International Journal of Environment and Climate Change

Several regions in the world are currently experiencing changes in their climatic conditions. However, these changes are not reflected in how the Intensity-Duration-Frequency curve modelling is done, as “stationarity” is still assumed. Stationarity assumes that the statistical parameters do not change over time.  This study aims to present a comparative analysis of stationary and non-stationary Intensity-Duration-Frequency models for Abakaliki, Nigeria. Thirty-one years of rainfall data (1992-2022) for Abakaliki obtained from the Nigerian Meteorological Agency (NIMET) were used to calibrate both the stationary and non-stationary IDF curves. The analysis employs the Generalised Extreme Value distribution with different parameterisations to calibrate both stationary and non-stationary models across precipitation durations ranging from 5 minutes to 24 hours. Results demonstrate that non-stationary models consistently outperform stationary approaches, especially at shorter durations and return periods. Percentage differences between rainfall intensity predictions range from -2.54% to +26.08%, with the most substantial underestimation by stationary models occurring for short-duration events. The 10-minute duration, 2-year return period event shows a 26.08% underestimation when using stationary assumptions, highlighting critical implications for urban drainage design. These findings indicate that continued reliance on stationary assumptions poses significant risks to infrastructure adequacy in Abakaliki under changing climatic conditions.