Historically, soil water retention curve (SWRC) measurements have focused on the wet end of the curve (suctions below pF 4.2 or 1,500 kPa). The dry end of the SWRC (suctions above pF 4.2), which is highly important for modeling hydraulic conductivity, plant survival, and soil-atmosphere gas exchange, is often missing in large datasets due to the historical challenges of collecting this data. This lack of dry-end measurements makes it difficult to properly calibrate SWRC models and develop comprehensive pedotransfer functions that accurately predict the entire range of soil moisture.

The wet part of the SWRC is controlled by both soil texture and structure, while the dry part is mainly governed by texture and specific surface area. Knowing this, we hypothesize that the dry end of the SWRC can be accurately estimated using the wet-end measurements together with readily available basic soil properties.

To test this hypothesis, the student will work with a large Danish soil dataset and develop predictive models using statistical and machine learning methods, such as linear regression or simple neural networks. The ultimate goal of the project is to develop a framework for balancing large SWRC datasets, which will enhance various aspects of vadose zone hydrology that depend on accurate hydraulic property measurements.