Response to the comments of Prof. Geoff Pegram (Reviewer2)

Lionel Benoit

doi:10.5194/hess-2021-453-ac2

Abstract

Stochastic rainfall generators are probabilistic models of rainfall space-time behavior. During parameterization and calibration, they allow the identification and quantification of the main modes of rainfall variability. Hence, stochastic rainfall models can be regarded as probabilistic conceptual models of rainfall dynamics. As with most conceptual models in Earth Sciences, the performance of stochastic rainfall models strongly relies on their adequacy in representing the rain process at hand. On tropical islands with high elevation topography, orographic rain enhancement challenges most existing stochastic models because it creates localized rains with strong spatial gradients, which break down the stationarity of rain statistics. To allow for stochastic rainfall modeling on tropical islands, despite non-stationarity, we propose a new stochastic daily rainfall generator specifically for areas with significant orographic effects. Our model relies on a preliminary classification of daily rain patterns into rain types based on rainfall space and intensity statistics, and sheds new light on rainfall variability at the island scale. Within each rain type, the spatial distribution of rainfall through the island is modeled following a meta-Gaussian approach combining empirical spatial copulas and a Gamma transform function, which allows us to generate realistic daily rain fields. When applied to the stochastic simulation of rainfall on the islands of O‘ahu (Hawai‘i, United States of America) and Tahiti (French Polynesia) in the tropical Pacific, the proposed model demonstrates good skills in jointly simulating site specific and island scale rain statistics. Hence, it provides a new tool for stochastic impact studies in tropical islands, in particular for watershed water resources management and downscaling of future precipitation projections.

Highlights

Stochastic rainfall generators are probabilistic tools aiming at simulating synthetic rains that mimic as closely as possible the statistical signature of rain observations [Richardson, 1981] [Wilks and Wilby, 1999] [Ailliot et al, 2015]
To this first order quasi-static picture is added the important variability of daily rainfall patterns associated with processes ranging from synoptic-scale disturbances [Hopuare et al, 2018] [Longman et al, 2021] to large-scale atmospheric circulations [Hopuare et al, 2015] [Frazier et al, 2018] [Brown et al, 2020]. This variability brings stochasticity on top of the relatively deterministic long-term patterns of orographic rain enhancement. To account for both the long-term quasi-static patterns of rain accumulation and the day-to-day fluctuations of the rainfall spatial distribution, this paper proposes a new stochastic rainfall model dedicated to high tropical islands with significant and complex topography
To better identify the main modes of rainfall variability over O‘ahu, rain types are pooled into three hyperclasses

Summary

Introduction

Stochastic rainfall generators are probabilistic tools aiming at simulating synthetic rains that mimic as closely as possible the statistical signature of rain observations [Richardson, 1981] [Wilks and Wilby, 1999] [Ailliot et al, 2015]. Stochastic rainfall modeling consists of statistical learning (i.e., inference) of the joint space-time probability density function (pdf) of rainfall at all sites and times of interest, and sampling this pdf to generate synthetic rains. The probabilistic approach followed by stochastic rainfall generators enables a comprehensive study of rainfall variability and, in turn, the assessment of uncertainty propagation along the whole modeling chain [Gabellani et al, 2007]. When conditioned to climate model outputs, stochastic rainfall generation can be used for the downscaling of future precipitation projections, resulting in local-scale and high-resolution scenarios of the possible evolution of rainfall in the context of climate change [Jha et al, 2014] [Volosciuk et al, 2017]

Results

Discussion

Conclusion