Jamie Anderson December 9, 2014 Climate Modeling 101 Jamie Anderson December 9, 2014

Climate Monitoring vs Climate Modeling Monitoring tells us how the current climate has/is changing Modeling tells us how the future climate might change Lower future GHG emissions Higher future GHG emissions Surface Temperature Change 2100 vs present Precipitation Change 2100 vs present N. Hemisphere Sept. Sea Ice 2100 vs present Ocean Surface pH Change 2100 vs present Adapted from IPCC 5th Assessment Summary for Policy Makers

Why are climate models important for DWR? Why are climate models important to DWR? Many of the management decisions that DWR needs to make depend on understanding what current and future stream flows will be like. It turns out, not surprisingly, that those stream flows are determined by climate processes that manifest themselves in the amount of rain we get, the amount of snow we get, where that rain and snow fall, and the temperature of the air throughout the year- which determines when the snow melts and how warm water in channels gets, but also is connected to how much water evaporates off of the vegetation, and out of the soil. And also how much water is needed for agriculture and landscaping.

Climate scientists and modelers explore possible future climate conditions DWR staff assess potential impacts and risks that those climate change conditions could have on California’s water resources http://www.water.ca.gov/climatechange/articles.cfm

What is a model? A model is a simplified representation of something– usually something really complex or something too big or two important to experiment with. Show two “cars” from the goodie bag. First show simple train car…little detail but you can tell what it is. Show other car and highlight extra details, hubcaps, spoiler, doors, etc. More detail. Climate models have different levels of detail in them.

Mathematical representation of the climate system GCM- Global Climate Model or General Circulation Model A Global Climate model takes the complex ocean and atmospheric interactions of the earth and converts them to mathematical equations that can be solved to provide information about how the atmosphere is expected to evolve in the future. Assumptions and simplifications in the equations incorporates uncertainty and potential errors. Complex ocean and atmospheric interactions represented by equations Solves equations to show how climate may evolve in the future

Building a Global Climate Model National Earth Science Teachers Association http://www.windows2universe.org/earth/climate/climate_modeling.html

Why so hard to do? Natural climate variability Future Uncertainty Data collection Physical processes and Feedback mechanisms Spatial Resolution Per Phil Duffy May 2013…in 1 day GCMs can simulate between 1-5 years, thus it take 1-6 months to simulate 150 years (1950-2100) Math Equations = Parameterization PHYSICS not terribly hard Some processes poorly understood, handled Especially CLOUDS Can’t model things we don’t understand Feedback Mechanisms Long-range trends, Larger-scale answers Resolution Too Coarse Models have Bias You need to quantify UNCERTAINTY Real World responding faster How to make DECISIONS? Climate models do not SCALE well Regional Impacts vary Can they give local information? Trivia Time: How long does it take a modern global climate model to simulate 150 years?

Climate models are used to explore possible future changes in climate Climate models don’t predict future changes in climate IPCC modeling development showing their improvements over the years. Better parameterizations since the 70’s, but even those crude models got some of it right for CO2 forcing. 70’s, 80’s, various ASSESSMENT Reports versions.

Weather vs Climate Modeling Weather Forecasting Future Climate Simulations What will the weather be on a specific day in a specific place Short term (7-10 days) What will the climate trends be in a region in the future Long term (100 years)

Natural Climate Variability The climate fluctuates. The trace on top shows nearly 3000 years of global temperature and precipitation variation. It shows how the earth moves back an forth between warm and cool, wet and dry driven by an array of physical processes (both anomalous (volcanoes) and cyclical (solar flairs and global wobble). Even an the local level at daily time scales we have variability. The tan graph shows average daily high and low temperatures for Glendale, CA. The other graph shows actual daily highs and lows for January 2013 in Glendale. You can see some days are warmer, some are cooler. For a given month the average may be higher or lower than long term averages. Global models can’t predict this level of variability. But they can forecast changes in monthly averages over long periods of time.

Climate Variability vs Climate Change Card Game California has 5 water year types: 1 card = 1 water year Critically dry and dry years: Ace-10 black cards Normal-wet years: all red cards & all face cards Joker represents an extreme event Dry period: 2 or more black cards Ace to 10 Shuffle your deck and deal cards 1 at a time. Count # dry periods Length of longest dry period Did extreme event occur during a dry period?

Climate Variability vs Climate Change Card Game Example Dry period: 2 or more black cards Ace to 10 # dry periods Length of longest dry period Did extreme event occur during a dry period? 3 dry years with extreme event 4 wet years 2 dry years

Lessons Learned Climate Variability vs Climate Change Card Game Current climate and future modified climate are both highly variable Each “sample” gave a different result An increased probability of dry conditions did not guarantee dry conditions Several “samples” are needed to understand possible trends

Climate models are useful for exploring possible future changes in climate Climate models don’t predict future weather or when extreme events will occur Scientists & planners use information from many models to make decisions IPCC modeling development showing their improvements over the years. Better parameterizations since the 70’s, but even those crude models got some of it right for CO2 forcing. 70’s, 80’s, various ASSESSMENT Reports versions.

Climate models are used to explore possible future changes in climate. IPCC modeling development showing their improvements over the years. Better parameterizations since the 70’s, but even those crude models got some of it right for CO2 forcing. 70’s, 80’s, various ASSESSMENT Reports versions. Global Climate Model Estimate of Future Greenhouse Gases

Which climate models are best for CA water resources? There is no “best” climate model for CA Several global climate models represent climate processes important to CA well DWR uses information from several of these climate models to assess potential impacts

Climate Change Technical Advisory Group CCTAG meets quarterly. Info available on website. http://www.water.ca.gov/climatechange/cctag.cfm

Downscaling Converting Global model output into Regional or Local information This demo illustrates both resolution and downscaling. Show lego blocks from goodie bag. First show MegaBlock. Say rectangle represents CA at a course resolution. Each bump represents a model node where data is available. Can’t see Sierra or coastal range. Then switch to the set of Duplo blocks. There are more bumps indicating that data is available in more locations. The raised section represents the Sierra Nevada (be sure to orient the blocks correctly). Then show the lego block set. The small bumps represent even more locations where data are available. The higher block set is the Sierra and the other raised block area is the coastal range. Indicate that if these topographic features aren’t represented in a Global Climate Model (less likely with newer models), downscaling tries to make adjustments that take those types of local effects into account. Set blocks on a table/podium for viewing or pass them around the class. More specific, Higher resolution, More useful Demonstration: Climate Modeling and Legos Images from Mike Dettinger, Scripps/USGS 19

Projected CA Climate Changes Temperatures OBSERVED Summer warming higher than winter Interior warming greater than coastal/marine Nighttime warming has exceeded daytime in last few decades PROJECTED Heat wave incidence: more frequent, intense, durable Precipitation change uncertain Probably will remain extremely variable Consensus of newest models shows little change in Northern Cal Continued potential for extremely heavy events Global climate models produce results that are two course to be useful at the local or even statewide scale. We have to take the GCM information and downscale it to a useful level. Mention database of downscaled projections

Future Climate Projections National Center for Atmospheric Research http://www.youtube.com/watch?v=U9YAzrmeOGI Show second movie clip.

Changes in societal values UNCERTAINTY Climate Projections Population Water cycle changes Future water demands Technology changes $$$$ Sea level rise Ecosystem response Adaptation Strategies Land use changes Changes in societal values

Take Home Message Global climate models explore possible future climate conditions, they don’t predict specific future conditions or events DWR uses results from several global climate models to evaluate possible impacts of climate change on California’s water resources Current modeling indicates that climate change with increase stress on water resources systems

Thank You! Jamie.Anderson@water.ca.gov

Extra Slides

Physical Processes and Feedback Mechanisms Mercury Venus Earth Mars Venus because it has clouds Which planet is the warmest?

Physical Processes and Feedback Mechanisms How quickly does the Earth’s temperature react to increases in GHG concentrations? www.ecy.wa.gov/climatechange/whatis.htm

Greenhouse Gases in Climate Models Colored lines – greenhouse gas emissions scenarios Black dots – actual greenhouse gas emissions Fossil Fuel CO2 Emissions Greenhouse Gases in Climate Models Actual greenhouse gas emissions were higher than future scenarios used in 2007 IPCC studies Future greenhouse gas concentrations cover a higher range in 2013 IPCC studies Called RCP or Representative Concentration Pathway Number at end indicates radiative forcing value at the end of the century 2.6, 4.5, 6.0, and 8.5 W/m2 in 2100 Concentration CO2 equivalent from all greenhouse gases RCP8.5 RCP6 RCP4.5 RCP2.6 IPCC = Intergovernmental Panel on Climate Change, RCP = Representative Concentration Pathway

Warming is substantial in climate projections July Temperature Change 14 GCMs X 3 RCP emissions Scenarios IPCC 5th Assessment (CMIP5) models Cayan et al. Scripps/USGS 2013

Spatial Scale and “downscaling” Regional features such as the Central Valley and Sierra Nevada are visible after downscaling Global climate models produce results that are two course to be useful at the local or even statewide scale. We have to take the GCM information and downscale it to a useful level. Mention database of downscaled projections IPCC = Intergovernmental Panel on Climate Change

Least total relative error  Most total relative error What model best represents California’s historical climate? Adapted from Cayan 2013 No clear best model, depends on criteria Models that didn’t perform well in SW US may be excluded in future analysis Ranking Global Climate Models by Rupp and Mote using a multi-factor historical climate evaluation scheme for the Southwestern United States High relative error in SW United States Evaluation Parameter Relative Error Low relative error in SW United States The point of showing this slide is to show that a simple question “what is the best model” has a very complicated answer. Models can be ranked relative to each other for a single metric, but when multiple metrics are used, there is no clear winner. Each model has its own strengths and weaknesses. No model or models have been shown to be superior for CA, however it looks like some don’t simulate the processed important to CA well, and the CCTAG will likely recommend dropping those models from our analyses. Climate Models Least total relative error  Most total relative error Figure 19. Relative error of the ensemble mean of each metric for each CMIP5 GCM. Models are ordered from least (left) to most (right) total relative error, where total relative error is the sum of relative errors from all metrics, excluding the diurnal temperature range (DTR) metrics. For 7 GCMs, the diurnal temperature range (DTR) metrics were not calculated (white squares).

Building Coalitions SCRIPPS Institute of Oceanography U.S. Geological Survey Lawrence Livermore Lab Santa Clara University Lawrence Berkeley Lab UC Davis I made this slide 10 years ago when we were tasked with getting to know the climate change community. DWR has done a great job connecting with the climate change community. California Energy Commission UC Berkeley

Climate Change Information Climate Information Providers: “Here’s the info… use it wisely.” Decision-Makers: “Keep it simple.” Decision Requirements Emission Scenarios Climate Change Information Who is involved? What do they want? Climate Simulations Modeling to support Decision Making This is so complex that we need outside guideance. Land Cover Changes Spatial Downscaling Modelers and Data Analyzers: “Keep it manageable.” Societal Changes Adapted from Levi Brekke USBR

Future Uncertainty- Atmospheric Composition

Other Future Uncertainties Population Technological Development

That’s a lot of Uncertainty… What do we really know? Global Climate Models (GCMs) from the 90’s did a reasonable job predicting warming for the last 20 yrs. GCMs also predicated the increase in extreme events for last decade. At the global and super-regional scale, these models do a good job GCM’s aren’t designed to provide more local information, so an additional step called downscaling is needed GCMs give us information that improves our future decision making capabilities Model in gray Data in red/blue Model in gray/black Data in red