A Stochastic Framework for Meteorological Data Imputation Under Varying Missingness Conditions

The presence of missing records due to sensor failures in weather stations compromises data quality, environmental research, and machine learning applications. Evidence remains limited on how missingness conditions, target variables, covariates, and dataset characteristics jointly affect imputation performance. This study aimed to investigate how these factors influence imputation performance. A complete 30-year database (1996–2025) with daily data from Piracicaba, Brazil, was employed to construct a stochastic framework for evaluating linear interpolation (LI), daily average imputation (DAI), KNN, ARIMAX, and Random Forest imputation (RFI). Performance was assessed by averaging R2, RMSE, and KGE over 600 iterations. Subsequently, the techniques were applied to daily datasets from two locations in the United States for 2021–2025. Results from the stochastic framework showed that LI effectively filled one-day gaps in air temperature, whereas multivariate techniques performed better for consecutive gaps of five days or longer. For global radiation, multivariate methods provided the best estimates. Furthermore, the results revealed that RFI and KNN were less sensitive to increasing missingness conditions than the other techniques. The application to the two U.S. locations further showed that climatic and dataset characteristics affected method performance. Notably, LI efficiently filled all target variables under most evaluated missingness conditions at the location with marked interannual climatic stability. DAI exhibited unstable performance and was generally inferior to the remaining techniques. Collectively, these results can guide method selection for researchers requiring complete weather records and support future research toward more reliable and efficient imputation strategies.

Authors

Institutions

Publication Details

Journal
Climate
Published
2026-11-16
DOI
https://doi.org/10.3390/cli14110251
Primary Topic
Precipitation Measurement and Analysis
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A Stochastic Framework for Meteorological Data Imputation Under Varying Missingness Conditions

Patrícia Angélica Alves Marques, Vinícius Villa e Vila, Marcos Roberto Benso, Tiago Alexandre Rozinholli
1 links
Climate
Precipitation Measurement and Analysis
article

A Stochastic Framework for Meteorological Data Imputation Under Varying Missingness Conditions

Patrícia Angélica Alves Marques, Vinícius Villa e Vila, Marcos Roberto Benso, Tiago Alexandre Rozinholli
article en

Abstract

The presence of missing records due to sensor failures in weather stations compromises data quality, environmental research, and machine learning applications. Evidence remains limited on how missingness conditions, target variables, covariates, and dataset characteristics jointly affect imputation performance. This study aimed to investigate how these factors influence imputation performance. A complete 30-year database (1996–2025) with daily data from Piracicaba, Brazil, was employed to construct a stochastic framework for evaluating linear interpolation (LI), daily average imputation (DAI), KNN, ARIMAX, and Random Forest imputation (RFI). Performance was assessed by averaging R2, RMSE, and KGE over 600 iterations. Subsequently, the techniques were applied to daily datasets from two locations in the United States for 2021–2025. Results from the stochastic framework showed that LI effectively filled one-day gaps in air temperature, whereas multivariate techniques performed better for consecutive gaps of five days or longer. For global radiation, multivariate methods provided the best estimates. Furthermore, the results revealed that RFI and KNN were less sensitive to increasing missingness conditions than the other techniques. The application to the two U.S. locations further showed that climatic and dataset characteristics affected method performance. Notably, LI efficiently filled all target variables under most evaluated missingness conditions at the location with marked interannual climatic stability. DAI exhibited unstable performance and was generally inferior to the remaining techniques. Collectively, these results can guide method selection for researchers requiring complete weather records and support future research toward more reliable and efficient imputation strategies.

ClimateVol. 14(11)
Universidade de São Paulo (BR)
Climate action
Precipitation Measurement and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon