Error Estimation for Soil Moisture Measurements With Cosmic Ray Neutron Sensing and Implications for Rover Surveys

Jannis Jakobi,Martin Schrön,Harry Vereecken,Johan A Huisman,Justus Fiedler,Heye R Bogena,Cosimo Brogi

doi:10.3389/frwa.2020.00010

Jannis Jakobi, Martin Schrön + Show 5 more

Open Access

https://doi.org/10.3389/frwa.2020.00010

Copy DOI

Abstract

Cosmic ray neutron (CRN) sensing allows for non-invasive soil moisture measurements at the field scale and relies on the inverse correlation between aboveground measured epithermal neutron intensity (1 eV−100 keV) and environmental water content. The measurement uncertainty follows Poisson statistics and thus increases with decreasing neutron intensity, which corresponds to increasing soil moisture. In order to reduce measurement uncertainty, the neutron count rate is usually aggregated over 12 or 24 h time windows for stationary CRN probes. To obtain accurate soil moisture estimates with mobile CRN rover applications, the aggregation of neutron measurements is also necessary and should consider soil wetness and driving speed. To date, the optimization of spatial aggregation of mobile CRN observations in order to balance measurement accuracy and spatial resolution of soil moisture patterns has not been investigated in detail. In this work, we present and apply an easy-to-use method based on Gaussian error propagation theory for uncertainty quantification of soil moisture measurements obtained with CRN sensing. We used a 3rd order Taylor expansion for estimating the soil moisture uncertainty from uncertainty in neutron counts and compared the results to a Monte Carlo approach with excellent agreement. Furthermore, we applied our method with selected aggregation times to investigate how CRN rover survey design affects soil moisture estimation uncertainty. We anticipate that the new approach can be used to improve the strategic planning and evaluation of CRN rover surveys based on uncertainty requirements.

Highlights

Soil moisture is an essential variable of the terrestrial system as it governs the transfer of both water and energy between the land surface and the atmosphere (Vereecken et al, 2015)
The present study investigates the statistical uncertainty of cosmic ray neutron sensing (CRNS) soil moisture estimates, which depends on the detector configuration, i.e., the number of counts in a given period of time
We found that the analytical expressions for measurement uncertainty underestimated the standard deviation for high soil moisture content (> ∼0.3 m3/m3) when the 1st and 2nd order Taylor expansions were used

Summary

Introduction

Soil moisture is an essential variable of the terrestrial system as it governs the transfer of both water and energy between the land surface and the atmosphere (Vereecken et al, 2015). Recent advances in non-invasive monitoring techniques enable continuous and contactless measurements of soil moisture dynamics at the field scale (Bogena et al, 2015). Stationary CRNS probes are used to obtain continuous information on field scale soil moisture dynamics (Zreda et al, 2012; Andreasen et al, 2017; Schrön et al, 2018a). Mobile applications of CRNS probes (i.e., CRN roving) have been introduced, which enable to measure spatial soil moisture variability at the larger catchment scale (Chrisman and Zreda, 2013; Dong et al, 2014; Franz et al, 2015; Avery et al, 2016; McJannet et al, 2017; Schrön et al, 2018b)

Objectives

Methods

Results

Conclusion