Regionalization of Daily Soil Moisture Dynamics Using Wavelet-Based Multiscale Entropy and Principal Component...

Area of Science:

• Hydrology
• Environmental Science
• Data Science

Background:

• Hydrological regionalization is crucial for modeling and prediction but hindered by data scarcity and complex variability.
• Understanding multiscale soil moisture variability is key to improving regionalization techniques.

Purpose of the Study:

• To examine multiscale soil moisture variability using simulated data.
• To cluster grid-cell soil moisture data for effective hydrological regionalization.
• To discuss the implications for hydrological applications in ungauged basins.

Main Methods:

• Utilized simulated soil moisture data from a large-scale hydrological model for the Xijiang River basin (2002-2010).
• Applied wavelet-based multiscale entropy and principal component analysis for data clustering.
• Regionalized 169 grid cells (0.5° × 0.5° resolution) into homogeneous groups.

Main Results:

• Effective regionalization was achieved by clustering based on wavelet-based entropy patterns.
• Four distinct modes explained 80.14% of the total embedded variability across different timescales.
• Identified homogeneous groups within the Xijiang River basin soil moisture data.

Conclusions:

• The developed regionalization method effectively groups hydrological data based on multiscale variability.
• Results offer practical implications for parameter estimation in ungauged catchments.
• Supports the design of uniform prediction strategies for large-scale basins.