Numerical Simulations of Snowpack Augmentation for Drought Mitigation Studies in the Colorado Rocky Mountains

1
Numerical Simulations of Snowpack Augmentation
for Drought Mitigation Studies in the Colorado
Rocky Mountains San Diego, CA January 12,
2005 Colorado WDMP Project Overview Presented by
Curt Hartzell, CCM, Project Consultant Joe
Busto, Colorado Water Conservation Board Ray
McAnelly, Colorado State University Gustavo
Carrio, Colorado state University PI for CSU
Research Team Dr. Bill Cotton
2
INTRODUCTION The Colorado WDMP research project
was joined with the Denver Water Departments
operational cloud seeding program in the central
Colorado Rocky Mountains for the 2003-2004 winter
season.
DWDs cloud seeding program was operated by
Western Weather Consultants, LLC - Larry
Hjermstad
Up to 56 ground-based seeding generators were
used to release silver Iodide over a mountainous
target area of about 3700 sq. km.
WWC seeding generator site V1 (Ellison),
elevation 7,088 ft, looking SE
3
GOAL To provide a physical evaluation of the
operational cloud seeding using the
well-established Colorado State University
Regional Atmospheric Modeling System (RAMS), with
a fine 3-km horizontal grid spacing covering the
entire seeding area.
RAMS 3-km grid with target area boundary, towns,
seeding generator locations, Snotel sites
4
OBJECTIVES Use RAMS to develop a better
understanding of the transport and dispersion of
seeding materials and to provide guidance as to
what meteorological conditions are most favorable
for augmenting snowfall from orographic clouds
over the Colorado mountains.
ESRIs ArcView 3D Analyst was used to create a 3D
model of the region immediately surrounding and
including the cloud seeding target area (looking
northwest)
5
PROJECT WEBSITE CSU implemented a website for the
Colorado WDMP at http//rams.atmos.colostate.edu/c
lseeding/

• menu
• Rear-time Forecasts (no-seed model runs based on
  00Z Eta data)
• Networks (towns, seeding generators, snowfall
  observational sites)
• Daily Precipitation Maps (control no-seed and
  after-the-fact simulated seed)
• Data (Snotel Snow Course Snow Water Equivalent
  24-hr precipitation)
• Evaluation Studies (simulated seeding,
  particle transport, statistical)
• GIS Maps (particle concentration and simulated
  precipitation)
• Progress Reports (required quarterly technical
  progress reports)
• Meetings Conference Calls (Colorado WDMP
  related)
• Conferences Workshops (related to the Colorado
  WDMP)
• Related Publications

6
OBSERVATIONS Snotel Sites For detailed
statistical evaluation 12 in target area 18
non-target area Also included in some
analyses 31 other non-target area sites
7
RAMS REAL-TIME FORECAST

• Real-time no-seed model runs based on 00Z Eta
  initialization data were posted on CSU website.
• Model output on website were used by DWDs
  seeding contractor for daily cloud seeding
  operations.
• 2-hr forecast products through the 48-hr forecast
  period provided guidance for the seeding
  operations.
• Evaluation of model performance was done
  throughout the 2003-2004 winter operational cloud
  seeding period.
• Simulated no-seed precipitation over-prediction
  bias and a low-level warm temperature bias were
  noted.
• Model fixes were implemented in mid-February
  2004.
• After-the-fact control no-seed simulations were
  rerun for 152 days (November 2003 - March 2004).

8
SIMULATED PRECIPITATION OVER-PREDICTION BIAS
Simulated 24-hr precipitation from the daily
control runs was used to establish no-seed
(control) precipitation for individual events
and monthly and seasonal totals. 30 operational
cloud seeding days were selected from November
2003 through March 2004 for evaluation
studies. Comparison with 61 Snotel observations
shows that the forecast runs generally simulated
the spatial distribution well, but with an
over-prediction bias for precipitation amounts
(factor of 1.88).

• Possible sources of model precipitation biases
  are
• Inadequate resolution of atmospheric dynamics and
  terrain, especially when embedded convection is
  prevalent.
• The Meyers formula used in the project for
  crystal concentrations over-predicts
  concentrations of natural ice crystals.

9
SIMULATED PRECIPITATION OVER-PREDICTION BIAS
Plot of 30-day total control-run precipitation
extracted at 61 Snotel sites vs. the 30-day
total observed precipitation.
10
LOW-LEVEL WARM TEMPERATURE BIAS

• Real-time use of RAMS output revealed a
  significant warm-temperature bias from the
  surface to above mountaintop level.
• Model fixes implemented February 14, 2004 reduced
  but did not eliminate this warm bias. Two case
  studies were subsequently done to evaluate the
  magnitude of the warm temperature bias . The
  model forecast temperatures at the 700 mb level
  were compared to 3 sounding stations in the 12km
  Grid 2 (DNR, GJT, RIW)
• Case 1 (040129.00) - model had about a 1.0C
  bias
• Case 2 (040205.00) - model had about a 1.8C
  bias
• The case studies suggest that there is still a
  slight warm temperature bias of 1 to 2 C in the
  prime seeding levels near mountain ridge tops.

11
RAMS EXTENDED TO INCLUDE SEEDING EFFECTS
Post-season RAMS seed simulations were preformed
for each of the 86 days on which seeding
operations were conducted during November 2003 -
March 2004. Simulated sources of silver iodide
(AgI) were at specified low-level model grid
points in accordance with the timing and
magnitude of AgI release at each seeding
generator as recorded in WWCs operational
seeding logs. The AgI was treated as a second
predictive IFN field with its own activation
characteristics. All other aspects of the seed
simulation runs were identical to the control
(no-seed) simulation design.
12
Seeding Activation Curve for AgI
WWC used a 4 silver iodide (AgI) solution with
sodium iodide (NaI) as a carrier in acetone along
with about 1 moth balls to improve nuclei
activation between -2.5C and -8.0C
13
COMPARISON OF SIMULATED NO-SEED SEED
PRECIPITATION
The differences in 24-hr precip between the seed
and no-seed control simulations are within 1
averaged over the target area. The daily
difference patterns are organized in bands,
resembling longitudinal rolls that extend upwind,
downwind, and laterally of the seeding target
area. Dr. Cotton commented that these roll
patterns suggest a possible weak dynamic response
to seeding.
Difference in total precip (Seed-Control) for the
30 selected days. Seeding generator sites are
plotted.
14
COMPARISON OF SIMULATED NO-SEED SEED
PRECIPITATION

• The small differences between RAMS seed and
  no-seed simulated precipitation could be because
• The background CCN and IFN concentrations are
  unknown therefore, the results are at the mercy
  of specified background concentrations.
• The model under-predicts supercooled liquid
  water content in the lower portion of clouds over
  the target area, thereby reducing seedability.
• An unforeseen dynamic response that appears to
  result in large areas of slightly suppressed
  precipitation in the target area and small
  regions of slightly enhanced precipitation.
• The low-level warm temperature bias results in
  delayed AgI nuclei activation, fewer activated
  nuclei, and less time for crystals to grow and
  snow to fall in the target area.
• The transport and diffusion of seeding material
  from generator sites is getting into the clouds
  too far downwind of the generator sites.

15
LAGRANGIAN PARTICLE DISEPSION ANALYSIS
The Lagrangian particle dispersion analysis for
the Colorado WDMP project has not been completed.
The 30 seeding days selected for evaluation
span the full range of meteorological regimes
that are conducive to snowfall in the project
target area. Six of the 30 cases are being
analyzed in detail to determine if and how
efficiently the seeding material is getting into
the simulated clouds under different wind and
stability regimes.
Example of a seeding simulation - available 20-hr
AgI concentration in lowest model layer.
16
MULTIPLE REGRESSION BLOCK PERMUTATION ANALYSIS
The MRBP analysis code was provided by Dr. Paul
Mielke, and modified by Gustavo Carrió at CSU for
specific use on the Colorado WDMP project. This
code computes the test statistic and associated
P-value of a randomized block experiment. Model
skill for predicting precipitation was evaluated
for the 30 selected days using MRBP statistics.
Analyses were conducted for observed daily
precipitation at 30 Snotel sites vs. 24-hr
control run precipitation extracted from those
locations, and for observed vs. seed run
simulated precipitation at the same. Mielkes
comments on the results of the MRBP analysis are
that the model is describing the no-seed and seed
simulations equally well. While the signal of
the fits is strong (all P-values about 1.0E-6 or
less), the agreement measures are not outstanding
(all fall between 0.18 and 0.26).
17
SUMMARY
The Colorado WDMP project applied the CSU RAMS to
simulate cloud seeding operations supported by
the DWD. Daily RAMS forecast simulations
initialized with 00Z Eta data were posted on a
CSU website developed for the research project.
RAMS was extended to include seeding effects
except for the added AgI release, the seed and
control simulations were identical. Model
precipitation biases are much greater than
differences between seed and no-seed
amounts. Model low-level warm temperature biases
were estimated to be about 1 to 2 C near
mountain ridge tops (700 mb level). 24-hr seed
minus control no-seed simulated precipitation
differences are consistently small.
18
RECOMMENDATIONS FOR FUTURE RESEARCH
Joined research and cloud seeding operations
should include measurements of background IN,
CCN, giant CCN concentrations. RAMS as used on
the operational winter orographic cloud seeding
project needs the development of a sub-grid
cumulus scheme.

• Colorado WDMP research areas needing further
  study
• Defining meteorological conditions favorable for
  cloud seeding.
• Lagrangian particle dispersion analyses.
• Seed vs. no-seed simulated precipitation
  difference analyses.
• RAMS forecast simulated precipitation
  over-prediction biases.
• RAMS low-level warm temperature biases during
  cold conditions.
• Add 4th grid (e.g. 1km) for seed control runs.
• Understand and correct RAMS deficiencies.

19
The End