Publication Date
5-2024
Date of Final Oral Examination (Defense)
12-6-2023
Type of Culminating Activity
Thesis
Degree Title
Master of Science in Hydrologic Sciences
Department Filter
Geosciences
Department
Geosciences
Supervisory Committee Chair
Jennifer L. Pierce, Ph.D.
Supervisory Committee Member
David P. Huber, Ph.D.
Supervisory Committee Member
Linda Reynard, Ph.D.
Abstract
The importance of dust deposition in the biogeochemical cycling of nutrients, sediment redistribution, and soil formation is well-known. Anthropogenic activities such as changing land use and agricultural practices can influence dust flux deposition rates, dust organic matter content, and the chemical composition of dust. This study examines the influence of agricultural activities on aeolian deposition in a the xeric intermontane westerns USA. We investigate seasonal and spatial variations in dust deposition rates, dust organic matter, dust inorganic matter, and dust geochemistry from passive dust traps. The Reynolds Creek Experimental Watershed (RCEW; Owyhee Mountains, Idaho) and the Northwest Irrigation and Soils Research Lab (Snake River Plain, Idaho) are considered non-agricultural (native) and agricultural (heavily managed) sites respectively. We use aeolian deposition flux, dust organic and inorganic content, and geochemical analysis to assess the impact of human activities on aeolian processes in these semi-arid study areas. Our results show high variability in monthly and annual aeolian mass flux. Notably, the annual dust flux at the agricultural site (50.2 g/m2/year) is 7.6 times higher than the annual dust flux rate at the non-agricultural site (6.6 g/m2/year). We compare measured loess accumulation rates with loess accumulation on millennial timescales inferred from loess thickness and loess ages from Optically Stimulated Luminescence (OSL). Based on the annual dust flux measured in this study, the dust accumulation rates for soils in the Snake River Plain 4 cm ka-1. OSL ages between 50.14 ± 9.19 ka to 8.32 ± 1.88 ka from the Snake River Plain site indicate average accumulation rates from ~ 50 ka to 34 ka and ~ 34 ka to 18 ka are 6 ± 4 cm ka-1 and 6 ± 3 cm ka-1 respectively. However, dust accumulation rates increase to 17 ± 4 cm ka-1 in the interval between 18 ka to 8 ka. We infer increased rates of deposition may reflect increased sediment availability following the ~18 ka Bonneville Flood on the ancestral Snake River. Over the last 8,000 years, the rate decreased to 7 ± 2 cm ka-1.
Dust organic matter analysis indicates a relatively high percentage of organic matter in dust. While aeolian deposits in both agricultural and non-agricultural areas show organic matter content of around 18 g/m2/year (~26%) and 4 g/m2/year (~38%), respectively, the soil in both locations exhibits a low organic matter percentage of 1.3% and 1.8%, respectively. Geochemical analysis shows the concentration of most elements is generally lower than the average upper continental crust concentration for both study areas; however, dust deposition is enriched in some trace elements such as Co, Cd, and Zn at both the agricultural and non-agricultural sites.
Dust is a well-known source of calcium, which can lead to the formation of soil inorganic carbon (SIC) in arid and semi-arid areas. SIC plays a vital role in shaping global carbon cycles, influencing hydrological processes, and affecting climate models. However, it is somewhat surprising that there has been a notable dearth of modeling studies dedicated to comprehensively understanding the formation, dynamics, and distribution of SIC in North America. Numerous surveys have characterized the distribution of SIC in dryland soils, showing accumulation of SIC usually occurs below 50 cm in the Bk horizon, with the depth or presence of SIC appears to be influenced by various factors, including precipitation, slope, and parent material. Nonetheless, high variation in the distribution of SIC is a significant challenge to mapping and spatio-temporal modeling of SIC as uncertainty increases with distance from the sampled area. In this study, we look for alternative land surface parameters, paired with satellite imagery and machine learning methods, to lower the uncertainty in unsampled regions compared to traditional spatial interpolation methods. We utilize the Random Forest Regressor (RFR) and Random Forest Classifier (RFC) models, identified as the top-performing models, to predict SIC across the continental United States (CONUS). To manage the high number of samples with zero SIC values the models are divided into two pieces by training a classifier to distinguish zero from non-zero SIC, and then training a regressor/classifier (RFC and RFR) on the non-zero target dataset.
The results indicate the moderate effectiveness of the RFR model with a coefficient of determination (R) of 0.37, Root Mean Square error (RMSE) of 3.97, and Mean Absolute Error (MAE) of 3.1 in predicting SIC. The RFC model with an overall accuracy of 0.61 classified the SIC into three classes. We fed several soil properties, climate variables, land surface parameters, and geographical information (latitude and longitude) as SIC controlling factors to the ML models. Then, we used the developed model in the form of Google Earth Engine to produce distribution maps of SIC.
Soil pH influences SIC accumulation and crystallization, with higher soil pH resulting in precipitation of SIC. Both RFR and RFC models highlighted soil pH as the primary controlling factor in SIC accumulation. It is important to note that the relationship between pH and SIC is indeed complex; pH can both influence and be influenced by SIC levels, creating a dynamic relationship where each factor can affect the other. In addition to pH, geographical data (latitude and longitude) are primary factors which control SIC in both models, suggesting these variables might encapsulate essential information related to regional climate variations, terrain characteristics, or soil properties strongly associated with SIC content. Prior studies identify precipitation and soil parent materials as important variables in SIC formation. We identified a precipitation threshold of 1700 mm for SIC accumulation across the CONUS, differing from the threshold of approximately 500 mm annual precipitation identified in other arid and semi-arid regions. Despite utilizing lithology as a substitute for parent material, the model did not recognize it as a significant factor in SIC formation, likely due to the limited resolution of the lithology data.
This thesis provides two comprehensive chapters related to dust deposition and the role of dust in the development of calcic soils. Chapter 1 examines rates and seasonality of dust accumulation in managed and native sites in southwestern Idaho, and inferred changes in dust accumulation over modern to millennial timescales. Chapter 2 provides a new analysis and map of calcic soil distribution in the continental United States and predicts controls on SIC distribution using a novel combination of models.
DOI
https://doi.org/10.18122/td.2213.boisestate
Recommended Citation
Ghahremani, Zahra, "Investigating the Distribution of Soil Inorganic Carbon, and the Role of Dust Deposition in Dryland Soils" (2024). Boise State University Theses and Dissertations. 2213.
https://doi.org/10.18122/td.2213.boisestate
