* Corresponding author address :

Submitted to the Journal of Applied Meteorology, 1998
 
 

A warm cloud modification experiment was carried out in an area of 4800 Km² in the Pune region (18^O 32'N, 73^O 51'E, 559 m asl) during the 11-summer monsoon (June-September) seasons (1973-74, 1976, 1979-86). A double-area cross-over design with area randomization was adopted and an instrumented aircraft was used for seeding and cloud physical measurements. Finely pulverised salt (sodium chloride) particles were released into the monsoon clouds (cumulus and stratocumulus) during aircraft penetrations into the clouds at a height of 200-300 m above the cloud-base. The results of the Indian Experiment have clearly emphasized the need for the physical understanding, sequential development (stepwise programmes to test the applicability of the warm cloud modification hypothesis), predictor variables, model simulations for obtaining conclusive results. The warm cloud responses to salt seeding are found to be critically dependent on the cloud physical characteristics e.g. , vertical thickness and liquid water content. Clouds with vertical thickness 1 km, LWC 0.5 gm m^-3 when seeded with salt particles (modal diameter 10 m m, concentration 1 per litre of cloud air) produced increase in rainfall of 24 per cent significant at 4 per cent level. Shallow clouds (vertical thickness < 1 km, LWC < 0.5 gm m^-3) when seeded showed tendency for dissipation. The cloud physical observations made in not-seeded (control) and seeded (target) clouds have provided some useful evidence to test the applicability of the warm cloud modification hypothesis. The results of the cloud model computations suggested that moderate convergence at the cloud-base is essential for the cloud growth and development of precipitation in the real world. Hygroscopic particle seeding of warm clouds under favourable dynamical conditions (convergence at the cloud-base level) may result in the acceleration of the collision-coalescence process resulting in the enhancement of rainfall.

The results of the 11-year Indian warm cloud modification experiment are useful for the planning of good field experiments in future. For obtaining conclusive results it is essential to prove the persistence of the seeding result through atleast two phases with the supporting stepwise cloud physical programmes to test the applicability of the warm cloud modification hypothesis and cloud simulation studies.
 
 

1. Rationale

Two major randomized warm cloud seeding experiments were carried out in India. The first experiment (from hereafter called as Exp-1) was carried out during 1957-66 in the Delhi, Agra and Jaipur regions located in the plains of northwest India. Seeding was carried out during the summer monsoon months of July to September when the prominent clouds were cumulus and stratocumulus with their tops not exceeding the freezing level. The results of the statistical analysis of Exp-1, indicated increase in rainfall on seeded days, on the average, by about 20 per cent significant at less than 0.5 per cent level. (Biswas et al., 1967, Ramanamurty and Biswas, 1968). Radar observations of the precipitation development in the clouds in the target and control areas were also made during the later part (1961-65) of Exp-1. The results of the radar observed cloud areal echo coverage indicated an overall positive result for seeding (Chatterjee et al., 1969). A fixed control - target design with day randomization was adopted for Exp-I. A brief summary of the design of Exp-I and the details of the seeding methodology used in the experiment are described in the following.

The areas of the target and control sectors in the three regions of Exp-I varied between 450 and 1270 km² and the density of the raingauge network varied from 1 gauge per 50 - 300 km². Seeding was carried out either by spraying from the ground a dilute salt solution using power sprayers and air compressors, or by dusting a finely powdered mixture of salt and soapstone in the ratio 10:1 (Biswas et al., 1967). The model radius of the salt particles was 5m m. The estimated dispersal rate at the source was approximately 2 x 10¹⁰ salt particles (radius 5m m) per second. The control and target areas were defined upwind and downwind of the central seeding locations and comparisons were made between the rainfall in these two areas for seeded (Target) and not-seeded (Control) days. Seedable days were selected on the basis of certain meteorological criteria, particularly in respect of low cloud amount, wind shear and humidity in the lower levels. Days on which rain occurred frequently or continuously were not considered as seedable days. Hence, it is unlikely that the rainfall recorded in the control and target areas could be from the tall convective clouds extending well above the freezing level which involve ice phase. The above hypothesis is further corroborated from the results of the analysis of 7287 aircraft reports of the meteorological observations of monsoon clouds collected during 1948 - 1951 which indicated that more than 90 per cent of the low cloud-tops lie below the freezing level during the year in India (Pramanik and Koteswaram, 1955; Devara and Ramanamurty, 1982).

The Exp-I has apparently provided the statistical evidence to show that salt seeding may have modified the precipitation in spite of other limitations, e.g.,. ground-based generators used for seeding, lack of the physical evidence in support of the seeding hypothesis persuasive of the statistical evidence of increases in precipitation over an area. The limitations have been discussed by some (Mason, 1971; Warner, 1973; Cotton, 1982). Warner (1973) argued that the results of Exp-I are ambiguous particularly due to the lack of the physical evidence in support of a hypothesis that precipitation from warm clouds can be increased through salt seeding technique.

In order to verify the statistical results obtained from Exp-I and for obtaining the requisite physical evidence for the warm cloud seeding hypothesis, a well designed randomized "Warm Cloud Modification Experiment" with good cloud physical measurements programme was carried out in Maharashtra State during the 11-summer monsoon seasons (1973-74, 1979-86). From hereafter this second Indian cloud seeding experiment is referred to as Exp-II. A DC-3 aircraft instrumented for cloud physical measurements was used for seeding. The physical measurements carried out in not-seeded (Control) and seeded (Target) clouds were used for documenting the warm cloud responses to seeding (physical evaluation). The results of the various studies carried out as a part of Exp-II are presented in this paper.
 
 

2. Design

A cross-over design having two sectors with a buffer in-between has been adopted. The three sectors have been designated as North (N), South (S), and Buffer (B) sectors (Figure 1).

The area of each sector is 1600 km². In the crossover design paired target areas are set-up and either area is seeded at random (area randomization), in each test event, the unseeded area serving as the control for that event. The data are obtained in the form of two series. One of the two areas is kept as target in a series and the other acts as control and vice-versa for the other series. The affect of seeding can be obtained from the root-double ratio (RDR) which can be expressed as

where N and S denote the average rainfall in the North and South sectors and the subscripts S and NS denote the seeded and not-seeded days respectively. When the North area (N) is allocated for seeding (Target) correspondingly the south area (S) is allocated for not-seeding (Control). Before the commencement of the experiment in each year a series of random numbers (Fisher and Yates, 1953) was taken and used for the allocation of the seeding of the North and the South sectors. Each series used for the experiment in any year was subjected to randomization tests for avoiding any possible bias due to the repetition of the series. In an experiment of sufficient duration the root-double-ratio provides an estimate of the factor by which the mean rainfall has been increased by seeding. The expected value would be close to 1.0 if the seeding has no effect.

The cross-over design minimises the noise of the natural variability because the fluctuations of the rainfall in the seeded area, to some extent, get neutralised by the parallel fluctuations in the highly correlated control areas. Pairwise randomization scheme is employed with the cross-over design for preventing possible chain of seeding events over the same area, to mitigate the persistence effect and thus prove its sensitivity and efficiency (Moran, 1959). This design is considered to be the most efficient and requires a high correlation between the rainfall of the target and control areas. The provision of the buffer area of the same size as the target and control areas would ensure any possible effects due to contamination.

3. Experiment Area

The experimental area is located on the Lee-side of the Western Ghats in the Deccan Plateau region at about an altitude of 550 m. It is about 40 km east of Pune (18^O32' N, 73^O 51'E, 559 m asl) and about 120 km from the west coast at Bombay. The experimental area is perpendicular to the westerly monsoon flow and it consists of three sectors North, South and a Buffer in between the target and control sectors (Figure 1). The dimensions of the North, South and the Buffer sectors are identical. The total area of the three sectors is 4800 km².

4. Meteorological Conditions

The experimental area is located in the path of the monsoon westerlies. Prominent weather developments take place in the region when there is a trough of low pressure off the west-coast. The region also experiences rainfall when the axis of the monsoon trough in the mid-troposphere (2.5 to 3.5 km) is situated along a more southerly latitude (19^O - 20^O N). The experimental area is situated in the semi-arid zone on the lee-side of the Deccan Plateau with the average annual rainfall less than 60 cm. About 80 per cent of the annual rainfall is received during the summer monsoon season (June - September).

Rain seems to fall primarily from the clouds below 3 to 4 km. Once the monsoon is established, the cumulonimbus clouds are practically absent. The freezing level in the experimental area during the summer monsoon months is at about 6 km and a large majority (more than 90 per cent) of the clouds do not reach higher than 5 km (Pramanik and Koteswaram, 1953). Hence, the dominant rain-forming process in these clouds is the collision-coalescence process. There are apparently a number of occasions when the warm cumulus clouds forming in the region do not give any rain.

5. Raingauge Network

In the Experimental area 90 standard type meteorological raingauges were installed and their distributions in the three sectors of the Experimental area are as follows : North Sector (36), South Sector (34) and Buffer (20). In the North and South sectors of the Experimental area, the density of the raingauge network is about 1 per 40 km² and in the Buffer sector it is about 1 per 80 km². The above raingauge network was installed and maintained by the India Meteorological Department (IMD). The 24-hour daily rainfall data recorded by these raingauges were obtained by the IMD. After scrutiny checks for the reliability of the rainfall data by the IMD, the data were supplied to the Institute for the statistical analysis and evaluation of the results of the Experiment. The 24-hour rainfall measured from 0800 AM of the given day (seeded) to 0800 AM of the next day was used in the analysis. Various investigators envisage the possible after effects of seeding and therefore, the inclusion of the night following a day with seeding does not seem objectionable (Neyman, 1980).

6. Rainfall Correlations

Historic rainfall data were available for 6 raingauge stations located three each in the North and South sectors of the Experimental area prior to the commencement of Exp-II. The six raingauge stations are part of the national network of the raingauge stations maintained by the India Meteorological Department. The monthly rainfall data obtained from these 6 raingauge stations for the 24-summer monsoon (June-September) seasons (1946-62 and 1964-70) were used for the computation of the correlation coefficients (Table 1) and also for the numerical simulation of the cloud seeding experiments (Twomey and Robertson, 1973, Mary Selvam et.al., 1978) carried out for evaluating the chances of detection of the prescribed increases in the rainfall due to seeding with a specified degree of confidence. The results of these numerical simulation experiments are presented in Section 12 below.

Also, the daily 24-hour rainfall data obtained from the 90 raingauge stations located in the experimental area (North sector 36, South sector 34, and Buffer sector 20) on the 284 days of the cloud seeding experiment carried out during the 11-summer monsoon seasons (1973, 1974, 1976-86) were utilised to compute the correlation coefficients and the results are furnished in Table 1.

7. Seedable Days

The classification of the seedable days has been based on the following criteria : (i) Forecast amount of low clouds (ii) forecast winds, (iii) special radiosonde observations carried out at Pune a few hours before the commencement of the actual seeding, (iv) current weather conditions particularly in respect of cloud formation and development as inferred from the synoptic charts, (v) meteorological debriefing reports obtained from the aircraft reconnaissance flights in the region, (vi) special current weather observations recorded at two stations (one each in the North and South sectors, viz, Ahmednagar in the North sector and Baramati in the South sector) in the Experimental area, a few hours (4-6 hours) before the commencement of the actual seeding. A day has been considered as seedable when the forecast is 3 Okta or more of the low clouds in case of (1) westerly wind with speed not exceeding 20 knots up to a height of 3 km a.s.l. in case of (ii), when the relative humidity is more than 75 per cent in the lower atmosphere (up to 700 mb) in case of (iii) and the synoptic conditions are favourable for the formation of low-clouds in one or more of the cases at (iv), (v) and (vi).

8. Seeding Aircraft and Instrumentation

The aircraft used for seeding was a Dakota (DC-3) which was fitted with the seeding equipment (Figure 2) and several instruments for obtaining the cloud physical observations.

The seeding equipment consisted of a funnel fitted inside the aircraft. The funnel is coupled through a venturi, to a dispensing duet assembly which is fitted to the fuselage of the aircraft (Figure 3). The funnel ends in an adjustable slit which can be operated by a calibrated mechanical gate valve arrangement fitted inside the aircraft. The funnel can accommodate at a time 150 Kg of the seeding material.

The seeding equipment operates due to the pressure developed inside the venturi during the aircraft flight. Also, at the base of the funnel, just above the slit, an agitator which operates at 300 r.p.m., was fitted for facilitating free flow of the salt seeding material. The rate of dispersal of the salt mixture can be adjusted to any value between 0 and 30 kg per minute or 0 to 30 kg per 3 km of the aircraft flight path. The cruising speed of the aircraft was about 180 kmph. A photograph of the plume of the seeding material released from the aircraft into the clear air is shown in Figure 4.

The details of the instruments fitted to the aircraft for making cloud physical measurements in seeded (target) and not-seeded (Control) clouds are furnished in Table 2.
 
 

Table 2 : Details of aircraft instruments used for the measurement of the cloud physical
                 parameters  during the experiment
 
 
 

                                              physical parameters during the experiment
 
 

The cloud electrical and physical parameters were recorded using a data logger consisting of the electronic equipment and a multichannel recorder. In addition to the above visual observations and cloud photography were also carried out to document important cloud conditions during the experiment.

9. Seeding Material and Seeding Methodology

The seeding material consisted of finely pulversied mixture of sodium chloride (salt) and soapstone (hydrated silicates and carbonates of magnesium) in the ratio 10:1. The median particle mass is approximately 10^-9 gm corresponding to a dry particle diameter of 10 m m. Analysis of the salt particle size distribution indicated that about 75 per cent of the particles had diameters less than 10 micrometer. The salt seeding material was released into the clouds during the aircraft penetrations at a height of 200 - 300 m above the cloud-base. The level of aircraft flights was maintained constant during the seeding operations. The rate of seeding as stated in section 7 varied between 0 and 30 kg per 3 km aircraft flight path.

10. Physical Hypothesis

The physical processes involved in the initiation and development of rain in warm clouds are condensation collision - coalescence, and break-up. The concept of warm cloud modification to increase rainfall is based on the modification of rain processes through seeding the clouds with either a hygroscopic material thereby tapping the potential precipitation efficiency of the cloud systems.

11. Seeding Techniques and Evaluation Methodologies

Simpson (1978) stated that successful weather modification experiments share three outstanding features, namely persistence through at least two phases, often requiring more than a decade, some type of predictive tool or stratification and a relatively uncomplicated cloud and / or evaluation situation, or by strong target control correlations. The classical approach to obtain statistical significance in the face of high natural variability is to increase the sample size. However, programmes where either the understanding of the complex processes and / or the experimental design were poor, even a hundred years of unevolving randomized experimentation would produce merely additional inconclusive or uninterpredictable statistics. A supplementary alternative which can mitigate the sample size requirement is to make use of information relating to the concomitant variables, i.e. early identification of stratifications, covariates or predictors. Without early identification of concomitant variables, the randomization or operational evaluation would probably fail. Incorporation of a stratification in experiment design can often make the difference between significant versus inconclusive results in a fixed time limit experiment, or alternately can save some years of expensive experimentation.

Most of the successes in weather modification owe a large part of their achievement to identification of concomitant variables, either at the outset or after an exploratory phase of the experiment. It would be satisfying if the predictor or stratification variables arise either from clearly understood physics or model simulations. For the physical understanding and testing of the warm cloud modification hypothesis, seeding and evaluation methodologies consisting of sequential stepwise programmes to test the applicability of warm cloud modification hypothesis, predictor variables and model simulations are to be adopted.

In view of the factors mentioned above the following two types of seeding techniques have been adopted depending on the type of distribution of clouds present in the experimental area on any day of the experiment. The details of the two types of seeding techniques are described in the following.

(i) Area Seeding Technique (Total Target)

On any seedable day when the experimental area is covered with a large number of stratocumulus and cumulus clouds with vertical thickness of 1 km or more and cloud liquid water content of 0.5 gm m^-3, the area seeding technique was adopted. On these days of the experiment the seeding material was released into the clouds at a slow rate, (10 kg per 3 km flight path) at a height of about 200 - 300 m above the cloud base, so as to treat as many clouds as possible. The seeding material used on any experimental day for the area seeding covering the whole target area was about 1000 kg. As all the clouds in the target area were seeded it is designated as the 'Area Seeding Technique'. The estimated concentrations of the salt particles artificially released into the clouds during their seeding on the area seeding days could be 1-10 per litre of cloud air.

The flight path followed for this type of seeding is in the form of a loop covering the 40 km width of the target area in about 12 longitudinal tracks viz., 6 tracks during the forward direction and 6 tracks during the return direction of the aircraft flight covered in the target area (Figure 5). The seeding operation commences a few kilometers (about 5 to 10 kms) upwind of the western border of the target area. This distance is determined by computing the time required for the transport of the seeded clouds into the target area under the prevailing westerly winds on any seeded day. Similarly the seeding was terminated at a similar distance ahead of the eastern border of the target area in the downwind of the experimental area.

(ii) Isolated Single Cloud Seeding Technique Floating Target

On any seedable day when the experimental area is covered with a few isolated cumulus clouds, a pair of clouds of nearly same physical characteristics (vertical thickness of 1 km or more and liquid water content of 0.5 gm or more) was selected for the experiment. Out of the two clouds of the pair, one of the clouds was selected by random choice (target cloud) and repeatedly seeded. The neighbouring cloud was designated as the control cloud. Identical cloud physical measurements were carried out in both the target and control clouds by making the same number of repeated aircraft penetrations at a height of about 200 - 300 m above the cloud-base. The number of traverses made in such pairs of clouds varied from 5 to 10. The target cloud was seeded with massive doses of salt. The maximum amount of seeding material used for seeding these individual target clouds varied from 700 to 1000 kg.

12. Numerical Simulation of the Cloud Seeding Experiment

Numerical simulation of the cloud seeding experiments using the historic rainfall data of any region are important for evaluating the probability of detecting the prescribed increases in rainfall due to seeding with a specified degree of confidence. (Twomey and Robertson, 1973; Mary Selvam et al., 1978). Such experiments were carried out using the historic rainfall data of 32 raingauge stations located in the region of Exp-II in Maharashtra State. The historic rainfall data were obtained from the India Meteorological Department for the period 1951-60. The details of the simulation technique were described elsewhere (Mary Selvam et al., 1979). The simulation experiments were carried out for the cloud seeding experiment with the double-area cross-over design and area randomization which is based on Exp-II. The results of the numerical simulation experiments of 5, 8 and 10-year duration are furnished in Table 3.

The results presented in Table 3 suggest that 20 per cent increases in rainfall due to seeding could be detected, with 80 percent or more probability in 5 years. It is considered that for a successful detection of the seeding effect, the probability of detection should be more than 80 per cent (Smith and Shaw, 1976).
 
 

13. Statistical Evaluation of Rainfall

13.1 Area Seeding Days

The rainfall data relating to (I) area cloud seeding technique have been analysed and the results of the 11-year experiment are shown in Table 4.

The results relating to the 160 days of the area seeding experiment indicate 24 per cent increase in rainfall significant at 4 per cent level. The statistical significance of the root double ratio values was evaluated using the Man-Whitney test (Siegel, 1956). Different results of Exp-II indicate that the warm cloud responses to salt seeding depend on the physical characteristics of the warm clouds particularly in respect of the vertical thickness (greater than 1 km or more) and the cloud liquid water content (greater than 0.5 gm m^-3 or more). When the experimental area is covered with such clouds and all of them are seeded their response to seeding is found to be positive. Shallow clouds (vertical thickness < 1 km and LWC < 0.5 gm m^-3) when seeded showed tendency for dissipation.

A photograph depicting the cloud distribution in the Experimental area on a typical area seeding day is shown in Figure 6. One of the clouds shown by an arrow in Figure 6 developed rain in 20 minutes following seeding and the photograph of the raining cloud is shown in Figure 7. The fallstreak of the raining cloud is seen clearly in the photograph.

The results presented in Table 4 suggest that the value of the root double ratio during the year of the lowest rainfall, namely, 1982, is very high which needs to be explained. The monsoon rainfall in the Indian region during 1982 was deficient and it has been classified as the drought year by the India Meteorological Department. When there is a drought, the rainfall variability would be very large particularly in the semi-arid region where the experimental area is located. The anomalous value of the root double ratio observed during 1982 is due to the large variability in rainfall caused by the drought.

The results shown in Table 4 indicate that in 5-years of Exp-II the rainfall increase was about 24 per cent and thereafter remained stable during the remaining 6-years of the experiment. The results of the numerical simulation of the cloud seeding experiment shown in Table 2, of Section 12 also suggest that 20 per cent increase in rainfall due to seeding could be detected during a 5-year Experiment with 83 per cent probability of detection. The results of the rainfall analysis of Exp-II and the results of the numerical simulation of the cloud seeding experiment are in agreement.

(a) Numerical Simulation of Cloud Microphysical Processes Including
      Hygroscopic Particle Seeding

Many investigators used cloud models to predict and to help to understand the effects of cloud seeding (Simpson and Wiggert, 1969, Kopp et al., 1983, Tzivion et al., 1994). Numerical simulation experiments were carried out using the 2DTD cloud model (Orville and Kopp, 1977). The results of the cloud model computations have indicated that the vertical growth of the clouds and precipitation development are markedly influenced by the convergence at the cloud-base level (Vijayakumar, 1997). The results of Exp-II have clearly indicated that the warm cloud responses to salt seeding are critically dependent on the cloud physical characteristics e.g., vertical thickness greater than 1 km and liquid water content greater than 0.5 gm m^-3. As per the well known cloud physical processes the convergence at the cloud-base level could influence the cloud vertical thickness and the liquid water content. Hence, it is evident that convergence at the cloud-base level could influence the warm cloud responses to salt seeding. The cloud model computations were also carried out simulating the hygroscopic particle seeding (1-10 particles of sodium chloride per litre of cloud air) and the results suggested enhancement of rainfall (maximum up to 1-2 times in ideal cases). A typical case of the cloud model computations carried out using the aerological data for 1 July 1980 of the Exp-II is shown in Figure 8.

As seen from the figure the rain depth in the case of seeded cloud is higher than that of the not-seeded cloud. In both the cases the value of the convergence at the cloud-base level was kept constant (convergence 0.0005 / Sec^-1.).

The results of the (i) cloud model computations presented above and the (ii) numerical simulation of cloud seeding experiments carried out using the historic rainfall data presented in Section 12 are in agreement with the results of the rainfall analysis of the Exp-II. For establishing the warm cloud responses to salt seeding, it is essential to prove the persistence of the seeding result through atleast two phases with the stepwise programmes to test the applicability of the warm cloud modification hypothesis and cloud simulation studies.

13.2 Isolated Single Cloud Seeding Days

On the days when "isolated single cloud seeding technique" was adopted only one pair of target control clouds was chosen for the experiment as described in Section 11(ii). Contrary to the area seeding days only one cloud (target) was repeatedly seeded on the isolated single cloud seeding days. Hence, it is not feasible to detect the seeding effects resulting from a single seeded cloud when the 24-hr average rainfall of the entire target area (1600 km²) was considered in the analysis. For the detection of the rainfall from a single cloud (floating target) direct measurement of rain volume using a suitable radar preferably a doppler weather radar with rainfall measuring capability would be required. In the absence of such a remote sensing technique which can detect the rain volumes from the treated clouds on real time basis, it is impossible to detect the seeding effects from the 24 hr rainfall recorded by the raingauge network in the experimental area. However, in spite of the above serious limitation an attempt has been made to analyse the rainfall relating to the 124 days of the experiment when isolated cloud seeding technique was adopted. For this analysis identical statistical techniques were utilised including the area randomization for seeding as adopted in the case of the 160 days of the experiment when the area seeding technique was adopted. The results of the rainfall analysis relating to the isolated cloud seeding are presented in Table 5.

As seen from the results the rainfall on the days of isolated single cloud seeding showed a decrease of 35 per cent which is not statistically significant even at 30 per cent level. The negative result may be due to (i) inability to detect the seeding effect, (ii) dissipation of the target clouds due to over seeding of a single cloud with massive doses of salt (700 - 1000 kg), (iii) aircraft penetrations into the clouds 5-10 times which may disturb the cloud life cycle of the cloud and it may even result in the dissipation of the cloud due to the entrainment of the dry air from the peripheries of the cloud, (iv) inability to collect the rainfall from the single seeded cloud by the raingauges located in the experimental area. The probability of collection of rainfall will be very low unless the cloud accidentally happens to be located overhead of one of the raingauges in the experimental area. The aircraft cloud physical observations in the seeded and not-seeded clouds are perhaps more useful for the understanding of the basic physical processes responsible for precipitation formation in warm clouds and their responses to salt seeding rather than the statistical analysis of the rainfall recorded in the target and control sectors of the experimental area.

14.0 Physical Evaluation of Aircraft Cloud Physical Observations

Identical cloud physical observations were carried out during the repeated aircraft penetrations into the control and target clouds at a height of about 200-300 m above the cloud-base and the flying altitude of the aircraft was maintained constant within the limits of the aircraft operational capabilities. The variations noticed in the cloud physical parameters recorded in the target and control clouds were used for the physical evaluation of the warm cloud responses to salt seeding. The results of the physical evaluation would serve as valuable observational evidence for the verification of the physical hypothesis of the warm cloud modification technology. Also, the results of the physical evaluation would provide the basis for better understanding of the results of the statistical analysis of the rainfall data. The motivation of this experimental series is to determine whether and under what conditions rainfall can be increased by seeding warm clouds with hygroscopic particles and also to improve the physical understanding of the warm cloud modification hypothesis. The results of the analysis of different cloud physical observations are presented in the following.

14.1 Giant Size Condensation Nuclei (GCN)

The concentration of Giant Size Condensation Nuclei (GCN) with diameter greater than 1-10 m^-6 were measured using the Casella cascade impactor since these hygroscopic particles would influence the precipitation formation in the warm clouds through the collision-coalescence process. The background concentration of GCN in the region is about 2 l ^-1. The aircraft observations made in the control and target clouds were analysed and mean concentrations of GCN in 46 pairs of the control and target clouds were respectively 2.8 (standard deviation 1.2) and 5.0 (standard deviation 2.3) per litre. Thus there is a 79 per cent increase in the concentration of GCN which is statistically significant at 1 per cent level. These observations indicate that the concentration of GCN in the target clouds is higher by about 2 particles per litre which could facilitate the formation of large size cloud drops resulting in the acceleration of the precipitation formation through the collision-coalescence process.
 
 

14.2 Cloud Droplet Spectra

The aircraft observations of the cloud droplet spectra obtained from 50 pairs of the control and target clouds were analysed and the time variations noticed in the control and target clouds are shown respectively in Figures 9 and 10. The cloud droplet spectra obtained in the first aircraft penetration were compared to those sampled after 15-20 minutes following seeding. As seen from the Figure there is a marked increase in the concentration of the large size cloud drops (radius > 20m m) in the case of the target clouds as compared to that observed in the control clouds. This observational evidence corroborates the warm cloud seeding hypothesis that salt particles released artificially during the seeding of the target clouds have transformed into large size cloud drops which facilitate precipitation formation through the collision-coalescence process. The average concentration of large size cloud drops with diameter greater than 50m m observed in the control clouds decreased from 0.187 cm^-3 to 0.073 cm^-3 in about 15-20 minutes. Similarly the average concentration of large drops in the target clouds, increased from 0.063 cm^-3 to 0.200 cm^-3 in about 15-20 minutes. The average Median Volume Diameter (MVD) in the control clouds increased from 9.8 m m to 10.0 m m (2% increase). Similarly in target clouds, the average MVD increased from 8.7 m m to 11.7 m m (increase of 34.8%). The above results suggest that the hygroscopic particles released into the target clouds have transformed into large size cloud drops (diameter greater than 50m m in about 15-20 minutes) following seeding and could enhance the collision - coalescence process leading to the early onset of the precipitation / increase in the precipitation efficiency / rainfall.

14.3 Liquid Water Content

The aircraft observations of the cloud Liquid Water Content (LWC) obtained from the control and target clouds using the JW-hot wire instrument were analysed. The observations of the LWC obtained from 60 pairs of control and target clouds were considered in the study. The maximum values of the LWC obtained from the control clouds were compared with the maximum values of the LWC obtained from the target clouds of identical physical characteristics. As already mentioned earlier, the target clouds of identical physical characteristics were selected on random basis and repeatedly seeded. The maximum values of the LWC were recorded in about 10-15 minutes following seeding. The average values of the LWC in the control clouds was 0.6 gm m^-3 (standard deviation 0.11) and that in target clouds was 1.0 gm m^-3 (standard deviation 0.17). The above results suggest an increase in the LWC of 40 per cent in the target clouds which is significant at less than 1 per cent level. The increases noticed in the LWC are consistent with the preliminary results of individual cases of seeded and not-seeded clouds reported earlier (Murty et al., 1975).

14.4 Corona Current

The electrical characteristics of clouds are closely associated with the convective activity vis-a-vis the developments of the cloud and the precipitation. A study of the variations noticed in the electrical properties of seeded and not-seeded clouds indicated that they could beneficially be used for the physical evaluation of the warm cloud responses to salt seeding (Murty et al., 1976). Observations of the corona discharge current made in about 60 pairs of seeded and not-seeded clouds were utilised for evaluating the warm cloud responses to seeding. The average maximum values of the corona current in control clouds was 1.1m A (standard deviation 0.16) and that observed in the target in was 1.7m A (standard deviation 0.15). The above results suggest an increase in the corona discharge current of 55 per cent (significant at less than 1 per cent level) following seeding. The increase in the corona discharge current in the target clouds indicate enhancement in the cloud development / precipitation processes.
 
 

14.5 Vertical Velocity

The observations of the vertical air velocity obtained from the 50 pairs of control and target clouds were analysed. The aircraft flight altitude was maintained constant while collecting the above observations and may be representative of the conditions at the cloud-base level. The maximum positive values of the draft vertical air velocity obtained from the target and the control clouds were compared. The average value of the vertical velocity in the control clouds was 3 m sec^-1 (standard deviation 1.0) and that in target clouds was 4 m sec^-1 (standard deviation 1.1). The above results suggest an increase of 33% in the vertical velocity of the target clouds following seeding.

14.6 Cloud Dynamical Characteristics

It is known that one dimensional spectra of the clear air temperatures obey the -5/3 power law in the inertial sub-range (Corrsin, 1951). Also, it is shown that the averaged spectra of motions inside the cloud tend to follow the -5/3 power law (Almeida, 1979). When the slope of the temperature spectra follows the -5/3 power law, the turbulence in the cloud is indicated to be isotropic with balance between the energy input and dissipation. An increase in the slope beyond - 5/3 indicates higher energy dissipation rates. The longer wave lengths of the temperature spectra are generated as a result of the latent heat of the condensation and the shorter wave lengths primarily represent the small scale turbulence (Warner, 1970). Therefore, it would be valuable to study the variations noticed in the in-cloud temperature spectra in clouds during cloud seeding experiments and such studies would be beneficial for the physical understanding of the dynamical characteristics of the seeded warm cumulus clouds.

High resolution aircraft observations of the temperatures along with other cloud physical observations were obtained during the aircraft penetrations at a single level in isolated warm cumulus clouds before and after they were seeded with massive doses of salt. The preliminary results of the spectral analysis of the above high resolutions temperature observations were reported (Parasnis et al., 1982). The temperature spectra showed a significant wavelength of 2 km. The slope of the temperature spectra relating to the not-seeded traverses followed the -5/3 power law. The slope of the spectra relating to the seeded traverses increased when the cloud liquid water content increased and the rain formed. The temperature spectra of seeded traverses showed a net energy gain in the larger wave-length ( 540 m) and a net energy loss in the shorter wave-lengths. It was suggested that the net energy gain could be due to condensation of water vapour on the salt particles. The net energy loss in the shorter wave lengths could be due to the decrease in the small scale turbulence resulting from the invigoration of the updraft. These features may manifest the alteration of the dynamics of the warm clouds following their seeding. Typical temperature spectra obtained from clear - air and from a cloud before and after its seeding are shown respectively in Figures 11 and 12. The temperature spectra in the clear air followed the -5/3 power law (Figure 11). The spectra obtained from the aircraft traverses I to VII made in a cloud of initial vertical thickness of 1.5 km are shown in Figure 12. The cloud was seeded with 1500 kg of salt mixture during traverses II to VII. The LWC showed a progressive increase from 0.2 to 0.9 gm m^-3 following seeding and the maximum value was recorded in traverse IV. The in-cloud temperature varied between 14 and 15° C during the period of observations. Light rain was observed during traverse VII. The slope of the temperature spectra followed the -5/3 power law in traverse I (not-seeded). The slope steeply increased in traverses II to V (seeded). The slope in traverses VI and VII showed tendency towards the -5/3 power law. The LWC also progressively increased during traverses II to V and steeply decrease during traverses VI and VII. The variations noticed in the slope of the temperature spectra of the cloud following seeding may manifest the alteration of the dynamics of the cloud following seeding.

14.7 Chloride and Sodium Ion Concentrations in Cloud and Rain water

Information on the chloride and sodium ion concentrations in cloud / rain water samples collected from the seeded and not-seeded clouds could be used for evaluating the warm cloud responses to salt seeding. Cloud and rain water samples have been collected using a specially designed high quality stainless steel gadget filled on the top of the aircraft (Khemani et al., 1982). The cloud and rain drops were collected by impaction on the specially designed honeycomb arrangement in the gadget. The cloud and rain drops were sucked into the gadget from the air stream and the cloud and rain water samples were collected into clean polyethylene bottles. The gadget is cleaned by flushing with double distilled water repeatedly before the flight every day. The concentrations of Cl and Na ion concentrations in the cloud and the rain water samples collected from the not-seeded (Control) and the seeded (Target) clouds are furnished in Table 6.

The Cl ion concentrations in the cloud and rain water samples collected from the seeded clouds were found to be higher respectively by 373 and 319 per cent. Similarly the Na ion concentrations in the cloud and rain water samples collected from the seeded clouds were higher respectively by 409% and 237%. The above results are statistically significant at less than 1 per cent level and they suggest that the giant size salt particles released into the seeded clouds may have entered the cycle of the warm rain process. The giant size salt particles released into the clouds would transform into large size cloud drops which could facilitate acceleration of the collision - coalescence process. The results of the analysis of the observations of Giant Size Condensation Nuclei (GCN) and Cloud Droplet Spectra presented in Sections 14.1 and 14.2 respectively also indicated increases in the concentrations of the GCN and large size cloud drops (diameter 50 m m) and the Median Volume Diameter. The above results suggest acceleration of the collision - coalescence process in the warm clouds following seeding.
 

15.0 Conclusion
 

The status of modification of precipitation from warm clouds has been reviewed (Cotton, 1982; Czys and Bruintjes, 1994). Rainfall enhancement from the hygroscopic seeding remains to be demonstrated (Czys and Bruintjes, 1994). The research in the area of weather modification has moved away from the statistical evaluations for seeding effects to studies oriented towards monitoring processes under natural and seeded conditions. The development of a scientifically acceptable weather modification technology is probably many years away, barring of course some major breakthrough (Czys, 1995). However, some experiments have provided evidence to convingly establish that seeding has worked as expected in at least a few steps of the physical chain of events hypothesized in the conceptual models. Most of the success in weather modification owe a large part of their achievement to identification of concomitant variables, either at the outset or after an exploratory phase of the experiment (Simpson, 1978). It would be satisfying if the predictor or stratification variables arise either from clearly understood physics or model simulations.

The results of the 11-year Indian warm cloud modification and the model simulation studies presented in this paper have clearly emphasized the need for the physical understanding, sequential development (stepwise programmes to test the applicability of the warm cloud modification hypothesis), predictor variables, model simulations. An interdisciplinary approach would be essential for the successful warm cloud modification experiments. The results of the Indian field experiment suggested that warm cloud responses to seeding are critically dependent on the cloud physical characteristics e.g., vertical thickness and liquid water content (LWC). Clouds with vertical thickness greater than 1 km, LWC greater than 0.5 gm m^-3 when seeded with salt particles (modal size 10 m m; concentration 1 per litre of cloud air) produced increase in rainfall of 24 per cent significant at 4 per cent level. Shallow clouds (vertical thickness less than 1 km, LWC less than 0.5 gm m^-3) when seeded showed tendency for dissipation. The cloud physical observations made in not-seeded (control) and seeded (target) clouds have provided some useful evidence to test the applicability of the warm cloud modification hypothesis. Results of the cloud model computations suggested that moderate convergence at the cloud-base is essential for the cloud growth and development of precipitation in the real world. Hygroscopic particle seeding of warm clouds under favourable dynamical conditions (convergence at the cloud-base level) may result in the acceleration of the collision-coalescence process resulting in the enhancement of rainfall.

The results of the 11-year Indian warm cloud modification experiment are useful for the planning of good field experiments in future. For obtaining conclusive results it is essential to prove the persistence of the seeding result through at least two phases with the supporting stepwise programmes to test the applicability of the warm cloud modification hypothesis and cloud simulation studies.

16.0 Acknowledgements

Several individuals and numerous organizations have contributed for the successful conduct of the Experiment. The India Meteorological Department provided invaluable co-operation and support in several areas particularly in respect of raingauge network, special pilot balloon and radiosonde observations, weather forecasts and meteorological data. Several organisations of the Government of India, particularly the Department of Civil Aviation, Indian Air Force, Department of Science and Technology, Agriculture, the Government of Maharashtra State have provided valuable support for the Experiment. The Governing Council of the Indian Institute of Tropical Meteorology (IITM) and the successive Directors of the IITM have provided valuable encouragement for the Experiment. The aircraft and related technical support for the Experiment were provided by M/s Air Works India, Bombay.

The authors gratefully acknowledge the guidance of Dr. Bh.V. Ramana Murty and the dedicated effort of the staff of the Institute for the Experiment.
 
 

REFERENCES
 
 

Biswas, K.R., Kapoor, R.K., Kanuga, K.K. and Ramana Murty, Bh.V., 1967 : Cloud seeding experiment using common salt. J. Appl. Meteor., 6, 914-923.

Chatterjee, R.N., Biswas, K.R. and Ramana Murty, Bh.V., 1969 : Results of cloud seeding experiment at Delhi as assessed by radar. Ind. J. Meteor. Geophys., 22, 11-16.

Corrsin, S., 1951 : On the spectrum of isotropic temperature fluctuations in an isotropic turbulence. J. Appl. Phys., 22, 469-473.

Cotton, W.R., 1982 : Modification of precipitation from warm clouds-A review. Bull. Amer. Meteor. Soc., 63, 146-160.

Czys, R.R., 1995 : Progress in planned weather modification research : 1991-94. Rev. Geophys. Suppl., 33, 823-832.

Czys, R.R. and Bruintjes, R.T., 1994 : A review of hygroscopic seeding experiments to enhance rainfall. J. Wea. Modification, 26, 41-51.

Devara, P.C.S. and Ramana Murty, Bh.V., 1984 : Physics of monsoon rain processes. Mausam, 35, 435-452.

Fisher, R.A. and Yates, F., 1974 : Statistical tables for Biological Agricultural and Medical Research. London, Oliver and Boyd, 114-119.

Khemani, L.T., S.K. Sharma, A.S.R. Murty and Ramana Murty, Bh.V., 1982: A simple gadget for collection of cloud and rain water from aircraft. Proc. Conference on Cloud Physics, 15-18 November 1982, Chicago Illinois, American Meteor. Soc., 303-305.

Kopp, F.J., Orville, H.D., Farley, R.D. and Hirsch, J.H., 1983 : Numerical simulation of dry ice cloud seeding experiments. J. Climate Appl. Meteor., 22, 1542-1566.

Mary Selvam, A., Murty, A.S.R. and Ramana Murty, Bh.V., 1978 : A numerical simulation technique for simulation of cloud seeding experiments. Proc. Indian Academy of Sciences, 87A, 179-191.

Mary Selvam, A., Murty, A.S.R. and Ramana Murty, Bh.V., 1979 : Numerical simulation of cloud seeding experiments in Maharashtra State, India. J. Wea. Modification, 11, 116-140.

Mason, B.J., 1971 : The Physics of Clouds, 2nd ed. Oxford, Clarendon Press, 671 pp.

Moran, P.A.P., 1970 : The methodology of rainmaking experiments. Rev. Int. Statist. Inst., 38, 105-119.

Murty, A.S.R., Mary Selvam, A. and Ramana Murty, Bh.V., 1975 : Summary of observations indicating dynamic effect of salt seeding in warm cumulus clouds. J. Appl. Meteor., 14, 629-637.

Murty, A.S.R., Mary Selvam, A., Paul, S.K., Vijayakumar, R. and Ramana Murty, Bh.V., 1976 : Electrical and microphysical measurements in warm cumulus clouds before and after seeding, J. Appl. Met.,16, 1295-1301.

Neyman, J., 1980 : "Comments on the discussion at the Workshop on the statistical design and analysis of weather modification experiments" in Statistical Analysis of Weather Modification Experiments. Eds. Wegman E.J. and DePriest, D.J., Lecture Notes in Statistics, vol. 3, Marcel Dekkar Inc., New York, 131-137.

Orville, H.D. and Kopp, F.J., 1977 : Numerical simulation of the history of a hailstorm. J. Atmos. Sci., 34, 1596-1618.

Parasnis, S.S., Selvam, A.M., Murty, A.S.R. and Ramana Murty, Bh.V. 1982 : Dynamic responses of warm monsoon clouds to salt seeding. J. Wea. Modification, 14, 35-47.

Pramanik, S.K. and Koteswaram, P., 1955 : Heights of tops of low clouds over India. Artificial Rain, Ed. Basu, S. Council of Scientific and Industrial Research, New Delhi, 104-111.

Ramana Murty, Bh.V. and Biswas, K.R., 1968 : Weather modification in India. J. Meteor. Soc. Japan, 46, 160-165.

Siegel, S., 1956 : Non-parametric Statistics for the Behavioral Sciences, New York, McGraw Hill, 312 pp.

Simpson, J. and Wiggert, V., 1969 : Models of precipitating cumulus towers. Mon. Wea. Rev., 99, 87-113.

Simpson, J., 1978 : What weather modification needs? - A Scientist's View. J. Appl. Meteor., 17, 858-866.

Smith, E.J. and Shaw, D.E., 1976 : Some aspects of the assessment of a site for cloud seeding experiments. Second WMO Scientific Conference on Weather Modification. Boulder, Colorado, USA, 2-6 August 1976, World Meteorological Organization, Geneva, Switzerland, 165-173.

Twomey, S. and Robertson, I., 1973 : Numerical simulation of cloud seeding experiment in selected Australian areas. J. Appl. Meteor.,12, 473-478.

Tzivion, S., Reisin, T. and Levin, Z., 1994 : Numerical simulation of hygroscopic seeding in a convective cloud. J. Appl. Meteor.,33, 252-267.

Vijayakumar, R., 1997 : Observations and numerical simulation of the microphysical processes in monsoon clouds. Ph.D. Thesis, University of Pune, Pune, India, 227 pp.

Warner, J., 1970 : The microstructure of cumulus clouds. Part III : The nature of updraft. J. Atmos. Sci., 27, 682-688.

Warner, J., 1973 : Rainfall Enhancement - A Review. First WMO/IAMAP Scientific Conference on Weather Modification, Tashkent, USSR, 1-7 October 1973, World Meteorological Organization, Geneva, Switzerland, 43-50.
 
 
 

LEGEND FOR FIGURES
 

Figure 1 : Experimental area.

Figure 2 : Instrumented aircraft (DC-3) used for seeding and cloud physical measurements.

Figure 3 : Seeding equipment (rear view) fitted to the fuselage of the aircraft.

Figure 4 : Plume of the seeding material released from the aircraft into the clear air.

Figure 5 : Aircraft flight path followed on area seeding days.

Figure 6 : Cloud distribution in the experimental area on a typical area seeding day.
            One of the seeded clouds  which developed rain following seeding is shown by the arrow.

Figure 7 : Seeded cloud which developed rain following seeding in 20 minutes.
            Fallstreak of the raining cloud is clearly seen in the photograph.

Figure 8 : Distributions of rain depth in the cloud model domain relating to seed and no-seeded cloud cases.

Figure 9 : Average cloud drop size distribution in not-seeded (control) clouds.

Figure 10 : Same as Fig. 9 for seeded (target) clouds.

Figure 11 : Temperature spectra in clear-air.

Figure 12 : Temperature spectra in cloud-air.

This page hosted by Get your own Free Home Page