Discrimination of Biological Scatterers in Polarimetric Weather Radar Data: Opportunities and Challenges

Abstract

For radar aeroecology studies, the identification of the type of scatterer is critically important. Here, we used a random forest (RF) algorithm to develop a variety of scatterer classification models based on the backscatter values in radar resolution volumes of six radar variables (reflectivity, radial velocity, spectrum width, differential reflectivity, correlation coefficient, and differential phase) from seven types of biological scatterers and one type of meteorological scatterer (rain). Models that discriminated among fewer classes and/or aggregated similar types into more inclusive classes classified with greater accuracy and higher probability. Bioscatterers that shared similarities in phenotype tended to misclassify against one another more frequently than against more dissimilar types, with the greatest degree of misclassification occurring among vertebrates. Polarimetric variables proved critical to classification performance and individual polarimetric variables played central roles in the discrimination of specific scatterers. Not surprisingly, purposely overfit RF models (in one case study) were our highest performing. Such models have a role to play in situations where the inclusion of natural history can play an outsized role in model performance. In the future, bioscatter classification will become more nuanced, pushing machine-learning model development to increasingly rely on independent validation of scatterer types and more precise knowledge of the physical and behavioral properties of the scatterer.

The Role of Weather Radar in Rainfall Estimation and Its Application in Meteorological and Hydrological Modelling—A Review

“…This influences the QPE to the highest degree. The echoes are mainly caused by (i) ground clutter (echoes from high objects close to the radar site, including wind farms) [39][40][41]; (ii) electromagnetic interference with the sun or external microwave emitters, which are usually visible in a radar image in the form of spikes pointing towards the radar site [42]; (iii) speckles caused by measurement noise; and (iv) biological objects, such as birds or insects [43].…”

A Review of Insect Monitoring Approaches with Special Reference to Radar Techniques

“…Classical threshold-based filtering remains a popular technique for this. Among novel solutions, machine learning approaches are becoming more popular [ 143 ]. Applying modern deep-learning algorithms enables the distinguishing of a tremendous amount of insect information in weather radar data [ 144 , 145 ].…”

Classification of Bird and Drone Targets Based on Motion Characteristics and Random Forest Model Using Surveillance Radar Data

“…1. Unlike other machine learning techniques which operate in a complete feature space, feature selection and tree construction strategies in random forest make it possess good noise tolerance capability and lower chance of overfitting [29]. In our problem, even though radar system could detect and track birds accurately, environmental interference and tracking algorithm's uncertainty make training database partially contain low quality non-bird tracks which are categorized as noise interference.…”