Mapeo y predicción de subsidencia a través de modelos de regresión espacio-temporal de observaciones de extracción de aguas subterráneas y de subsidencia

Translated title of the contribution: Mapping and predicting subsidence from spatio-temporal regression models of groundwater-drawdown and subsidence observations

Muhammad Zeeshan Ali, Hone Jay Chu, Thomas J. Burbey

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Many regions of the earth are experiencing land subsidence owing to aquifer-system compaction, a consequence of groundwater depletion manifesting as excessive groundwater drawdown. The relation between groundwater drawdown and land subsidence caused by aquifer-system compaction is nonstationary in space and time due to the highly heterogeneous aquifer material, hydraulic and mechanical properties, and spatio-temporal variations in aquifer recharge and groundwater extraction. Annual land subsidence maps are developed using geographical time-slice weighted regression (GTSWR) and geographical temporal weighted regression (GTWR). Considering these spatiotemporal regressions, groundwater drawdown is used as the input parameter to estimate spatial and temporal patterns of land subsidence in both Changhua and Yunlin counties, Taiwan, for an 8-year period. Results indicate that the GTSWR or GTWR models yield greater accuracy with a lower root mean square error (RMSE) than linear regression (LR). The correlation between the predicted and observed data for LR, GTSWR and GTWR is 0.31, 0.93 and 0.94, respectively. In the spatiotemporal models, areas with smaller model coefficients represent over-consolidated sediments, whereas the areas with larger coefficients represent where sediments are normally consolidated. Normally consolidated sediments tend to produce the greatest amount of land subsidence. Annual subsidence patterns reveal that greater levels of subsidence are progressing inland. The greatest level of subsidence occurs in central Yunlin (7 cm/year) due to groundwater extraction. The spatio-temporal regression model is used to predict the effects of reduced groundwater extraction for different areas based on two scenarios of 30 and 50% reductions in groundwater drawdown.

Translated title of the contributionMapping and predicting subsidence from spatio-temporal regression models of groundwater-drawdown and subsidence observations
Original languageSpanish
Pages (from-to)2865-2876
Number of pages12
JournalHydrogeology Journal
Volume28
Issue number8
DOIs
Publication statusPublished - 2020 Dec

All Science Journal Classification (ASJC) codes

  • Water Science and Technology
  • Earth and Planetary Sciences (miscellaneous)

Fingerprint Dive into the research topics of 'Mapping and predicting subsidence from spatio-temporal regression models of groundwater-drawdown and subsidence observations'. Together they form a unique fingerprint.

Cite this