Root zone soil moisture estimation with Random Forest

Coleen Carranza,Corjan Nolet,Michiel Pezij,Martine Van Der Ploeg

doi:10.1016/j.jhydrol.2020.125840

Coleen Carranza, Corjan Nolet + Show 2 more

Open Access

https://doi.org/10.1016/j.jhydrol.2020.125840

Copy DOI

Abstract

Accurate estimates of root zone soil moisture (RZSM) at relevant spatio-temporal scales are essential for many agricultural and hydrological applications. Applications of machine learning (ML) techniques to estimate root zone soil moisture are limited compared to commonly used process-based models based on flow and transport equations in the vadose zone. However, data-driven ML techniques present unique opportunities to develop quantitative models without having assumptions on the processes operating within the system being investigated. In this study, the Random Forest (RF) ensemble learning algorithm, is tested to demonstrate the capabilities and advantages of ML for RZSM estimation. Interpolation and extrapolation of RZSM on a daily timescale was carried out using RF over a small agricultural catchment from 2016 to 2018 using in situ measurements. Results show that RF predictions have slightly higher accuracy for interpolation and similar accuracy for extrapolation in comparison with RZSM simulated from a process-based model combined with data assimilation. RF predictions for extreme wet and dry conditions were, however, less accurate. This was inferred to be due to infrequent sampling of such conditions that led to poor learning in the trained RF model and to incomplete representation of relevant subsurface processes at the study sites in the RF covariates. Since RF does not depend on parameters required to estimate subsurface water flow, it is more advantageous than a process-based model in data-poor regions where soil hydraulic parameters are incomplete or missing, especially when the primary goal is only the estimation of soil moisture states.

Highlights

Root zone soil moisture (RZSM) is an important environmental variable that impacts hydrological processes relevant for agriculture and climate-related studies
The Random Forest (RF) models generated using different training sets indicate that the highest and lowest root mean square error (RMSE) are based on 50% and 80% of the total data for interpolation and 80% and 60% of the total data for extrapo lation (Fig. 6a and b)
The runtime is fastest with a 50% training set, decreasing the computation time of the best performing model by at least 44%

Summary

Introduction

Root zone soil moisture (RZSM) is an important environmental variable that impacts hydrological processes relevant for agriculture and climate-related studies. It is one of the main drivers for agricultural productivity (Rigden et al, 2020) and serves as an indicator for crop water stress, which is valuable for drought monitoring (Bolten et al, 2009). Direct RZSM measurements can be obtained from in situ sensors installed along the soil profile or at specific depths (Vereecken et al, 2008; Dobriyal et al, 2012). It has become relatively common to extract RZSM from surface soil moisture (SSM), which may be in situ or satellite-derived (Ulaby et al, 1996), since they are more obtained. Satellitederived SSM has the advantage of providing spatially distributed soil moisture while in situ measurements offer higher temporal frequency (second or minutes) compared to satellites, which only provide snap shots at regular time intervals (days or weeks)

Methods

Results

Conclusion