High resolution studies of rainfall on Norfolk Island, Part III: A model for rainfall redistribution S. G. Bradley a , *, K. N. Dirks a , C. D. Stow a a Department of Physics, University of Auckland, Auckland, New Zealand Received 20 March 1997; accepted 24 April 1998 Abstract A dense network of high resolution rain gauges has provided data on spatial scales of about 1.5 km and time scales of 15 sec on Norfolk Island, where the topography is dominated by a 300 m hill that is found to modify the spatial distribution of rainfall depending on wind direction. A correlation method is used to track coherent storm motion so as to identify the dominant wind direction and speed. Transects are selected across the island in the wind direction and the rainfall interpolated along each transect. It is found that there is often a decrease in rainfall upwind of the hill, followed by an increase downwind of the peak. The ratio of downwind to upwind catches increases linearly with wind speed. These observations do not accord with the often accepted explanation using a seeder–feeder mechanism. A simple wind drift model is proposed, in which raindrops follow trajectories modified by the perturbed wind flow over the hill. The transects are approximated by bell-shaped hill profiles. Analytic solutions for potential flow over a cylinder are then used to find a streamline with a bell shape similar to a topography transect through the hill and aligned with the wind direction. This streamline is used as the solid surface boundary, thereby giving a direct analytic flow field for the bell-shaped hill. A streamfunction is found for drop motion, allowing rainfall variations to be predicted. A parameterisation is developed that explains the shape and magnitude of the observed rainfall variations along a transect. Agreement between the rudimentary model and observations is found possible to within a few percent. This indicates that the wind drift process is viable as the dominating mechanism for determining rainfall distribution in the presence of low hills when low level moisture is absent.  1998 Elsevier Science B.V. All rights reserved. Keywords: Rainfall; Mesoscale; Enhancement; Wind flow; Orographic modelling 1. Introduction The presence of a hill causes lifting of incident air. This in turn can lead to condensation and eventual growth of droplets to precipitation size. However, such autoconversion is not a viable mechanism for increasing rainfall near short hills since, for common wind speeds, there is not enough time for large drops to grow while the air passes over the hill. Alternatively, condensation in lifted air can cause instability and rapid vertical growth leading to active convective rainfall. Such a mechanism is less effec- tive for low hills, since the vertical perturbations of the incident air are also usually too small, for low hills, to initiate convective instability leading to active convective storms. Therefore, an alternative mechan- ism is needed to explain the enhancement observed over small hills. Bergeron (1965) observed that hills of even a few tens of metres high could produce rainfall increases of up to 50%. Such enhancements can be explained by a Journal of Hydrology 208 (1998) 194–203 HYDROL 3576 0022-1694/98/$ - see front matter  1998 Elsevier Science B.V. All rights reserved. PII S0022-1694(98)00156-5 * Corresponding author. Fax: +64 9 3737445; e-mail: s.bradley @auckland.ac.nz