TL;DR
This study investigates the potential of Cosmic-Ray Neutron Sensing (CRNS) for estimating soil water content (SWC) in both farmland and mountainous environments. The research demonstrates that CRNS is a viable technology for both, with improved accuracy in farmland when spatial weighting is applied, but highlights the challenges posed by complex terrain and vegetation in mountainous regions.
What Problem Does This Paper Address?
Accurately measuring soil water content (SWC) is crucial for various fields like agriculture, hydrology, and forestry. Traditional methods often fall into two extremes: point-scale measurements (like TDR or FDR) which are labor-intensive and provide limited spatial coverage, or remote sensing which offers broad coverage but can lack the necessary resolution or accuracy for certain applications. Cosmic-Ray Neutron Sensing (CRNS) has emerged as a promising "mesoscale" technology, offering non-contact, relatively large-area measurements that bridge this gap. However, the applicability and accuracy of CRNS in diverse ecological conditions, particularly in complex terrains like mountainous areas, require further investigation. This paper aims to address this by evaluating CRNS performance in both farmland and mountainous settings, and by exploring the influence of environmental factors on its accuracy.
Methods
The study employed a Cosmic-Ray Neutron Sensing Probe (CRNP) to measure neutron intensity, which is inversely correlated with hydrogen content (and thus soil water). To understand the spatial characteristics of CRNS measurements, the researchers used the Ultra-rapid adaptable neutron-only simulation (URANOS) model. URANOS was used to simulate the CRNS footprint (the area from which detected neutrons originate) and to derive spatial weighting functions. These functions were then used to calculate a weighted average of soil moisture from sampled points, which was compared to the CRNS measurements. The study also incorporated standard meteorological equipment to record atmospheric pressure, humidity, and precipitation.
Key methodological aspects include:
- URANOS Simulation: This Monte Carlo-based code was used to simulate neutron transport and derive the effective detection radius and depth of the CRNS probe under various conditions. This is an advancement over simpler linear models for footprint estimation.
- Spatial Weighting: The study applied simulated spatial weights to point-scale soil moisture measurements. This approach aims to better represent the CRNS footprint by giving more importance to samples closer to the probe or within the simulated footprint, improving the correlation with CRNS data.
- Calibration and Correction: CRNP counts were corrected for atmospheric pressure and air humidity, standard practices in CRNS.
- Accuracy Assessment: Measurement accuracy was quantified using the coefficient of determination ($R^2$) and Root Mean Square Error (RMSE) by comparing CRNS-derived SWC with in-situ measurements.
Datasets
The study utilized data collected from two experimental sites:
- Fengqiu Station: Located in a typical warm-temperate, semihumid monsoon climate farmland in northern China. This site featured flat topography and was cultivated with winter wheat and summer maize.
- Hemuqiao Station: Located in a humid hilly area of southeast China, characterized by steep headwater hillslopes and covered by moso bamboo.
Soil moisture data was collected using both the CRNS probe and soil-mounted FDR (Frequency Domain Reflectometry) auto soil moisture monitors. Additionally, soil samples were collected at various radial distances and depths from the CRNP for calibration and validation.
Key Insights and Ideas
- CRNS is a viable tool for both farmland and mountainous areas: The study confirms that CRNS can effectively monitor soil water content in both contrasting environments.
- Spatial weighting significantly improves CRNS accuracy in farmland: Applying simulated spatial weights to point measurements enhanced the correlation between CRNS and ground truth in the flat farmland environment, suggesting that this method better captures the CRNS footprint in homogeneous terrain.
- Mountainous terrain complicates CRNS footprint and accuracy: The complex topography of the mountainous site led to a deviation between the theoretical and actual neutron footprint, limiting the benefit of spatial weighting and reducing overall estimation accuracy compared to farmland.
- Farmland retains more water despite less precipitation: The study observed that the farmland site (Fengqiu) retained more soil water, even with lower precipitation, likely due to its flat topography and soil characteristics that facilitate water storage.
- Non-soil hydrogen sources can influence CRNS readings: The paper notes that CRNS is sensitive to other hydrogen sources besides soil water (e.g., vegetation), which can lead to SWC estimates being slightly higher than those from soil-mounted detectors.
Key Results
- Farmland (Fengqiu):
- With spatial weighting, $R^2$ improved to 0.645 and RMSE was 0.046 cm³·cm⁻³.
- The weighting calculation was crucial for improving accuracy in this flat terrain.
- Mountainous Area (Hemuqiao):
- $R^2$ was 0.773 and RMSE was 0.049 cm³·cm⁻³.
- Spatial weighting did not significantly improve accuracy in this complex terrain.
- Footprint Differences: The detection radius in farmland was generally smaller (127-139 m) than in mountainous areas (125-153 m) under different humidity conditions. Detection depth was also shallower in farmland (16.9-31.2 cm) compared to mountainous areas (19.1-45.9 cm).
- Diurnal Variation: CRNS captured daily SWC variations well in farmland, but struggled to represent diurnal changes in the mountainous site, showing a low determination coefficient (0.059) with FDR data.
Limitations and Future Work
- Influence of Non-Soil Hydrogen Sources: The study acknowledges that CRNS is sensitive to hydrogen from sources other than soil water (e.g., vegetation, atmospheric water vapor). Future work should focus on correcting for these non-soil hydrogen sources, especially in vegetated areas.
- Vegetation Effects in Mountainous Areas: The dense moso bamboo canopy in the mountainous site significantly impacted CRNS measurements. The authors recommend further studies on calibrating CRNS for forest vegetation to improve accuracy.
- Topographical Obstructions: The complex terrain in mountainous areas obstructs neutron transmission, leading to a narrowed actual footprint and reduced precision. Future research should explore methods to account for these topographical effects.
- Sampling Frequency: While the study assumed daily or 12-hour sampling was sufficient, it notes that increasing sampling frequency could improve statistical stability and potentially reduce errors, especially in dynamic environments.
- Further Investigation of Mountainous Terrain: The paper concludes that while CRNS is suitable for mountainous areas, more attention needs to be paid to the specific effects of topography and vegetation for more accurate estimations.
Summary updated