that the used calibration was not successful, probably due to shallow electrical conductivity changes that were not captured by the direct current measurements. The calibration procedure needs to be re-evaluated and different direct current measurements and setups should be considered.Civil Engineering and GeosciencesGeoscience & EngineeringApplied Earth Scienc