Compound hot–dry events can strongly affect water resources, agriculture and ecosystems, yet their stochastic structure and dependence on time-scale remain incompletely understood in many regions. This study develops a copula-based framework to quantify multiscale dependence between temperature, precipitation and a simple water-balance deficit across Australia using high quality station observations. Daily maximum and minimum temperatures from the ACORN-SAT network, daily precipitation from the Australian high quality rainfall network and annual evaporation from selected high quality pan stations are aggregated to annual, seasonal and monthly indices. Station‑level dependence is first explored using Kendall’s tau between temperature and precipitation and, where available, between temperature and a precipitation–evaporation‑based water‑balance deficit. Bivariate copulas are then fitted to pseudo observations of (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\({T}_{\text{max}},P\)</EquationSource> </InlineEquation>), (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\({T}_{\text{min}},P)\)</EquationSource> </InlineEquation> and, at evaporation sites, (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\({T}_{\text{max}},D)\)</EquationSource> </InlineEquation> and (<InlineEquation ID="IEq4"> <EquationSource Format="TEX">\({T}_{\text{min}},D)\)</EquationSource> </InlineEquation> at each time-scale, with families selected by information criteria and uncertainty evaluated via nonparametric bootstrap. Annual τ(<InlineEquation ID="IEq5"> <EquationSource Format="TEX">\({T}_{\text{max}},P\)</EquationSource> </InlineEquation>) is moderately negative at most stations (median ≈ − 0.33), whereas annual τ(<InlineEquation ID="IEq6"> <EquationSource Format="TEX">\({T}_{\text{min}},P\)</EquationSource> </InlineEquation>) is near zero, indicating much weaker coupling of night-time temperature and rainfall. At seasonal and monthly scales, median τ(<InlineEquation ID="IEq7"> <EquationSource Format="TEX">\({T}_{\text{max}},P\)</EquationSource> </InlineEquation>) weakens and often spans zero, while τ(<InlineEquation ID="IEq8"> <EquationSource Format="TEX">\({T}_{\text{min}},P\)</EquationSource> </InlineEquation>) becomes weakly positive on average, revealing pronounced time-scale and diurnal asymmetry. At two dry, water limited case-study sites, copula-based probabilities of annual hot–dry years (90th-percentile <InlineEquation ID="IEq9"> <EquationSource Format="TEX">\({T}_{\text{max}}\)</EquationSource> </InlineEquation> and 10th-percentile precipitation) are around 4–5%, whereas probabilities of joint hot and high‑deficit years (90th‑percentile temperature and 90th‑percentile precipitation–evaporation‑based deficit) are below 1%, quantifying the rarity of truly hot–dry–high demand combinations under present-day climatology.

Multi scale copula analysis of hot–dry dependence across Australia

期刊论文
发表日期：

2026年5月2日

查看全文

Khaled Haddad

摘要

Compound hot–dry events can strongly affect water resources, agriculture and ecosystems, yet their stochastic structure and dependence on time-scale remain incompletely understood in many regions. This study develops a copula-based framework to quantify multiscale dependence between temperature, precipitation and a simple water-balance deficit across Australia using high quality station observations. Daily maximum and minimum temperatures from the ACORN-SAT network, daily precipitation from the Australian high quality rainfall network and annual evaporation from selected high quality pan stations are aggregated to annual, seasonal and monthly indices. Station‑level dependence is first explored using Kendall’s tau between temperature and precipitation and, where available, between temperature and a precipitation–evaporation‑based water‑balance deficit. Bivariate copulas are then fitted to pseudo observations of ( \({T}_{\text{max}},P\) ), ( \({T}_{\text{min}},P)\) and, at evaporation sites, ( \({T}_{\text{max}},D)\) and ( \({T}_{\text{min}},D)\) at each time-scale, with families selected by information criteria and uncertainty evaluated via nonparametric bootstrap. Annual τ( \({T}_{\text{max}},P\) ) is moderately negative at most stations (median ≈ − 0.33), whereas annual τ( \({T}_{\text{min}},P\) ) is near zero, indicating much weaker coupling of night-time temperature and rainfall. At seasonal and monthly scales, median τ( \({T}_{\text{max}},P\) ) weakens and often spans zero, while τ( \({T}_{\text{min}},P\) ) becomes weakly positive on average, revealing pronounced time-scale and diurnal asymmetry. At two dry, water limited case-study sites, copula-based probabilities of annual hot–dry years (90th-percentile \({T}_{\text{max}}\) and 10th-percentile precipitation) are around 4–5%, whereas probabilities of joint hot and high‑deficit years (90th‑percentile temperature and 90th‑percentile precipitation–evaporation‑based deficit) are below 1%, quantifying the rarity of truly hot–dry–high demand combinations under present-day climatology.