A Numerical Evaluation of the Impact of Operational Ground-Based Glaciogenic Cloud Seeding on Precipitation over the Wind River Range, Wyoming

Thomas Mazzetti aUniversity of Wyoming, Laramie, Wyoming

Search for other papers by Thomas Mazzetti in
Current site
Google Scholar
PubMed
Close
,
Bart Geerts aUniversity of Wyoming, Laramie, Wyoming

Search for other papers by Bart Geerts in
Current site
Google Scholar
PubMed
Close
, and
Lulin Xue bNational Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Lulin Xue in
Current site
Google Scholar
PubMed
Close
Free access

We are aware of a technical issue preventing figures and tables from showing in some newly published articles in the full-text HTML view.
While we are resolving the problem, please use the online PDF version of these articles to view figures and tables.

Abstract

This study evaluates an operational glaciogenic cloud-seeding program using ground-based generators of silver iodide (AgI), with a total of 190 seeded storms over 10 cold seasons, using the Weather Research and Forecasting Weather Modification (WRF-WxMod) scheme at 900-m grid spacing. This study examines both the quantitative change in precipitation and the ambient and cloud conditions impacting seeding efficacy. An ensemble approach is used, with differing model boundary conditions, ice nucleation physics, concentrations of cloud condensation nuclei, and boundary layer schemes. This is intended to provide an envelope of uncertainty of natural clouds and seeding impacts. The simulations are validated against radiosonde, snow gauge, and microwave radiometer observations, and the seeding impact is inferred from simulations with/without AgI seeding. The seeding-induced precipitation enhancement (“yield”) varies greatly between storms. A small portion of the cases produces the majority of the yield. Overall, the precipitation in the target area (the Wind River Range in Wyoming) increased by 1.10% ± 0.13% in the 10 years of operational seeding. This rather low fractional increase is related to the frequent seeding at unsuitable times, primarily because of low-level flow blocking. The flow and cloud structure for select cases are examined to provide better insight into the variability of yield. Cases with unblocked surface flow and abundant cloud liquid water tend to be the most productive. The technique presented here can be readily adapted to evaluate the seeding impact of other long-term glaciogenic seeding operations and to improve their operational efficiency.

Significance Statement

In the United States and elsewhere, there are several operational programs to enhance cold-season precipitation through glaciogenic seeding of orographic clouds. The impact of such activity on seasonal precipitation has always been difficult to quantify. Recent observational and numerical modeling studies indicate that orographic cloud seeding can increase precipitation, although the amounts and optimal seeding conditions remain uncertain. Operators lack guidance about the seeding efficacy and about the most suitable environmental conditions. In recent years a model parameterization, called Weather Research and Forecasting Weather Modification (WRF-WxMod), has been tested against detailed measurements. This sets the stage for our work, a well-designed numerical evaluation of 10 years of operational cloud seeding over the Wind River Range, a mountain range in Wyoming that feeds the Colorado River basin and other watersheds. The WRF-WxMod based simulation experiment presented here, one of the most computationally expensive numerical experiments on this subject to date, quantifies seeding impact and its uncertainty. It is demonstrated with a high degree of confidence that over this 10-yr period, suitable seeding conditions were rare over this mountain range, that most seeding events were unproductive, and that, as a result, the overall yield over 10 years was a mere 1.1%.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Bart Geerts, geerts@uwyo.edu

Abstract

This study evaluates an operational glaciogenic cloud-seeding program using ground-based generators of silver iodide (AgI), with a total of 190 seeded storms over 10 cold seasons, using the Weather Research and Forecasting Weather Modification (WRF-WxMod) scheme at 900-m grid spacing. This study examines both the quantitative change in precipitation and the ambient and cloud conditions impacting seeding efficacy. An ensemble approach is used, with differing model boundary conditions, ice nucleation physics, concentrations of cloud condensation nuclei, and boundary layer schemes. This is intended to provide an envelope of uncertainty of natural clouds and seeding impacts. The simulations are validated against radiosonde, snow gauge, and microwave radiometer observations, and the seeding impact is inferred from simulations with/without AgI seeding. The seeding-induced precipitation enhancement (“yield”) varies greatly between storms. A small portion of the cases produces the majority of the yield. Overall, the precipitation in the target area (the Wind River Range in Wyoming) increased by 1.10% ± 0.13% in the 10 years of operational seeding. This rather low fractional increase is related to the frequent seeding at unsuitable times, primarily because of low-level flow blocking. The flow and cloud structure for select cases are examined to provide better insight into the variability of yield. Cases with unblocked surface flow and abundant cloud liquid water tend to be the most productive. The technique presented here can be readily adapted to evaluate the seeding impact of other long-term glaciogenic seeding operations and to improve their operational efficiency.

Significance Statement

In the United States and elsewhere, there are several operational programs to enhance cold-season precipitation through glaciogenic seeding of orographic clouds. The impact of such activity on seasonal precipitation has always been difficult to quantify. Recent observational and numerical modeling studies indicate that orographic cloud seeding can increase precipitation, although the amounts and optimal seeding conditions remain uncertain. Operators lack guidance about the seeding efficacy and about the most suitable environmental conditions. In recent years a model parameterization, called Weather Research and Forecasting Weather Modification (WRF-WxMod), has been tested against detailed measurements. This sets the stage for our work, a well-designed numerical evaluation of 10 years of operational cloud seeding over the Wind River Range, a mountain range in Wyoming that feeds the Colorado River basin and other watersheds. The WRF-WxMod based simulation experiment presented here, one of the most computationally expensive numerical experiments on this subject to date, quantifies seeding impact and its uncertainty. It is demonstrated with a high degree of confidence that over this 10-yr period, suitable seeding conditions were rare over this mountain range, that most seeding events were unproductive, and that, as a result, the overall yield over 10 years was a mere 1.1%.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Bart Geerts, geerts@uwyo.edu

1. Introduction

In arid regions, there has long been an interest in the enhancement of precipitation through cloud seeding (Haupt et al. 2019). A method commonly used in the western United States and elsewhere is the glaciogenic seeding of orographic clouds in the cold season (http://www.nawmc.org/). These mountains are targeted because the region is relatively arid and water supply is largely derived from the high-elevation snowpack. Interest in cloud seeding to enhance precipitation has increased as the Colorado River basin is experiencing a multidecade water shortage (Udall and Overpeck 2017), and as the mountain snowpack is declining in a globally warming climate (Mote et al. 2018).

The efficacy of glaciogenic cloud seeding has repeatedly been called into question in authoritative reports (e.g., National Research Council 2003). The challenge has been that, even though there is much evidence for the presence of supercooled liquid water (SLW) in orographic clouds (Rauber and Grant 1987; Heymsfield and Miloshevich 1993; Kusunoki et al. 2005; Sassen et al. 1990; Lohmann et al. 2016), quantitative precipitation impact assessment is extremely difficult, because precipitation measurements are uncertain and no two storms are the same (Rauber et al. 2019).

The most promising progress in recent years has come through physical process studies, examining properties of natural orographic clouds and microphysical changes resulting from the injection of seeding material, usually silver iodide (AgI), combining observations with process-resolved numerical simulations (Geerts and Rauber 2022). First, two recent field experiments in the interior western United States have provided strong evidence that under the right conditions, glaciogenic cloud seeding from the ground (Pokharel et al. 2017, 2018, and references therein) and from an aircraft (French et al. 2018) enhance ice crystal production and surface precipitation. These campaigns are the 2011/12 AgI Seeding Cloud Impact Investigation (ASCII; Geerts et al. 2013; Pokharel and Geerts 2016) and the 2017 Seeded and Natural Orographic Wintertime Precipitation: The Idaho Experiment (SNOWIE; Tessendorf et al. 2019), respectively. Observational evidence has not established the range of seedable orographic clouds: for instance, the observational evidence mentioned above focused on relatively shallow clouds. In essence, adequate SLW needs to be present, and the seeding material needs to be dispersed such that it blends with the SLW at sufficiently low temperatures.

Second, several numerical schemes have been developed recently that capture the cloud microphysical impact of AgI seeding. One AgI scheme, the Weather Research and Forecasting Weather Modification (WxMod) scheme (Xue et al. 2013a,b), uses the Thompson bulk microphysics scheme (Thompson and Eidhammer 2014). A more recent AgI scheme is built on a bin-resolved cloud parameterization (Geresdi et al. 2017, 2020). The validity of these schemes is difficult to ascertain, but Xue et al. (2022), using the WxMod scheme, for the first time simulated the observed quantitative seeding impact for a case in SNOWIE (Tessendorf et al. 2019) with minimal natural precipitation and with measurable seeding-enhanced precipitation (Friedrich et al. 2020).

Based on this success, the WxMod scheme is used here to examine the impact of 10 years of operational ground-based cloud seeding in the Wind River Range in Wyoming on surface precipitation. Seeding impact uncertainty is quantified by means of an ensemble approach, whereby model boundary conditions and some physics choices are altered, following the technique pioneered by Rasmussen et al. (2018). Also, the large number of cases simulated (190 cases) is interrogated for cloud and environmental conditions that affect seeding efficacy.

In section 2, the focus region and the seeding program are described. Section 3 outlines the model configuration. Section 4 evaluates the ensemble model output against observations. Section 5 presents the simulation results, section 6 discusses some nuances and ramifications of this work, and section 7 lists the conclusions.

2. Focus region and operational seeding program

The target mountain of this study is Wyoming’s Wind River Range (WRR) (Fig. 1), a rather linear, northwest–southeast-oriented barrier that stands out above the high plains in all directions except to the north, where it continues into the Absaroka Mountains and other ranges surrounding Yellowstone National Park. The WRR serves as headwaters to four major rivers (Fig. 1b): The Green River on the southwest, a tributary to the Colorado River; the Wind River to the northeast and Sweetwater River to the southeast, both tributaries to the Missouri River; and the Snake River to the northwest, a tributary to the Columbia River. Because of this placement, the WRR’s seasonal snowpack and runoff has an impact on the water resources in a large region. The WRR has had an operational seeding program for many years. The WRR’s significance as headwater region, plus data collected as part of the operational seeding program, make this range a good candidate to examine cloud-seeding potential. Tessendorf et al. (2015) analyze cloud-seeding potential over the WRR using 8 years of 4-km-resolution model output (the CONUS dataset; Liu et al. 2017). This analysis showed that while the prevailing crest-level flow is along the WRR spine (from the west to northwest), cold-season seeding opportunities most frequently occur on the southwestern slopes of the WRR, when Pacific moisture is advected from the west to southwest. They also found that low-level stability may limit the dispersion of ground-released AgI into cloud. Both findings were corroborated by Tessendorf et al. (2020), a study that examines seeding feasibility over several Wyoming mountain ranges. They were confirmed also by Mazzetti et al. (2021, their Figs. 5 and 6d, respectively) who used 10 years of hourly data from another 4 km resolution regional climate model (the IWUS simulation; Wang et al. 2018). These analyses of climatological frequency of environmental and cloud conditions suitable for glaciogenic seeding do not quantify seeding impact. In contrast, the present study simulates the impact of AgI nuclei on cloud microphysical processes and precipitation. Tessendorf et al. (2015) also report on cloud-seeding simulations run retrospectively for the actual ground-based seeding events over the WRR in one cold season (2014/15), using WxMod. These simulations indicate a less than 1% increase in seasonal precipitation over the WRR. This small increase was attributed to the prevalence of near-surface stable conditions at the locations of the AgI generators (Tessendorf et al. 2015). The present study builds on this work, using 10 years of seeding events and a newer version of WxMod.

Fig. 1.
Fig. 1.

(a) Elevation map of the focus region, which is also the truncated simulation domain. The map is annotated with SNOTEL sites (black stars), the Pinedale (red cross) and Riverton (red star) radiosonde sites, the Boulder microwave radiometer site (red dot), and 10 AgI ground generators (white dots); U.S. state borders are shown in gray. (b) The same domain as in (a) but colored by Hydrologic Unit Code 6 river basins (RBs) and delineated by high vs low elevation at the 2500 m MSL threshold (black contour). The target area is defined as the high-elevation WRR (above 2500 m) and is divided into four watersheds as shown. The two red lines in (b) locate cross sections used later, and the two blue quadrilaterals define the west and east basins’ grid boxes that were used for analysis.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Weather Modification, Inc. (WMI), has been operating 10 AgI generators around the WRR (Fig. 1a) since 2007. These generator units are located in the foothills of the WRR and are operated remotely with a satellite connection and solar power. A solution containing AgI is injected into a propane flame to produce microscopic AgI salts that are then dispersed by the wind (Fig. 2).

Fig. 2.
Fig. 2.

A photograph of a ground-based AgI generator like the ones deployed in the WRR, with annotations for components (the image is provided through the courtesy of Weather Modification, Inc.).

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

This study uses the 2007–17 seeding seasons. There were 190 seeding events in these 10 cold seasons (15 November–15 April). The seeding cases were selected by human forecasters with the guidance of available observations and models. Both tools evolved somewhat during those 10 years. General seeding criteria relate to the presence of SLW, and the likelihood that ground-released AgI will mix into SLW clouds at temperatures of −6°C or colder. The temperature threshold from −6° to −8°C is commonly used for AgI seeding-case calling (e.g., Breed et al. 2014; Manton et al. 2011; Geerts et al. 2013) and in seeding feasibility studies (e.g., Tessendorf et al. 2020; Mazzetti et al. 2021), since significant nucleation of ice through immersion or contact freezing by AgI nuclei requires temperatures at least this low (DeMott 1997). The forecasters generally had access to data from a passive microwave radiometer to provide real-time liquid water path (LWP). In general, it was cold enough over the WRR, so forecasters primarily monitored low-level wind and LWP. WMI forecasters launched their own radiosondes in advance of some cases. Model guidance generally came from the National Center for Atmospheric Research (NCAR) Real-Time Four-Dimensional Data Assimilation (RT-FDDA) system, running at sufficient grid resolution (2 km) to capture orographic flow and clouds (Breed et al. 2014; Tessendorf et al. 2015). The dataset of 190 cases used in this study is the result of case calling judgments made by WMI forecasters. The present study cannot quantify “missed” seeding opportunities, because it only simulates these 190 cases. Mazzetti et al. (2021), using a temporally continuous 4-km regional climate simulation, demonstrated that suitable seeding opportunities are more numerous but shorter in duration than these 190 seeding events suggest.

In any given seeding event, between one and nine generators were activated depending on flow direction and stability. Up to nine generators, all located on the southwest foothills of the WRR (Fig. 1a), were activated during storms with (south) westerly winds. These are referred to as westerly cases (141 in total). The one generator on the east slope of the WRR was activated during storms with easterly low-level winds, referred to as easterly cases (49 in total). The imbalance in number of generators on opposite sides is a reflection of the fact that most precipitation over the WRR occurs when crest-level winds have a component from southwest to northwest (Tessendorf et al. 2015, 2020). WMI provided detailed records of their seeding operations, including the “on” times of each generator as well as the flow rate of seeding solution. The WRR seeding program disbursed ∼230 kg of seeding material over the course of ∼2200 h of seeding in 10 years. Most of the 190 cases are shorter than ∼12 h; the longest event lasts nearly 2 days (Fig. 3a), resulting in a broad variation of AgI mass dispersed per case (Fig. 3d). The time of day of AgI dispersal is well distributed around the diurnal clock (Fig. 3b), and there is no bias toward weekdays (as opposed to weekends) (not shown). The middle of the season is seeded only slightly more than the shoulder periods (Fig. 3c). We define the target region for the WRR as land above 2500 m (Fig. 1b), where a seasonal snowpack typically is present.

Fig. 3.
Fig. 3.

Histograms characterizing the 190 seeding cases: (a) duration, (b) time of day (UTC), (c) time of year (by week, starting on 15 Nov), and (d) AgI dispersed.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

3. Model experiment design

The primary tool of this research is the Advanced Research version of the Weather Research and Forecasting model (WRF-ARW) (Skamarock et al. 2008) with the Noah land surface model with multiphysics options (Niu et al. 2011). The model experimental design broadly follows Tessendorf et al. (2015) and Rasmussen et al. (2018). The simulation of each seeding case has a spinup period of 12 h before the start of the case. All simulations run at 900-m grid spacing and 81 vertical levels over a domain that corresponds to the area plotted in Fig. 1a, plus 10 grid points (9 km) in all four directions. This model resolution is determined from recent modeling work focusing on orographic clouds and precipitation (Xue et al. 2022). Convection is permitted and not parameterized.

This study uses the WxMod module (Xue et al. 2013a,b), which is built on top of the aerosol-aware Thompson microphysics scheme (Thompson and Eidhammer 2014) for use in WRF-ARW. The Thompson scheme was chosen because of its ability to accurately capture SLW (Thompson et al. 2008) and orographic precipitation (Ikeda et al. 2010; Jing et al. 2017). The aerosol-aware version of this scheme allows for the addition of aerosol sources, such as AgI released from point sources. WxMod parameterizes four AgI nucleation modes: deposition, condensation freezing, contact freezing, and immersion freezing (Xue et al. 2013a). The AgI-salt particles dispersed by the generators also are effective cloud condensation nuclei (CCN). WxMod originally was used in limited-area simulations (Xue et al. 2013b) but has since been used also in large-eddy simulations (Xue et al. 2016, 2017; Chu et al. 2014, 2017a,b) because of improved aerosol dispersal from point sources and better representation of flow in complex terrain. These studies also showed that model initial conditions may need to be tuned to accurately capture an orographic cloud system before evaluating seeding impact, which is not surprising given that the reanalysis driver dataset has a relatively coarse resolution in time (1, 3, or 6 hourly) and space. Comparisons between non-LES and LES approaches and comparisons with observed precipitation changes led to improvements in WxMod (Xue et al. 2022). Observationally constrained WxMod simulations have been shown to be able to represent the timing, geographical extent, and amount of seeding-induced precipitation in a case that produced very little natural precipitation (Xue et al. 2022).

Because of the uncertainties associated with (i) the initial and boundary conditions, (ii) low-level stratification and the dispersal of aerosol in the planetary boundary layer (PBL), (iii) the natural background cloud-active aerosol concentration, and (iv) the microphysical processes involving AgI, an ensemble approach is used, following Rasmussen et al. (2018). The ensemble set contains 28 members: 12 control simulations and 16 seeded simulations (Table 1). The range of parameters used is intended to bracket the reality. Rasmussen et al. (2018) used many more seeded and control ensemble members, but some members were found to be virtually identical to others. Therefore, this reduced set is used, which still should capture the full range of uncertainty.

Table 1.

Summary of the 28 ensemble simulations, run for each case. These simulations are identified with numbers from 0 to 27 in the last three columns.

Table 1.

Uncertainty in the initial and boundary conditions is addressed with different reanalysis datasets: the fifth major global reanalysis produced by ECMWF (ERA5) (Hersbach et al. 2020), the North American Regional Reanalysis (NARR) (Mesinger et al. 2006), and the Climate Forecast System, version 2 (CFSv2), Reanalysis (CFSR) (Saha et al. 2014). Uncertainty in the depth and mixing strength of the PBL is addressed with two different PBL schemes: YSU (Hong et al. 2006) and MYNN (Nakanishi and Niino 2009). The former (YSU) uses a nonlocal turbulent mixing coefficient and explicit PBL top entrainment processes, whereas MYNN focuses on eddy-diffusivity and mass-flux processes in which the turbulent kinetic energy (TKE) can be advected as other scalars.

The default background CCN concentration is a spatially resolved monthly mean from the Goddard Chemistry Aerosol Radiation and Transport model (GOCART) climatology (Chin et al. 2002). The alternative background CCN concentration is 20% of that. This choice is informed by our WxMod modeling experience in Idaho (Xue et al. 2022). There is also geographically specific circumstantial evidence for this choice. In cold-season fair-weather conditions, inversions are common in the upper Green River basin, and this implies air quality deterioration and aerosol buildup (e.g., Oltmans et al. 2014). Since these conditions prevail, they contribute significantly to the GOCART climatology. Frontal and especially postfrontal shallow orographic clouds (which are common in westerly cases) tend to have low relatively CCN (e.g., Reynolds and Dennis 1986; Rasmussen et al. 2002; Ikeda et al. 2007).

Uncertainty in the parameterization of ice initiation as a function of ambient temperature and ice supersaturation is represented by alternately using the Cooper (1986) and the Meyers et al. (1992) schemes. Ice initiation also depends on the concentration and properties of ice nucleating particles (INP), including AgI. The main uncertainty in WxMod involves the rate of AgI depletion by (i) scavenging by hydrometeors, (ii) AgI self-coagulation, and (iii) AgI dry deposition (Xue et al. 2013a). We use both the default AgI nucleation and AgI depletion rate coefficients, along with 5.0 times the default rates (members 24–27 in Table 1), again based on previous experience (Rasmussen et al. 2018).

The seeding impact is quantified as a difference of pairs within each row in Table 1: there are 12 unseeded (control) simulations, 12 seeded simulations matched to the 12 unseeded, and 4 more seeded members with the higher AgI production and sequestration rate. Absolute yield is defined as area-integrated volume of liquid-equivalent extra precipitation (m3) within the target area (highlighted in Fig. 1b). Relative yield is the yield divided by the natural precipitation (%). AgI-weighted yield is the yield per kilogram of AgI dispersed (m3 kg−1). The AgI-weighted relative yield is the relative yield per kilogram of AgI dispersed (% kg−1).

4. Model validation

Model sensitivity studies often are conducted to evaluate what model physics and architecture best capture a phenomenon of interest. We did evaluate the performance of the three reanalysis driver datasets and the two PBL schemes in capturing the observed precipitation and thermodynamic/wind profiles around the WRR for the 190 cases, but this analysis did not reveal significant systematic performance differences. Instead, we use available observations to evaluate the aggregate of simulations, that is, all seeded ensemble members for all cases. As mentioned before, the ensemble members capture uncertainties in the driver datasets, in PBL processes, and (to some extent) in cloud processes.

a. Thermodynamic and wind profiles

Ensemble-mean model output is compared with data from 340 rawinsondes: 51 launched by WMI from Pinedale, Wyoming (upstream for westerly cases), and 289 operational sondes launched by the U.S. National Weather Service (NWS) Riverton, Wyoming, office (upstream for easterly cases) (locations shown in Fig. 1a), during the 190 cases used in this study. The NWS Riverton office used Sippican Microsonde IIA sondes in 2006–13 and Lockheed Martin LMS-6 sondes in 2013–17. These systems are processed with a robust quality control system (Gutman et al. 2005; Brown and Fitzgibbon 2016). The Pinedale soundings were launched by WMI for seeding decision making. WMI used Vaisala RS92 SGP sondes, which have the following manufacturer-stated accuracies at 95% confidence: temperature ±0.5°C, relative humidity ±5%, pressure ±1 hPa, and wind speed 0.6 m s−1 (Dirksen et al. 2014). Rawinsonde observations were linearly interpolated to the model levels.

The case- and ensemble-mean model temperature bias (Fig. 4a) is relatively small in comparison with the instrument accuracy, with the largest bias and standard deviation (Fig. 4e) near the surface. The slight near-surface warm bias and the significant positive wind speed bias is an indication that the model mixes too much momentum toward the surface. These biases are most pronounced in the lee of the WRR (westerly cases in the Riverton soundings, i.e., the most common type; Figs. 5a,c): during westerly cases the model tends to produce winds ∼5 m s−1 too strong on average at ∼700 hPa, which is close to the average crest level in the WRR, over Riverton (Fig. 5c), and a negative speed bias aloft. Both PBL schemes have this same pattern in wind speed bias and large wind speed errors, especially on the leeside (not shown). There are small clockwise and counterclockwise low-level wind direction biases on the upwind and lee sides, respectively (Fig. 5d). This may be a result of model overrepresentation of flow deflection by the WRR causing the simulated winds in Pinedale to turn more parallel to the range and possibly a similar effect in Riverton with flow wrapping around the southern tip of the WRR, an indication of blocked flow. Consistent with this, the model appears to slightly overestimate near-surface stratification near Pinedale during westerly cases (Fig. 5a): it has a small (<1 K) cold bias within ∼70 m from the surface. This may be real, but it may also be a contamination due to insufficient aspiration of the sonde in preparation for manual launch, an issue the NWS soundings do not have. The model is slightly too moist, especially at upper levels (Fig. 4c), and especially at Riverton (Fig. 5b). The case- and ensemble-based standard deviation (Figs. 4e–h) is of a similar magnitude as the spread of the ensemble mean errors (Figs. 4a–d). This supports the usage of ensemble means to characterize any of the 190 cases (see below).

Fig. 4.
Fig. 4.

Vertical profiles of (a)–(d) the distribution of ensemble mean error (model minus observation) and (e)–(h) ensemble standard deviation for 340 soundings for temperature [in (a) and (e)], dewpoint [in (b) and (f)], wind speed [in (c) and (g)], and wind direction [in (d) and (h)]. Shown are the mean (thick line), the median (thin line), and the interquartile range (shaded).

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Fig. 5.
Fig. 5.

As in Figs. 4a–d, but showing only the mean and distinguishing between upwind and leeside soundings, for westerly and easterly cases.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Effective ground-based seeding of orographic clouds requires flow over the target mountain (“unblocked” flow). To assess how well the model predicts flow blocking (Markowski and Richardson 2010, chapter 13), we examine the Froude number Fr of the upstream flow (Table 2). We define Fr as U/(NH)(Fr=U/NH), that is, the layer-mean WRR-normal wind speed U divided by the layer depth H (surface to crest level) and the layer-mean Brunt–Väisälä (B-V) frequency N, at the rawinsonde launch location. The layer-mean N is calculated as the dry B-V frequency below the lifting condensation level (LCL) and as the moist B-V frequency above the LCL. The generators are located not much higher than the sounding site (238 m on the southwest side on average and 341 m on the northeast side); therefore, we compute Fr from the surface. If U < 0 (downslope flow), we set Fr < 0 as well. The Froude number could not be calculated for 8 of the 340 radiosondes because of missing data (Table 2). Table 2 shows that, during seeding cases, (i) the layer-mean flow sometimes was from the wrong direction (17% and 26% of the soundings according to observations and model, respectively); (ii) the flow was blocked about 50% of the time (Fr < 1.0); (iii) unblocked flow (Fr > 1.0), required for effective dispersal of ground-released AgI into orographic cloud, was somewhat uncommon (27%); and (iv) the critical success index (the correct predictions divided by the total) was somewhat low (0.43). The high frequency of blocked flow (Fr < 1.0) is due mainly to the prevalence of highly stratified conditions: N values below 0.5 × 10−2 s−1 were relatively uncommon, in only 29% of the 332 rawinsondes (23% of the time–space-matched model output). The wind speed also tends to be low during highly stratified conditions, that is, |U| and N are negatively correlated (not shown).

Table 2.

Confusion matrix for the Froude number of the upstream flow, for N = 332 rawinsondes.

Table 2.

b. Precipitation

The most important output variable in this study is precipitation, so we evaluate modeled against observed precipitation. Snowfall is a challenging measurement (Rasmussen et al. 2012). The WRR hosts 17 Snowpack Telemetry (SNOTEL) (Serreze et al. 1999) sites (annotated in Fig. 1a), which measure precipitation rate using a gauge, snow depth, and snow water equivalent (SWE) using a snow pillow. These sites surround the WRR but are not located above the tree line, where most of the precipitation falls. They are located at various elevations within the forest to capture the precipitation and snowpack with minimal impact of blowing and drifting snow. In the WRR, there are additional siting restrictions due to wilderness and reservation regulations. Given the distribution and sparsity of SNOTEL sites in the WRR, the precipitation distribution cannot be accurately validated in the WRR (Jing et al. 2017). The purpose of this evaluation is not to quantify seeding effect, but rather to generally evaluate the model in terms of orographic precipitation. This is in contrast with experimental seeding impact studies that deploy a network of gauges in target and control areas in an attempt to observationally quantify the seeding impact (e.g., Manton et al. 2011; Breed et al. 2014).

Here, we compare the difference in SWE between the end and the start of each case at all SNOTEL sites. The SWE difference is usually positive for the various cases and sites, but it can be negative when the snow ablation (sublimation, melt, or wind erosion) exceeds the fresh accumulation during a case (Fig. 6). The modeled and observed SWE changes align reasonably well along the 1:1 ratio line, although the model has a positive bias, and the spread from the 1:1 line is large. The model also significantly underestimates snowfall in a few extreme cases (Fig. 6). These discrepancies are not surprising: high-resolution simulations tend to capture seasonal precipitation better than precipitation totals from individual storms (e.g., Ikeda et al. 2010). In addition, seasonal 4-km WRF simulations disagree with the 4 km PRISM data (a gauge-based slope-interpolated gridded dataset) more over the WRR than over some other mountains in the interior western United States (Liu et al. 2017; Jing et al. 2017).

Fig. 6.
Fig. 6.

Normalized density plot of model vs SNOTEL SWE change over the duration of each case for each of the 17 SNOTEL sites surrounding the WRR and for the 16 seeded ensemble members. The 1-to-1 line is shown in red.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

c. Cloud liquid water

As part of the seeding-case calling process, WMI deployed a Radiometrics WVP1500 in Boulder, Wyoming (Fig. 1a). This five-channel radiometer has a manufacturer-listed brightness temperature accuracy of 0.5 K. A proprietary neural net algorithm is used to infer LWP. The radiometer was placed with an elevation angle of 8° to the east, to graze above the skyline of the WRR (its beamwidth is 2.5°). The radiometer was not always available or functioning: useful data were collected in 77 cases, all during the 2008–14 seasons. The radiometer data were extensively quality controlled by correcting the drifting baseline of the readings. Because radiometer-estimated LWP varies rapidly in time, it is averaged in 30-min windows around the half-hourly model output times, to attain a more robust and comparable value. Model LW was extracted along the 8° radiometer grazing angle, from which the LWP was computed, and expressed as a vertical water depth (mm, or in units of kilograms per meter squared when multiplied by water density). Before any time averaging, a threshold of 0.025 kg m−2 was applied to the radiometer and the model estimates to exclude noise (Liljegren et al. 2001; Crewell and Löhnert 2003). Half-hourly model and radiometer LWP estimates are both averaged over the duration of the cases (or radiometer data availability) (Fig. 7). The model has a high LWP bias of almost a factor of 2. This bias is independent of the thresholding or the averaging window for the observations. This bias is due both to cloud cover (the frequency of liquid water presence above the threshold) and to liquid water content (when cloud is present). This needs to be considered when interpreting the results in section 5. The positive LWP bias is consistent with the positive snowfall bias during the 190 cases (Fig. 6).

Fig. 7.
Fig. 7.

Scatterplot of LWP from the Boulder radiometer (location shown in Fig. 1) vs LWP from case average ensemble mean for the 77 cases during which the radiometer was in operation. Model mean (dots) and standard deviation (lines) are shown. The black line is the linear regression with intercept fixed to 0, weighted by inverse standard deviation. The red lines are 1-to-1, 1-to-2, and 2-to-1 ratios.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

5. Seeding impact evaluation

a. Overall seeding impact

Over the course of the 190 seeded storms, there was approximately 7.39 × 109 m3 of natural precipitation over the target area, with an uncertainty (ensemble standard deviation) of ±0.54 × 109 m3. According to our ensemble simulations, the WRR operational seeding program resulted in a yield of (81.2 ± 9.4) × 106 m3 [(65.8 ± 7.7) × 103 acre ft (a-f)] over the target area. This equates to a 1.10% ± 0.13% increase in the precipitation of these storms (Table 3). Interestingly, the ensemble standard deviation in the precipitation and the yield are both ∼11%.

Table 3.

Seeding impact on the three main river basins within the WRR target area (Fig. 1b).

Table 3.

The cases can be ranked by their ensemble average absolute yield, AgI-weighted yield, and relative yield (Fig. 8). A small portion of the cases is responsible for a majority of the extra snowfall (Figs. 8a,c). This conclusion was reached also for a randomized seeding study over two ranges in southern Wyoming (Fig. 14 in Rasmussen et al. 2018), and in the modeling study of the impact of the 2014/15 seeding events over the WRR (Tessendorf et al. 2015). The ensemble distribution of yield for each case looks fairly normal, with the mean and median close together centered in the interquartile range (IQR). There are only a few outliers that fall more than 1.5 IQR away from the 25th or the 75th percentile.

Fig. 8.
Fig. 8.

Box-and-whisker plots of the distribution of the ensemble mean yield in the target area for each of the 190 cases ordered by (a) absolute yield, (c) AgI-weighted yield, and (e) relative yield. Westerly and easterly cases are respectively shown in red and blue. The black points are outlier ensemble members. The red (westerly) and blue (easterly) dots are the t-test P values (right axis) against the null hypothesis. The dashed red line is P = 0.05. (b), (d), (f) Distribution of case yield (total count 190) for the ensemble median (black) percentiles, with yield expression matched horizontally to (a), (c), and (e). Also shown are the distributions for the 10th- (red), 25th- (yellow), 75th- (green), and 90th- (blue) percentile ensemble members.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

The ranking of the cases shifts for AgI-weighted yield (Fig. 8c) so that the easterly cases (with just one AgI generator in operation) are the most productive. An overwhelming majority of the cases have a statistically significant seeding impact (Figs. 9a,c), while the few cases that have a small negative impact on precipitation are not statistically significant. Here, we use the Student’s t test, with a null hypothesis that there is no seeding impact, and a threshold p value of p = 0.05.

Fig. 9.
Fig. 9.

Scatterplot of ensemble average absolute yield vs AgI-weighted relative yield for the 190 cases (westerly: red; easterly: blue), with error bars representing ensemble standard deviation.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

The yield expressed as an amount of natural precipitation is extremely skewed (Figs. 8e,f): only 3 cases have a yield above 10%, and 28 cases have a yield above 5%. A case that is very productive in absolute terms is not necessarily productive from an AgI-weighted perspective, and vice versa (Fig. 9). Westerly cases tend to yield more water, but when dividing by AgI mass disbursed, the best cases tend to be easterly (Fig. 9).

Most natural precipitation across all 190 cases is orographic. The precipitation total shown in Fig. 10a matches that of an entire winter season (e.g., Fig. 6 in Jing et al. 2017), with the caveat that the precipitation in Fig. 10a is centered on storm passage over the WRR and not the adjacent ranges. The ensemble standard deviation is highest where the precipitation is the highest, as seen by the less saturated hues for the highest precipitation values (Fig. 10a). The absolute yield (Fig. 10b) is the difference between the precipitation in corresponding seeded and unseeded members, within the rows of Table 1 (16 estimates, with both the mean and standard deviation shown in Fig. 10b). This yield is positive, except in a few small spots, and is essentially confined to the target area (the WRR). There is a slight impact over nearby mountain ranges. This extra-area effect is near zero, although the ensemble standard deviation is large: some ensemble members have a positive impact while others have a negative (Fig. 10b).

Fig. 10.
Fig. 10.

Maps of (a) the simulated total precipitation (m) during 190 seeded storms, (b) total yield (mm), (c) relative yield (%), and (d) WRF-WxMod-resolved total AgI wet deposited (by rain, snow, or graupel) at the surface (kg m−2). All color bars have a second dimension to represent ensemble uncertainty, except in (c), where the second dimension indicates total precipitation. The 2500 m MSL contour is shown as in Fig. 1, for georeferencing.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

The 10-yr-average percentage change due to seeding is shown in Fig. 10c; here the saturation of the color bar is determined by the total precipitation (Fig. 10a), such that arid plains regions (where the relative change can be large) are marked by a very light hue. Some small areas near the south end of the WRR experience an increase as high as 5% (Fig. 10c). On average (for the 190 cases), AgI is scavenged mainly by hydrometeors and deposited as part of the extra precipitation (“wet deposition”), mainly over the target area (Fig. 10d). Dry deposition of AgI occurs as well, especially close to the AgI generators during blocked flow conditions, and in the shape of a plume extending westward from the southern tip of the WRR (not shown). A very small fraction of the total AgI dispersed leaves the domain.

The changes shown in Fig. 10 are largely due to the more frequent westerly cases. This explains why most of the absolute yield in the target area falls in the Green River basin (Table 3). Westerly cases also result in spillover extra precipitation on the lee side (Bighorn basin) (not shown). The less common easterly cases contribute about 30% of the total precipitation during these 190 cases, and most of it falls east of the crest (Fig. 11a). Precipitation during easterly cases is more commonly convective than during westerly cases. The seeding signal for the easterly cases shows a positive response near the southeastern tip of the WRR, close to the lone east-side AgI generator (Fig. 11b). This response is rather strong, up to 5% (Fig. 11c), and associated with wet deposition of AgI (Fig. 11d). This explains why the Sweetwater River basin is the largest beneficiary of the seeding in a relative sense, with a 2.1% increase (Table 3). The widespread precipitation east of the WRR is hardly affected by the seeding, with rather noisy but weak and uncertain positive and negative changes well outside the target area (Fig. 11b).

Fig. 11.
Fig. 11.

As in Fig. 10, but for only the 49 easterly cases.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

b. Understanding the seeding impact: Some case studies

For a deeper understanding of the cloud-seeding mechanisms, three cases are shown here, although several more were examined, illustrating the possible scenarios of flow and AgI dispersal: a productive westerly case, a westerly case with low-level blocked flow, and an easterly case. Rather rapid synoptic, mesoscale, and microphysical changes typically occur during many seeding cases, as evident from half-hourly maps and transects for each case (not shown). Here we illustrate a few snapshots only. In each of the three cases, a model ensemble member with seeding is shown, specifically the one that has an absolute yield closest to the ensemble mean. The time of the snapshot is intended to highlight key characteristics of each of the three cases.

1) Most productive westerly case

An intense storm with strong west-southwest winds passed over the WRR on 28 December 2016, with upward of 22 mm of precipitation over the highest parts of the WRR (Fig. 12a). The cross-mountain flow produced a downslope windstorm in the lee of the WRR (Fig. 13a). The average radiometer LWP (thresholded as discussed in section 4c) was 0.12 kg m−2 during this case, one of the highest values observed (Fig. 7). Seven generators were used in this case, yielding nearly 5 × 106 m3 (4000 a-f) of extra precipitation in the target area, most of it falling over the central and southern portions of the WRR (Fig. 12b). The ensemble mean percentage increase is up to 15% over the southern WRR (Fig. 12c). The flow was generally unblocked (Figs. 12e and 13a), only near the very end of the case did the flow become blocked and started to flow around the southern tip of the WRR. The case-average Fr (averaged at the locations of the westerly AgI generators) was 0.87. At the start of seeding, high ice clouds were present, cloud-top temperatures (CTT) below −30°C. Their tops dropped consistently throughout the duration of the case to ∼−25°C (Fig. 13c). By the time that the seeding plumes reached the WRR crest (Fig. 13b), low orographic mixed phase clouds were present, with ice tops and consistent pockets of cloud water at their leading edges (Fig. 13c). This allowed for a broad region of AgI activation (magenta, Fig. 12e), over the target area. By the end of seeding, the clouds no longer had ice tops. The AgI dispersion remained close to the ground and did not get carried, in any considerable concentration, deeper in the SLW clouds. This suggests that airborne seeding can be more effective than the ground seeding in this environment. Some dry AgI were advected downstream, in the lee of the WRR (Fig. 12e).

Fig. 12.
Fig. 12.

(a)–(d) As in Fig. 10, but for case 181 (28 Dec 2016), the westerly case with the highest yield, for the ensemble member closest to the ensemble mean. (e) Map of PBL average winds and vertically integrated AgI number concentration (dry or wet, i.e., immersed in liquid or solid hydrometeors), and (f) map of vertically integrated mixing ratios (path) of graupel, snow, rain, ice, and cloud water, following the color key at the bottom, at 0400 UTC.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Fig. 13.
Fig. 13.

Transects for case 181, at the same time as Figs. 12e and 12f. (a),(c) Transects normal to the WRR; (b),(d) transects along the WRR (see Fig. 1b for location). Shown are AgI concentrations (colors) with potential temperature (K; red contours) [in (a) and (b)] and hydrometeor mass mixing ratios (colors) with temperature (°C; red) [in (c) and (d)].

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

2) Westerly case with low-level blocked flow

The next case illustrated (24 December 2016) had roughly the same amount of natural precipitation over the WRR as the first (Fig. 14a), but the surface flow over the upstream Pinedale plains was very weak (Fig. 14e) because of high low-level stratification there (Fig. 15a), with a case-average Fr = 0.14. All nine AgI generators on the southwest side were in operation, but almost no AgI was advected into the target area except at the north end (Fig. 14d), where the flow was less blocked and some AgI was advected into orographic cloud to produce extra precipitation (Figs. 14b,e). The stable layer was only a few hundred meters deep (Fig. 15a). The strong southwest winds above this layer advected much moist flow over the WRR. Initially the storm had widespread mid- to upper-tropospheric, ice-dominated clouds moving in from the north-west while descending and producing surface snowfall by 0700 UTC (Figs. 14f and 15c,d). As the mid- to upper-level clouds moved off, low-level orographic clouds became liquid-dominated upwind of the crest, until the clouds were purely liquid near the end of the case. While the WRR produced a high-amplitude trapped lee-wave train (Fig. 15a), the upstream range (the Wyoming range) did not, yet this range penetrates well above the shallow Pinedale plains inversion (Fig. 15a). This supports the notion that AgI generators could be placed atop the upstream mountain range instead of in the WRR foothills, a principle suggested by Rauber et al. (2019).

Fig. 14.
Fig. 14.

As in Fig. 12, but for case 180, a westerly case with blocked flow, orographic precipitation and little seeding effect; (e),(f) maps for 0700 UTC.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Fig. 15.
Fig. 15.

As in Fig. 13, but for case 180 at 0700 UTC.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

3) Easterly case

In many easterly cases the low-level flow near the south end of the WRR is deflected around the range, as for westerly cases. The example shown in Figs. 16 and 17 shows stagnant flow in the Wind River basin, and flow from the northeast, south of the WRR (Fig. 16e). Because of the weak winds, Fr at the location of the lone east-side generator was very low on average (Fr = 0.04). Yet there was a deep, well-mixed moist PBL at the location of the generator (Figs. 17c,d). Therefore, some seeding material was dispersed over the southern WRR (Fig. 16d) and was absorbed in cloud (Figs. 16e and 17a,b), resulting in enhanced snowfall (Fig. 16b). Upslope clouds were present east of the WRR; they were blocked at crest level and extended far to the east (Figs. 16f and 17c). They had liquid cloud tops close to −13°C and even produced some supercooled rain that glaciated to snow precipitation Fig. 17c). Heavy snowfall occurred over the southern foothills of the WRR, and also over the Wind River basin to the east (Fig. 16a), as is typical for easterly cases (Fig. 11a). No natural precipitation occurred on the southwest side of the WRR, and significant precipitation enhancement (>2%) was confined to the southern foothills (Figs. 16b,c), where the AgI was sequestered mostly as part of the snowfall (Fig. 16e).

Fig. 16.
Fig. 16.

As in Fig. 12, but for case 083 (19 Dec 2011), a typical easterly case; (e),(f) maps for 1930 UTC.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

Fig. 17.
Fig. 17.

As in Fig. 13, but for case 083 at 1930 UTC.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

c. Correlating yield with environmental parameters

Many of the cases in the WRR were relatively unproductive (Fig. 8). One way to improve case calling guidance in operational seeding programs is to run high-resolution simulation forecasts of the kind presented here, with seeding impact predicted by the WRF-WxMod, but ahead of time instead of retrospectively as is done here. This approach is still computationally expensive; therefore, operators use environmental and cloud criteria, as mentioned in section 2. These criteria are based on forecasts and observations, and they may not always be satisfied; for example, 17% of the rawinsondes in the 10 years of operational seeding over the WRR showed flow in the wrong direction (Table 2). The most-difficult-to-predict criterium involves the presence of SLW, which is why a radiometer is deployed, although radiometer data were available for only 77 of the 190 cases (section 4c).

Faced with uncertainties, operators tend to run AgI generators for the duration of the snowstorm. Is precipitation a good proxy for seedability? We explore this question using the dataset of 190 cases and 16 (seed–no-seed) ensemble member pairs. The scatter-density plot in Fig. 18a examines the relation between two cumulative quantities, (natural) case precipitation and absolute yield. There is one point for each of the 190 cases and the 16 ensemble member pairs. Figures 18a–c show the normalized count of scatter points in bins, that is, the color indicates the point density (fraction of points per bin). High total yield cases occur under both high and low natural precipitation totals (Fig. 18a). The SNOWIE project demonstrated that in Idaho high relative yield is possible under low natural precipitation (Friedrich et al. 2020). SNOWIE findings to date have not ruled out the possibility of high yield under high natural precipitation. Cases with low yield tend to be mostly dry (no precipitation). In short, the yield is usually high during significant snowstorms, but some of these storms have a slightly negative yield in the target area. In these cases, the surface flow is blocked (not shown), as illustrated in section 5b(2).

Fig. 18.
Fig. 18.

Scatter-density plots (a) between case-total absolute yield and case-total natural precipitation, (b) between case-total AgI-weighted yield and case-average Fr, (c) between case-total AgI-weighted yield and case-average target-area-average SLWP, and (d) between the same Fr and SLWP colored by AgI-weighted yield.

Citation: Journal of Applied Meteorology and Climatology 62, 4; 10.1175/JAMC-D-22-0132.1

In an effort to refine operational seeding criteria, we explored the relationship between yield and a number of environmental and cloud parameters (overall 130), for all 190 seeding cases and 16 ensemble member pairs. Yield is variably expressed as absolute (m3), relative (%), AgI-weighted [m3 (kg-AgI)−1] and AgI-weighted relative [% (kg-AgI)−1] yield. Examined parameters include PBL depth, N, U, Fr, SLW path (SLWP), and many others over various volumes including the target area, generator locations, and the basins as depicted in Fig. 1, the most insightful of which are shown in Fig. 18.

Ground-based seeding can only be effective if the AgI are carried into orographic clouds. Therefore, we examine Fr, averaged for all generator locations on the upstream side of the mountain, and between the surface and crest level (Fig. 18b). Fr is also averaged linearly over time, for the duration of each case. Case-average Fr < 0 (downslope flow) is less common (0.01% of all ensemble members of all cases) than the rawinsonde analysis suggests (Table 2), but many cases contain periods with low-level downslope flow, for instance case 180 (Fig. 14e). The vast majority of ensemble members/cases have low-level fully blocked flow (0 < Fr < 0.5) (92.7%) or partially blocked flow (0.5 < Fr < 1.0) (6.8%). This is the fundamental reason why so many cases are marginally productive (Fig. 8). Most of the low Fr cases have highly stratified PBL, with drainage flow in the valleys. In general, yields tend to decrease with PBL stratification: the highest absolute yields are at small static stability (not shown). It is rare for the upstream airmass to be unblocked (Fr > 1.0) (0.5%). Fr is a good predictor for yield: cases with Fr > 0.5 almost all have a yield exceeding 4.0 × 104 m3. The reason is not simply advection of ground-released AgI over the target mountain. Stronger normal wind speed U also implies (for a given PBL water vapor mixing ratio) more orographic SLW.

Yield (per unit AgI mass dispersed) correlates with the SLWP over the target area (Fig. 18c). Sometimes supercooled fog or low cloud is found in a highly stratified, blocked PBL, producing near zero, or even negative yield. Otherwise, cases with high SLW (orographic cloud over the high terrain) have the highest yield. Similarly, the highest yields are found when the upstream LCL is close to the ground, optimally in the 200–400 m AGL range, when the upstream PBL-mean mixing ratio is high, and when the crest-level temperature (700 hPa) is relatively high (from −6° to −10°C) (not shown). Higher crest-level temperatures (above −6°C) are almost never encountered. The relationship between yield and SLWP is rather insensitive to temperature threshold, below −6°C or merely below 0°C (not shown). AgI-weighted yield tends to be higher under higher SLWP and under higher Fr values (Fig. 18d), but these two variables do not reveal a clear separation between low and high yields. Note that Fr can be inferred from rawinsondes, and the SLWP from a radiometer (as long as the cloud base is above the freezing level), so these key observations can be monitored to guide seeding operations. The ensemble modeling dataset indicates that Fr and SLWP combined are good, but not unequivocal, predictors of seeding efficacy.

We also examined yield as a function of cloud parameters that cannot readily be observed, such as the fraction of hydrometeors that is frozen. The highest absolute yield occurs when the solid hydrometeor fraction is low, when the cloud tops are relatively shallow (∼600–800 m above the target area), and the CTT relatively high, in the range from −18° to −14°C (not shown). These correlations, based on 190 cases and 16 ensemble member pairs, are not very strong. This analysis suggests that real-time, well-tested high-resolution (∼1 km) simulations have some utility in guiding seeding operations, but ultimately simulations that quantify the seeding impact (e.g., using WxMod) will provide better quantitative guidance.

6. Discussion

This study has demonstrated that the WRR operational seeding program was rather inefficient, with a small overall yield on account of the many cases and many time periods within cases that were unproductive. A small fraction of cases was quite productive (Fig. 8). These cases are characterized by relatively high Fr (unblocked flow) and high SLWP (Fig. 18). The many cases that were marginally productive or unproductive had a rather high low-level stratification and weak or negative cross-mountain wind. In many cases, the surface flow at the AgI generators was away from the mountain. Thermally driven drainage flow typically is concentrated in valleys. Therefore, the generators were located as much as possible on ridges in the foothills around the WRR. Still, while the midmountain to crest-level flow may be upslope and produce orographic clouds, the flow at low levels may be katabatic (as was the case much of the time in the case illustrated in Figs. 15 and 16), resulting in AgI being advected the wrong way, and near-zero yield. In many westerly cases, the low-level blocked flow resulted in AgI eventually being advected around the southern tip of the WRR. In response to this observation, WMI moved the two southernmost generators on the west slope northward near Pinedale, starting with the 2021/22 seeding season. Cases with low-level blocked flow may be suitable for aerial seeding rather than ground-based seeding, as discussed in section 5b(2).

Our analysis has focused on time-average ambient and cloud conditions during seeding cases. An analysis of half-hourly model output (not shown) indicates that conditions are often far from steady-state, sometimes fluctuating between favorable and unfavorable seeding conditions multiple times through the course of a single case. Predicting the timing and intensity of these fluctuations is challenging: this is both a problem of model resolution and of imperfect initial and boundary conditions. In addition, 0.5–2 h typically lapse between AgI dispersal from a generator and downstream wet deposition on the ground. Because of these advective and microphysical time scales, and because of model uncertainty at short time scales, operators understandably operate AgI generators for longer than ultimately warranted, effectively reducing the relative yield.

One of the uncertainties mentioned in section 3 involves the background CCN and INP concentrations. These 190 cases were studied in isolation with no information about synoptically varying natural aerosol concentrations, and no consideration of AgI plumes that may have been advected in from upstream seeding programs, in Idaho for example, by Idaho Power, Inc. In the future, it could prove useful to incorporate aerosol concentration data, for example, from the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) reanalysis, and to simultaneously evaluate all of the seeding programs in the western United States with a regional climate simulation using WRF-WxMod.

The uncertainties listed in section 3 motivated an ensemble approach. Aside from the representativeness of the parameter space captured in this ensemble, there is a question of numerical noise influencing the precipitation differences between seed and no-seed simulations. This noise occurs not just in the immediate vicinity of the perturbation (in this case, the point injection of INP), but across the domain, as illustrated in Fig. 13b for an individual case and in Fig. 12b for all easterly cases. Because this noise is geographically random (not tied to the terrain), compositing over many cases effectively suppresses it, for example, Fig. 12b.

Conclusions derived from model output should be nuanced based on model validation. The model appears to have had ∼2× too much liquid water (Fig. 7), so the yield may be overestimated. On the other hand, the model may overestimate the frequency of drainage flow over the generators (Table 2) and thus may underestimate good seeding opportunities.

Stakeholders (those funding the seeding programs) will be interested in the conclusions about relative and absolute yield and about the scarcity of suitable conditions presented here. However, their next question will address implications of the extra snowfall on seasonal streamflow. The simulation experiment presented here can be used directly for hydrological modeling to address this question, as was done in Rasmussen et al. (2018). Stakeholders may also be interested in the simulated location of high LWP for westerly cases and for easterly cases, or at least for the few cases with unblocked flow. Maps of average LWP during good seeding periods may help them to optimize the location of radiometers to guide future seeding operations.

A key conclusion of this study is that the seeding impact on precipitation is modest (Table 3). This is consistent with the modeling study of the impact of the 2014/15 seeding events over the WRR (Tessendorf et al. 2015). Recent randomized seeding experiments (Manton et al. 2011; Rasmussen et al. 2018) and physical process studies (Friedrich et al. 2020; Xue et al. 2022) over different mountains suggest a larger relative increase in precipitation than is found in this study over the WRR. Comparing relative increases from different projects should be done with care, given differences not only in typical environmental, cloud, and precipitation properties that affect seedability (Tessendorf et al. 2020; Mazzetti et al. 2021), but also in seeding methods, snowfall measurement methods, and storm selection process. The finding that just a few cases are responsible for most of the extra snowfall (Figs. 8a,c) is consistent with other studies in this region (Tessendorf et al. 2015; Rasmussen et al. 2018).

A more effective targeting of suitable seeding conditions will increase relative yield, reduce operational cost, and also minimize the impact of AgI seeding on the environment and on water quality. A total of 230 kg of seeding material was dispersed during the 10 years of ground-based seeding. Some parts of the WRR received much more AgI than others (Fig. 10d). When AgI nuclei dissolve in water, they leave behind silver ions, which in sufficient concentration may have harmful ecological and toxicological effects (Klein and Molise 1975; Ćurić and Janc 2013; Fajardo et al. 2016). However, silver strongly adsorbs onto particulate matter in water and in soils, which may explain why residual silver in seeding areas appears similar to background levels (Tsiouris et al. 2002; Williams and Denhom 2009). Nevertheless, it may be wise for the WRR seeding operations to monitor seed material concentrations in snow, in soils, and in streams, for potential ecotoxicity. Such data also provide evidence of effective targeting.

7. Conclusions

This study evaluates the effectiveness of an operational cold-season glaciogenic cloud-seeding program targeting the Wind River Range (WRR) in Wyoming. A total of 190 cases were seeded over 10 cold seasons, using ground-based generators of AgI. This study uses the WRF model at 900 m grid spacing, with the aerosol-aware Thompson cloud microphysics scheme and the WxMod AgI physics scheme. To capture impact uncertainty, an ensemble approach is assumed with 28 members, of which 16 are seeded and 12 unseeded. The main conclusions are as follows:

  • The 10 years of seeding resulted in an estimated increase of precipitation (snowfall) by 1.10% ± 0.13% or (81.2 ± 9.4) × 106 m3 (65 800 ± 7700 a-f) of water over the WRR target area. The uncertainty range is based on ensemble simulations in which the following were varied: initial and boundary conditions, PBL mixing, background CCN concentration, ice initiation parameterization, and AgI sequestration rate.

  • The large majority of the extra precipitation (yield) came from a small fraction of the 190 cases. The highest yield results from cases with unblocked low-level flow (high Fr) and cases with much supercooled liquid water in the orographic clouds. Seeding was minimally effective in most cases. The main reason was low-level blocked flow, making it impossible for the ground-released AgI to reach the orographic clouds.

  • Most cases were westerly, targeted by up to nine generators. A few of these cases produce the largest absolute and relative yield, mostly over the high terrain of the central and southern WRR. Easterly cases, targeted with just one generator, can be very efficient in terms of yield per kilogram of AgI solution dispersed.

Acknowledgments.

This work was funded by the Wyoming Water Development Commission and the U.S. Geological Survey, under the auspices of the University of Wyoming Water Research Program, as well as the University of Wyoming. This paper does not constitute the opinions of the State of Wyoming, the Wyoming Water Development Commission, or the Wyoming Water Development Office. The National Center for Atmospheric Research is sponsored by the National Science Foundation. The numerical simulations were conducted on Cheyenne at the NCAR Wyoming Supercomputer Center using Grant WYOM0078. We acknowledge Weather Modification, Inc., for providing their records of the WRR seeding operations and Courtney Weeks for providing the quality-controlled and corrected radiometer LWP data. This work benefited from insights and comments from Sarah Tessendorf, Roy Rasmussen, Jeffrey French, Jefferson Snider, and two anonymous reviewers.

Data availability statement.

The WRF configuration files and all other input files needed to rerun all simulations are available upon request.

REFERENCES

  • Breed, D., R. Rasmussen, C. Weeks, B. Boe, and T. Deshler, 2014: Evaluating winter orographic cloud seeding: Design of the Wyoming Weather Modification Pilot Project (WWMPP). J. Appl. Meteor. Climatol., 53, 282299, https://doi.org/10.1175/JAMC-D-13-0128.1.

    • Search Google Scholar
    • Export Citation
  • Brown, R., and J. Fitzgibbon, 2016: A preliminary discussion on the national weather service’s upper air data continuity study. 18th Symp. on Meteorological Observation and Instrumentation, New Orleans, LA, Amer. Meteor. Soc., 6B.3, https://ams.confex.com/ams/96Annual/webprogram/Paper285988.html.

  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci., 59, 461483, https://doi.org/10.1175/1520-0469(2002)059<0461:TAOTFT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Chu, X., L. Xue, B. Geerts, R. Rasmussen, and D. Breed, 2014: A case study of radar observations and WRF LES simulations of the impact of ground-based glaciogenic seeding on orographic clouds and precipitation. Part I: Observations and model validations. J. Appl. Meteor. Climatol., 53, 22642286, https://doi.org/10.1175/JAMC-D-14-0017.1.

    • Search Google Scholar
    • Export Citation
  • Chu, X., B. Geerts, L. Xue, and B. Pokharel, 2017a: A case study of cloud radar observations and large-eddy simulations of a shallow stratiform orographic cloud, and the impact of glaciogenic seeding. J. Appl. Meteor. Climatol., 56, 12851304, https://doi.org/10.1175/JAMC-D-16-0364.1.

    • Search Google Scholar
    • Export Citation
  • Chu, X., B. Geerts, L. Xue, and R. Rasmussen, 2017b: Large-eddy simulations of the impact of ground-based glaciogenic seeding on shallow orographic convection: A case study. J. Appl. Meteor. Climatol., 56, 6984, https://doi.org/10.1175/JAMC-D-16-0191.1.

    • Search Google Scholar
    • Export Citation
  • Cooper, W. A., 1986: Ice initiation in natural clouds. Precipitation Enhancement—A Scientific Challenge, Meteor. Monogr., No. 43, Amer. Meteor. Soc., 29–32, https://doi.org/10.1175/0065-9401-21.43.29.

  • Crewell, S., and U. Löhnert, 2003: Accuracy of cloud liquid water path from ground-based microwave radiometry. Part 2: Sensor accuracy and synergy. Radio Sci., 38, 8042, https://doi.org/10.1029/2002RS002634.

    • Search Google Scholar
    • Export Citation
  • Ćurić, M., and D. Janc, 2013: Wet deposition of the seeding agent after weather modification activities. Environ. Sci. Pollut. Res., 20, 63446350, https://doi.org/10.1007/s11356-013-1705-y.

    • Search Google Scholar
    • Export Citation
  • DeMott, P. J., 1997: Report to North Dakota Atmospheric Resource Board and Weather Modification Incorporated on tests of the ice nucleating ability of aerosols produced by the Lohse airborne generator. Colorado State University Department of Atmospheric Science Rep., 15 pp.

  • Dirksen, R. J., M. Sommer, F. J. Immler, D. F. Hurst, R. Kivi, and H. Vömel, 2014: Reference quality upper-air measurements: GRUAN data processing for the Vaisala RS92 radiosondes. Atmos. Meas. Tech., 7, 44634490, https://doi.org/10.5194/amt-7-4463-2014.

    • Search Google Scholar
    • Export Citation
  • Fajardo, C., G. Costa, L. T. Ortiz, M. Nande, M. L. Rodríguez-Membibre, M. Martín, and S. Sánchez-Fortún, 2016: Potential risk of acute toxicity induced by AgI cloud seeding on soil and freshwater biota. Ecotoxicol. Environ. Saf., 133, 433441, https://doi.org/10.1016/j.ecoenv.2016.06.028.

    • Search Google Scholar
    • Export Citation
  • French, J. R., and Coauthors, 2018: Precipitation formation from orographic cloud seeding. Proc. Natl. Acad. Sci. USA, 115, 11681173, https://doi.org/10.1073/pnas.1716995115.

    • Search Google Scholar
    • Export Citation
  • Friedrich, K., and Coauthors, 2020: Quantifying snowfall from orographic cloud seeding. Proc. Natl. Acad. Sci. USA, 117, 51905195, https://doi.org/10.1073/pnas.1917204117.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and R. M. Rauber, 2022: Glaciogenic seeding of orographic clouds to enhance precipitation: Status and prospects. Bull. Amer. Meteor. Soc., 103, E2302E2314, https://doi.org/10.1175/BAMS-D-21-0279.1.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and Coauthors, 2013: The AgI Seeding Cloud Impact Investigation (ASCII) campaign 2012: Overview and preliminary results. J. Wea. Modif., 45, 2443, https://doi.org/10.54782/jwm.v45i1.121.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., L. Xue, and R. Rasmussen, 2017: Evaluation of orographic cloud seeding using a bin microphysics scheme: Two-dimensional approach. J. Appl. Meteor. Climatol., 56, 14431462, https://doi.org/10.1175/JAMC-D-16-0045.1.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., L. Xue, N. Sarkadi, and R. Rasmussen, 2020: Evaluation of orographic cloud seeding using a bin microphysics scheme: Three-dimensional simulation of real cases. J. Appl. Meteor. Climatol., 59, 15371555, https://doi.org/10.1175/JAMC-D-19-0278.1.

    • Search Google Scholar
    • Export Citation
  • Gutman, S. I., J. Facundo, and D. R. Helms, 2005: Quality control of radiosonde moisture observations. Ninth Symp. on Integrated Observing and Assimilation Systems for the Atmosphere, Oceans, and Land Surface, San Diego, CA, Amer. Meteor. Soc., 11.7, https://ams.confex.com/ams/pdfpapers/86707.pdf.

  • Haupt, S. E., R. M. Rauber, B. Carmichael, J. C. Knievel, and J. L. Cogan, 2018: 100 years of progress in applied meteorology. Part I: Basic applications. A Century of Progress in Atmospheric and Related Sciences: Celebrating the American Meteorological Society Centennial, Meteor. Monogr., No. 59, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-18-0004.1.

  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., and L. M. Miloshevich, 1993: Homogeneous ice nucleation and supercooled liquid water in orographic wave clouds. J. Atmos. Sci., 50, 23352353, https://doi.org/10.1175/1520-0469(1993)050<2335:HINASL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 23182341, https://doi.org/10.1175/MWR3199.1.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., R. M. Rasmussen, W. D. Hall, and G. Thompson, 2007: Observations of freezing drizzle in extratropical cyclonic storms during IMPROVE-2. J. Atmos. Sci., 64, 30163043, https://doi.org/10.1175/JAS3999.1.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., and Coauthors, 2010: Simulation of seasonal snowfall over Colorado. Atmos. Res., 97, 462477, https://doi.org/10.1016/j.atmosres.2010.04.010.

    • Search Google Scholar
    • Export Citation
  • Jing, X., B. Geerts, Y. Wang, and C. Liu, 2017: Evaluating seasonal orographic precipitation in the interior western United States using gauge data, gridded precipitation estimates, and a regional climate simulation. J. Hydrometeor., 18, 25412558, https://doi.org/10.1175/JHM-D-17-0056.1.

    • Search Google Scholar
    • Export Citation
  • Klein, D. A., and E. M. Molise, 1975: Ecological ramifications of silver iodide nucleating agent accumulations in a semi-arid grassland environment. J. Appl. Meteor. Climatol., 14, 673680, https://doi.org/10.1175/1520-0450(1975)014<0673:EROSIN>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kusunoki, K., and Coauthors, 2005: Observations of quasi-stationary and shallow orographic snow clouds: Spatial distributions of supercooled liquid water and snow particles. Mon. Wea. Rev., 133, 743751, https://doi.org/10.1175/MWR2874.1.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., E. E. Clothiaux, G. G. Mace, S. Kato, and X. Dong, 2001: A new retrieval for cloud liquid water path using a ground-based microwave radiometer and measurements of cloud temperature. J. Geophys. Res., 106, 14 48514 500, https://doi.org/10.1029/2000JD900817.

    • Search Google Scholar
    • Export Citation
  • Liu, C., and Coauthors, 2017: Continental-scale convection-permitting modeling of the current and future climate of North America. Climate Dyn., 49, 7195, https://doi.org/10.1007/s00382-016-3327-9.

    • Search Google Scholar
    • Export Citation
  • Lohmann, U., J. Henneberger, O. Henneberg, J. P. Fugal, J. Bühl, and Z. A. Kanji, 2016: Persistence of orographic mixed-phase clouds. Geophys. Res. Lett., 43, 10 51210 519, https://doi.org/10.1002/2016GL071036.

    • Search Google Scholar
    • Export Citation
  • Manton, M. J., L. Warren, S. L. Kenyon, A. D. Peace, S. P. Bilish, and K. Kemsley, 2011: A confirmatory snowfall enhancement project in the Snowy Mountains of Australia. Part I: Project design and response variables. J. Appl. Meteor. Climatol., 50, 14321447, https://doi.org/10.1175/2011JAMC2659.1.

    • Search Google Scholar
    • Export Citation
  • Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. John Wiley and Sons, 407 pp.

  • Mazzetti, T. O., B. Geerts, L. Xue, S. Tessendorf, C. Weeks, and Y. Wang, 2021: Potential for ground-based glaciogenic cloud seeding over mountains in the interior western United States and anticipated changes in a warmer climate. J. Appl. Meteor. Climatol., 60, 12451263, https://doi.org/10.1175/JAMC-D-20-0288.1.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, https://doi.org/10.1175/BAMS-87-3-343.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., P. J. DeMott, and W. R. Cotton, 1992: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor. Climatol., 31, 708721, https://doi.org/10.1175/1520-0450(1992)031<0708:NPINPI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Mote, P. W., S. Li, D. P. Lettenmaier, M. Xiao, and R. Engel, 2018: Dramatic declines in snowpack in the western US. npj Climate Atmos. Sci., 1, 2, https://doi.org/10.1038/s41612-018-0012-1.

    • Search Google Scholar
    • Export Citation
  • Nakanishi, M., and H. Niino, 2009: Development of an improved turbulence closure model for the atmospheric boundary layer. J. Meteor. Soc. Japan, 87, 895912, https://doi.org/10.2151/jmsj.87.895.

    • Search Google Scholar
    • Export Citation
  • National Research Council, 2003: Critical Issues in Weather Modification Research. National Academies Press, 123 pp.

  • Niu, G.-Y., and Coauthors, 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res., 116, D12109, https://doi.org/10.1029/2010JD015139.

    • Search Google Scholar
    • Export Citation
  • Oltmans, S., and Coauthors, 2014: Anatomy of wintertime ozone associated with oil and natural gas extraction activity in Wyoming and Utah. Elementa, 2, 000024, https://doi.org/10.12952/journal.elementa.000024.

    • Search Google Scholar
    • Export Citation
  • Pokharel, B., and B. Geerts, 2016: A multi-sensor study of the impact of ground-based glaciogenic seeding on clouds and precipitation over mountains in Wyoming. Part I: Project description. Atmos. Res., 182, 269281, https://doi.org/10.1016/j.atmosres.2016.08.008.

    • Search Google Scholar
    • Export Citation
  • Pokharel, B., B. Geerts, X. Jing, K. Friedrich, K. Ikeda, and R. Rasmussen, 2017: A multi-sensor study of the impact of ground-based glaciogenic seeding on clouds and precipitation over mountains in Wyoming. Part II: Seeding impact analysis. Atmos. Res., 183, 4257, https://doi.org/10.1016/j.atmosres.2016.08.018.

    • Search Google Scholar
    • Export Citation
  • Pokharel, B., B. Geerts, and X. Jing, 2018: The impact of ground-based glaciogenic seeding on a shallow stratiform cloud over the Sierra Madre in Wyoming: A multi-sensor study of the 3 March 2012 case. Atmos. Res., 214, 7490, https://doi.org/10.1016/j.atmosres.2018.07.013.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, R. M., I. Geresdi, G. Thompson, K. Manning, and E. Karplus, 2002: Freezing drizzle formation in stably stratified layer clouds: The role of radiative cooling of cloud droplets, cloud condensation nuclei, and ice initiation. J. Atmos. Sci., 59, 837860, https://doi.org/10.1175/1520-0469(2002)059<0837:FDFISS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, R. M., and Coauthors, 2012: How well are we measuring snow? The NOAA/FAA/NCAR Winter Precipitation Test Bed. Bull. Amer. Meteor. Soc., 93, 811829, https://doi.org/10.1175/BAMS-D-11-00052.1.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, R. M., and Coauthors, 2018: Evaluation of the Wyoming Weather Modification Pilot Project (WWMPP) using two approaches: Traditional statistics and ensemble modeling. J. Appl. Meteor. Climatol., 57, 26392660, https://doi.org/10.1175/JAMC-D-17-0335.1.

    • Search Google Scholar
    • Export Citation
  • Rauber, R. M., and L. O. Grant, 1987: Supercooled liquid water structure of a shallow orographic cloud system in southern Utah. J. Appl. Meteor. Climatol., 26, 208215, https://doi.org/10.1175/1520-0450(1987)026<0208:SLWSOA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rauber, R. M., and Coauthors, 2019: Wintertime orographic cloud seeding—A review. J. Appl. Meteor. Climatol., 58, 21172140, https://doi.org/10.1175/JAMC-D-18-0341.1.

    • Search Google Scholar
    • Export Citation
  • Reynolds, D. W., and A. S. Dennis, 1986: A review of the Sierra Cooperative Pilot Project. Bull. Amer. Meteor. Soc., 67, 513523, https://doi.org/10.1175/1520-0477(1986)067<0513:AROTSC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 21852208, https://doi.org/10.1175/JCLI-D-12-00823.1.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., A. W. Huggins, A. B. Long, J. B. Snider, and R. J. Meitín, 1990: Investigations of a winter mountain storm in Utah. Part II: Mesoscale structure, supercooled liquid water development, and precipitation processes. J. Atmos. Sci., 47, 13231350, https://doi.org/10.1175/1520-0469(1990)047<1323:IOAWMS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., M. P. Clark, R. L. Armstrong, D. A. McGinnis, and R. S. Pulwarty, 1999: Characteristics of the western United States snowpack from snowpack telemetry (SNOTEL) data. Water Resour. Res., 35, 21452160, https://doi.org/10.1029/1999WR900090.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://dx.doi.org/10.5065/D68S4MVH.

  • Tessendorf, S. A., L. Xue, D. Breed, C. Weeks, K. Ikeda, D. Axisa, and R. Rasmussen, 2015: Evaluation of weather modification modeling in the Wind River Range, WY. U.S. Bureau of Reclamation Final Rep., 124 pp., https://www.usbr.gov/research/docs/WRR.pdf.

  • Tessendorf, S. A., and Coauthors, 2019: A transformational approach to winter orographic weather modification research: The SNOWIE project. Bull. Amer. Meteor. Soc., 100, 7192, https://doi.org/10.1175/BAMS-D-17-0152.1.

    • Search Google Scholar
    • Export Citation
  • Tessendorf, S. A., K. Ikeda, C. Weeks, R. Rasmussen, J. Wolff, and L. Xue, 2020: An assessment of winter orographic precipitation and cloud-seeding potential in Wyoming. J. Appl. Meteor. Climatol., 59, 12171238, https://doi.org/10.1175/JAMC-D-19-0219.1.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., and T. Eidhammer, 2014: A study of aerosol impacts on clouds and precipitation development in a large winter cyclone. J. Atmos. Sci., 71, 36363658, https://doi.org/10.1175/JAS-D-13-0305.1.

    • Search Google Scholar
    • Export Citation
  • Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 50955115, https://doi.org/10.1175/2008MWR2387.1.

    • Search Google Scholar
    • Export Citation
  • Tsiouris, S. E., and Coauthors, 2002: Soil silver content of agricultural area subjected to cloud seeding with silver iodide. Fresenius Environ. Bull., 11, 697702.

    • Search Google Scholar
    • Export Citation
  • Udall, B., and J. Overpeck, 2017: The twenty-first century Colorado River hot drought and implications for the future. Water Resour. Res., 53, 24042418, https://doi.org/10.1002/2016WR019638.

    • Search Google Scholar
    • Export Citation
  • Wang, Y., B. Geerts, and C. Liu, 2018: A 30-year convection-permitting regional climate simulation over the interior western United States. Part I: Validation. Int. J. Climatol., 38, 36843704, https://doi.org/10.1002/joc.5527.

    • Search Google Scholar
    • Export Citation
  • Williams, B. D., and J. A. Denhom, 2009: An assessment of the environmental toxicity of silver iodide-with reference to a cloud seeding trial in the Snowy Mountains of Australia. J. Wea. Modif., 41, 7596.

    • Search Google Scholar
    • Export Citation
  • Xue, L., and Coauthors, 2013a: Implementation of a silver iodide cloud-seeding parameterization in WRF. Part I: Model description and idealized 2D sensitivity tests. J. Appl. Meteor. Climatol., 52, 14331457, https://doi.org/10.1175/JAMC-D-12-0148.1.

    • Search Google Scholar
    • Export Citation
  • Xue, L., S. A. Tessendorf, E. Nelson, R. Rasmussen, D. Breed, S. Parkinson, P. Holbrook, and D. Blestrud, 2013b: Implementation of a silver iodide cloud-seeding parameterization in WRF. Part II: 3D simulations of actual seeding events and sensitivity tests. J. Appl. Meteor. Climatol., 52, 14581476, https://doi.org/10.1175/JAMC-D-12-0149.1.

    • Search Google Scholar
    • Export Citation
  • Xue, L., X. Chu, R. Rasmussen, D. Breed, and B. Geerts, 2016: A case study of radar observations and WRF LES simulations of the impact of ground-based glaciogenic seeding on orographic clouds and precipitation. Part II: AgI dispersion and seeding signals simulated by WRF. J. Appl. Meteor. Climatol., 55, 445464, https://doi.org/10.1175/JAMC-D-15-0115.1.

    • Search Google Scholar
    • Export Citation
  • Xue, L., and Coauthors, 2017: WRF large-eddy simulations of chemical tracer deposition and seeding effect over complex terrain from ground- and aircraft-based AgI generators. Atmos. Res., 190, 89103, https://doi.org/10.1016/j.atmosres.2017.02.013.

    • Search Google Scholar
    • Export Citation
  • Xue, L., and Coauthors, 2022: Comparison between observed and simulated AgI seeding impacts in a well-observed case from the SNOWIE field program. J. Appl. Meteor. Climatol., 61, 345367, https://doi.org/10.1175/JAMC-D-21-0103.1.

    • Search Google Scholar
    • Export Citation
Save
  • Breed, D., R. Rasmussen, C. Weeks, B. Boe, and T. Deshler, 2014: Evaluating winter orographic cloud seeding: Design of the Wyoming Weather Modification Pilot Project (WWMPP). J. Appl. Meteor. Climatol., 53, 282299, https://doi.org/10.1175/JAMC-D-13-0128.1.

    • Search Google Scholar
    • Export Citation
  • Brown, R., and J. Fitzgibbon, 2016: A preliminary discussion on the national weather service’s upper air data continuity study. 18th Symp. on Meteorological Observation and Instrumentation, New Orleans, LA, Amer. Meteor. Soc., 6B.3, https://ams.confex.com/ams/96Annual/webprogram/Paper285988.html.

  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci., 59, 461483, https://doi.org/10.1175/1520-0469(2002)059<0461:TAOTFT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Chu, X., L. Xue, B. Geerts, R. Rasmussen, and D. Breed, 2014: A case study of radar observations and WRF LES simulations of the impact of ground-based glaciogenic seeding on orographic clouds and precipitation. Part I: Observations and model validations. J. Appl. Meteor. Climatol., 53, 22642286, https://doi.org/10.1175/JAMC-D-14-0017.1.

    • Search Google Scholar
    • Export Citation
  • Chu, X., B. Geerts, L. Xue, and B. Pokharel, 2017a: A case study of cloud radar observations and large-eddy simulations of a shallow stratiform orographic cloud, and the impact of glaciogenic seeding. J. Appl. Meteor. Climatol., 56, 12851304, https://doi.org/10.1175/JAMC-D-16-0364.1.

    • Search Google Scholar
    • Export Citation
  • Chu, X., B. Geerts, L. Xue, and R. Rasmussen, 2017b: Large-eddy simulations of the impact of ground-based glaciogenic seeding on shallow orographic convection: A case study. J. Appl. Meteor. Climatol., 56, 6984, https://doi.org/10.1175/JAMC-D-16-0191.1.

    • Search Google Scholar
    • Export Citation
  • Cooper, W. A., 1986: Ice initiation in natural clouds. Precipitation Enhancement—A Scientific Challenge, Meteor. Monogr., No. 43, Amer. Meteor. Soc., 29–32, https://doi.org/10.1175/0065-9401-21.43.29.

  • Crewell, S., and U. Löhnert, 2003: Accuracy of cloud liquid water path from ground-based microwave radiometry. Part 2: Sensor accuracy and synergy. Radio Sci., 38, 8042, https://doi.org/10.1029/2002RS002634.

    • Search Google Scholar
    • Export Citation
  • Ćurić, M., and D. Janc, 2013: Wet deposition of the seeding agent after weather modification activities. Environ. Sci. Pollut. Res., 20, 63446350, https://doi.org/10.1007/s11356-013-1705-y.

    • Search Google Scholar
    • Export Citation
  • DeMott, P. J., 1997: Report to North Dakota Atmospheric Resource Board and Weather Modification Incorporated on tests of the ice nucleating ability of aerosols produced by the Lohse airborne generator. Colorado State University Department of Atmospheric Science Rep., 15 pp.

  • Dirksen, R. J., M. Sommer, F. J. Immler, D. F. Hurst, R. Kivi, and H. Vömel, 2014: Reference quality upper-air measurements: GRUAN data processing for the Vaisala RS92 radiosondes. Atmos. Meas. Tech., 7, 44634490, https://doi.org/10.5194/amt-7-4463-2014.

    • Search Google Scholar
    • Export Citation
  • Fajardo, C., G. Costa, L. T. Ortiz, M. Nande, M. L. Rodríguez-Membibre, M. Martín, and S. Sánchez-Fortún, 2016: Potential risk of acute toxicity induced by AgI cloud seeding on soil and freshwater biota. Ecotoxicol. Environ. Saf., 133, 433441, https://doi.org/10.1016/j.ecoenv.2016.06.028.

    • Search Google Scholar
    • Export Citation
  • French, J. R., and Coauthors, 2018: Precipitation formation from orographic cloud seeding. Proc. Natl. Acad. Sci. USA, 115, 11681173, https://doi.org/10.1073/pnas.1716995115.

    • Search Google Scholar
    • Export Citation
  • Friedrich, K., and Coauthors, 2020: Quantifying snowfall from orographic cloud seeding. Proc. Natl. Acad. Sci. USA, 117, 51905195, https://doi.org/10.1073/pnas.1917204117.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and R. M. Rauber, 2022: Glaciogenic seeding of orographic clouds to enhance precipitation: Status and prospects. Bull. Amer. Meteor. Soc., 103, E2302E2314, https://doi.org/10.1175/BAMS-D-21-0279.1.

    • Search Google Scholar
    • Export Citation
  • Geerts, B., and Coauthors, 2013: The AgI Seeding Cloud Impact Investigation (ASCII) campaign 2012: Overview and preliminary results. J. Wea. Modif., 45, 2443, https://doi.org/10.54782/jwm.v45i1.121.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., L. Xue, and R. Rasmussen, 2017: Evaluation of orographic cloud seeding using a bin microphysics scheme: Two-dimensional approach. J. Appl. Meteor. Climatol., 56, 14431462, https://doi.org/10.1175/JAMC-D-16-0045.1.

    • Search Google Scholar
    • Export Citation
  • Geresdi, I., L. Xue, N. Sarkadi, and R. Rasmussen, 2020: Evaluation of orographic cloud seeding using a bin microphysics scheme: Three-dimensional simulation of real cases. J. Appl. Meteor. Climatol., 59, 15371555, https://doi.org/10.1175/JAMC-D-19-0278.1.

    • Search Google Scholar
    • Export Citation
  • Gutman, S. I., J. Facundo, and D. R. Helms, 2005: Quality control of radiosonde moisture observations. Ninth Symp. on Integrated Observing and Assimilation Systems for the Atmosphere, Oceans, and Land Surface, San Diego, CA, Amer. Meteor. Soc., 11.7, https://ams.confex.com/ams/pdfpapers/86707.pdf.

  • Haupt, S. E., R. M. Rauber, B. Carmichael, J. C. Knievel, and J. L. Cogan, 2018: 100 years of progress in applied meteorology. Part I: Basic applications. A Century of Progress in Atmospheric and Related Sciences: Celebrating the American Meteorological Society Centennial, Meteor. Monogr., No. 59, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-18-0004.1.

  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., and L. M. Miloshevich, 1993: Homogeneous ice nucleation and supercooled liquid water in orographic wave clouds. J. Atmos. Sci., 50, 23352353, https://doi.org/10.1175/1520-0469(1993)050<2335:HINASL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 23182341, https://doi.org/10.1175/MWR3199.1.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., R. M. Rasmussen, W. D. Hall, and G. Thompson, 2007: Observations of freezing drizzle in extratropical cyclonic storms during IMPROVE-2. J. Atmos. Sci., 64, 30163043, https://doi.org/10.1175/JAS3999.1.

    • Search Google Scholar
    • Export Citation
  • Ikeda, K., and Coauthors, 2010: Simulation of seasonal snowfall over Colorado. Atmos. Res., 97, 462477, https://doi.org/10.1016/j.atmosres.2010.04.010.

    • Search Google Scholar
    • Export Citation
  • Jing, X., B. Geerts, Y. Wang, and C. Liu, 2017: Evaluating seasonal orographic precipitation in the interior western United States using gauge data, gridded precipitation estimates, and a regional climate simulation. J. Hydrometeor., 18, 25412558, https://doi.org/10.1175/JHM-D-17-0056.1.

    • Search Google Scholar
    • Export Citation
  • Klein, D. A., and E. M. Molise, 1975: Ecological ramifications of silver iodide nucleating agent accumulations in a semi-arid grassland environment. J. Appl. Meteor. Climatol., 14, 673680, https://doi.org/10.1175/1520-0450(1975)014<0673:EROSIN>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kusunoki, K., and Coauthors, 2005: Observations of quasi-stationary and shallow orographic snow clouds: Spatial distributions of supercooled liquid water and snow particles. Mon. Wea. Rev., 133, 743751, https://doi.org/10.1175/MWR2874.1.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., E. E. Clothiaux, G. G. Mace, S. Kato, and X. Dong, 2001: A new retrieval for cloud liquid water path using a ground-based microwave radiometer and measurements of cloud temperature. J. Geophys. Res., 106, 14 48514 500, https://doi.org/10.1029/2000JD900817.

    • Search Google Scholar
    • Export Citation
  • Liu, C., and Coauthors, 2017: Continental-scale convection-permitting modeling of the current and future climate of North America. Climate Dyn., 49, 7195, https://doi.org/10.1007/s00382-016-3327-9.

    • Search Google Scholar
    • Export Citation
  • Lohmann,