Continuous rainfall generation for a warmer climate using observed temperature sensitivities

Continuous rainfall sequences are often used as inputs in hydrologic modeling, particularly where a probabilistic assessment is required. Continuous rainfall sequences provide a means for accounting of all aspects of rainfall that produce flooding, for example, not just the design rainfall event but also the rainfall prior to the extreme rainfall event. With the advent of climate change, higher temperatures have been associated with changes in rainfall, in particular intensifying rainfall extremes with less uniform temporal patterns. Given these demonstrated changes to extreme rainfall with temperature rise, there is a need to modify continuous rainfall generators to account for current and likely future changes in temperature. In this work we propose a novel method for simulating continuous rainfall sequences for a future warmer climate by conditioning parameters on their historical sensitivity with temperature. To demonstrate the proposed technique we use a one-dimensional Neyman-Scott Rectangular Pulses model at two locations across Australia. The statistics used in the parameter estimation are conditioned on their historical sensitivity to average monthly temperature to simulate rainfall for a change in temperature. The results are validated by comparing the simulated rainfall against observations originating from differing temperatures and it is shown that the model captures the relative difference in the mean monthly rainfall and monthly maxima. Encouraged by these results we simulate rainfall for higher temperatures and capture expected changes to annual maxima and design temporal patterns for a warmer climate. While we demonstrate our methodology in the simulation of sub-daily rainfall using a specific model, the approach presented here can be applied to all weather generation schemes for projection in a warmer climate. (C) 2016 Elsevier B.V. All rights reserved.