Neural-estimator for the surface emission rate of atmospheric gases

AbstractThe emission rate of minority atmospheric gases is inferred by a new approachbased on neural networks. The new network applied is the multi-layer per-ceptron with backpropagation algorithm for learning. The identi cation ofthese surface uxes is an inverse problem. A comparison between the newneural-inversion and regularized inverse solutions is performed. The resultsobtained from the neural networks are signi cantly better. In addition, theinversion with the neural networks is faster than regularized approaches, aftertraining.Key words: Neural networks; inverse problems; surface emission rate ofatmospheric gases.1. IntroductionThe enhancing of the concentration of greenhouse e ect gases is a centralissue nowadays, meanly regarding the most important anthropogenic gases,such as methane (CH 4 ) and carbon dioxide (CO 2 ). Despite the rati cationof the Kyoto Protocol, the forecast is that the releases of CO 2 and CH 4 inthe atmosphere continue to increase in next decade [16].One mandatory strategy is to monitoring the concentration of these gasesin the atmosphere. However, in order to understand the bio-geochemicalcycle of these gases, it is necessary to estimate the surface emission rates.One procedure for this is to employ inverse problem methodology.