Deseasonalized anomaly time series of temperature ((a), (b)) and precipitation minus evaporation (<em>P</em> − <em>E</em>) changes over the tropical oceans and the tropical land for ((c), (d)) wet and ((e), (f)) dry regions

2013-07-03T00:00:00Z (GMT) by Chunlei Liu Richard P Allan
<p><strong>Figure 1.</strong> Deseasonalized anomaly time series of temperature ((a), (b)) and precipitation minus evaporation (<em>P</em> − <em>E</em>) changes over the tropical oceans and the tropical land for ((c), (d)) wet and ((e), (f)) dry regions. The wet and dry regions are defined as the 30% highest and 70% lowest <em>P</em> − <em>E</em> grid points each month. Data are from the CMIP5 historical and RCP 4.5 simulations (1850–2100) and the AMIP simulations (1979–2008). Anomalies are calculated with respect to the 1860–1950 period for CMIP5 historical data, 1961–1990 for HadCRUT4 and 1988–2005 for AMIP5 data sets; HadCRUT4 and AMIP anomalies are adjusted to agree with the CMIP5 ensemble mean over the period 1988–2005. All lines are 48 month running means. The shaded area is the ensemble mean ± 1 standard deviation.</p> <p><strong>Abstract</strong></p> <p>Global warming is expected to enhance fluxes of fresh water between the surface and atmosphere, causing wet regions to become wetter and dry regions drier, with serious implications for water resource management. Defining the wet and dry regions as the upper 30% and lower 70% of the precipitation totals across the tropics (30° S–30° N) each month we combine observations and climate model simulations to understand changes in the wet and dry regions over the period 1850–2100. Observed decreases in precipitation over dry tropical land (1950–2010) are also simulated by coupled atmosphere–ocean climate models (−0.3%/decade) with trends projected to continue into the 21st century. Discrepancies between observations and simulations over wet land regions since 1950 exist, relating to decadal fluctuations in El Niño southern oscillation, the timing of which is not represented by the coupled simulations. When atmosphere-only simulations are instead driven by observed sea surface temperature they are able to adequately represent this variability over land. Global distributions of precipitation trends are dominated by spatial changes in atmospheric circulation. However, the tendency for already wet regions to become wetter (precipitation increases with warming by 3% K<sup>−1</sup> over wet tropical oceans) and the driest regions drier (precipitation decreases of −2% K<sup>−1</sup> over dry tropical land regions) emerges over the 21st century in response to the substantial surface warming.</p>