170 likes | 296 Views
Climate model results for the watershed of the Delaware Estuary. Raymond Najjar Department of Meteorology The Pennsylvania State University May 2009. Goals. Assess the performance of climate models for the watershed of the Delaware Estuary
E N D
Climate model results for the watershed of the Delaware Estuary Raymond Najjar Department of Meteorology The Pennsylvania State University May 2009
Goals • Assess the performance of climate models for the watershed of the Delaware Estuary • Provide climate projections under two greenhouse gas emissions scenarios for this region
Methods • 14 climate models from the Coupled Model Intercomparison Project, Phase 3 (CMIP3) • Model resolution ~1.5-4.5º • Simulations using observed forcing for 20th century, and B1 and A2 emissions scenarios for 21st century • Daily output for temperature and precipitation • Observations from U. Delaware (monthly, from 1901) and North American Regional Reanalysis (daily, from 1979) • Model output and observations gridded to 1º resolution
Analysis domain Results presented will be averaged over this 1º× 3º domain
Model evaluation: monthly means • Models tend to be slightly too cool and wet, on average. • Models differ substantially in their precipitation simulations.
Model evaluation: interannual variability • Models slightly too variable in temperature • Models capture annual cycle in temperature variability • Models miss annual cycle in precipitation variability
Model evaluation:intramonthly variability Model-mean does a good job at simulating variability within a month, though underestimates precipitation variability in summer.
Overall Model Evaluation • An error index was computed based on the monthly means, interannual variability, and intramonthly variability of temperature and precipitation (six metrics) • Each symbol represents a model. The 14-model average and the observations are also shown. • Bootstrapping was used to determine the 95% confidence bounds on the error index • A perfect model would have an error index of zero; an average model would have an error index of 1 • Results show that the model average is superior to any individual model
Summary of climatic extremes evaluation • Models, on average, provide a good representation of climatic extremes • NARR output may underestimate heat waves due to 3-hour resolution of output • Too many frost days reflects cool bias of models
21st Century Climate Projections • Projections shown as changes with respect to 1980-1999 for three future time periods: 2011-2030 (early century), 2046-2065 (mid century), and 2080-2099 (late century). • B1 (lower emissions) and A2 (higher emissions) scenarios are shown. • Changes shown using box-and-whisker plots, which present 14-model maximum, minimum, median, 25th percentile, and 75th percentile.
Temperature change • More warming in summer than in winter • Full range of change among models is about a factor of two • Scenario differences minor until late century
Precipitation change • Most models predict annual precipitation increase • Greater increase and agreement among models in winter than in summer • Less agreement among models for precipitation change than for temperature change
Annual frost days and growing season length changes • Defined using temperature thresholds • Warmer spring and fall means fewer frost days and longer growing seasons
Changes in precipitation extremes • Most models project increases in heavy precipitation • Large spread among models
Changes in heat waves • All models project increases in heat waves • Large spread among models
Summary of climate projections • Early-century results independent of emissions scenario—means that some additional human-induced climate change is unavoidable • All models warm; factor of two range due to model choice. • Precipitation projected to increase, particularly in winter and spring • Extreme precipitation and extreme heat are projected to increase