Paper Title
Inter Comparison Of Gleam, Gldas And Modis Based Actual Evapotranspiration In Asia Using Triple Collocation Analysis

Abstract
Actual Evapotranspiration (AET) is the most critical component of hydrological cycle because it links the energy exchanges between land surface and atmosphere and it is directly associated with climate change. High spatio-temporal variability of AET makes it difficult to be calculated on large scale. Recently Remote Sensing (RS) data is being used to calculate large scale AET datasets like MODerate resolution imaging spectroradiometer (MOD16), Global Land Evaporation Amsterdam Model (GLEAM) and Global Land Data Assimilation System (GLDAS). These Global AET products are disseminated and being used for drought monitoring, flood control, irrigation management and climate change prediction. Before using these global AET products it is imperative to estimate the errors associated with these global AET datasets. For an accurate measurement of AET at point scale, several ground based techniques have been deployed e.g. scintillometers and lysimeters but these are unable to provide spatial distribution of hydro meteorological datasets on large or global scale due to their limited foot print size and sparse network density. In order to circumvent this issue, we implemented the Triple collocation (TC) error estimation technique to calculate the Root Mean Square Error (RMSE) of three global AET datasets (MOD16, GLEAM, GLDAS) using 11 years of datasets (2000-2010). TC was only applied to pixels with at least 500 triplets in order to limit the estimation error to within 10%. We also implemented the point based error estimation and validated these AET products against 10 Eddy Covariance (EC) based flux tower measurements and different statistical metrics were applied (i.e. Index of Agreement (IOA), Correlation Coefficient (R), Bias and RMSE). The result obtained from TC was further validated against different land cover types in Asia using MODIS land cover product (MCD12Q1). Different land cover types which were considered for analysis were mixed forest, Rice paddy, Grassland and Urban canyon. TC has never been applied to GLDAS, GLEAM and MOD16 AET datasets. Previously TC was only applied to point based AET datasets therefore this research is also unique in a sense that TC is applied on both point and grid based dataset and then the results were further evaluated against different land cover types in Asia to get the best agreement. Keywords- Evapotranspiration, Triple collocation analysis, MOD16, GLEAM, GLDAS, MODIS, Taylor diagram, Eddy covariance