Fine-scale rainfall over New Caledonia under climate change

Abstract

Global climate models projections indicate no clear future rainfall changes over the Southwestern Pacific islands in response to anthropogenic forcing. Yet, these models have low (~ 100–200 km) spatial resolution and suffer from large systematic biases, such as the trademark “double ITCZ”. Here, 4 km-resolution simulations were used with a nested regional atmospheric model, which resolves the New Caledonian mountainous topography. The resulting present-day rainfall amount, spatial structure, seasonal cycle, and extremes compare much better with observations than at 20 km resolution. We applied projected changes from global climate models under the RCP8.5 scenario at the boundaries, following an approach that corrects both their present-day and projected sea surface temperature biases. Unlike climate models, our refined projections reveal an 18% decrease in annual mean rainfall over New Caledonia by 2080–2100, especially on the leeward side of the island (up to 30%) and during the hot season (that accounts for ~ 80% of the rainfall decrease). This drying is robust without bias correction, and much stronger than at ~ 20 km resolution. It is mainly driven by circulation changes. A weather regime classification further demonstrates that ~ 80% of the hot season drying relates to a strong anti-cyclonic and air subsidence anomalies centred on the north of NC, which reduces moisture convergence over the archipelago. The frequency and intensity of extreme precipitation events is also reduced by ~ 20% by 2080–2100. This drastic projected drying would have dramatic impacts on water resources and terrestrial ecosystems, pleading for carefully-planned adaptation policies for New Caledonia.

Forecasting the future risk of dengue epidemics facing climate change in New Caledonia, South Pacific

“…It is worth to note however, that a limit of our study is the use of these CMIP5 climate models, because they have a poor horizontal resolution (~100-200km), that is not a priori adapted to simulate the climate of a mountain island such as New-Caledonia. While the amplitude of temperature changes simulated by these CMIP5 models is robust among models ( Fig S1) and reasonable compared with projections performed with an atmospheric regional model at the island scale [30], it is known that projections of precipitation in the South Pacific and at island scales in CMIP5 models are not reliable according to Dutheil et al [31]. These authors used a very high resolution (~4km) atmospheric model to show, a possible strong reduction of precipitation over New Caledonia at the end of st century in the RCP8.5, confirming doubts on CMIP5 rainfall projections.…”

Convection‐permitting modeling with regional climate models: Latest developments and next steps

“…However, some experiments took advantage of convection‐permitting resolutions to study other variables such as temperature (Expósito et al, 2015; Zhang, Wang, et al, 2016a, 2016b), and wind (Nawri et al, 2014; Argüeso & Businger, 2018; Dayal et al, 2020, 2021) that also require fine spatial detail. Most studies agree that precipitation extremes are more realistic in CPRCMs (Kendon et al, 2012; Zhang, Wang, et al, 2016a; Dutheil et al, 2021), despite the fact that they tend to overestimate rainfall amounts compared to observations (Murata, Sasaki, Kawase, & Nosaka, 2017). The overestimation is typically present with steep topography (Morel et al, 2014; Hassim et al, 2016; Vincent & Lane, 2016, 2017; Argüeso et al, 2020), indicating that the orographic forcing may be too strong in CPRCMs.…”

“…The vast majority of studies using CPRCMs over islands have focused on rainfall (Love et al, 2011;Chan et al, 2013;Chan, Kahana, et al, 2018;Morel et al, 2014;Kendon et al, 2014;Fosser, Kendon, Stephenson, & Tucker, 2020;Dutheil et al, 2021) because it is directly affected by the way convection is represented in models. However, some experiments took advantage of convection-permitting resolutions to study other variables such as temperature (Exp osito et al, 2015;Zhang, Wang, et al, 2016a, 2016b, and wind (Nawri et al, 2014;Argüeso & Businger, 2018;Dayal et al, 2020Dayal et al, , 2021) that also require fine spatial detail.…”

“…Modeling studies at convection‐permitting scales that span climate‐relevant periods (≥10 years) include the British Isles (Chan et al, 2013; Kendon et al, 2012, 2014, 2017; Fosser, Kendon, Chan, et al, 2020; Fosser Kendon, Stephenson, et al, 2020), Svalbard (Dobler, 2019; Dobler et al, 2020), Iceland (Nawri et al, 2014), Japanese Islands (Murata, Sasaki, Kawase, & Nosaka, 2017; Murata, Sasaki, Kawase, Nosaka, Aoyagi, et al, 2017), Hawaii (Zhang et al, 2012; Zhang, Wang, et al, 2016a, 2016b; Argüeso & Businger, 2018; Xue et al, 2020), Puerto Rico (Bhardwaj et al, 2018), Canary Islands (Expósito et al, 2015), Maritime Continent (Vincent & Lane, 2017, 2018), and New Caledonia (Dutheil et al, 2021). Other have focused on shorter periods, such as La Réunion (Morel et al, 2014), Cyprus (Zittis et al, 2017), Hainan (Zhu et al, 2017), Puerto Rico (Wootten et al, 2016), Fiji (Dayal et al, 2020, 2021), and Corsica (Barthlott et al, 2014).…”

“…Some studies have used idealized configurations to investigate the role of islands and their topography on rainfall (Coppin & Bellon, 2019a, 2019b; Tan et al, 2020). Studies of future climate projections were limited to single realizations (Murata, Sasaki, Kawase, Nosaka, Aoyagi, et al, 2017; Bhardwaj et al, 2018; Chan, Kahana, et al, 2018; Dobler, 2019; Dobler et al, 2020; Fosser, Kendon, Stephenson, & Tucker, 2020) or have generally used a PGW approach to reduce computational costs (Expósito et al, 2015; Zhang, Wang, et al, 2016b; Dutheil et al, 2021; Xue et al, 2020).…”