Multi-scale comparison of rainfall measurement in Paris area between two optical disdrometers of different working principles

A disdrometer is an instrument designed to assess both the size and velocity of descending hydrometeors. The applications of rainfall measurements retrieved with the help of disdrometers are diverse, spanning areas such as traffic control, scientific research, airport observation systems, and hydrology. Modern disdrometers leverage microwave or laser technologies that have increased the accuracy of the measurements with each iteration. Still, the quality of measurements fluctuates depending on factors such as raindrop size, wind velocity, and rain rate. A comprehension of these variations is needed to better understand the level of reliability of each device depending on the specific rain conditions.

In this study, we compare the performance of two optical disdrometers : 3D Stereo disdrometer (manufactured by Thies Clima) and Parsivel2 (manufactured by OTT). Both devices provide size resolved measurement of rainfall along with velocity of falling drops. Parsivel is set to record data every 30 seconds over a sampling area of 54 cm² and arranges the information in 32 x 32 classes of drop size and velocity. Unlike the Parsivel, 3D Stereo does not discretize measurements, and directly provides the diameter and velocity of each falling drop in a sampling area of 100 cm² with a measuring resolution of 0.08 mm and 0.2 m/s respectively, and a temporal resolution of 1 millisecond. This finer resolution data enables us to study rainfall variability at very small scales which are not usually available.

Here, we used continuously and simultaneously measured data since 21/08/2023, from TARANIS observatory of ENPC (https://hmco.enpc.fr/portfolio-archive/taranis-observatory/). The initial comparison of the data was done using a time series of rain-rate for rainfall events in between a dry period of at least 15 minutes and total depth >0.7 mm. This revealed an unexpected disparity in the water volume collected between the devices. Parsivel collected more than 3D Stereo on every instance, and the disparity got bigger as the rain rate increased. With the purpose of studying the source of this disparity, the sampling area of the 3D Stereo was divided into 8 sections and compared with each other. This showed that the estimate of rainfall parameters such mean diameter, mean velocity of the drops (which were expected to be uniform over long periods regardless of the section where drops are measured) were not the same for the sections studied, and exhibited clear trends. To understand this discrepancy in a scale invariant way, and to evaluate the performance of devices across scales and not only at a single scale, the widely used framework for studying variability of geophysical fields – Universal Multifractals (UM) was employed for assessing the scaling behavior of fields. Rainfall from both devices showed previously reported average scaling behavior from 30 s to 30 min. The difference between rain events and also the behavior at finer scales, which can be accessed from 3D stereo disdrometer were also studied using the UM framework and will be discussed.

Authors acknowledge the Ra2DW project (supported by the French National Research Agency - ANR-23-CE01-0019), for partial financial support.

Keywords: rainfall; disdrometer; multifractals;