PUMA toolbox Philip Mote OCCRI.net Oregon State University Philip Mote
CMIP3 CMIP5 NARCCAP regcpdn WRF delta BCSD CA BCCA MACA etc. observations reanalysis PRISM
eractive chemical or biochemical components. nationality model A parallel evolution toward increased complexity and olution has occurred in the BCCR domain of numerical weather ediction, and has resulted in a large and verifiable improvement operational weather forecastccsm3 quality. This example alone ows that present models are more realistic than were those of CGCM (T47) decade ago. There is also, however, a continuing awareness t models do not provide acgcm perfect (T63) simulation of reality, cause resolving all important spatial or time scales remains CNRM beyond current capabilities, and also because the behaviour CSIRO such a complex nonlinear system may in general be chaotic. It has been known since the work of Lorenz (1963) that even ECHAM5 mple models may display intricate behaviour because of their nlinearities. The inherent nonlinear behaviour of the climate ECHO-G stem appears in climate simulations at all time scales (Ghil, FGOALS 89). In fact, the study of nonlinear dynamical systems has come important for a wide range of scientific disciplines, and GFDL-CM2.0 corresponding mathematical developments are essential to GFDL-CM2.1 erdisciplinary studies. Simple models of ocean-atmosphere eractions, climate-biosphere interactions or climate-economy GISS-AOM eractions may exhibit a similar behaviour, characterised by GISS-ER rtial unpredictability, bifurcations and transition to chaos. In addition, many of the keyhadcm3 processes that control climate nsitivity or abrupt climate changes (e.g., clouds, vegetation, eanic convection) depend on HADGEM1 very small spatial scales. They nnot be represented in full detail in the context of global INMCM odels, and scientific understanding of them is still notably omplete. Consequently, thereipsl is a continuing need to assist the use and interpretation of complex models through models MIROC t are either conceptually simpler, or limited to a number of ocesses or to a specific region, therefore enabling a deeper MIROC-hires derstanding of the processes at work or a more relevant PCM mparison with observations. With the development of Figure 1.4. Geographic resolution characteristic of the generations of climate models used in the IPCC Assessment Reports: FAR (IPCC, 1990), SAR (IPCC, 1996),
PNW changes Mote and Salathé 2010 - graphics by Jeremy Littell
Worst Climate variable Best Mean Median Gleckler, Taylor, and Doutriaux, Journal of Geophysical Research (2008) Model used in IPCC Fourth Assessment
cm 25 20 15 10 20th century seasonal cycle Mean precipitation gfdl_cm2_1 giss_aom giss_er hadcm hadgem1 inmcm3_0 ipsl_cm4 miroc3_2_hi miroc_3.2 pcm1 multi-model avg NCEP 5 0 Jan Mar May Jul Sep Nov Jan Mote and Salathé 2010
2080s PNW summer precipitation change 40 20 % change 0-20 -40-60 0.0 0.2 0.4 0.6 0.8 1.0 Skill factor Mote, Brekke, Duffy, Maurer 2011 10
NARCCAP narccap.ucar.edu
NARCCAP
climateprediction.net (CPDN)
Simulated July Tmax Observed (PRISM) July Tmax
Multivariate Adaptive Constructed Analogs (MACA) Late 20 th 20C3M Mid 21 st Meeting the needs of fire management Needs 1. Full meteorology: TMAX/TMIN, ppt, winds, RH, solar 2. Daily timescales TEMP, RH, PPT, 10-m winds 8km resolution Methods 1. Combines attributes of CA and BCSD methods similar to BCCA (Maurer and Hidalgo, 2010) 2. Accounts for disappearing analogs, new variables 3. Compared to daily BCSD method Abatzoglou and Brown, submitted
MAC advice Considerations in design of climate scenarios Which variables and at what time/space res. how to characterize uncertainty model evaluation? Other considerations: paleo, time-to-detection,
MAC advice Dynamical vs statistical always some statistical (remove biases, get large ensemble); dynamical for physical relationships, extremes; evaluate approaches according to needs at hand