Abstract
Abstract. An inverse error variance weighting of the anomalies of three terrestrial evaporation (ET) products from the WACMOS-ET project based on FLUXNET sites is presented. The three ET models were run daily and at a resolution of 25 km for 2002–2007, and based on common input data when possible. The local weights, derived based on the variance of the difference between the tower ET anomalies and the modelled ET anomalies, were made dynamic by estimating them using a 61-day running window centred on each day. These were then extrapolated from the tower locations to the global landscape by regressing them on the main model inputs and derived ET using a neural network. Over the stations, the weighted scheme usefully decreased the random error component, and the weighted ET correlated better with the tower data than a simple average. The global extrapolation produced weights displaying strong seasonal and geographical patterns, which translated into spatiotemporal differences between the ET weighted and simple average ET products. However, the uncertainty of the weights after the extrapolation remained large. Out-sample prediction tests showed that the tower data set, mostly located at temperate regions, had limitations with respect to the representation of different biome and climate conditions. Therefore, even if the local weighting was successful, the extrapolation to a global scale remains problematic, showing a limited added value over the simple average. Overall, this study suggests that merging tower observations and ET products at the timescales and spatial scales of this study is complicated by the tower spatial representativeness, the products' coarse spatial resolution, the nature of the error in both towers and gridded data sets, and how all these factors impact the weights extrapolation from the tower locations to the global landscape.
Highlights
The surface latent heat flux governs the interactions between the Earth and its atmosphere (Betts, 2009), is an essential component of the water and energy cycles (Sorooshian et al, 2005), and plays a key role in the climate system and in the linking of biochemical cycles (Wang and Dickinson, 2012)
As the biome-based calibration of PM-MOD was derived with MODerate resolution Imaging Spectroradiometer (MODIS) products, any errors introduced by this simple rescaling can propagate to the PM-MOD estimates and can be responsible for some ET patterns differing from the official use of the Mu et al (2011) algorithm for the MODIS ET product
A simple average (SA-merger) and an inverse error variance weighting (WA-merger) of the three global ET products generated during the WACMOS-ET project is presented
Summary
The surface latent heat flux governs the interactions between the Earth and its atmosphere (Betts, 2009), is an essential component of the water and energy cycles (Sorooshian et al, 2005), and plays a key role in the climate system and in the linking of biochemical cycles (Wang and Dickinson, 2012). Monteith, 1965; Priestley and Taylor, 1972) to derive global estimates at different timescales and spatial scales This has led to the rise and proliferation of satellite observation-based retrieval models (and subsequent data sets) of ET over the last few years (for an overview see Wang and Dickinson, 2012; Zhang et al, 2016). Far from discouraging the use of these ET data sets, the inter-product differences have been perceived as an opportunity to foster research and find new means to combine these data sets in an optimal manner These efforts have ranged from averaging a number of ET products (Mueller et al, 2013) to more complex approaches, such as weighted averages (Hobeichi et al, 2018), fusion algorithms where the original ET products are combined to reproduce flux observations (Yao et al, 2017), or integration methodologies that seek consistency between ET products and related products of the water cycle (Aires, 2014; Munier et al, 2014). Here we explore a local flux tower-based weighting of GLEAM, PT-JPL, and PM-MOD and compare it with the more typical simple average, followed by an appraisal of the potential to globally extrapolate the resulting merging framework
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
