Rao Kotamarthi
Science Director, Center of Climate Resilience and Decision Sciences (CCRDS)
Argonne National Laboratory
Abstract:
There is an increasing need to develop estimates of expected changes in climate impacts/hazards and understand the evolving climate risk to communities, installations, and scales where decisions are usually made about adaptation. Global scale climate models are typically having spatial resolution of 100km or more and they can’t represent the impacts at these spatial scales and are also not very good at modeling extreme scale phenomena that is typically associated with estimating risk. In this presentation, we will report on our work using a method known as dynamic downscaling over North America at a spatial resolution of 12km and 4km. The design of the ensemble and the numerical model simulations for current climate and future projections, model biases and value added will be discussed. The application of this downscaled data archive for estimating climate impacts, climate risks from these impacts at spatial resolutions that could be used for informing decisions at infrastructure and installation scales will be discussed. For example, high resolution wind data, temperature and atmospheric water vapor will serve better for drought and wildfire risk estimates, simulating tropical convective storms provide guidance on coastal flooding and coastal flood risks; high resolution precipitation will provide much better dataset for hydrological modeling (such as WRF-Hydro®) that is useful for estimating flood risks and stream flow changes. We will present the use of these impact data to generate return periods, estimate extremes event risks and their uncertainties using and approach based on Generalized Extreme Value methodology. Some recent applications of these various impact and climate risk estimates to private utility sector partners and public sector entities will be discussed.
Bio:
Dr. Kotamarthi has nearly 30 years of experience in climate research. He serves as the Science Director for the Center for Climate Resilience and Decision Sciences and as the Chief Scientist for the Environmental Science Davion at the Argonne National Laboratory. He is Principal Investigator for projects ranging from climate modeling to wind energy for the Department of Energy. He has led a team of scientists in developing climate projections suitable for addressing climate impacts, resilience and adaptation challenges at local and regional scales and its application to utilities serving power and telecommunications industries, among others. He authored a book published recently titled “Downscaling Techniques for High-Resolution Climate Projections - From Global Change to Local Impacts”, published by Cambridge University Press, UK.
Summary:
Climate change
CO2 concentrations have never increased this rapidly and are highest in 20m years
Global surface temperatures have been rising steadily since 1850 and when baselined to 1850-1900, we see the most rapid increases starting around 1950.
Simulations show that the human activities are the cause (difference between model predictions for natural vs natural+human causes)
What to do?
Mitigate: Largest industrial emitters (31% power, 19% Industry (concrete, steel, etc.), 17% transport, 8% buildings, 17% Ag and land use)
Adapt:
Risk=hazards+Vulnerability+Exposure
Cost of protecting cities can be very high
Intervene:
Carbon Dioxide removal (permanent but misses all the other pollutants
Geoengineering, radiative forcing (short-term delay in warming but doesn’t affect the key drivers)
Focus: Adaptation at various spatial scales
Installation (building, power plant, campus)
Infrastructure (sewage system, local power system)
Urban (all systems in a city)
Operations (continental power grids)
Example: nuclear power plant risk
Extreme weather accounts of significant fraction of nuclear power disruptions (e.g. hurricanes/strong storms, availability of water for cooling)
Drought frequency is increasing (cooling water)
Approaches to estimating climate change risk
Possible maximum precipitation: assumes maximum in stationary climate
Storyline approach: propose possible climate scenarios
Downscaling climate models to make concrete predictions about local impacts
Focus is on downscaling using the WRF model and climate models for boundary conditions
Have 12km resolution North American simulation, working on 4km.
We are performing flood and stream flow calculations for all Continental US at 200m spatial resolution (current and mid-century). We have completed these calculations for most of the CONUS and expect to be done in a few months.
Ensemble of multiple simulations 1995-2004
Each member is a 10 year run with different boundary conditions
Measured error between each member’s predictions and observations
Bias: systematic difference between ensemble and measurements
Uncertainty:
Variability within climate dynamics
Uncertainty about model parameters
Uncertainty about society’s emissions in the future
Ensembles allow us to identify major differences between different models and parameters
Applications
Can locally predict
Temperature peaks
Droughts (midwest vulnerable)
Precipitation intensity
Has been increasing over the past few decades in US East
Warmer air holds more water
Storms are fewer and more wet
Extreme event analysis for local disasters, safety of cities and infrastructure
Uncertainty analysis of models and scenarios to establish once-in-a-decade/century event predictions, confidence intervals
Predict streamflow in rivers, which determines floods frequency and area
Coastal flood, storm surge projections
Wildfire (midwest especially vulnerable)
Hurricane distribution
Adaptation by different neighborhoods in Chicago (power consumption, green roof, flood risk, air quality); implications for equity
Predict power use (cooling/heating degree days)
Working on developing ML-based emulator models
Reduced computational cost with similar accuracy
Can enable many localized analyses and large ensembles that can evaluate the probability of a given climate event happening with/without pollution