Recent extreme drought events in the Amazon rainforest: assessment of different precipitation and evapotranspiration datasets and drought indicators

[摘要] Over the last decades, the Amazon rainforest has been hit by multiple severedrought events. Here, we assess the severity and spatial extent of the extreme drought years 2005, 2010 and 2015/16 in the Amazon region andtheir impacts on the regional carbon cycle. As an indicator of droughtstress in the Amazon rainforest, we use the widely applied maximum cumulative water deficit (MCWD). Evaluating nine state-of-the-artprecipitation datasets for the Amazon region, we find that the spatialextent of the drought in 2005 ranges from 2.2 to 3.0 (mean  =2.7 )  ×10 6  km 2 (37 %–51 % of the Amazon basin, mean  =45  %), where MCWD indicates at least moderate drought conditions (relative MCWD anomaly - 0.5 ). In 2010, the affected area was about 16 % larger, ranging from 3.0 up to 4.4 (mean  =3.6 )  ×10 6  km 2 (51 %–74 %, mean  =61  %). In 2016, the mean area affected by drought stress was between 2005 and 2010 (mean  = 3.2 × 10 6  km 2 ; 55 % of the Amazon basin), but the general disagreement between datasets was larger, ranging from 2.4 up to 4.1×10 6  km 2 (40 %–69 %). In addition, we compare differences and similarities among datasets using the self-calibrating Palmer Drought Severity Index (scPDSI) and a dry-season rainfall anomaly index (RAI). We find that scPDSI shows a stronger and RAI a much weaker drought impact in terms of extent and severity for the year 2016 compared to MCWD. We further investigate the impact of varying evapotranspiration on the drought indicators using two state-of-the-art evapotranspiration datasets. Generally, the variability in drought stress is most dependent on the drought indicator (60 %), followed by the choice of the precipitation dataset (20 %) and the evapotranspiration dataset (20 %).Using a fixed, constant evapotranspiration rate instead of variableevapotranspiration can lead to an overestimation of drought stress in theparts of Amazon basin that have a more pronounced dry season (for examplein 2010). We highlight that even for well-known drought events the spatialextent and intensity can strongly depend upon the drought indicator and the data sources it is calculated with. Using only one data source and droughtindicator has the potential danger of under- or overestimating drought stressin regions with high measurement uncertainty, such as the Amazon basin.

[学科分类] 大气科学