Collection of Historical Weather Data: Issues with Missing Values

TL;DR

This paper addresses the challenge of missing data in historical weather datasets by proposing and evaluating multiple imputation methods to improve data quality for agricultural and forestry decision-making.

Contribution

The study introduces new missing data imputation techniques tailored for weather data and demonstrates their effectiveness through experimental evaluation.

Findings

Two proposed methods show high promise for large-scale application.

Imputation methods significantly improve data completeness.

Effective handling of missing values enhances decision-making in agriculture and forestry.

Abstract

Weather data collected from automated weather stations have become a crucial component for making decisions in agriculture and in forestry. Over time, weather stations may become out-of-order or stopped for maintenance, and therefore, during those periods, the data values will be missing. Unfortunately, this will cause huge problems when analysing the data. The main aim of this study is to create high-quality historical weather datasets by dealing efficiently with missing values. In this paper, we present a set of missing data imputation methods and study their effectiveness. These methods were used based on different types of missing values. The experimental results show that two the proposed methods are very promising and can be used at larger scale.