AVHRR data for real-time operational flood forecasting in Malaysia

Abstract

Flash floods strike quickly and in most cases without warning. They are usually observed before any warning can be issued and usually persons and property have been affected before the warning reaches them. Such are the conditions prevalent in Malaysia's extreme monsoon weather that occasionally causes floods and results in the extensive damage to property and sometimes loss of lives. Over the years variously hydrological and structural engineering measures have been implemented for flood monitoring and forecasting. These measures have only yielded limited success as may be seen in the recurring flood situation. Yearly financial and property loss estimates have increased and an estimated cost of over 2.5 billion RM is projected for the year 2004 according to sources from the drainage and irrigation department of Malaysia. It has thus become apparent that Malaysia institutes an effective operational flood forecasting to arrest the persisting flood problem. In this paper we will expound on current flood management and forecasting system being implemented in the country, particularly the Klang Valley that includes Kuala Lumpur where there has been tremendous urban growth and development in the last one and half decades. The paper further discusses where current flood management systems have been lacking in the absence of real-time hydro-meteorological forecasts. Where as hydrodynamic simulations and structural control measures have been emphasized in many flood management systems in Malaysia, the integration of real-time hydro-meteorological forecasts have been conspicuously absent, rendering most in-situ flood forecasts and early warnings ineffective in address the flood problem in the country. Malaysia is a tropical country that lies along the path of the northeast and southwest monsoon. Although satellite image based NWP have proved useful for the tropical and equatorial regions of the world in flood forecasting, they have yet to be applied in Malaysia. Observations have generally shown heavy cumulonimbus clouds formation and thunderstorms precede the usual heavy monsoon rains that cause floods in the region. This makes quantitative precipitation forecast a must be input to any flood early warning design. Numerous empirical studies have determined that cloud top temperatures less that 235k in the tropics are generally expected to produce convective rainfall at the rate of 3mm/hr. In this study we thus investigate monsoon cloud formation that has the propensity to precipitate using NOAA-AVHRR data for real-time operational flood early warning in Malaysia. The AVHRR data has been preferred for its relatively high temporal resolution of at most 6/hours, its easy acquisition and cost effectiveness and its ability for automated geometric rectification when compared to GEOS and GMS data. Cloud cover and types are processed using cloud indexing and pattern recognition techniques on the AVHRR data. The cloud indexing technique was initially developed for NOAA but was later also adapted for Geostationary satellite images. The technique assigns rainfall levels to each cloud type identifies in an image based on the relationship between cold and bright clouds top temperature and the high probability of precipitation. We discuss how visible (VIS) and infrared (IR) techniques are applied to bi-spectral cloud classification and rain areas are determined by classifying pixel clusters in the VIS/IR histogram. Precipitation probability is evaluated based on the relationship between cold and brightness temperature of clouds. The near infrared (NIR) and infrared (IR) channels 3, 4, and 5 of the data are processed for temperature and brightness. Cold clouds with temperature below 235k threshold value are taken as indication of rain. Rainfall is estimated based on the assumption that every cloud pixel has a constant unit rain-rate of 3mmh-1, which is appropriate for tropical precipitation over 2.5° × 2.5° areas around the equator. The paper finally discusses current developments in nowcasting that utilizes latest satellite observations together with numerical weather prediction models and how this system can be adapted to the needs of very short term forecast for flood early warnings in Malaysia. © 2005 Springer-Verlag Berlin Heidelberg.