Random Forest Groundwater Time Series Imputation

Knowledge graph centered on Random Forest Groundwater Time Series Imputation with 69 nodes and 141 connections. Top connected: Crested Butte, East River, Winter snowpack, evapotranspiration, Soil characteristics.

Method synopsis

A methodology using random forest algorithms to fill missing values in sub-hourly groundwater monitoring data with entropy-based uncertainty quantification.

Synthesized from method descriptions across 2 papers using this protocol.

Procedure as described in the canonical source

Steps below were extracted from the paper that introduces this protocol — Imputation of contiguous gaps and extremes of subhourly groundwater time series using random forests (2022), Journal of Machine Learning for Modeling and Computing. Implementations in other papers (listed below) may differ.

1. Apply single-well imputation
   Missing values were imputed using linear interpolation (SWI-LI) assuming equally spaced data points, and Random Forest (SWI-RF) using temporal predictors including Unix timestamps, days, months, and years from the same well
   Quantities: Applied to individual wells with varying gap percentagesDuration: Applied to synthetic gap scenariosConditions: Used discrete set of available data points for interpolation
   Equipment: RandomForestRegressor from Python's sklearn package
2. Apply multiple-well imputation
   Random Forest imputation using time-series data from neighboring wells as predictors, including timestamps, days, months, years, and water levels from other wells located in proximity to the target well
   Quantities: Feature space included temporal variables plus neighboring well dataDuration: Applied to synthetic gap scenarios for comparison with SWIConditions: Leveraged spatial correlation between neighboring wells
   Equipment: RandomForestRegressor from Python's sklearn package

Typical Equipment

• computational software for random forests
• computational analysis software

Output Measurements

• imputed groundwater levels
• uncertainty estimates

Papers Using This Protocol (2)

Species Studied With This Protocol

Places Where Used