IPCC Report.pdf - Adam Curry

Chapter 3Changes in Climate Extremes and their Impacts on the Natural Physical Environmentmajor soil moisture drought, as could be inferred from satellitemeasurements (Andersen et al., 2005), model simulations (Fischer et al.,2007a,b), and impacts on ecosystems (Ciais et al., 2005; Reichstein etal., 2007).There is low confidence in dryness trends in South America (Table 3-2),partly due to lack of data and partly due to inconsistencies. For theAmazon, repeated intense droughts have been occurring in the lastdecades but no particular trend has been reported. The 2005 and 2010droughts in Amazonia are, however, considered the strongest in the lastcentury as inferred from integrating precipitation records and waterstorage estimates via satellite (measurements from the GravityRecovery and Climate Experiment; Chen et al., 2009; Lewis et al., 2011).For other parts of South America, analyses of the return intervalsbetween droughts in the instrumental and reconstructed precipitationseries indicate that the probability of drought has increased during thelate 19th and 20th centuries, consistent with selected long instrumentalprecipitation records and with a recession of glaciers in the Chilean andArgentinean Andean Cordillera (Le Quesne et al., 2006, 2009).Changes in drought patterns have been reported for the monsoon regionsof Asia and Africa with variations at the decadal time scale (e.g., Janicot,2009). In Asia there is overall low confidence in trends in dryness bothat the continental and regional scale, mostly due to spatially varyingtrends, except in East Asia where a range of studies, based on differentindices, show increasing dryness in the second half of the 20th century,leading to medium confidence (Table 3-2).In the Sahel, recent years have been characterized by greater interannualvariability than the previous 40 years (Ali and Lebel, 2009; Greene et al.,2009), and by a contrast between the western Sahel remaining dry andthe eastern Sahel returning to wetter conditions (Ali and Lebel, 2009).Giannini et al. (2008) report a drying of the African monsoon regions,related to warming of the tropical oceans, and variability related to ENSO.In the different subregions of Africa there is overall low to mediumconfidence regarding regional dryness trends (Table 3-2).For Australia, Sheffield and Wood (2008a) found very limited increasesin dryness from 1950 to 2000 based on soil moisture simulated usingexisting climate forcing (mostly in southeastern Australia) and somemarked decreases in dryness in central Australia and the northwesternpart of the continent. Dai (2011), for an extended period until 2008 andusing different PDSI variants as well as soil moisture output from a landsurface model, found a more extended drying trend in the eastern halfof the continent, but also a decrease in dryness in most of the westernhalf. Jung et al. (2010) inferred from a combination of remote sensingand quasi-globally distributed eddy covariance flux observations that inparticular the decade after 1998 became drier in Australia (and parts ofAfrica and South America), leading to decreased evapotranspiration, butit is not clear if this is a trend or just decadal variation.Following the assessment of observed changes in the AR4 (Chapter 3),which was largely based on one study (Dai et al., 2004), subsequentwork has drawn a more differentiated picture both regionally andtemporally. There is not enough evidence at present to suggest highconfidence in observed trends in dryness due to lack of direct observations,some geographical inconsistencies in the trends, and some dependenciesof inferred trends on the index choice. There is medium confidence thatsince the 1950s some regions of the world have experienced moreintense and longer droughts (e.g., southern Europe, west Africa) butalso opposite trends exist in other regions (e.g., central North America,northwestern Australia).Causes of the Observed ChangesThe AR4 (Hegerl et al., 2007) concluded that it is more likely than notthat anthropogenic influence has contributed to the increase in thedroughts observed in the second half of the 20th century. This assessmentwas based on several lines of evidence, including a detection study thatidentified an anthropogenic fingerprint in a global PDSI data set withhigh significance (Burke et al., 2006), although the model trend wasweaker than observed and the relative contributions of natural externalforcings and anthropogenic forcings were not assessed.There is now a better understanding of the potential role of landatmospherefeedbacks versus SST forcing for meteorological droughts(e.g., Schubert et al., 2008a,b), and some modeling studies have alsoaddressed potential impacts of land use changes (e.g., Deo et al., 2009),but large uncertainties remain in the field of land surface modeling andland-atmosphere interactions, in part due to lack of observations(Seneviratne et al., 2010), inter-model discrepancies (Koster et al., 2004b;Dirmeyer et al., 2006; Pitman et al., 2009), and model resolution oforographic and other effects. Nonetheless, a new set of climate modelingstudies show that US drought response to SST variability is consistentwith observations (Schubert et al., 2009). Inferred trends in drought arealso consistent with trends in global precipitation and temperature, andthe latter two are consistent with expected responses to anthropogenicforcing (Hegerl et al., 2007; X. Zhang et al., 2007). The change in the patternof global precipitation in the observations and in model simulations isalso consistent with the theoretical understanding of hydrologicalresponse to global warming that wet regions become overall wetterand dry regions drier in a warming world (Held and Soden, 2006; seealso Section 3.1.6), though some regions also display shifts in climateregimes (Section 3.1.6). Nonetheless, some single events have beenreported as differing from projections (Seager et al., 2009), though this isnot necessarily incompatible given the superimposition of anthropogenicclimate change and natural climate variability (Section 3.1). For soilmoisture and hydrological drought it has been suggested that thestomatal ‘antitranspirant’ responses of plants to rising atmospheric CO 2may lead to a decrease in evapotranspiration (Gedney et al., 2006). Thiscould mean that increasing CO 2 levels alleviate soil moisture andstreamflow drought, but this result is still debated (e.g., Piao et al.,2007; Gerten et al., 2008), in particular due to the uncertainty inobserved runoff trends used to infer these effects (e.g., Peel andMcMahon, 2006; see also Section 3.2.1).171