Projected increase in droughts over the Arabian Peninsula and associated uncertainties

Abstract

The Arabian Peninsula (AP) has been reported to experience increasing drought in recent decades. With this background, this study evaluates best performing Climate Model Intercomparison Project 6 (CMIP6) Global Climate Models (GCMs) for historical (1985-2014) simulations and future drought projections across the AP until 2100, using the standardized precipitation index (SPI) and standardized precipitation-evapotranspiration index (SPEI). We assess uncertainties from model differences, scenarios, timescales, and methods. Our findings reveal the limitations of most models in reproducing precipitation, leading to uncertainties in SPI projections. Nonetheless, CMIP6-GCMs better capture the increase in the current-day potential evapotranspiration (PET) and therefore the SPEI, which is dominated by PET. The Hargreaves based PET is identified as the most suitable method for SPEI projections. The rate of increase in PET surpasses that of precipitation in all scenarios by fivefold. Consequently, SPEI indicates projected increase in future droughts with greater changes in SSP585 and SSP370 scenario compared to SSP245 and SSP126. In general, drought will exacerbate in the AP despite uncertainties from indices selection, scenarios, and inter-model variability, followed by methods and timescales which predominantly impacts drought magnitude. Over findings emphasize the need for more reliable representation of the AP precipitation in climate models for improved drought projection over the AP to enhance planning and adaptation strategies.

Fig. 1

Performance of CMIP6-GCMs for the period 1985–2014 over the AP in replicating various observational climate statistics for area-averaged precipitation and temperature.

Fig. 2

Spatial climatology of MMM and MSWX for the period 1985–2014 and their biases for precipitation and PET over the AP. Black hatches in difference figures show the significance value at the 95% confidence level. Unit is mm/day for both variables.

Fig. 3

(a, b) Kernel density estimation of SPI and SPEI based on observation, CMIP6 individual models, and MMM for 1985–2014. (c, d, e, f) Spatial distribution of DD based on SPI and SPEI for MMM and MSWX for the period 1985–2014.

Fig. 4

Boxplots denote projected changes in precipitation and PET (both in mm/day) for each CMIP6 model against the reference period for each scenario and subperiod. Horizontal black line in each box represents the median of models. Lower and upper boxes designate the quartile range, and whiskers denote the boundaries of the confidence intervals. Horizontal line in all plots at 0.1 mm/day is indicated for comparison.

Fig. 5

Area-averaged SPI and SPEI temporal variability over the AP from observations and MMM of CMIP6 models for the reference period and for the SSP126, SSP245, SSP370 and SSP585 emission scenarios. Dotted lines of each time series represent the corresponding trend.

Fig. 6

(a) MMM of projected changes in drought frequency (per year) with respect to the historical period (1985–2014) for SPEI in the SSP126, SSP245, SSP370 and SSP585 emission scenarios for the near, mid, and far future. (b) Projected changes in MDD (months) with respect to the reference period for each scenario and subperiods based on the SPEI dataset. The ranges for the characteristics define the inter-model differences. Box plot inside each violin plot shows the data distribution, while outer violin plot displays the density distribution of the dataset. The width of the plot in each case denotes data density. Black dots in each figure denote individual models.