Lesson 3

Spatially advised Machine Learning models to fill spatiotemporal data gaps in hydrological monitoring networks

Description: The Lecture series provides a comprehensive introduction to spatiotemporal data, machine learning (ML), and interactive visualization techniques in R, with a focus on groundwater level monitoring. The first session introduces participants to challenges in groundwater data, handling and analyzing large spatiotemporal datasets, covering data exploration, and creating static and interactive visualizations. The second session delves into ML techniques, specifically random forest models, for filling gaps in spatiotemporal hydrological data, using a case study on groundwater depth interpolation in Brandenburg. By the end of the series, participants will gain practical skills in spatiotemporal data handling, ML model development, and advanced visualization, equipping them to address complex hydrological challenges using cutting-edge tools and methodologies.

• Understand why groundwater monitoring data are essential for assessment and management.
• Recognize common structural limitations of groundwater monitoring networks (spatial coverage, representativeness, and bias).
• Identify typical causes and consequences of temporal discontinuity and data gaps in groundwater records.
• Explain the major challenges in achieving complete and accurate groundwater assessments.
• Differentiate between physically based, conceptual, geostatistical, and machine learning models used in groundwater modeling.
• Identify the sources of uncertainty in hydrological models and understand the concept of equifinality.
• Discuss the “scale problem” and its implications for model design and performance.
• Evaluate the limitations of non-spatial machine learning models and describe how spatially explicit methods can improve predictions.
• Relevance of spatial temporal data in spatial predictions

Contributors:

Dr. Ahsan Raza (Contact: )
Prof. Dr. Claas Nendel
Dr. Gohar Ghazaryan