Use Of Satellite Data To Estimate Areal Evapotranspiration From A Tropical Watershed
Amin, M.S.M1, Amjad Nabi1 and Shattri Mansor2
1Department of Biological and Agricultural Engineering
2Department of Civil engineering
faculty of engineering, University Putra Malaysia
43400, UPM Serdang, Selangor, Malaysia
E-mail:amin@eng.upm.edu.my
Abstract
A physically based scheme was used to model evapotranspiration (ET) from a tropical watershed using weather data satellite imagery. The land cover classes over the watershed were interpreted from Landsat TM imagery. The watershed was subdivided into hydologicallly homogeneous ground response units (GRUs). A geographic information system was used to integrate topographic, soils and vegetation data for each GRU. Leaf are index (LAI) as the most important vegetation structural variable was estimated form Landsat TM data. A simple form of LAI as a function of TM data was established. ET estimates were made over the whole year of 1993. the derived ET compared well with the other regional studies.
Introduction
The evapotranspiration (ET) is a key process controlling the exchange of energy and mas by vegetation. Forest hydrology studies suggest, the change in vegetation type and density re likely to alter ET significantly, and so does the eater yield ( Bosch and Hewlett, 1982 ). The role of vegetation in controlling energy and mass exchange rate from land surfaces depends on physiological characteristics of vegetation as well as climate ( Jarvis and McNaughton, 1986). In humid tropical regions, the substitution of tropical rain forest by agricultural crops and modify energy and mass exchange and alter atmospheric characteristics. In Malaysia generally, single species of trees such as Hevea, Oil Palm and Cocoa are used for afforestation. In particular, the water use of rain forests is of major interest on a local scale where a change of land use will affect nearly every water resources aspect. The aim of this study is the use of the remote sensing data s input to estimate and model ET and examining the relative magnitudes of watershed water budgets and input variables needed by the model that can be estimated from the satellite remote sensing data . the bernam River basin in the state of Perak, west Malaysia was chosen for this study where the recession of water resources since 1970's is the main impact of environmental changes in this regions ( Amin and Amjad, 1996).
Remote sensing provides the needed dynamic temporal view of vegetation, and complete spatial coverage. For this study Landsat TM data were used to identify the dominant categories of land cover characteristics and surfacial features of watershed to aid in partitioning of the watershed into ground respones units 9 GRUs) in conjunction with topographic data. Leaf are index ( LAI ) s an important structural variable to quantify energy and mass exchange of palnt canopies was also estimated from the TM data. A vegetation-indices based model was used to relate field measured LAI from different vegetation stands to vegetation index 9 NDVI) from the Landsat TM data. GIS applications are becoming popular in application of hydrologic modeling for parameter estimation and watershed partition. Data overlays were used in estimating selected model parameters such as vegetation type and density, and soil physical properties.
Area description and Climate
Trolak watershed covers 66 km2 of the north0east of Bernam River basin in the state of Perak. The topography is hilly ot mountainous country rising to heights of 1067 m ( a.m.s.i) in the north-east corner. The are is composed principally of sedimentary deposits. All soils are well drained. Textrual classes mostly lie between loam to clay with moderate to average soil moisture holding capacity. Most of the lowland is plantation of 15 to 25 year-old Oil palm and Hevea. On higher evaluations the dominant vegetation is dipterocarp forest. Phydically and hydro logically the Trolak watershed is typical hilly tropical watershed with 2794 mm average annual rainfall. The annual distribution of rainfall is influenced by the north-east and south-west monsoons. The rainfall is seasonally distributed with the maximum mean monthly rainfall of 339 mm occurring in November and the minimum of 149 mm in July. The temperature and humidity seasonal variation is little. Tj9e daily temperature varies form 180C to 360C and annual mean is 27.50C.
Evapotranspiration (ET) Modeling
A physical distributed modeling approach is used to model ET from tropical vegetated watershed using daily rainfall, and meterological data to adequately reflect the effects on ET of changes in vegetation, climate and soil moisture. For the purpose of this study, vegetation is broadly categorized as rainforest , plantation of Oil palm and Heva and mixed vegetation. The concept of division of the watershed into subunits ( GRUs) of homogeneous vegetation is used to provide capability of the distributed parameters. In order to compute average responses. Evaporation form vegetation canopy, and transpiration from plants re modeled separately and used to compute actual ET at the end of each day.
As a brief summary of the model function, canopy vegetation is treated as proportional to the LAI. Interception is computed as a reflection of LAI and evaporation form the canopy is limited by the potential radiation energy. Net rainfall is les an amount for intercepted rainfall is added to the ground surface where it is available for transpiration. The available potential radiation energy is used for evaporating canopy intercepted water and transpiration energy based on the LAI and the species dependent light extinction coefficient. Transpiration is compound using he Penman-Monteith ( PM) equation incorp9orating radiation and vapour pressure deficit drivers for evaporation, and linked with the soil moisture balance model to account of environmental variables. The humidity, temperature, and radiation correction are also done sequentially in computing canopy resistances, which in turn affects transpiration. Aerodynamic conductance was considered 0.2 ms-1 in the PM equation based on forest ET modeling experiments by calder et al., (1986) in this region. The values of light extinction coefficient ranging from 0.35 to 6 were used for Beers aw attenuation of incident radiation to produce canopy average radiation. Soil moisture model based on field capacity is considered. Soil moisture is updated at the end of each day after correcting for transpiration, infiltration, an drainage. As the completer modedl is rather large , it is not possible in this paper to cover all the equations used in the model.
