Winter Weather

Background

Figure 1. A map of Wyoming with coarse representation of topography and major river basins. Yellow areas denote the five mountain ranges under recent study related to winter orographic cloud seeding programs: Medicine Bow, Sierra Madre, Wind River, Salt River/Wyoming, and Big Horn Ranges.  Current NCAR studies focus on the Bighorns, Medicine Bow, and Sierra Madre mountain ranges.
Figure 1. A map of Wyoming with coarse representation of topography and major river basins. Yellow areas denote the five mountain ranges under recent study related to winter orographic cloud seeding programs: Medicine Bow, Sierra Madre, Wind River, Salt River/Wyoming, and Big Horn Ranges.  Current NCAR studies focus on the Bighorns, Medicine Bow, and Sierra Madre mountain ranges.

Since the completion of the Wyoming Weather Modification Pilot Project (WWMPP) in 2014, the Wyoming Water Development Commission (WWDC) has broadened its focus for cloud seeding research and operations to encompass additional mountain ranges in the State (see Figure 1).  The WWMPP had focused on research in the Medicine Bow, Sierra Madre, and Wind River Ranges. In May 2015, the WWDC awarded two new studies to NCAR.  The two NCAR studies were a Final Design and Permitting study to design an operational cloud seeding program in the Medicine Bow and Sierra Madre Ranges, and a Feasibility Study to assess the potential for cloud seeding in the Big Horn Range in north central Wyoming.  This report summarizes the key accomplishments during FY17 for each of these two studies (covering three mountain ranges, the Bighorns, Medicine Bow, and Sierra Madre, see Figure 1), as well as plans for the next year.

FY2017 Accomplishments

Medicine Bow/Sierra Madre Final Design and Permitting Study

Given the results of the WWMPP, the WWDC decided to fund a study to develop a final design for operational cloud seeding in the Medicine Bow and Sierra Madre Ranges in southeast Wyoming.  NCAR/RAL leads this program with collaboration from WMI and Heritage Environmental Consultants (HEC). This project began in June 2015 and highlights from the analyses conducted in FY17 are presented below.  The results were submitted in a draft report to the WWDC in early FY17.  After an external review, revisions to the report were made.

A study to evaluate the cloud seeding model by simulating all 118 Randomized Statistical Experiment (RSE) cases from the WWMPP was completed in FY17. This effort, which began in FY16, established a design plan featuring an ensemble modeling approach to better address uncertainties in the model. A set of 96 total simulations were conducted in FY17 for each RSE case: 24 served as control simulations and 72 as seeding simulations. The 72-member seeding ensemble was used to address the seeding process uncertainties in the model.  This ensemble resulted in the creation of 8,496 seeded simulations and 2,832 non-seeded simulations.  The members were designed to encompass uncertainties arising from the use of different initialization data sets, boundary layer schemes, cloud condensation nuclei and ice nuclei background levels, silver iodide (AgI) activation and removal functions, as well as spatial and temporal uncertainties of the simulated precipitation.

Figure 2. Distribution of RSE case paired errors by subtracting the ensemble mean of the 24 ensemble members run using each re-analysis from the 4-hour observed precipitation accumulation at each gauge. Each box and whisker plot represents the paired differences of the 118 RSE Experimental Units (EUs). This plot provides an example of the results for the target snow gauge in the Sierra Madre (HY = “Highway 47” gauge site). The mean difference is represented by the horizontal red line.   Results are shown in each plot for the three different initializations (CFSR, ERA and NARR) and for seeding onset lag times of 30 and 60 minutes (CFSR30, CFSR60, etc.).  The statistical design specified that the lag to estimate seeding effect at a gauge site would be 30 minutes from the start of seeding.
Figure 2. Distribution of RSE case paired errors by subtracting the ensemble mean of the 24 ensemble members run using each re-analysis from the 4-hour observed precipitation accumulation at each gauge. Each box and whisker plot represents the paired differences of the 118 RSE Experimental Units (EUs). This plot provides an example of the results for the target snow gauge in the Sierra Madre (HY = “Highway 47” gauge site). The mean difference is represented by the horizontal red line.   Results are shown in each plot for the three different initializations (CFSR, ERA and NARR) and for seeding onset lag times of 30 and 60 minutes (CFSR30, CFSR60, etc.).  The statistical design specified that the lag to estimate seeding effect at a gauge site would be 30 minutes from the start of seeding.

An ensemble approach allows one to better account for initial condition and model biases and random errors in the model simulations.  A prerequisite to using a model, however, is that the simulations reasonably represent reality.  As such, an inter-comparison between the model ensemble simulations and the WWMPP snowgauge data per RSE case (or Experimental Unit, EU) showed reasonable agreement in terms of the model mean and median errors being close to zero, with the distribution of errors in the paired comparison symmetric about zero error in most cases (see Figure 2). The results of this study will be finalized in FY18.

Big Horn Mountains Feasibility Study

The WWDC funded a feasibility study to assess the potential for cloud seeding in the Big Horn Mountains in north-central Wyoming.  NCAR/RAL leads this program in collaboration from private sector partners, WMI and HEC. This project began in June 2015 and highlights from a few tasks completed in FY16 are provided below. A draft report summarizing the results of this study was submitted to the WWDC in August 2016.  The report underwent external review, was subsequently revised, and a final report was submitted to the WWDC in March 2017.

FY2018 Plans

  • Complete the model evaluation of the WWMPP RSE study and submit a manuscript on these results
  • Submit Medicine Bow/Sierra Madre Final Design and Permitting Study final report

Idaho Power Project

Background

Figure 3.  Map of the Snake Watershed in Idaho (large red outline) and existing ground generator and observational facility sites on a map of terrain height (m; color shading).  The Payette River Basin, Boise Basin, and Woods Basin target areas are located (north to southeast) in the western Snake watershed, north of Boise and each is also outlined in red.  The Upper Snake River Basin target area is located in Eastern Idaho, also outlined in red.  Ground generator locations are identified as circle and triangle symbols, and color-coded as red-filled circles (Payette, Boise, Woods), blue (north Eastern Idaho), and green (south Eastern Idaho).  The circles are Idaho Power owned generators, and the triangles are generators operated by Let it Snow. Grey lines indicate flight tracks used by the cloud seeding aircraft. Black squares indicate the location of microwave radiometers, black x’s for atmospheric sounding sites, open blue and red circles are high-resolution snow gauge sites (located in the western Snake Basin).
Figure 3.  Map of the Snake Watershed in Idaho (large red outline) and existing ground generator and observational facility sites on a map of terrain height (m; color shading).  The Payette River Basin, Boise Basin, and Woods Basin target areas are located (north to southeast) in the western Snake watershed, north of Boise and each is also outlined in red.  The Upper Snake River Basin target area is located in Eastern Idaho, also outlined in red.  Ground generator locations are identified as circle and triangle symbols, and color-coded as red-filled circles (Payette, Boise, Woods), blue (north Eastern Idaho), and green (south Eastern Idaho).  The circles are Idaho Power owned generators, and the triangles are generators operated by Let it Snow. Grey lines indicate flight tracks used by the cloud seeding aircraft. Black squares indicate the location of microwave radiometers, black x’s for atmospheric sounding sites, open blue and red circles are high-resolution snow gauge sites (located in the western Snake Basin).

The Idaho Power Company (IPC) conducts a winter cloud seeding program to augment snowfall along the Snake River Basin and its tributaries for hydroelectric generation.  The program has been focused in Payette River basin in western Idaho and the upper Snake River system in eastern Idaho (Figure 3), and has recently expanded into the Boise and Wood basins in western Idaho.

In FY17, RAL completed a numerical modeling “Phase Six” study to provide real-time and retrospective model-based guidance on the effectiveness of cloud seeding using ground generators and aircraft tracks.  The primary goal of Phase Six was to continue the development of a real-time cloud seeding forecast guidance system using the WRF model developed in Phases Three–Five, and to conduct a field project that was funded by the National Science Foundation (NSF) in partnership with IPC, entitled “Seeded and Natural Orographic Wintertime clouds: the Idaho Experiment”, or SNOWIE. Since the completion of Phase Six, a Phase Seven study was begun with goals to continue research and improvements to the cloud-seeding module, especially utilizing SNOWIE data.

FY2017 Accomplishments

In 2017, RAL provided real-time and retrospective model-based guidance on the effectiveness of cloud seeding using ground generators and aircraft tracks.  Components of this effort included:

  • refining the real-time cloud-seeding decision algorithm;
  • collaborating with the University of Arizona (UofA) to incorporate the cloud-seeding module into the UofA real-time WRF model;
  • running a research version of WRF on the UofA computing cluster that provided tailored precipitation and cloud-seeding forecasting relevant to the Idaho Power cloud-seeding operations during the 2016-2017 winter season, along with updates to the web-based display;
  • completing a prototype new case-calling algorithm that uses fuzzy logic and running it alongside the real-time modeling system run during the 2016-2017 season;
  • comparing model simulation results with observations when available, such as measurements of silver in the snow;
  • testing a bin microphysics seeding parameterization against the seeding parameterization in the bulk microphysics scheme;
  • analyzing atmospheric river events that impact Idaho; and,
  • conducting the SNOWIE field project
Figure 4. Box and whisker plots of the simulated accumulated Ag areal density (10-12 kg m-2) at all sampling sites (DWXX) for two cases: 29 January 2016 (top) and 18 February 2016 (bottom).  The mean or averaged of the model results are represented by the red circle with cross inside. The converted observations from two snow pit profiles are indicated by blue triangles.
Figure 4. Box and whisker plots of the simulated accumulated Ag areal density (10-12 kg m-2) at all sampling sites (DWXX) for two cases: 29 January 2016 (top) and 18 February 2016 (bottom).  The mean or averaged of the model results are represented by the red circle with cross inside. The converted observations from two snow pit profiles are indicated by blue triangles.

 

Given that the dispersion of AgI is a key process that determines the targeting efficiency and ultimately the seeding effect on the ground, improving how the model simulates this process is of paramount importance for the model to realistically simulate cloud seeding impacts. Therefore, simulations were performed to understand the behavior of AgI dispersion in cases with trace chemistry (i.e. silver) measurements in the snow for comparison with the model simulations. The trace chemistry sampling experiment to measure silver in the snow was carried out by Boise State University. The measurements were collected in target and downwind areas to serve as validation data for the AgI dispersion simulated by the model. However, the original AgI cloud seeding parameterization (ASPEN) did not consider some physical processes related to AgI removal that can impact the simulated silver deposition and downwind seeding effect. To address these issues, ASPEN was modified to include the missing physics: AgI self-coagulation, AgI scavenging by precipitating particles, and AgI dry deposition due to surface roughness and turbulence and these new updates were evaluated compared to the silver measurements in snow. An example analysis comparing the silver in snow measurements with the model simulations is presented in Figure 4.

Figure 5. Precipitation accumulation (inches) from 1500 to 1800 UTC for each control (no seeding) model simulation compared to SNOTEL observations on 12 February 2014.
Figure 5. Precipitation accumulation (inches) from 1500 to 1800 UTC for each control (no seeding) model simulation compared to SNOTEL observations on 12 February 2014.

NCAR collaborated with the University of Pecs to develop a bin microphysics scheme coupled with the ASPEN (UPNB-ASPEN, Geresdi et al. 2017). Three seeding experiments were simulated in FY17 using actual seeding times in a pre-SNOWIE case (1612 to 1622 UTC 12 February 2014) to compare the results of the UPNB-ASPEN versus the bulk ASPEN.  The bin scheme was used to simulate one of the experiments (BIN). The Thompson-Eidhammer bulk scheme coupled with the ASPEN (TE-ASPEN) was used to simulate the other two experiments (BULK and BULK_HYD). The experiment ‘BULK_HYD’ used boundary conditions that include hydrometeors from the outer domain outputs. The idea of including the hydrometeors in the boundary condition is to make the pre-existing clouds outside of the inner (nested) domain available to advect into the inner domain, which is supposed to be more realistic than without hydrometeors as part of the boundary conditions. The results of the different schemes were compared and an example of the different precipitation produced are shown in Figure 5.  This analysis and collaboration with University of Pecs will continue in FY18, with focus shifting to cases from SNOWIE.

The 12 February 2014 pre-SNOWIE case was determined to be influenced by an atmospheric river (AR) impacting the western U.S. coast.  As such, the characteristics of atmospheric conditions and precipitation when ARs impact Idaho were investigated.  ARs were determined to frequently impact the western coast of the U.S., however only a small subset had an impact as far inland as Idaho.  Conditions during ARs that impacted Idaho, however, did exhibit some unique features compared to the normal winter climatology.  Primarily, the precipitation that falls during ARs is disproportionately high compared to the number of precipitating ARs that occur.

The SNOWIE program was funded by NSF in the spring of 2016 and conducted between January-March of 2017.  The project was led by Principal Investigators (PIs) Dr. Jeff French (Univ. Wyoming) and Dr. Sarah Tessendorf (NCAR/RAL), with involvement from Drs. Bart Geerts (Univ. Wyoming), Bob Rauber (Univ. Illinois), and Katja Friedrich (Univ. Colorado), Roy Rasmussen (NCAR/RAL), and Lulin Xue (NCAR/RAL), and IPC staff as co-PIs.   RAL was heavily involved in the planning of this field project, from the proposal writing to planning for the operations and deploying instruments (i.e., Geonor and ETI snow gauges, radiometers, a microwave rain radar, and snow depth sensors) to the field (Figure 6-Figure 7). The project aims to evaluate ground and airborne cloud seeding using physical and numerical modeling approaches, as well as to validate the cloud seeding module.

Figure 6. Topography map of SNOWIE ground-based instrumentation sites.  The Payette River Basin is overlaid in dark gray.
Figure 6. Topography map of SNOWIE ground-based instrumentation sites.  The Payette River Basin is overlaid in dark gray.

The SNOWIE field campaign was very successful in that it collected data in 24 Intensive Observing Periods (IOPs), including some that had very clear evidence of impacts from cloud seeding. After the conclusion of the SNOWIE field campaign, data from the ground-based instruments was consolidated and the quality control (QC) process was begun.  This process was completed and the final data set uploaded to the data archive hosted by NCAR’s Earth Observing Lab (EOL) in September 2017.  NCAR led this process for the radiometer and snow gauge data from SNOWIE, as well as assisting with the sounding data and the Snowbank microwave rain radar data.

Figure 7. Photo of the Silver Creek Plunge precipitation gauge site (top) and the Snowbank radiometer site (bottom) deployed by NCAR.
Figure 7. Photo of the Silver Creek Plunge precipitation gauge site (top) and the Snowbank radiometer site (bottom) deployed by NCAR.

 An article summarizing the SNOWIE field campaign is being written, led by Dr. Sarah Tessendorf, for submission to the Bulletin of the American Meteorological Society (BAMS).  Another manuscript, led by Dr. Jeff French, is also being prepared that provides the first evidence of seeding effects on ice nucleation and precipitation formation as observed in a few of the IOPs. These articles will be submitted in early FY18.

Figure 7. Photo of the Silver Creek Plunge precipitation gauge site (top) and the Snowbank radiometer site (bottom) deployed by NCAR.

Work is currently underway on Phase Seven of the IPC program. Efforts are focused on analyzing the SNOWIE field project data, conducting model simulations, and continuing to make improvements to the cloud seeding forecast guidance system, especially the new fuzzy logic version of the case-calling algorithm. Thus far, the Phase Seven effort has updated the software for the new fuzzy-logic case-calling algorithm so that it can run more efficiently in the real-time modeling system run in the 2017–2018 winter season.

FY2018 Plans

  • Run the newly created fuzzy-logic seeding case-calling algorithm in the real-time model for the 2017-2018 season, and make improvements to the code so it runs more efficiently with reduced latency.
  • Conduct data analysis and model simulations from the SNOWIE field project and collaborate with SNOWIE university PIs to perform analysis on high priority cases
  • Run simulations of cases with silver in snow samples and collaborate with Boise State University to compare model results with measurements
  • Perform detailed case study simulations and analyses to improve the cloud seeding model, using cases from SNOWIE and pre-SNOWIE
  • Publish journal papers on the major findings from these studies.