INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 27: 541–553 (2007) Published online 25 October 2006 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/joc.1414 Short Communication Observation and characterisation of rainfall over Hawaii and surrounding region from the Tropical Rainfall Measuring Mission Chris Kidd a and Glenn McGregor a, * a School of Geography, Earth and Environmental Sciences, University of Birmingham, UK Abstract: Satellite observations of clouds and precipitation have a history stretching back over 40 years. The observations of clouds initially led to the inference of precipitation from cloud top features and subsequently more direct methods derived from passive microwave observations. More recently, the Tropical Rainfall Measuring Mission (TRMM) has provided unprecedented information from the first active microwave instrument specifically designed to measure rainfall, the precipitation radar (PR). This paper utilises data from the TRMM PR for the period December 1997 to November 2005 to investigate rainfall over Hawaii and the surrounding ocean. The PR instrument not only provides spatial information on rainfall, but also vertical profiles of precipitation. Data from the PR instrument is used to map rainfall at a spatial resolution of 5 km and, utilising the information from the vertical profiles, map the distribution of trade-wind and non-trade-wind precipitation. This paper provides quantitative estimates of rainfall over Hawaii and the surrounding ocean that generally match those of surface observations and models. It is found that the islands exert a significant influence on the distribution of the amount and occurrence of precipitation over this region. Copyright  2006 Royal Meteorological Society KEY WORDS rainfall; Hawaiian Islands; TRMM Received 28 July 2005; Revised 14 July 2006; Accepted 15 July 2006 INTRODUCTION A multitude of techniques now exist for the estimation and retrieval of rainfall from satellite sensors. Algorithms based upon visible and infrared data have been in use for over 40 years, although limitations in the usefulness of the visible channels have concentrated most efforts upon the formulation of infrared-based techniques. Such techniques rely upon the simple premise that cold cloud tops are more likely to be associated with rainfall than warmer cloud tops. Algorithms, such as the GOES Precipitation Index (GPI) (Arkin and Meisner, 1987), utilise a simple cloud top temperature threshold to delineate rain areas, which when aggregated over long time periods and large areas produce good results. Despite the simplicity of the GPI technique it is often seen as a benchmark algorithm (Spencer, 1993). However, the variability of precipitation processes does not permit the use of a simple threshold to be universally applicable, particularly at fine temporal and spatial resolutions. More complex visible and/or infrared techniques have been * Correspondence to: Glenn McGregor, Department of Geography, King’s College London, London. E-mail: ijclimatology@kcl.ac.uk developed, but are limited by the inference of rainfall from cloud top characteristics. Algorithms based upon passive microwave informa- tion have been developed since the 1970s with much progress made since the launch of the first Special Sen- sor Microwave/Imager in 1987. Algorithms using passive microwave radiometry are more direct due to the fact that precipitation-sized particles are the dominant source of atmospheric attenuation at these frequencies. Techniques using frequencies below 30 GHz rely upon the emission of radiation from water droplets to identify precipitation, while above 50 GHz scattering from raindrops and ice becomes pronounced and may be used to estimate pre- cipitation. Although passive microwave techniques are more direct than those of visible/infrared techniques they too suffer from a number of drawbacks: emission tech- niques may only be used over the oceans, and while scattering techniques can be usefully employed over land and ocean they respond primarily to ice hydrometeors rather than rain alone. Furthermore, passive microwave data are gathered by polar-orbiting based instrumenta- tion and therefore provide data less frequently than that of infrared sourced algorithms. It should also be noted that the passive microwave information is derived over an integrated column rather than a near-surface sample. Copyright  2006 Royal Meteorological Society