Adverse conditions for cosmic-ray neutron sensing: high water content low bulk density – can we still infer soil moisture over the full moisture range?

Near-surface soil moisture variation is an important variable in peatlands, controlling chemical processes and peat development or degradation. Cosmic-ray neutron sensing (CRNS) provides an area average soil moisture over a support volume of > 150 m radius and down to 50 cm depth by relating the abundance of secondary fast neutrons above ground to soil moisture. However, standard calibration and weighting functions for CRNS were developed and tested for mineral soils with dry bulk densities above 1 g cm^-³ and only up to 55 % of volumetric soil moisture. Peat soils, in contrast, are characterized by high organic matter content, low bulk densities, and high soil moisture when saturated. This makes peatlands a challenging environment for any soil moisture monitoring, including CRNS. In such adverse conditions, questions remain on the appropriate CRNS calibration approach and therefore the accurate determination of soil moisture.

This study presents lessons learned from operating a CRNS at a fen site with extensively used grassland in Northeast Germany (nature conservation area “Kremmener Luch”) for 3.5 years. The CRNS was complemented with point-scale soil moisture sensor profiles down to 1 m (FDR and TDR) in several locations of its footprint as well as groundwater level observations to identify periods of ponding that occur frequently at the site. Measuring soil moisture with the dielectric point-scale sensors showed challenges on its own. We increased the precision of point-scale data by a local soil specific calibration relating sensor permittivity to soil moisture. However, strong jumps and unreliable values remained, presumably due to swelling and shrinking of the organic-rich soil and loss of contact with the sensor. FDR and TDR time series showed large differences in absolute values as well as spatially different soil moisture regimes due do effects of microtopography. This is opposed to the CRNS, which senses average water content independent of small-scale heterogeneities. To derive a CRNS soil moisture time series we tested calibrating the CRNS using data from dedicated soil moisture sampling campaigns or the point-scale time series. We obtained unrealistically high CRNS-soil moisture regardless of which calibration function we chose – the standard “Desilets’ equation” or the recently proposed advanced “Universal Transport Solution”. Following the suggestion in previous CRNS studies conducted at peaty sites, we adjusted the parameters of the Desilets’ equation, which lead to a more realistic soil moisture range. However, the estimation of the CRNS integration depth with the standard procedures is very sensitive to the low bulk density of the organic soil and remains largely uncertain. This data set serves as a valuable testbed for extending the validity of existing calibration and weighting functions, and we will utilize neutron simulations to enhance our understanding of the vertical footprint of CRNS under conditions of low bulk density and high soil moisture.

Improved understanding and precision of CRNS soil moisture in peatlands can support peatland restoration efforts by providing insights into near-surface soil moisture variations allowing the evaluation of water level management success.