No abstract

No abstract

The existing medium-resolution land cover time series produced under the European Space Agency's Climate Change Initiative provides 29 years (1992-2020) of annual land cover maps at 300-m resolution, allowing for a detailed study of land change dynamics over the contemporary era. Because models need 2D parameters rather than 2D land cover information, the land cover classes must be converted into model-appropriate plant functional types (PFTs) to apply this time series to Earth system and land surface models. The first generation cross-walking table that was presented with the land cover product prescribed pixel-level PFT fractional compositions that varied by land cover class but lacked spatial variability. Here we describe a new ready-to-use data product for climate modelling: spatially explicit annual maps of PFT fractional composition at 300 m resolution for 1992-2020, created by fusing the 300 m medium-resolution land cover product with several existing high-resolution datasets using a globally consistent method. In the resulting data product, which has 14 layers for each of the 29 years, pixel values at 300-m resolution indicate the percentage cover (0-100 %) for each of 14 PFTs, with pixel-level PFT composition exhibiting significant intra-class spatial variability at the global scale. We additionally present an updated version of the user tool that allows users to modify the baseline product (e.g., re-mapping, re-projection, PFT conversion, and spatial sub-setting) to meet individual needs. Finally, these new PFT maps have been used in two land surface models -ORCHIDEE and JULES -to demonstrate their benefit over the conventional maps based on a generic cross-walking table. Regional changes in the fractions of trees, short vegetation, and bare soil cover induce changes in surface properties, such as the albedo, leading to significant changes in surface turbulent fluxes, temperature, and vegetation carbon stocks.

This paper proposes the utilization of hybrid models of supervised neural networks for the modelling of dynamic systems. Particularly, as an example of a system, a autonomous helicopter or UAV is identified in both attitude and position systems. The evaluation of the model is done by comparing the radial basis and multilayer perceptron with the real system.

No abstract