Imputing Missing Values in Time Series Data (March 2025)

In this webinar, we explored various techniques for imputing missing values in time series data using GoldSim software. The presentation covered several key methods:

1. Imputation Using Reconstructed Data from Regression Analysis: We discussed how to use data from neighboring sites to fill in missing streamflow records through correlation and regression methods.
2. Imputation Using Stochastic Models: We examined the application of stochastic models to impute missing climate data, such as temperature, wind, and precipitation, ensuring realistic simulations of environmental processes.
3. Simpler Methods for Shorter Missing Value Periods: We reviewed the built-in interpolation methods in GoldSim for handling small gaps in data, including linear interpolation and "constant over next or previous" interpolation.
4. Cubic Spline Interpolation: Additionally, we introduced cubic spline interpolation as a robust method for imputing missing values in a time series, which can be more effective than linear or constant interpolation in certain cases.

By the end of this webinar, participants gained a better understanding of various imputation techniques and how to apply them in GoldSim to maintain the integrity and continuity of their time series data. 

Below is a video recording of the webinar presentation:

Download the example models used for this presentation:

• Markov Process Precipitation Simulator
• Temperature Simulator (TempGen)
• Cubic Spline Interpolation of Time Series Data