Premium
Sampling Strategies for Estimation of Parameters Related to Ground Water Quality
Author(s) -
Crane Pamela E.,
Silliman Stephen E.
Publication year - 2009
Publication title -
groundwater
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.84
H-Index - 94
eISSN - 1745-6584
pISSN - 0017-467X
DOI - 10.1111/j.1745-6584.2009.00578.x
Subject(s) - sampling (signal processing) , sampling design , statistics , monte carlo method , ranging , mean squared error , quality (philosophy) , range (aeronautics) , computer science , sample size determination , sampling bias , set (abstract data type) , stratified sampling , sample (material) , mathematics , engineering , population , telecommunications , philosophy , chemistry , demography , filter (signal processing) , epistemology , chromatography , sociology , computer vision , programming language , aerospace engineering
Multiple theoretical sampling designs are studied to determine whether sampling designs can be identified that will provide for characterization of ground water quality in rural regions of developing nations. Sampling design in this work includes assessing sampling frequency, analytical methods, length of sampling period, and requirements of sampling personnel. The results answer a set of questions regarding whether using innovative sampling designs can allow hydrogeologists to take advantage of a range of characterization technologies, sampling strategies, and available personnel to develop high‐value, water‐quality data sets. Monte Carlo studies are used to assess different sampling strategies in the estimation of three parameters related to a hypothetical chemical observed in a ground water well: mean concentration (MeanC), maximum concentration (MaxC), and total mass load (TML). Five different scenarios are simulated. These scenarios are then subsampled using multiple simulated sampling instruments, time periods (ranging from 1 to 10 years), and sampling frequencies (ranging from weekly to semiannually to parameter dependent). Results are analyzed via the statistics of the resulting estimates, including mean square error, bias, bias squared, and precision. Results suggest that developing a sampling strategy based on what may be considered lower quality instruments can represent a powerful field research approach for estimating select parameters when applied at high frequency. This result suggests the potential utility of using a combination of lower quality instrument and local populations to obtain high frequency data sets in regions where regular monitoring by technicians is not practical.