Transcript DAYMET vs prism

Evaluating gridded
observations:
questions from the
NC-CSC
NCPP “QED”Workshop
24 August 2013
Boulder, CO
Andrea J. Ray, Ph.D.,NOAA Earth System Research Lab
Thanks to: P. Shafroth, L Perry, Amy Symstad, J. Morisette, North Central Climate
Science Center
1
Many different “gridded observational datasets”
•
•
•
•
•
•
•
PRISM
CRU (Climate Research Unit, Univ East Anglia)
Daymet
WorldClim
Maurer BCCA & BCSD
Rehfeldt – USFS
NARR = NOAA/NCAR North American Regional Reanalyses
• Compare results from different p?
• Compare results from different
Riparian corridors (Shafroth, Perry,
Hay)
• Research Question: Will warming decouple the timing of
seed dispersal from spring floods?
• Cottonwood & willow seed dispersal typically occurs during
or just after snowmelt peak flows; flowering determined by
accumulated chilling/forcing temperature units
• Seed dispersal during or just after the spring flood =
• exposed, bare, moist soil in the recruitment band
• Too high: dessicate
• Too low, scoured away
Too high – drought stress
• “just right”
Receding
spring flood
Recruitment band
Too low – ice and flood scour
Base flow
Riparian corridors
Need to calculate “chilling” and “forcing” units for budburst from temperature
– Need daily data, not (readily) available from PRISM, so choose DAYMET
– Streamflow studies using PRISM and Maurer
– How does DAYMET compare to PRISM?
– How will streamflow results calculated with PRISM compare?
Streamflow
based on
PRISM
Phenology
based on
Daymet
????
4
Daymet vs PRISM
• Daymet generally warmer,
but not consistently across
these sites
WY
NE
CO
Mean Monthly Runoff in the
South Platte 1982-1999
EPA logo
Fort Collins, Colorado
March 7, 2013
Gridded observational data
• Multiple gridded
observational datasets
exist
• They’re all slightly
different representations
of reality
• Perry found a nearly 2C
difference between
DAYMET and PRISM for
her sites (not shown)
• Solution? Plot them for
your area, compare to
your data – know what
you’re working with
• Related question: what if
your observations don’t
match?
BLM Wyoming Basin Rapid Ecoregional Assessment
Overall objective: to develop “Reasonably Foreseeable Climate
Scenarios” (RFCS), based on analysis and comparison
among several climate projection datasets and to compare
the RFCS to the historic period for the region
•
Support development of risks of climate change to Wyoming
ecoregions, conservation elements and key plant and animal
species
•
Two year time line for report, take advantage of existing
studies, little funding for new work.
– BLM-WY wants to take advantage of ecological modeling
done by Rehfelt et al (USFS) done based on their own
statistical downscaling
– National BLM requires consideration of Hostetler
dynamical downscaling
•
Question: How do the statistical downscaling by Rehfeldt and
the dynamical downscaling by Hostetler
–
Start with how Rehfeldt’s gridded obs compare to Hostetler’s simulation
Questions on “gridded observational datasets”
• Need consistency – or at least an understanding of bias to
integrate results from different parts of studies?
• Riparian forests example
• Need consistency – or at least an understanding of bias
compare existing results from different studies being
aggregated into an assessment
• Wyoming Basin REA example
Challenge: Comparing results from different GCMs &
their downscaling
• Many ecological and hydrological studies published – the basis for work like REAs
• Each have made choices, sometimes arbitrary -- but how to compare results
• WY REA, for example wants to use Rehfeldt, but he downscaled different GCMs from
Hostetler, which they’re required to consider
• Solution?
• We’re comparing Hostetler & Rehfeldt downscaling for particular variables
• If the variables themselves have differences, e.g. warmer/cooler/drier, we’ll have
documented the source
• Another solution: Consider products that have been used/evaluated in many projects
• Statistical products:
“Maurer” – the basis for the Reclamation “SECURE Water Act” report,and extensively analyzed
– IMHO the gold standard for now – downscaled many GCMS and ensemble members/GCM;
has been run thru a hydrologic model, so available for hydroclimate variables; already being
used in DOI policy & planning. Available from several portals, with visualization tools Climate
Wizard, GeoDataPortal
Other options: Hayhoe’s downscaling to stations & other products, used in the National
Climate Assessment; WorldClim – widely used, less evaluated
• Dynamical products covering North America
NARCCAP, North American Regional Climate Change Assessment Program,
http://www.narccap.ucar.edu. IMHO the gold standard for now – 6 GCMs and multiple RCMs;
used in the 2013 National Climate Assessment
 Other options: Hostetler, aka “USGS Dynamical downscaling,”
http://pubs.usgs.gov/of/2011/1238/; http://regclim.coas.oregonstate.edu; caveat: only 3 GCMs,
1RCM; BLM requiring its use in REAs*
Down-scaled climate simulations -- recent conditions
EPA logo
Fort Collins, Colorado
March 7, 2013
Delivering climate
projections on
regional scales to
support adaptation
planning: ESRL/PSD
activities
Climate Predictions Applications Workshop
24 April 2013
Logan, UT
Andrea J. Ray, Ph.D., Jeffrey T. Morisette, Dennis Ojima
NOAA Earth System Research Lab, NC-CSC, CSU.
Thanks to: Amy Symstad, Donimique Bachelet, P. Shafroth, L Perry, Max Post Van
der Berg, Brant Liebmann, Joe Barsugli, Jeff Morisette, Dennis Ojima
14
EXTRAS
“Tiles”
• Ecologists
• Edge effects
• Solution????
“DAYMET vs prism”
• Can’t copy laura’s figure?
• Edge effects
• Solution????
“DRAFT”
• Ecologists
• Edge effects
• Solution????
General questions/needs across the projects
 Climate projections & gridded observational data are widely available from
multiple sites
 However, little consistent information on evaluations, guidance on use, or “translational information”
 Give them the same critical eye as ecological data you use – PLOT & explore the data just as
you would biological data {GRAPHIC from daymet and from Laura}
 Climate scientists need to do a better job of explaining what different products should and
shouldn’t be used for.
 Results of objective & quantitative evaluation
 Narratives, which may include qualitative and quantitative aspects of data use; e.g., expert guidance
on the suitability of the data for an application; also narratives that provide summary information of how
the climate has changed or how it will change
 Guidance on appropriate uses & interpretation
 Characterize & interpret uncertainty
 “Time series” data into the future, using the projections as “predictions”
 The sequence of events matters for a lot of ecological studies -- Neither ensemble means or simply
selecting one or a few runs is a good choice
 GRAPHIC
Hostetler/USGS Dynamical Downscaling – being used a lot in ecological studies
 how do the 3 GCM selected runs compare {GRAPHIC}
 Consider this for any GCMS your project is using
 Comparing results among available analysis projects
Ecologist’s questions
Scale – temporal and spatial
• Downscale to very fine resolution, 1km – but is the fine scale actually adding
value? Meaningful?
• Bioclim indices – widely used, but do the 30 year averages “wash out” extremes
and variability that’s an important feature in the
• Climate often used“uncritically”
• Table with hostetler, rehfeldt, mauerer, MACA/Abatzoglu,Stamm etc and whith
models downscaled