Angelica Jaconi

On Dec. 16, 2019, Angélica Jaconi defended and completed her doctoral thesis on "Determination of soil properties using NIRS and chemometrics analysis on a regional data set" at the Humboldt University of Berlin. She wrote her dissertation within the framework of the German Agricultural Soil Inventory. The aim of the thesis was to explore the analytical possibilities of near-infrared spectroscopy on the scale of Germany and to develop new calibrations for different soil parameters.

Angélica Jaconi tested new algorithms in this methodologically oriented thesis, including memory-based learning, a machine learning technique. Using the data set of near-infrared spectra from the Agricultural Soil Inventory, she was able to show that the algorithms could hardly cope with the wide range of soil carbon contents in German soils and that stratification resulted in much better results. With improved algorithms, near-infrared spectroscopy can be successfully used as a rapid screening method for soils if the analytical result does not need to be as accurate. In particular, Angélica Jaconi developed calibrations that replace costly laboratory analyses, e.g., for fractionation of soil organic matter, and thus can be used for larger data sets. Through her work, particulate organic matter was predicted for nearly 3000 soils of the national soil inventory and unexpected regional differences emerged.

The dissertation has significantly advanced the development of calibrations for near-infrared spectra, reducing calibration errors and thus making them useful for routine laboratory applications.