Optimization Design of Groundwater Level Monitoring Network Based on the Geo-Statistics

Jing Chen2013 · DOI

Abstract

The current groundwater monitoring network has many problems caused by history and poor techniques, providing incomplete groundwater information and hindering the management of groundwater resources. This study takes the deep pore confined water monitoring network in Ningbo Plain as an example, using geostatistics—specifically variograms and Kriging interpolation—to adjust the number and position of the current 29 wells. The optimization resulted in a more reasonable monitoring network consisting of 19 wells after removing 17 redundant wells and adding 7 new wells.

Introduction

Managing groundwater resources requires accurate, real-time data, typically gathered through a network of monitoring wells. However, many existing networks are "legacy" systems—built over decades using outdated techniques or haphazard placement. This leads to two major problems: redundancy, where too many wells in one area waste resources, and data gaps, where large unmonitored areas leave decision-makers with incomplete information. This paper treats monitoring network design as an optimization problem, aiming to find a balance: enough wells for a complete picture of the groundwater level without creating an economic burden.

Methodology: The Geostatistical Approach

The study relies on Geo-Statistics to model spatial patterns. The workflow involves two primary pillars:

  1. Variograms: This tool measures the spatial correlation between data points, determining how far apart wells can be before data from one no longer helps predict the state of another. A Gaussian model was identified as the best fit for the Ningbo Plain data.
  2. Kriging Interpolation: This method provides the "Best Linear Unbiased Prediction." Beyond estimating water levels, Kriging provides a Theoretical Variance (or Standard Deviation, Ksd), which acts as a quantitative measure of uncertainty.

The optimization process utilized a trial-and-error adjustment method across two software platforms: Surfer for statistical analysis and contour mapping, and Mapgis for spatial adjustments of well locations.

Flowchart of the optimization process using geostatistical tools and software synergy.

Case Study: Ningbo Plain

The study focuses on the Ningbo Plain in Zhejiang province, China, covering 850 km². The target is the "second pore confined aquifer," a deep layer critical for the city's water supply. Input data consisted of groundwater level observations from October 2008 taken from the original 29-well network. The specific optimization area covered 365.74 km².

Optimization Results

The paper adopts a precision standard where a Kriging standard deviation (Ksd) between 0.5 and 0.77 is considered optimal for management needs. By mapping the Ksd, "over-monitored" zones (Ksd < 0.5) and "under-monitored" zones (Ksd > 0.77) were identified.

The network was successfully reduced from 29 wells to 19 wells. This was achieved by removing 17 redundant wells, primarily in the crowded northern sector, and adding 7 new wells in previously unmonitored areas. The final configuration brought the majority of the study area into the target Ksd range, ensuring a more uniform and reliable data stream.

The optimized monitoring network showing wells kept and new wells added to achieve target precision.

Conclusion and Future Work

This geostatistical approach effectively thins out redundant monitoring points while filling critical data gaps. However, this specific Kriging approach is best suited for stable regions with natural groundwater flow. In more complicated regions—where human activity, heavy industrial pumping, or complex environmental factors disturb the aquifer—standard Kriging may fall short. Future research should focus on building "constrained" Kriging models that incorporate these external variables to maintain accuracy in highly disturbed urban or industrial environments.

Summary updated

Abstract

The current groundwater monitoring network has many problems caused by the history and poor techniques, which has provided incomplete groundwater information and it is adverse to the management of groundwater resources. So it is necessary to optimize it. The writer takes the deep pore confined water monitoring network in Ningbo Plain as an example ,using the main two parts of geo-statistics-variogram and Kriging interpolation to adjust the number and position of the current 29 wells on it, finally getting a new more reasonable monitoring network consists of 19 wells after removing 17 redundant wells and adding 7 new wells.
BibTeX
@misc{chen2013optimization,
  author = {Jing Chen},
  title = {{Optimization Design of Groundwater Level Monitoring Network Based on the Geo-Statistics}},
  year = {2013},
  doi = {10.2991/icssr-13.2013.22},
}
Figure