1. Gabriela Bucini & AIM Team
Climate Variables for Water Quality Modeling: Downscaling and a Weather Generator
Gabriela Bucini & AIM Team
May 27, 2015

2. IAM TEAM Asim Zia Carol Adair Ibrahim Mohammed Patrick Clemins
Lesley-Ann Dupigny-Giroux
Steve Scheinert
Christopher Koliba
Gillian Galford
Linyuan Shang
Andrew Schroth
Scott Turnbull
Jody Stryker
Brian Beckage
Gabriela Bucini
Yushiou Tsai
Arne Bomblies
Sarah Coleman
Kristen Underwood
Donna Rizzo
Justin Guilbert
Yaoyang Xu
Jonathan Winter
Peter Isles
Ahmed Hamed
Alan Betts
Scott Hamshaw

3. Study area
General circulation models

4. IAM GENERAL CIRCULATION MODELS REGIONAL CLIMATE CHANGE SCENARIOS
DOWNSCALING
GRASS & RHESSYS
HYDROLOGIC MODEL
FLOW
LimnoTech
LAKE MODEL
Integrated Assessment Model
Downscaling: from 100 km to 1 km
LAND USE MODEL
LAND USE CHANGE
SCENARIOS
WATER QUALITY
LAND USE

5. PEGASUS WORKFLOW Processing of vast amounts of data
Organizing the process
Allows users to automate multi-step computational tasks
Defines a sequence of operations. Pegasus workflow system is used to connect of the elements of the model and automate the sequence of operations. It connects the data files to thei models and runs the oprations under the various initial conditions (scenarios).
Source:
Pegasus workflow management system
University of Southern California

6. Challenge:
GCM projections only provide T and P, but our lake model needs other climate variables as well.

7. GENERAL CIRCULATION MODELS
P, Tmin, Tmax
REGIONAL CLIMATE CHANGE SCENARIOS
DOWNSCALING
P, Tmin, Tmax
GRASS & RHESSYS
HYDROLOGIC MODEL
FLOW
LimnoTech
LAKE MODEL
Integrated Assessment Model
Daily Tmin, Tmax and precipitation
LAND USE MODEL
LAND USE CHANGE
SCENARIOS
WATER QUALITY
LAND USE

8. GENERAL CIRCULATION MODELS
P, Tmin, Tmax
P, Tmin, Tmax,
solar radiation,
relative humidity,
pressure,
cloud cover
u, v
REGIONAL CLIMATE CHANGE SCENARIOS
DOWNSCALING
P, Tmin, Tmax
WEATHER GENERATOR
GRASS & RHESSYS
HYDROLOGIC MODEL
FLOW
LimnoTech
LAKE MODEL
Integrated Assessment Model
LAND USE MODEL
LAND USE CHANGE
SCENARIOS
North American Regional Reanalysis (NARR)
32 km grid resolution, daily from 1979 – 2014
WATER QUALITY
LAND USE

9. General Circulation Models
Climate downscaling
General Circulation Models
Climate
~100 Km
CMIP5 Intermediate
Downscaling
Climate
~12 Km
General circulation models, intermediately downscaled climate, and fine-resolution downscaling
We take an intermediately downscaled product called CMIP5
Fine
Downscaling
Climate
~ 1km

10. Downscaling to <1km cell size
Cell size:~ 12 km
Lake Champlain

11. Adding climatic information and spatial detail
Lapse rate relationships:
to adjust T and P for variations in elevation
Interpolation:
to increase spatial resolution of the grid
Run on Yellowstone high-performance computing system:
Goal 112 ensembles:
28 GCMs
2 interpolation methods
2 climate scenarios
RCPs: climate scenarios

12. Nov. 21, 1970
access1-0 RCP 8.5

13. GENERAL CIRCULATION MODELS
P, Tmin, Tmax
P, Tmin, Tmax,
solar radiation,
relative humidity,
pressure,
cloud cover
u, v
REGIONAL CLIMATE CHANGE SCENARIOS
DOWNSCALING
P, Tmin, Tmax
WEATHER GENERATOR
GRASS & RHESSYS
HYDROLOGIC MODEL
FLOW
LimnoTech
LAKE MODEL
Integrated Assessment Model
LAND USE MODEL
LAND USE CHANGE
SCENARIOS
WATER QUALITY
LAND USE

14. Weather generator resampling
Downscaled GCM
T and P
Daily,
NARR
T, P, u, v, RH, Pressure, Solar Radiation, Cloud Cover
Daily,
North American Regional Reanalysis (NARR)
The methods resamples data from the historical records provided by NARR reanalysis data
Generate daily weather sequences as inputs for the lake model
Time-series resampling algorithm capable of generating daily values of weather variables for future time ranges.
The sampling of historical values is based on nearest-neighbor concept. The search is conditioned on proximity to both temperature and precipitation
For each day of future projections, precipitation and temperature from downscaled data are compared to the NARR historical dataset. The most similar temperature and precipitation values (closest neighbors) are extracted. One of the P&T closest neighbors is randomly selected for the weather temporal sequence.
The set of NARR variables for that date is used as input for the lake model.

15. daily weather sequence of all climate variables
For each future daily T and P
Find a matching pair of T and P in historic NARR
Steps:
Search the historic data under two conditions:
time: near selected date
value: close T and P values
Collect a set of nearest T and P neighbors
Randomly select one neighbor
daily weather sequence of all climate variables

16. NN Searching methods
1st method: Conditional on smallest Euclidean distances of both T and P. 2nd method: Conditional on smallest temperature differences first and then on smallest precipitation differences.
How we solved it... 2 resampling methods:
2nd: we have more control over the search: better control in the selection of temperature and precipitation

17. SUMMARY
Downscaling:
Climate projections of P and T into regional fine-resolution grids
Resampling Weather Generator:
Large set of climate variables tracking P and T projections

18. THANK YOU!

19. Gabriela Bucini & AIM Team
Measuring the Climate Change Impact on Water Quality using a Weather Generator Pegasus Workflow
Gabriela Bucini & AIM Team
May 27, 2015

20. Our accomplishment:
Estimating the suite of climate variables under GCM projections by
downscaling and
sampling from historical NARR reanalysis data.
Data source: North American Regional Reanalysis (NARR)

21. Integrated Assessment Model
High Resolution Forecasting of Global Climate Change Impacts on Fresh Water Lakes:
Integrating Climate, Land-Use, Hydrological and Limnology Models.

22. Integrated Assessment Model
OVERARCHING QUESTION
How will the interaction of climate change and land use alter hydrological processes and nutrient transport from the landscape, internal processing and eutrophic state within the lake and what are the implications for adaptive management strategies?
High Resolution Forecasting of Global Climate Change Impacts on Fresh Water Lakes:
Integrating Climate, Land-Use, Hydrological and Limnology Models.
To simulate impacts of adaptive interventions on water quality

23. Interactive Land Use Transition Agent-based Model
Climate
Change
Modeling
Global
Climate
Change
Regional
Climate
Changes
Downscaling
Interactive Land Use Transition Agent-based Model
NLCD
Land use
Simulated
World File
GRASS
Forest Elaboration
Module
RHESSYS
Lake model input variables
Land Use Model
Flow
Hydrologic Model
Water
Quality
Lake
Model

24. Lake Champlain Management
Climate Change
Social &
Economic
behavior
Hydrology
Land
Use
Water
Quality
Architecture of IAM
Lake Champlain Management
Policy Decisions

25. Pegasus Workflow MGMT SYSTEM
NSF Funded since 2001 in collaboration with USC + ISI + HTCondor UW-Madison
Built on top of HTCondor DAGMan
(Directed Acyclic Graph Manager) is a meta-scheduler for HTCondor
Abstract Workflows - Pegasus input workflow description
Workflow “high-level language”
Python, Java, and Perl
Pegasus is a workflow “compiler” (plan/map)
Target is DAGMan DAGs and HTCondor submit files
Transforms the workflow for performance and reliability
Automatically locates physical locations for both workflow components and data
Collects runtime provenance

26. Pegasus WMS Architecture

27. Derive Lapse Rates with an Assumed Functional form and MLE
Calculate the relationship between elevation (z), temperature (T), precipitation (P), and latitude (f)
Maximum Likelihood Estimation (MLE) with station observations to find α, β, and χ

28. Yellowstone high-performance computing system
Goal 112 ensembles: 28 GCMs 2 interpolation methods 2 climate scenarios
RCPs: climate scenarios

29. A2EM Progress Background: A2EM (Advanced Aquatic Ecosystem Model)
Wind (speed, dir.)
Temp RH
Pressure
Solar Radiation
Cloud Cover
Phytoplankton growth and nutrient uptake parameters
Initial nutrients, phytoplankton, zooplankton
Bathymetry
River Inputs,
Main lake level
Not implemented in this version
Initial water levels, temp
Initial sediment nutrient concentrations, bulk density, sediment diagenesis parameters

30. A2EM Lake model
Wind (speed, dir.), Precipitation Temperature, Relative Humidity, Pressure, Solar Radiation, Cloud Cover
A2EM (Advanced Aquatic Ecosystem Model)

31. Hydrodynamic Model Calibration

32. A2EM Model Grid