id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-024552-hgowgq41	Zhang, Ruixi	Hydrological Process Surrogate Modelling and Simulation with Neural Networks	2020-04-17		.txt	text/plain	3564	226	53	Moreover, we argue that the neural network model, although trained on some example terrains, is generally capable of simulating terrains of different sizes and spatial characteristics. We propose to learn a flood surrogate model by training a neural network with pairs of inputs and outputs from the numerical model. With the trained model from a given data set, the neural network is capable of simulating directly spatially different terrains. Moreover, while a neural network is generally constrained to a fixed size of its input, the model that we propose is able to simulate terrains of different sizes and spatial characteristics. In Case 2, the network is trained and tested with 200 different synthetic DEMs. The data set is generated with Landlab. We propose a neural network model, which is trained with pairs of inputs and outputs of an off-the-shelf numerical flood simulator, as an efficient and effective general surrogate model to the simulator.	./cache/cord-024552-hgowgq41.txt	./txt/cord-024552-hgowgq41.txt