A global assessment of change in flood volume with surface air temperature

Highlights

• In most of the world, flood volume declines as temperatures (T) rise.

• Large catchments (> 1000 km²) in tropics face water quantity reductions with rising T.

• In contrast, small catchments in the tropics face more severe flooding due to rising T.

Abstract

Climate change is expected to have a significant impact on water security, with higher temperatures causing both enhanced droughts and flood extremes. Here, using global flow data from non-urban catchments, we investigate the sensitivity of flood volume to changes in concurrent surface air temperature. We find most of the world shows decreases in flood volumes with increasing temperature. To understand why this correlation exists, we assess the sensitivity of the above result to mean daily temperatures (climate region), catchment size, and severity of the flood event. Our results indicate that most of the world shows decreases in flood volume with rising temperature for frequent events (50th percentile in this study) and a lesser decrease for rarer events. Changes in the flood volume in tropical regions show the greatest sensitivity to flood percentiles and catchment size. Large catchments in the tropics (≥ 1000 km²) have considerable sensitivities of flood volume with temperature at rates of -10 to -5%/ °C for frequent events (< 90th percentile) whereas small catchments (1000 km²) have sensitivities that only -5%/ °C or greater (i.e., less magnitude). On the other hand, when temperature increases, smaller catchments in the regions are likely to be exposed to more severe flooding at rates up to 15%/ °C for the most severe floods (99.99 percentile in this study) while the rate for large catchments approach zero. Although this study does not seek to find a causality between air temperature and flood runoff, the results suggest a possible decrease in water security with climate change, particularly in large tropical catchments.

Introduction

The impact of climate change on the water cycle continues to be a topic of debate (Donat et al., 2016; Koutsoyiannis, 2020). It is generally expected that precipitation extremes will increase (Fowler et al., 2021), but across large parts of the world flood magnitudes have declined (Do et al., 2017). Meanwhile, on average mean rainfalls are predicted to increase but so is the occurrence of drought and hence water scarcity. Most increases in rainfalls have been observed over the tropics (Beck et al., 2019) with decreases elsewhere, while most warming is occurring over the land relative to the oceans, leading to climate change affecting different parts of the world disproportionally.

Rising temperatures will intensify heavy rainfall events (Fowler et al., 2021; Craig, 2010; Trenberth et al., 2003; Kim et al., 2022) as the atmospheric moisture‐holding capacity increases as per the Clausius‐Clapeyron (C‐C) relation (Westra et al., 2014; Magan et al., 2020). Due to the CC relationship there is a 6–7%/ °C increase in the saturation vapor pressure, creating a corresponding increase in the maximum amount of low-level moisture in the atmosphere (Held and Soden, 2006). Assuming constant relative humidity (Lenderink and Attema, 2015), this can be translated to a similar increase in extreme precipitation with higher temperatures (Sun et al., 2021), a measure that has been termed scaling or sensitivity (expressed in%/ °C units) in the literature (Trenberth et al., 2003).

As long-lasting or intense precipitation is often the main cause of pluvial flooding (Smith and Ward, 1998), such positive scaling estimates along with precipitation trends have been used to suggest future increases in flooding (Ivancic and Shaw, 2015; Seneviratne et al., 2012; Kundzewicz et al., 2014; Field et al., 2012a). This is despite the considerable uncertainty associated with future precipitation projections (Kim et al., 2020; Hosseinzadehtalaei et al., 2017), and little observational evidence to suggest that flood magnitudes have increased (Field et al., 2012b; Hirabayashi et al., 2013). In fact, observational records often present more evidence for a decrease in annual flood maxima (Do et al., 2017; Wasko and Nathan, 2019; Gudmundsson et al., 2019), despite increases in precipitation with climate change being well documented (Westra et al., 2014; Sun et al., 2021; Alexander et al., 2006).

For example, Shen and Chui (2021a) found that both scaling in extreme precipitation and streamflow across the Continental United States (CONUS) are not always on the same direction, depending on whether the extreme streamflow is more correspondent to extreme precipitation or other factors/processes like soil moisture and snowmelt. The inconsistency in both precipitation and streamflow scaling was also identified in South and Central Asia (Ghausi and Ghosh, 2020) where daily scaling of extreme precipitation becomes negative at high temperature while extreme streamflow scaling against temperature remains strongly positive. However, Shen and Chui (2021b) found that flood-temperature scaling is often similar to that of extreme precipitation-temperature from an analysis for three different hydroclimatic regions in the United States for short durations and impervious surfaces. Indeed, an evaluation of historical trends across Australia, where smaller floods have been decreasing and larger floods increasing found good agreement with peak-flow temperature sensitivities (Wasko, 2021).

In addition to flood control (Milly et al., 2008), water resource management has been argued to be under threat as temperatures rise worldwide (Karr, 1991; Lenzen and Foran, 2001). While some studies have found increases in streamflow globally (Labat et al., 2004; Probst and Tardy, 1987), an overall drying trend was found using the largest 925 rivers around the world (accounting for 73% of global runoff) by Dai et al. (2009), consistent with Milliman et al. (2008) who used a sample representing approximately half of the world's discharge. Indeed, higher annual temperatures have been linked to lower annual streamflows (Milly et al., 2018) and higher seasonal temperatures to greater water insecurity (Hettiarachchi et al., 2022).

Although Wasko and Sharma (2017) found negative peak flood scaling with temperature for frequent flood events, whether this translates into a reduction of water availability and water supply reliability, remains an open question. Given the importance of flood volume in ensuring water supply reliability in reservoir systems worldwide, this study builds off the above-mentioned negative flood scaling results to focus on flood volume and address the following questions: (1) Does there exist consistency in the scaling relationship for flood peak and flood volume? (2) If inconsistent, what factors can be correlated to the different pattern of change for flood volumes compared to peaks? and (3) What implications do these scaling relationships have on water security, both in the context of flood insecurity, and more importantly, given the focus on flood volumes, on water supply insecurity worldwide?