Generating Synthetic Climate Data for Design that Match Natural Trends
December 10, 2020
Closure design of mining facilities is dependent on future climate conditions. Often, during the design stage, long-term records for precipitation, temperature and potential evapotranspiration are not available. This paper presents a case study where synthetic climate data that mimics historic trends were generated for a proposed tailings facility closure cover design. The project site has 22 years of precipitation, temperature, relative humidity (RH) and solar radiation records, and six years of wind speed records. Precipitation was found to have a strong correlation to the Southern Oscillation Index (SOI), but not necessarily to the other climate parameters. Precipitation statistics were evaluated on an annual, monthly and daily basis. Statistical terms derived using best-fit distributions such as log-normal or gamma were programmed into a GoldSim™ model and used to generate a long sequence of daily synthetic precipitation (GoldSim 2016). The statistics and trends of the synthetic precipitation were compared to the historical record to confirm the annual and seasonal trends were captured. Following generation of the synthetic precipitation sequence, the remaining climate parameters were selected from the historical record and assigned to the synthetic sequence based on a match for daily precipitation. The match was selected by using an objective function comprised of the weighted relative percent difference (RPD) for daily precipitation, the occurrence of rain the day before and after, and the time of year. The day from the historical record with the minimum weighted RPD was selected to represent the corresponding synthetic day, and the climate parameters assigned accordingly.
Patterson, K. and C. Wang. 2017. “Generating Synthetic Climate Data for Design that Match Natural Trends,” in CDA 2017 Annual Conference, October 16-18, 2017. Kelowna, BC, Canada.