Fast, sensitive, quantitative, and low-cost analysis of environmental pollutants is highly valuable for environmental monitoring. Due to its single-molecule sensitivity and fingerprint specificity, surface-enhanced Raman spectroscopy (SERS) has been widely employed for heavy metal, organic compound, and pathogen detection. However, SERS quantitation is challenging because 1) analytes do not stay in the strongest enhancing region ("hot spots") and 2) SERS reproducibility is poor. In this dissertation, gold nanoparticle/bacterial cellulose (AuNP/BC) substrates were developed to improve SERS sensitivity by increasing hot spot density within the laser excitation volume. Environmentally relevant organic amines were fixed at "hot spots" by lowering solution pH below the analyte pKa and thus enabling SERS quantitation. In addition, a new SERS internal standard was developed based upon the electromagnetic enhancement mechanism that relates Rayleigh (elastic) and Raman (in-elastic) scattering. Rayleigh scattering arising from the amplified spontaneous emission of the excitation laser was employed as a normalization factor to minimize the inherent SERS signal variation caused by the heterogeneous distribution of "hot spots" across a SERS substrate. This highly novel technique, hot spot-normalized SERS (HSNSERS), was subsequently applied to evaluate the efficiency of SERS substrates, provide in situ monitoring of ligand exchange kinetics on the AuNP surface, and to reveal the relationship between the pKa of aromatic amines and their affinity to citrate-coated AuNPs (cit-AuNPs). Finally, colloidally stable stable pH nanoprobes were synthesized using co-solvent mediated AuNP aggregation and subsequent coating of poly(ethylene) glycol (PEG). These nanoprobes were applied for pH detection in cancer cells and in phosphate buffered aerosol droplets. The latter experiments suggest that stable pH gradients exist in aerosol droplets.