I. Precipitation Enhancement
Scientists in RAP continued to pursue research that strengthens the scientific basis of rainfall enhancement efforts. In FY04, several countries have contacted RAP scientists for guidance and advice on potential weather modification programs. Figure I-1 shows countries or regions where the RAP group has either had projects or provided an assessment and/or advice on weather modification efforts in FY04.
Major efforts in FY04 have involved field projects in the United Arab Emirates (UAE) and Saudi Arabia, with other efforts being short-term or relatively small in scope. The hygroscopic flare test facility was improved, and is ready for more extensive testing. Numerical modeling efforts also increased, particularly in implementing the WRF model for tracer studies in complex terrain. Much of the work is still in progress, particularly since the field projects were only completed in late Summer 2004. Primary scientists and managers in the field projects are Daniel Breed, Tara Jensen, and Vidal Salazar. The UAE project statistician is Tressa Fowler. Technical assistance has been provided by Nancy Rehak and Tres Hofmeister, as well as by engineers in ATD.
The projects and assessments throughout the world have incorporated scientists and research groups in each country as well as U.S. university researchers. Both formal (e.g., seminars, workshops) and informal (e.g., one-on-one interactions) training opportunities allowed RAP scientists and engineers to transfer knowledge and responsibilities for data collection, analyses, and archival related to weather modification research and operations.
Over the last several years, a hygroscopic flare test facility was developed and constructed to characterize the output of various hygroscopic flares in a controlled and reproducible environment. Critical aspects of the facility design are: to allow adequate cooling of the flare output to closely simulate the conditions encountered during airborne operations and thereby simulate the initial particle formation processes and to dilute the flare smoke to concentration levels that are measurable by the instrumentation (i.e., Particle Measuring Systems probes). One of the most important characteristics of a hygroscopic flare is the relative concentration of large particles, yet this is the most difficult to measure. V. Salazar investigated the theoretical performance and limitations of the original facility design, and redesigned the ductwork, airflow, and dilution control of the facility to address potential losses of large particles. The original facility design involved ductwork that incorporated 180° bends in small (diluting) pipe sections, as seen in Figure I-2a. Construction of the improved design eliminated pipe bends and permitted a free travel path for all particle sizes. The only loss in this sampling procedure should be due to gravity. The changes that V. Salazar made, shown in Figure I-2b, also improved the cooling rate of the flare to more closely simulate that in an aircraft environment and matched the dilution airflow speeds to the sample airflow. This redesigned facility has been used to test various flare compositions.
Figure I-2a. Original design of the flare testing facility with two-stage dilution. Note the 180° bends (1 and 2) that potentially cause losses of larger particles.
Figure I-2b. Redesigned flare testing facility schematic above and photo of setup below. Measuring station is at the end of the duct (on the right in photo) after two stages of dilution. [Top]
2.2 Preliminary results of flare testing
Earlier tests of flare characteristics, mostly involving particle size distributions, established that the facility succeeded in providing a reproducible environment for the combustion of flares and for the measurement of the resultant particles. However, the relative lack of large particles produced by the flares led to further investigation into the cause: either there were no large particles or they were not being measured adequately. With the redesigned facility, flares are now being re-tested. An example of one test is presented in Figure I-2c and shows that large particles (>1 µm) are being produced by the flare. The combination of measuring instruments, a PCASP (Passive Cavity Aerosol Spectrometer Probe) and a modified FSSP (Forward Scattering Spectrometer Probe) (SPP 100), is slightly different than previous tests, and so a direct comparison between the original facility design and the redesign requires more tests. However, other flare characteristics are consistent with previous tests. Burn time for the Ice Crystal Engineering (ICE) flare is approximately 3 min and has some transient (and interesting) spectral changes at the end of the burn. The majority of the particles are in the 0.2-0.4 µm range. Deviations occur during the burn, possibly due to temperature changes as the flare casing burns away, that cause a decrease in small particles and a correlated increase in larger particles (>0.5 µm). This latter point is particularly evident in the SPP-100 measurements.
Figure I-2c. Time-series of particle spectra measurements from the combustion of an ICE flare containing 70% potassium perchlorate: left, from a PCASP, size range 0.1-3.0 µm; and right, from a SPP-100, size range 0.5-8.0 µm.
1) winter clouds in the UAE rarely produced conditions amenable to hygroscopic cloud seeding;
2) summer convective clouds developed often enough to justify a randomized seeding experiment;
3) collecting quantitative radar observations continues to be a complex but essential part of evaluating a cloud seeding experiment;
4) successful flight operations would require solving several logistical problems; and
5) several scientific questions be studied to fully evaluate the efficacy and feasibility of hygroscopic cloud seeding, including cloud physical responses (e.g., formation of drizzle drops), radar-derived rainfall estimates as related to rainfall at the ground, and hydrological impacts (e.g., groundwater recharge, comparative economic analyses).
Based on these results, the Department of Water Resources Studies (DWRS) in the UAE and NCAR/RAP proceeded with Phase II, which involved designing and implementing a randomized hygroscopic cloud seeding experiment during the summer season to quantify statistically the potential for cloud seeding to enhance rainfall, specifically over the UAE and Oman Mountains, while collecting measurements to support the statistical results and provide substantiation for the physical hypothesis. The University of Witwatersrand (WITS) in South Africa, along with the South African Weather Service, has been an integral part of the overall program by providing the operational components of the program (aircraft, instrumentation, pilots, technicians, etc.) and as scientific collaborators. For example, the close work with professors and students at WITS led to the appointment of Bruintjes as honorary professor at WITS and also resulted in a post-doc visit to NCAR for a WITS graduate, Kristy Ross. [Top]
An Experimental Design document and an Operations Plan were developed for the UAE project, which detailed procedures for choosing, treating, and evaluating storms (or cases) in the randomized seeding experiment. In addition, microphysical measurements to investigate the impact of seeding on precipitation development, cloud droplet broadening, and raindrop size distributions were designed to reveal critical physical links in the seeding conceptual model. Two aircraft were used: a seeding aircraft and a research aircraft (also capable of seeding). An AeroCommander 690A was instrumented with a suite of cloud microphysics instrumentation in addition to GPS and state variables (e.g., pressure, temperature, relative humidity, airspeed) with a telemetry system for real-time display of position in the operations center. The seeding-only aircraft, a Piper Cheyenne, was equipped with a GPS, state variables, and the telemetry system. Figure I-3 shows photos of the two aircraft in flight. The primary radar for operations and evaluation was a 5-cm Doppler radar located at the Al Ain airport. Working in conjunction with ATD, an NCAR receiver/processor was added to the radar which consisted of a digital receiver and a PC-based Doppler processor board. The processor is relatively easy to configure with parameters that are understandable (i.e., relevant to radar engineering) and regularly recorded (i.e., with each radar beam), and allowed for well-calibrated and consistent radar data to be collected. For example, the NCAR processor monitors the transmitted power of the radar and automatically adjusts the reflectivity calculation to account for transmitter fluctuations.
Figure I-3. The two aircraft used in the UAE project: left, the Aerocommander equipped for cloud microphysics measurements and seeding, and right, the Cheyenne equipped for seeding. Photos were taken during a formation flight for comparing state variables recorded on each aircraft.
University collaborations established during
Phase I of the UAE project continued with aerosol particle characterization
work performed by researchers at Arizona State University and University
of Arizona. Microphysical modeling parameterizations and studies continue
to be performed in conjunction with Richard Farley of South Dakota School
of Mines and Technology. During the field project, various training
opportunities were realized. Hamid Al Brashdi, a meteorologist with
the Oman Weather Service and a WITS student, actively participated in
the operations as a flight observer and radar operations coordinator.
Workshops and tutorials were given on the MM5 model (setup and maintenance
of the forecast version), aircraft instrumentation, radar meteorology,
and the Titan/CIDD software system. Occasional presentations were also
made for visiting officials from other agencies in the UAE and from
Oman. In 2004, an outstanding opportunity for collaboration developed
with the concurrent Unified
Aerosol Experiment-United Arab Emirates (UAE2). This experiment,
initiated between WITS, DWRS and NASA Goddard Space Flight Center (GSFC),
builds on and extends aerosol work performed in Phase I, allowing more
comprehensive studies (for the UAE project) on the role of aerosol interactions
with clouds. RAP scientists developed new collaborations with researchers
from NASA/GSFC, the Jet Propulsion Laboratory, the Naval Research Laboratory,
Scripps Institution of Oceanography, the University of Hawaii, and the
University of Maryland.
A total of fifty-nine randomized seeding experiments were performed during the 2003 summer season, and seventy-six cases were flown in 2004. Sixteen and eighteen coordinated microphysical experiments were performed in 2003 and 2004 respectively, although the 2004 cases were far better sampled than the 2003 cases. Analysis of the microphysical flights is ongoing, and details of the 2004 randomized seeding cases still need to be examined. However, preliminary results exist for the 2003 cases.
The locations of the randomized seeding cases in 2003 are shown in Figure I-4. As expected from the Phase I feasibility study, the cases are distributed primarily to the east and southeast of Al Ain. Of the fifty-nine randomized seeding cases selected in 2003, forty-five met the analysis criteria established in the Experimental Design. Fourteen cases did not qualify for four reasons:
1) ten never developed a Titan track (i.e., thresholds of 30 dBZ in reflectivity and 20 km3 in volume size);
2) two were tracked for only one volume scan (two volume scans are necessary for an integration to yield rain mass);
3) one was outside the coverage area (more than 140 km away from the radar); and
4) one track was treated twice, though more than one hour apart, and could only be counted once.
Figure I-4. Locations of the randomized seeding cases in 2003 flown by the two aircraft (called Research 1 and Research 2).
The distribution of rain mass from the forty-five cases in 2003 is similar to that from the 2001-2002 samples, which T. Fowler used to estimate the number of cases needed to show a significant outcome of the randomized hygroscopic seeding experiment. A useful plotting technique for displaying distributions is the box and whisker plot, which will simply be referred to here as a box plot. It is capable of showing several statistical attributes of the measurements. The rain mass distributions for the samples from 2001-2002 and 2003 are given in box plot form in Figure I-5 2003 for 45 Titan tracks and 2001-2002 for 715 Titan tracks. As expected, the 2003 cases consisted of fewer small storms than the general population since reasonably well developed clouds were targeted for treatment. Also, fewer large storms were in the 2003 sample due mostly to the smaller sample size. The rain mass amounts from storms in the 2003 season is slightly higher than the values from the 2001-2002 seasons, although not significantly so (i.e., the 95% confidence intervals overlap). More importantly, the variability in the rain mass amounts from the 2003 season is less. The required number of cases for detecting an expected result from hygroscopic seeding was estimated prior to the 2003 field project using the 2001-2002 data. So, the smaller variance in the rain mass measurements for 2003 suggests that they were selected consistent with the assumptions from the 2001-2002 samples, validating the original estimates. [Top]
Figure I-5. Box plots of the distributions of rain mass measurements (on a natural log scale in ktons) for storms from the 2003 season versus the 2001-2002 seasons combined. [Box plot explanation: The center of the box is the median. The notch in the box represents an approximate 95% confidence interval for the median. The top and bottom of the box represent the 75^th and 25^th percentiles, respectively. The capped whiskers extend to the most extreme values (i.e. maximum and minimum).]
Of the forty-five valid cases from 2003, twenty-three were seeded and twenty-two were left unseeded. The rain mass and duration of the storms were calculated by the Titan storm tracking software. Distributions of rain mass values for seeded and unseeded cases are shown by the box plots in Figure I-6. No significant difference exists between the seeded and unseeded cases.
Figure I-6. Box plots of rain mass for the 2003 unseeded and seeded storms. Rain mass is in kilotons on a natural log scale.
Consistent treatment of each case, whether seeded or unseeded, is critical to the success of a randomized experiment. One measure of this consistency is to compare the intervals between the start of the storm track and decision time (time of declaration of a case) for the seeded and unseeded groups. This should show at what stage of storm development the treatment occurred. The results show that the typical interval from the start of the track to the decision time did not differ between the seeded and unseeded storms. This indicates that the two groups were treated consistently as far as timing of storm development is concerned. [Top]
4.1 Program overview
The overall objective of the feasibility study for the augmentation of rain in the Kingdom of Saudi Arabia is to develop, test, implement and transfer the technology of hygroscopic seeding. The first step in achieving this broad objective was to characterize the development of convection and precipitation in the area of study (Figure I-7) and compare this to storm characteristics in other regions where cloud seeding has been successfully evaluated. V. Salazar, with assistance from Roelof Bruintjes, participated in a month-long field project (July into early August of 2004) to collect data on aerosols, cloud formation, and precipitation development near Abha. Interactions included training and collaboration with scientists from the Presidency of Meteorology and Environment.
Figure I-7. Map of the Kingdom of Saudi Arabia area of study.
The eastern slope of the mountain range in Asir, where the field project was focused, is gently sloped, melding into a plateau region that drops gradually into the Rub al Khali. Although rainfall is infrequent in this area, several fertile wadis, make oasis agriculture possible on a relatively large scale. The escarpment on the west slope of the Asir range is very abrupt as it falls away to the Red Sea. Rainfall is more plentiful along the escarpment due to strong orographic lifting in westerly flow regimes.
The research aircraft carried an instrument package designed to measure parameters needed to evaluate the seeding potential for Saudi Arabian clouds. These include microphysical instruments to assess the natural and seeded precipitation processes, and instruments to characterize aerosols and the local thermodynamic structure of the atmosphere in which clouds develop. The aircraft sampled both unseeded and seeded clouds.
4.2 Preliminary data analysis
A total of thirty-three research flights were conducted between 1 July and 5 August 2004, totaling fifty-five flight hours. Thirty-eight of those hours were spent conducting seeding experiments and collecting cloud microphysical properties as well as aerosol sampling, while seventeen hours were primarily used for aerosol sampling. Three-hundred-forty hygroscopic flares were consumed during the cloud seeding experiments. Cloud penetrations were made mainly over the Asir mountain range but with significant investigations in clouds over the plains, as seen in Figure I-8. [Top]
Figure I-8. Map of the Asir region in southwest Saudi Arabia showing locations of aircraft cloud penetrations from 3 July to 5 August 2004. Background shows elevation relief and red crosses show each cloud penetration.
Aerosol measurements occurred on nearly all the flights. Figure I-9 shows PCASP-measured concentrations of aerosols and sizes with height near the Kamis Mushait Air Base. The decrease in concentrations of small particles near 3000 m coincides with the approximate depth of the boundary layer. There is also an indication of an increase in larger particles just above the boundary layer. Other layering is evident in this composite plot, though some apparent changes are due to sampling and partitioning resolution. However, the small particle layer near 4000 m is real, and was observed visually.
Figure I-9. Aerosol number concentrations with height over Kamis Mushait Air Base as a function of size distribution measured with the PCASP. Note that the size scale on the x-axis is not linear they reflect each channel of the PCASP. Also, ZN = Log10 N(r) (per micron).
At the request of the Ministry of Regional Municipalities, Environment and Water Resources, R. Bruintjes and D. Yates developed a Review and Assessment of the Potential for Cloud Seeding to Enhance Rainfall in the Sultanate of Oman. The report emphasized the need for a climatology of cloud systems and precipitation along with a hydrological component to assess economic and social impacts of any potential cloud seeding experiment. Elements of a rainfall enhancement research effort were identified:
A preliminary study of the weather and rainfall characteristics established some parameters for further study. For example, a comparison of the winter and summer average rainfall totals suggest that summer convective rainfall over the Al Hajar Al Gharbi mountains (in northern Oman) is a slightly higher percentage of the total annual rainfall than that contributed during the winter months. The percentage of days in the month when it rained in July and August (Figure I-10 ) show a gradient from the high percentages in the south, where the highest mountains reside, to the north. During July and August, a higher percentage of rainfall occurred in the south than in the north. These and other data were fed into a hydrological model to investigate ranges of theoretical increases to ground water from rainfall enhancement. These studies were designed to show potential capabilities for further studies given more complete data sets.
Figure I-10. Percentage of days with rainfall in July and August from 1994 to 1999 over northern Oman. Gray dots represent rain gauge locations.
During late Summer of 2003, Weather Modification Inc. conducted an operational cloud seeding project to enhance rainfall using hygroscopic and silver iodide flares in the State of Karnataka in India. An aircraft instrumented with a cloud physics research package was used to collect cloud microphysical data in natural and seeded clouds. R. Bruintjes provided guidance and field project support for the cloud studies, and some preliminary analyses on the data collected with the aircraft and the project radar. He also presented a seminar/workshop to local scientists, in collaboration with the India Institute of Tropical Meteorology.
The airborne measurements in Karnataka, although limited, showed some clear evidence of precipitation processes in a polluted environment. The primary conclusions can be summarized as follows:
Figure I-11. Aerosol (red line) and CCN (blue line) concentrations as a function of altitude over Karnataka on 14 October 2003. Aerosol measurements are from the PCASP, and the CCN measurements were taken at 0.6% supersaturation.
Figure I-12. 2-D images of larger columnar ice crystals, originally produced through the Hallet-Mossop ice multiplication process. The distance between the vertical bars is 0.8 mm. The bottom panel provides text data for in-flight conditions, showing ice particle concentrations of greater than 120/liter.
In conjunction with Weather Modification Inc., a feasibility study of cloud seeding programs for winter snowpack augmentation in Wyoming is currently underway. The Wyoming Water Development Commission is specifically interested in two project areas, the Wind River Range and the Sierra Madre/Medicine Bow Ranges. RAP is responsible for the numerical modeling portion of this study, led by T. Jensen and R. Bruintjes.
Typically, a winter orographic cloud seeding operation consists of a network of generators upwind from the target area, which release silver iodide particles or liquid propane to enhance ice crystal growth. The objective is to intercept clouds containing supercooled liquid water when they are forecast to pass over the target area. The modeling effort is designed to verify this seeding protocol and enhance operations by revealing the influences of local circulations, by identifying optimal locations for seeding equipment, and by improving on the forecast guidance from coarse resolution models. The WRF model has been configured for Wyoming and incorporated a tracer to simulate seeding releases. The nested grid design is shown in Figure I-13 and case studies are being run to validate the model setup. Scenarios will then be run to address the objectives listed above.
R. Bruintjes visited the Puglia region
of Italy to conduct initial discussions with AEROTECH (a weather modification
contractor) and the relevant authorities in the Puglia region and to
survey several sites in the region. Discussions
The following components were identified as important in providing a scientific foundation for the Puglia precipitation enhancement program:
Particular emphasis was placed on upgrading
the radar capabilities, and on implementing a comprehensive numerical
Figure I-14. Mean annual rainfall in the Puglia province of Italy from the period 1991 through 2000. Puglia is in the southeast of Italy, and except for the Promontorio del Gargano mountain in the northwest corner, consists mostly of rolling plains. The Apennine Mountains make up the western boundary.
5.5 Burkina Faso
RAP assisted Weather Modification Incorporated in configuring and installing the Titan software on the Burkina Faso radar used in their weather modification operation. N. Rehak provided the software support and contributed to the training program for the Bukina Faso scientists.
At the request of the Chinese Academy of
Sciences, R. Bruintjes visited China to assess the countrys
many programs in weather modification. Presentations and workshops were
presented to numerous groups, such as the Chinese Meteorological Administration,
the Beijing Meteorological Bureau, and provincial meteorological offices.
The weather modification efforts in China are extensive, involving 35,000
people and covering nearly every province, as seen from the map in Figure
I-16 . A comprehensive assessment was proposed as a first step
toward coordinating and managing the various programs. R. Bruintjes
visit resulted in a collaborative visit to NCAR by Dr. Xueliang Guo
of the Chinese Academy of Sciences.
Figure I-16. Map of weather modification programs in China, as of 2002.