Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2016-07-22T15:28:33Z No. of bitstreams: 1 BoanergesVictorSoaresDeAraujo_DISSERT.pdf: 1038853 bytes, checksum: 6cb9d474725caa33946cdd3023ef86e4 (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2016-07-27T21:26:40Z (GMT) No. of bitstreams: 1 BoanergesVictorSoaresDeAraujo_DISSERT.pdf: 1038853 bytes, checksum: 6cb9d474725caa33946cdd3023ef86e4 (MD5) / Made available in DSpace on 2016-07-27T21:26:40Z (GMT). No. of bitstreams: 1 BoanergesVictorSoaresDeAraujo_DISSERT.pdf: 1038853 bytes, checksum: 6cb9d474725caa33946cdd3023ef86e4 (MD5) Previous issue date: 2015-08-17 / Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico (CNPq) / Gases-tra?o s?o importantes para nosso ambiente, mesmo se manifestando apenas em ?tra?os?, mas suas concentra??es devem ser monitoradas, para que quaisquer interven??es necess?rias sejam tomadas a tempo. Condi??es apropriadas para nossas vidas residem dentro de limites inferiores e superiores destes gases, por isso seu monitoramento ? essencial hoje em dia e diversas t?cnicas podem ser usadas para este fim. Uma delas ? a espectroscopia de absor??o ?ptica diferencial (DOAS), que consiste de uma regress?o - o m?todo cl?ssico usa m?nimos quadrados - para determinar a concentra??o dos gases-tra?o. Para melhorar os resultados, diversos estudos buscaram abordagens alternativas. Alguns tentaram pr?-processar o sinal, atrav?s de uma remo??o de ru?do, e.g., usando transformada wavelet discreta (DWT). Este trabalho apresenta um estudo semi-emp?rico para encontrar a DWT com o melhor desempenho para remo??o de ru?do. Dentre diversas fam?lias conhecidas, buscou-se aquelas cuja remo??o de ru?do foram melhor, preservando caracter?sticas principais do sinal original, ent?o com a redu??o do ru?do, o res?duo ap?s a regress?o tamb?m diminui. A an?lise leva em conta o n?vel de decomposi??o wavelet, o limiar aplicado nos coeficientes de detalhe e como aplic?-los - limiariaza??o abrupta ou suave. Os sinais usados prov?m de um banco de dados aberto e online com sinais caracter?sticos de gases-tra?o usualmente estudados. / Trace gases are important to our environment even though their presence comes only by ?traces?, but their concentrations must be monitored, so any necessary interventions can be done at the right time. There are some lower and upper boundaries which produce nice conditions for our lives and then monitoring trace gases comes as an essential task nowadays to be accomplished by many techniques. One of them is the differential optical absorption spectroscopy (DOAS), which consists mathematically on a regression - the classical method uses least-squares - to retrieve the trace gases concentrations. In order to achieve better results, many works have tried out different techniques instead of the classical approach. Some have tried to preprocess the signals to be analyzed by a denoising procedure - e.g. discrete wavelet transform (DWT). This work presents a semi-empirical study to find out the most suitable DWT family to be used in this denoising. The search seeks among many well-known families the one to better remove the noise, keeping the original signal?s main features, then by decreasing the noise, the residual left after the regression is done decreases too. The analysis take account the wavelet decomposition level, the threshold to be applied on the detail coefficients and how to apply them - hard or soft thresholding. The signals used come from an open and online data base which contains characteristic signals from some trace gases usually studied.