1 ESTIMATING LOCAL SUGARCANE EVAPOTRANSPIRATION USING LANDSAT TM IMAGERY * Xihua Yang, Qiming Zhou and Mike D. Melville School of Geography, University of New South Wales Sydney 2052, Australia ABSTRACT This paper presents an approach for evapotranspiration (ET) estimation using Landsat Thematic Mapper (TM) data for a sugarcane field in northern New South Wales, based on the concept of a vegetation index / temperature trapezoid (VITT). Landsat TM thermal band was used to extract surface temperature and the red and near infrared bands were used to derive the vegetation indices. Three vegetation indices (NDVI, SAVI and MSAVI) were compared, and the relationships between these indices and the surface temperature (T s ) were examined. A moisture availability index (M a ) and, subsequently, the ET rate were computed using T s and NDVI derived from remotely sensed data. The VITT concept was evaluated using the field measurements of surface reflectance and surface obtained temperature during sugar cane growth period. The estimation of M a was evaluated using a water balance model which uses well-established computer software and input data. Results shows that Landsat TM data can provide a practical means for studying the spatial variation in ET and making useful estimates of ET and related soil water status of the sugarcane field. Keywords: Evapotranspiration, Vegetation Index, Surface Temperature, Remote Sensing. INTRODUCTION The evapotranspiration (ET) is an important parameter in studying the water and energy balances on the earth’s surface. Understanding the distribution of surface moisture and ET is a key factor in successful applications of water balance models to optimal agricultural water management. To date, the only practical means of mapping the spatial distribution of ET on a regional or local scale is to use remotely-sensed multispectral data from satellite-based sensors (Moran et al., 1989). Satellite imagery has been widely used for large-area estimation of mass and energy fluxes and review of many of these applications is given by Carlson (1986), Jackson (1985), and Moran et al. (1989). The techniques traditionally involved combination of atmospheric corrections, models of resistance to mass and energy, and detailed spatial information of major surface and climate variables. Operational applications of this technology, however, is often limited due to the inherent complexity of this procedure. Many studies on radiometric surface temperature (T s ) have focused on the widely observed negative correlation between T s and remotely-sensed measurements of actively transpiring vegetation, such as Normalised Difference Vegetation Index (NDVI) (Hope et al., 1992; Nemani et al., 1989, and Moran et al., 1994). One hypothesis to explain this correlation is that the relationship between T s and NDVI is an indicator of evaporative cooling by plant * in Proceedings of 8 th Australasian Remote Sensing Conference, 25-29 March 1996, Canberra, Vol. 2, pp 262- 269.