Modeling of an impact transducer for in situ adaptive disdrometer calibration

Abstract

In situ calibration of a rainfall impact disdrometer using a collocated tipping bucket rain gauge has been previously reported. Several factors affect the accuracy of this technique. These include the specific mathematical model of the transducer, tipping bucket errors, clock/time accuracy, switch delay time and calibration reference window used to update the transducer model coefficients. In this work, two general forms of an empirical transducer model have been investigated: a simple two-coefficient power-law and an Nth order polynomial, where the input is a specific measure of the electrical signal from the transducer and the output is the hydrometeor diameter. Even though previous work allowed for a more general non-linear transducer model description, the subsequent implementation of an adaptive calibration algorithm required use of a steepest descent recursive gradient search algorithm, which can be difficult to implement in a realtime system. Linear transducer models allow an implementation suitable in a real-time signal processing system.