Drop size and rain rate characteristics of Indian monsoon rainwater

Abstract

Raindrops are an integral part of nature’s hydrological cycle. Information on their sizes in different seasons and climatic conditions is required in many practical applications. We report seasonal average properties of rainfall examined through raindrop sizes and rain rates measured with optical disdrometers in India. Disdrometers were operated at three sites having coastal, orographic and continental climates. Analysis has been carried out by clubbing data of each station into two groups, namely, all available and summer monsoon season-only data. The study presents contributions of different raindrop sizes and rain rates to the total rainfall. There are between 2.4 and 2.9 million raindrops (with size > 0.25 mm) in one litre of rainwater, with the orographic site having the maximum number. A sum of two exponential functions describes the size dependence of number of drops present in rainwater. Major contribution to the monsoon rainfall comes from drops having sizes in 1.5–2.5 mm range with a peak of around 2 mm. Average rain accumulation with rain rate (R) approximately varies as R^–b (for R > 0); b depends on location and lies between 1.1 and 1.3, the former corresponding to the continental site while the latter to the orographic.

Research highlights

There are between 2.4 and 2.9 million raindrops (with size > 0.25 mm) in one litre of rainwater. Major contribution to the monsoon rainfall comes from drops having sizes in 1.5–2.5 mm range with a peak around 2 mm. On the other hand, all values of rain rate (R) are important to rain accumulation; low rain rate contributes by its sheer number (i.e., their total long duration) while large rain rate by its magnitude. Average rain accumulation varies as R^–b, where b lies between 1.1 and 1.3 depending on location.

Appendix: LPM instrument details and data quality control

The sensing area (A) of the LPM is a horizontal laser sheet, approximately 234 mm long, 20 mm wide (A ~ 4680 mm²) and 0.75 mm thick. LPM measures size (equivalent diameter) and fall velocity of individual objects passing through A. The standard output is the number of drops N(D_i,j) passing through A every minute. Here, the subscripts i and j refer to size and velocity classes, respectively. There are 22 size classes (i.e., 0.125, 0.25, 0.375, 0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, and ≥8 mm). Drops in each size class are divided into 20 velocity classes (0.0, 0.2, 0.4, 0.6, 0.8, 1.0, 1.4, 1.8, 2.2, 2.6, 3.0, 3.4, 4.2, 5.0, 5.8, 6.6, 7.4, 8.2, 9.0, and ≥10 m s^–1). Number of objects having size and fall speed greater than and equal to and less than the next belong to a given (i, j) pair. There are many occasions when only a few tiny drops are present in a sample and contribute insignificantly to total rainfall. On the other hand, contribution of one big drop cannot be ignored. So, after examining the data, we imposed a threshold on the radar reflectivity factor Z (which is proportional to the sixth power of D, Marshall et al. 1947). One-minute samples having Z < 0.01 dBZ are not included in further analysis to remove time intervals that contribute insignificantly to total rainfall but significantly to total number of minutes.

The LPMs installed at AGB and IWC were manufactured in 2016, while that at Bengaluru were in 2012. Data quality control is very important in order to get reliable products from disdrometer measurements. Frasson et al. (2011) carried out a study installing four early versions of LPMs next to each other along with tipping bucket rain gauges (TBRGs). They report 5–10% variations in the accumulated rainfall among LPMs. The differences narrow down after calibrating LPM with spheres of known diameter. Consistency among LPM rainfalls is better compared to that between TBRG and LPM.

In our work, we take advantage of Joannieum Research’s 2DVD data to correct LPM data. Both 2DVD and LPM were operated at Bengaluru in 2012 and they share a common data period of about 2 months. The 2DVD consists of 2 high-speed video cameras and two horizontal light sheets separated vertically by about 6 mm (measurement area ~100 mm × 100 mm). Drop position is precisely measured in a 2D plane at a spatial resolution of ~0.18 mm (Kruger and Krajewski 2002). For every raindrop j that passes through the sensing area, equivalent diameter D_j, measurement area A_j (which is made a function of D_j), oblateness and fall velocity (V_fj) are recorded along with time. In both LPM and 2DVD, measurement area is enclosed inside a supporting structure. Some fragments of bigger drops, splashing after hitting the supports, may enter disdrometer’s measurement area. Dust particles and leaves may also pass through the measurement area. Such objects are to be identified and removed from the raindrop population. Fall velocity as well as shapes (e.g., oblateness) of splash fragments/dust are different from that of normal hydrometeors, and this information can be used to design filters. Taking all available 2DVD data at Bengaluru, statistics of V_f and oblateness as a function of drop size have been calculated. The average V_f is well described by the equation, V_f = 10.3 – 11.1exp(–0.6D). A drop is removed from the 2DVD data if its oblateness or fall velocity lies outside 2.5 times the standard deviation from the respective mean values. The expressions derived/used for terminal velocity of raindrops assume the atmosphere to be at rest (Gunn and Kinzer 1949; Beard 1976). Atmospheric flows are often turbulent near the surface, especially during the monsoon season. In a turbulent flow field, fall velocity of a drop could differ from its theoretical terminal velocity. Therefore, allowance is made for spread in raindrop fall velocities. LPM does not measure the shape of raindrops and only fall velocity information is available. LPM data is passed through a velocity filter where drops having fall velocity outside the permitted range for a given diameter (using 2DVD data derived V_f) are ignored. Further, all drops having diameters <0.25 mm have been removed for the following reasons. Flow in the atmospheric boundary layer (where disdrometers are installed) is often turbulent, and instantaneous vertical component of velocity of the order of 0.5–1 m s^–1 is not uncommon in the atmospheric boundary layer. Small drops tend to follow the wind (some passing through disdrometer may even be moving upwards in a highly turbulent boundary layer) and the right range for the velocity filter for D < 0.25 mm is not clear to us. Finite size resolution of disdrometer increases the relative uncertainty at small sizes. Therefore, we have ignored sizes <0.25 mm in our analysis. Application of velocity filter made the agreement between LPM and 2DVD or LPM and TBRG consistently good. Some examples are shown in figure 2.