|
Abstract:
|
The impacts of heat extremes are often dependent on the co-occurring humidity, motivating an examination of how joint temperature-humidity extremes may be changing with warming global temperatures. However, the covariance structure between temperature and humidity is spatially variable, and cannot generally be described by a single parametric form. Here, we propose a model based on non-crossing quantile smoothing splines for the distribution of humidity conditional on temperature, wherein the temperature-humidity relationship is permitted to change linearly with global mean temperature. The direction of conditioning is based on the fact that temperature provides a strong physical constraint on the maximum moisture content of the atmosphere, and the form of the model also allows for the enforcement of the physical constraint that dew point cannot exceed temperature. Based on analysis of synthetic data with similar properties and length to real data, we demonstrate that the model can provide an unbiased estimate of the quantiles of humidity as a function of temperature, including the ability to describe complex temperature-dependent changes in the covariance structure.
|
