ASL 761 Uncertainty Bias Correction and Downscaling Rainfall
ASL 761 Uncertainty, Bias Correction, and Downscaling
Rainfall change under global warming June-Aug. , 2070 -2099 avg minus 1961 -90 avg. CMIP 5 Multi-model Ensemble Mean (mm/day) Analysis: J. Meyerson CMIP 5
CMIP 5 examples of individual model rainfall change Rainfall change: June-Aug. , 2070 -2099 avg minus 1961 -90 avg. (mm/day) Analysis: J. Meyerson CMIP 5
CMIP 5 examples of individual model precip change Precipitation change: June-Aug. , 2070 -2099 avg minus 1961 -90 avg. Analysis: J. Meyerson (mm/day) CMIP 5
CMIP 5 examples of individual model precip change Precipitation change: June-Aug. , 2070 -2099 avg minus 1961 -90 avg. (mm/day) Analysis: J. Meyerson CMIP 5
IPCC AR 5
AR 5 – Projected Change in Annual Mean Surface Temperature • Hatching indicates regions where the multi-model mean climate change signal is small compared to natural internal variability (i. e. , less than one standard deviation of natural internal variability in 20 -year means) • Stippling indicates regions where: (i) (ii) the multi-model mean climate change signal is large compared to natural internal variability (i. e. , greater than two standard deviations of natural internal variability in 20 -year means)and where at least 90% of models agree on the sign of change
AR 5 – Projected Change in Average Annual Mean Rainfall • Hatching indicates regions where the multi-model mean climate change signal is small compared to natural internal variability (i. e. , less than one standard deviation of natural internal variability in 20 -year means) • Stippling indicates regions where: (i) (ii) the multi-model mean climate change signal is large compared to natural internal variability (i. e. , greater than two standard deviations of natural internal variability in 20 -year means)and where at least 90% of models agree on the sign of change
Bias Correction
Bias Correction
Bias Correction JJAS Rainfall (mm/day) 2006 -2013
Downscaling
Downscaling • The needs of decision makers to plan for CC at the regional level are much finer than what is provided by the global climate models • The coarse resolution climate change projections need to be translated into finer spatial resolution • Downscaling refers to this method of translating coarser scale climate information to finer scales • Although this is a standard practice there a variety of assumptions behind these techniques that are used to derive such finer scale information
Downscaling • Two principal methods: (1) Dynamical, and (2) Statistical • Dynamical – Physically-based RCMs forced by GCM data at a much higher resolution and over a limited domain of interest. Computationally intensive method.
Dynamical Downscaling Gutowski, et al. , 2016
CORDEX South Asia
Downscaling • Two principal methods: (1) Dynamical, and (2) Statistical • Dynamical – Physically-based RCMs forced by GCM data at a much higher resolution and over a limited domain of interest. Computationally intensive method. • Statistical – Finding statistical relationships between large -scale climate features (GCM scale) and local climate from observations, and using them for finer scale projections. They require minimal computing but are strongly dependent on: (1) the accuracy of historical observations, and (2) assumptions of stationarity of historical relationships between the large scale and finer scales.
Assumptions for SD • Strong relationship predictand between the predictor and the • GCMs accurately simulate the predictor • Statistical relationship between the predictand does not change over time predictor and
Climate Change Uncertainty due to Parameter Sensitivity
Precip change under global warming for control values JJA Prec. Anom. 2071 -2090 – 1976 -1995 (mm/day) Stippled for T-test at 5% level
CESM 1 param. sensitivity of RCP 8. 5 prec. change JJA Prec. Anom. 2071 -2090 – 1976 -1995 downdraft fraction across case 0. 75 minus case 0 (mm/day) Stippled for T-test at 5% level Downdraft fraction
CESM 1 param. sensitivity of RCP 8. 5 prec. change JJA Prec. Anom. 2071 -2090 – 1976 -1995 deep convective adjustment time across case 240 minus case 30 min (mm/day) Stippled for T-test at 5% level Deep convection adjustment time
CESM 1 param. sensitivity of RCP 8. 5 prec. change JJA Prec. Anom. 2071 -2090 – 1976 -1995 entrainment across case at 2 km -1 minus case 0 Stippled for T-test at 5% level Entrainment parameter dmpdz
- Slides: 23