Effects of Changes in Land Use Land Cover and Climate on Long-term Total Dissolved Solids Trends in the Mid-Atlantic Region of the United States

Abstract

The Mid-Atlantic region (MAR) of the U.S. is subjected to a variety of stressors that affect the headwaters of the major rivers. Some of these stressors are abandoned mine drainage, agriculture, municipal point sources, urban areas, out-of-basin diversions, competing water uses, rapid population growths in the lowlands, alterations in water availability due to climate change and habitat alteration. In addition to these regional stressors, the rapid population growths and energy sources shifting have resulted in changes in land use and land cover (LULC) over the last few decades. The interactive effects of LULC and interannual/long-term climate changes have resulted in water quality changes in the region. The goal of this research is to investigate long-term changes in stream total dissolved solids (TDS) under changing LULC and climate variability in the MAR. Also, this dissertation is intended to generate understanding of how predominant LULC features, interannual climate variability, and their pollution processes interact to influence receiving water conditions. This research consists of three complementary studies. The first study was to investigate the interactive effects of interannual climate variability and LULC on in-stream TDS trends at 27 sites in the north-central Appalachian region over a 20-year period (1990 – 2010). The second analysis was to characterize individual streams susceptibility to LULC changes and its effects on TDS changing rates at 29 monitoring sites in the MAR. The third study was to develop a modeling approach to predict the combined effects of LULC and interannual precipitation on stream TDS concentrations using data of 77 monitoring sites during 2008-2018. In these studies, traditional statistical non-parametric approaches (e.g., Mann Kendall test (MK), Theil-Sen slope estimator), principal component analysis (PCA), and advanced statistical modeling methods (structural equation modeling, SEM, and latent growth modeling, LGM) and geographical information system (GIS) techniques were used. Results of the first study showed varying TDS trends with 16 (60%) of the sites having an increasing trend, 7 (26%) having a decreasing trend, and 4 (14%) with no statistically significant trend during a time of major LULC and climate changes. The relationships between TDS and climate revealed that 55% of the sites had a negative TDS-precipitation (TDS-P) slope; 45% had a positive (TDS-P) slope; 32% of the sites had a negative TDS-T slope and 68% had a positive TDS-temperature slope (TDS-T). Principal component analysis revealed that watersheds with an increasing TDS trend were distributed along the vectors of barren, and agriculture lands