Este trabalho discute a composição de campos de precipitação a partir de duas fontes de dados: os pluviômetros e o radar meteorológico. Estudaram-se métodos baseados somente em dados de pluviômetros, somente em dados de radar, e técnicas que combinam as duas fontes de dados. O objeto de estudo é a bacia do rio Jacaré-Guaçu, que conta com 65 postos pluviométricos e um radar meteorológico, do IPMet-Bauru. Foi feita uma comparação direta entre os campos gerados pelas diversas técnicas, onde foi avaliado o comportamento do índice G (razão entre o registro pluviométrico e a média dos registros de radar dos 9 pixels que circundam o pluviômetro), a capacidade destas técnicas de fornecer a chuva pontual e a altura média diária e anual de chuva por área de integração. Os métodos compostos apresentaram valores pontuais de chuva muito elevados e foram introduzidos fatores limitantes para compensar estas super correções. Os resultados obtidos reproduziram qualitativamente os valores da literatura. Ao analisar a média das chuvas diárias para toda a bacia e para todo o período, utilizando como padrão de comparação o método do Inverso do Quadrado da Distância (IQD), constatou que o radar fornece valores 12% menores, e que os métodos mistos apresentam diferenças na faixa de -0,5 a +16%. Nesta forma de análise existe um ganho ao se utilizar as técnicas mistas, porém ao se trabalhar com valores diários, integrados em sub-bacias, as diferenças atingem valores de -45% até +70%. Estes campos de precipitação foram aplicados em um modelo hidrológico distribuído, de embasamento físico, com 19 parâmetros calibráveis. Trabalhou-se com 10 postos fluviométricos e com 6 anos de dados. A calibração foi feita com dois anos e o restante deles foi usado para validação. Para garantir a comparação entre os resultados usou-se rigorosamente a mesma metodologia de calibração, com apoio de algoritmo genético. Foram utilizadas três funções objetivo: uma para verificação dos picos, outra para recessão e a última para avaliar a diferença de volume. Verificou-se que os melhores resultados foram obtidos para os métodos IQD, Brandes com o maior limitador, Radar e Costa. Nestes métodos, o processo de calibração consegue compensar as diferenças dos campos de precipitação. As diferenças observadas nos campos de precipitação foram reproduzidas nos hidrogramas. Os hidrogramas resultantes da aplicação dos dados de radar não reproduziram bem a recessão e os hidrogramas resultantes dos campos gerados apenas por pluviômetros apresentam picos elevados. As técnicas mistas ora atenuam os picos ora intensificam-nos. Sugere-se mais pesquisa para o desenvolvimento de métodos mistos que explorem as vantagens dos dois equipamentos de medida de chuva / This work discusses the composition of precipitation fields using two data sources: rain gauges and weather radar. Methods based solely on rain gauges, on weather radar, and techniques that combine these two measurement instruments were studied. The study object is the Jacaré-Guaçu river basin, with 65 rain gauges and a meteorological radar (IPMet-Bauru). A direct comparison of these fields generated by diverse techniques was made to study the following subjects: G index (reason between the rain gauge register and the average of the 9 pixels radar registers that surround the rain gauge), the capacity of these techniques to supply the point rain and the daily and annual mean rain height over an integration area. The combined methods provides very high point values, therefore some limitations were introduced to compensate these super corrections. The literature results were qualitatively reproduced in this study. The daily mean rain height comparative analyses for all the basin, and for all the period, evidenced that the radar supplies to values 12% minors, and that the composed methods present differences from -0,5 up to +16%; the comparison pattern was the Inverse of Square Distance method (ISD). The study of mean rain height calculated over a great period and to the entire river basin shows a profit when using the combined techniques; however, when daily values integrated in sub-basins are used, the differences reach values from -45% until +70%. These precipitation fields had been applied in a distributed hydrologic model, physically based, with 19 calibrated parameters. There were 10 fluviometric stations and 6 years of data. The calibration was made with two years, and that remain data was used for validation. To guarantee the results comparison, the same calibration methodology was rigorously used, with support of genetic algorithm. Three objective functions were used: one for peaks verification, another for recession analyses and the last one for volume difference evaluation. The best results were achieved by the application of the precipitation fields gotten by ISD, Brandes with high limitation, Radar and Costa methods. For these methods, the calibration process compensated the differences on the precipitation fields. The differences observed in the precipitation fields had been reproduced in the hydrograms. The hydrograms of the radar data applications had not well reproduced the recession curve, and the hydrograms of the precipitation fields based only on rain gauges presented high peaks. Sometimes the composed techniques attenuate the peaks, however, sometimes they intensify them. More research is recommended to develop compoud methods that explore the advantages of the two equipments for rain measure

Numerical models often fail to accurately simulate and forecast a hydrological state in operation due to its inherent uncertainties. Data Assimilation (DA) is a promising technology that uses real-time observations to modify a model's parameters and internal variables to make it more representative of the actual state of the system it describes. In this thesis, hydrological DA is first reviewed from the perspective of its objective, scope, applications and the challenges it faces. Special attention is then given to nonlinear Kalman filters such as the Extended Kalman Filter (EKF). Based on a review of the existing studies, it is found that the potential of EKF has not been fully exploited. The Soil and Water Assessment Tool (SWAT) is a semi-distributed rainfall-runoff model that is widely used in agricultural water management and flood forecasting. However, studies of hydrological DA that are based on distributed models are relatively rare because hydrological DA is still in its infancy, with many issues to be resolved, and linear statistical models and lumped rainfall-runoff models are often used for the sake of simplicity. This study aims to fill this gap by assimilating streamflow and surface soil moisture observations into the SWAT model to improve its state simulation and forecasting capability. Unless specifically defined, all ‘forecasts’ in Italic font are based on the assumption of a perfect knowledge of the meteorological forecast. EKF is chosen as the DA method for its solid theoretical basis and parsimonious implementation procedures. Given the large number of parameters and storage variables in SWAT, only the watershed scale variables are included in the state vector, and the Hydrological Response Unit (HRU) scale variables are updated with the a posteriori/a priori ratio of their watershed scale counterparts. The Jacobian matrix is calculated numerically by perturbing the state variables. Two case studies are carried out with real observation data in order to verify the effectiveness of EKF assimilation. The upstream section of the Senegal River (above Bakel station) in western Africa is chosen for the streamflow assimilation, and the USDA ARS Little Washita experimental watershed is chosen to examine surface soil moisture assimilation. In the case of streamflow assimilation, a spinoff study is conducted to compare EKF state-parameter assimilation with a linear autoregressive (AR) output assimilation to improve SWAT’s flood forecasting capability. The influence of precipitation forecast uncertainty on the effectiveness of EKF assimilation is discussed in the context of surface soil moisture assimilation. In streamflow assimilation, EKF was found to be effective mostly in the wet season due to the weak connection between runoff, soil moisture and the curve number (CN2) in dry seasons. Both soil moisture and CN2 were significantly updated in the wet season despite having opposite update patterns. The flood forecast is moderately improved for up to seven days, especially in the flood period by applying the EKF subsequent open loop (EKFsOL) scheme. The forecast is further improved with a newly designed quasi-error update scheme. Comparison between EKF and AR output assimilation in flood forecasting reveals that while both methods can improve forecast accuracy, their performance is influenced by the hydrological regime of the particular year. EKF outperformed the AR model in dry years, while AR outperformed the EKF in wet years. Compared to AR, EKF is more robust and less sensitive to the length of the forecast lead time. A combined EKF-AR method provides satisfying results in both dry and wet years. The assimilation of surface soil moisture is proved effective in improving the full profile soil moisture and streamflow estimate. The setting of state and observation vector has a great impact on the assimilation results. The state vector with streamflow and all-layer soil moisture outperforms other, more complicated state vectors, including those augmented with intermediate variables and model parameters. The joint assimilation of surface soil moisture and streamflow observation provides a much better estimate of soil moisture compared to assimilating the streamflow only. The updated SWAT model is sufficiently robust to issue improved forecasts of soil moisture and streamflow after the assimilation is ‘unplugged’. The error quantification is found to be critical to the performance of EKF assimilation. Nevertheless, the application of an adaptive EKF shows no advantages over using the trial and error method in determining time-invariant model errors. The robustness of EKF assimilation is further verified by explicitly perturbing the precipitation ‘forecast’ in the EKF subsequent forecasts. The open loop model without previous EKF update is more vulnerable to erroneous precipitation estimates. Compared to streamflow forecasting, soil moisture forecasting is found to be more resilient to erroneous precipitation input.

Reusing stormwater is a sustainable approach that a lot of cities around the world, including cities in Canada, are developing to improve local and regional water resources. For this purpose, water is typically withdrawn from stormwater ponds (large urban infrastructure that retain stormwater) and used for applications that require less than pristine water quality. However, the large size of these ponds along with the heterogeneity in water quality internally, make the withdrawal location from these ponds for reusable stormwater critically important. Also due to the large sizes of these ponds, collecting data throughout the pond to determine the optimal location for withdrawal is not practical. Modelling however, can provide a more practical means of studying contaminant distribution within the pond over time in order to identify the withdrawal location, among other valuable information. In this dissertation, a modelling approach was developed that simulates fate and transport of bacteria in stormwater ponds after rainstorm events. The model was run to simulate bacteria in the Inverness stormwater pond, which is a large T-shaped pond located in southeast of the City of Calgary, Alberta, Canada. The model has two components: a hydrological component and a Computational Fluid Dynamics (CFD) component. The hydrological component calculates the stormwater runoff of the subbasins of the catchment draining into the pond. The results were compared with collected data and good agreement was observed. Then, the results were fed to the CFD component as input in order to simulate the distribution of contamination brought in by the local hydrology. The CFD component simulates the hydrodynamics of the pond 3-dimensionally. The model was run based on collected data from the pond and multiple versions of the model were developed with regard to free-surface and particulate-attached bacteria transport. In order to address a common issue with hydro-environmental models – being difficult to validate - the model was validated in two ways. First, an instrument was designed and built to measure fluid flow velocity magnitude and direction in the pond. Once calibrated, it was deployed to the pond and the flow field was measured at multiple locations for validation purposes. Second, a non-dimensional number was introduced allowing a comparison between the bacteria concentration data from collected data and that of modelling result in multiple locations of the pond. In both of the validations, good agreement with collected data was observed. A volume of Fluid model and sediment transport model were integrated into the model, which allowed consideration of free-surface effects and for modeling wider range of bacteria, respectively. The model was used to identify the optimal location for water withdrawal for reuse. The middle of the pond, where the three wings join and near the surface, was located as the optimal location due to the lowest bacteria concentration. In an attempt to improve the water quality in the optimal location, strategic tree planting on the north bank of the West wing was studied. It was shown that the trees can reduce the transport of bacteria from the most contaminated location to the withdrawal location. The model was also used to study the impact of some of the important assumptions and environmental factors, such as rain and wind, on bacteria distribution. Wind was found to play a crucial role in the bacteria distribution in the pond. / Graduate

Magíster en Ciencias de la Ingeniería, Mención Recursos y Medio Ambiente Hídrico. Ingeniero Civil / La representación del ciclo hidrológico es, hasta el día de hoy, difícil de modelar producto de diversos desafíos entre los que se encuentran la heterogeneidad de la cuenca, la representación de procesos físicos, las observaciones disponibles, la incertidumbre en las forzantes del modelo y la incertidumbre estructural, entre otros. Sin embargo, en los últimos años se ha puesto a dis-posición de la comunidad diversos productos e imágenes satelitales que buscan aumentar la disponibilidad de información mediante percepción remota. Algunas de las variables hidroló-gicas de interés que hay disponibles son la fracción de área cubierta por nieve, la humedad de suelo, la evapotranspiración, el equivalente en agua de nieve, índices de vegetación, precipita-ción, temperatura del suelo y el albedo. En este trabajo se busca incorporar información de percepción remota para analizar si su inclusión mejora la modelación hidrológica tradicional que se basa, principalmente, en el contraste de caudales modelados con los observados. La zona de estudio corresponde a tres sub-cuencas del río Elqui, en la Región de Coquimbo, que están definidas por estaciones fluviométricas pertenecientes a la DGA: río Cochiguaz en el Peñón, estero Derecho en Alcohuaz y río Toro antes junta río La Laguna. En estas cuencas se intenta (1) estimar los caudales usando el modelo CRHM (Cold Regions Hydrological Model) considerando que no existe información fluviométrica, bajo la hipótesis que es posible estimar el caudal de manera razonable si se calibran parámetros asociados a otros procesos del ciclo hidrológico, (2) calibrar el modelo usando sólo caudales y (3) incorporar la información de percepción remota junto con la fluviométrica para representar el ciclo hidrológico. Todo esto a una escala temporal horaria, con forzantes que se construyen a partir de información de la red meteorológica DGA en conjunto con la red CEAZA. Los resultados muestran que estimar la escorrentía en cuencas sin información fluvio-métrica sigue siendo un desafío, pues los caudales estimados a partir de la calibración de otros procesos entregan, para distintos sets de parámetros, índices NSE que, en promedio, son 0,56, 0,26 y -0,29 para Cochiguaz, Derecho y Toro respectivamente. No obstante, la modelación que considera percepción remota y caudales mejora los índices de 0,74 a 0,89 y de 0,75 a 0,8 para las cuencas Cochiguaz y Derecho respectivamente. En el caso del río Toro, el NSE se mantiene en 0,74.

In 1910, Paris experienced its biggest flood in the 20th century. In 2010, for the anniversary of this event – supposed to happen every 100 years ! – the flood prediction model that is now used on the Seine basin was tested on its simulation,… and failed to reproduce the observed flood volume. This paper will try to explain, and correct, such disappointing results. Many hypotheses have been tested and based on their results, it has been decided to develop a frost module in order to assess the influence of this phenomenon – that is not taken into account by the lumped hydrological model that is used – on the flood formation. A soil temperature model using air temperature as input data was also designed because soil temperature data were not available in 1910. The addition of the frost module did not, however, bring many improvements to the 1910 flood simulation because frost is a too rare phenomenon on the Seine basin for the module to be correctly calibrated. However, new perspectives are presented to continue the research on this phenomenon.

Earth system modeling, climate modeling, water resource research as well as integrated modeling (e.g., climate impact studies) require the coupling of hydrological and meteorological models. The paper presents recent concepts on such a coupling. It points out the difficulties to be solved, and provides a brief overview on recently realized couplings. Furthermore, a concept of a hydrometeorological module to couple hydrological and meteorological models is introduced. / Wasserresourcenforschung, Erdsystem- und Klimamodellierung sowie integrierte Modellierung (z.B. Klimafolgenforschung) erfordern das Koppeln von hydrologischen und meteorologischen Modellen. Dieser Artikel präsentiert Konzepte für eine solche Kopplung. Er zeigt die zu lösenden Schwierigkeiten auf und gibt einen kurzen Überblick über bisher realisierte Kopplungen. Ferner stellt er ein Konzept für einen hydrometeorologischen Moduls zur Kopplung von hydrologischen mit meteorologischen Modellen vor.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第22760号 / 工博第4759号 / 新制||工||1744(附属図書館) / 京都大学大学院工学研究科都市環境工学専攻 / (主査)教授 清水 芳久, 教授 田中 宏明, 教授 米田 稔 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Devido ao processo de crescimento das cidades e atividade agrícolas há impactos que afetam os recursos hídricos como a escassez, degradação da qualidade de água e consequente conflitos de usos. Neste contexto, os modelos hidrológicos são importantes ferramentas para avaliar o comportamento hídrico de bacias hidrográficas, além de também poderem ser utilizados na previsão de cenários a fim de verificar o impacto do uso e ocupação do solo. Assim, o presente estudo teve como finalidade avaliar o desempenho do modelo Soil and Water Assessment Tool (SWAT) na simulação da vazão na Bacia Hidrográfica do Ribeirão Taquaruçu. A calibração automática foi realizada com o programa SWAT-CUP, em passo diário, com dados de duas estações fluviométricas da BRK Ambiental no período de abril de 2013 a julho de 2014, enquanto os dados de agosto de 2014 a agosto de 2015 foram utilizados no período de validação. A análise de sensibilidade foi realizada com 14 parâmetros selecionados entre aqueles mais significativos para a simulação de vazão em região do bioma Cerrado. Os resultados da análise de sensibilidade indicaram que os parâmetros mais influentes são o SOL_K (condutividade hidráulica saturada do solo) e CN2 (curva número para a condição II). Para avaliação do desempenho do modelo foram utilizadas as funções objetivos Nash-Sutcliffe (NSE) e o coeficiente de determinação (R2) que indicaram, no período de calibração, os seguintes valores: -0,05 e 0,55 (sub-bacia 1); 0,51 (sub-bacia 3), respectivamente. Já no período de validação foram apresentados os seguintes resultados para NSE e R2: 0,44 e 0,54 (sub-bacia 1); 0,24 e 0,29 (sub-bacia 3), nesta ordem. Considerando que o objeto de estudo é importante por ser responsável por grande parte do abastecimento de água do município de Palmas – TO e que o software não conseguiu simular de forma adequada as vazões mínimas, os resultados são considerados insatisfatórios. / Due to the process of city growth and agricultural activity there are impacts that affect water resources such as scarcity, degradation of water quality and consequent conflicts of uses. In this context, hydrological models are important tools to evaluate the hydrological behavior of watersheds, and can also be used to predict scenarios to verify the impact of land use and occupation. Thus, the present study had as purpose to evaluate the performance of the Soil and Water Assessment Tool (SWAT) model in the flow simulation in the Ribeirão Taquaruçu Watershed. The automatic calibration was performed with the SWAT-CUP program, in a daily step, with data from two BRK Ambiental fluviometric stations from April 2013 to July 2014, while data from August 2014 to August 2015 were used in the period of validation. The sensitivity analysis was performed with 14 parameters chosen considering the most significant ones for the simulation of flow in the Cerrado region. The results of the sensitivity analysis indicated that the most influential parameters are the SOL_K (saturated hydraulic conductivity of the soil) and CN2 (number curve for condition II). The Nash-Sutcliffe objective functions (NSE) and the determination coefficient (R2) were used to evaluate the performance of the model, which indicated, in the calibration period, the following values: - 0.05 and 0.55 (sub-basin 1); 0.51 (sub-basin 3), respectively. In the validation period, the following results were presented for NSE and R2: 0,44 and 0,54 (sub-basin 1); 0.24 and 0.29 (sub-basin 3), in that order. Considering that the object of study is important because it is responsible for a large part of the water supply of the municipality of Palmas – TO and that the software was not able to simulate the minimum flows adequately, the results are considered unsatisfactory.

Studies from the Forestry Institute of São Paulo State have shown that in the end of the 20th century, the native forest cover of the state of SP reached the maximum level of forest loss. From that point on, a period of forest increase and expansion started. Industrialization, law enforcement, economic benefits, and social pressure experienced in recent years are believed to be contributing to the preservation and regrowth of the native vegetation cover in certain locations. This study proposed to model the dynamics of native vegetation cover in the Piracicaba River basin (12,500 km²) in the state of São Paulo, Brazil, to evaluate possible effects of these changes in ecosystem services related to river flow & regulation and landscape structure, linking to biodiversity & habitat supported by forest patches. To achieve the proposal set out in this research, dynamic models of native vegetation were established. Thematic land cover maps of the years 1990, 2000 and 2010, originated from Landsat 5 TM images, formed the spatiotemporal basis of this study. With the aid of Dinamica EGO (a dynamic modeling software), three future scenarios were created, called status quo (SQ), no deforestation (ND) and riparian restoration enforcement (RRE). An analysis using weights of evidence was done to identify forest transition drivers. The drivers are divided into two groups, (1) environmental & physical, consisting of soil types, hydrographic network, rainfall and presence of native forest fragments and (2) anthropic, consisting of population density, gross national product, road network, urban patches and predominant rural activities. Resulting scenarios were analyzed by means of landscape metrics to compare and qualify vegetation patches in relation to structure as proxy for supporting ecosystem services. Finally, Soil & Water Assessment Tool (SWAT), a hydrological model, was used to determine the influence of different forest scenarios in mean annual water yield and regulation processes throughout the basin, and, therefore, compare scenarios as to effects on regulating ecosystem services. Results show that forest transition is indeed occurring, with native vegetation cover parting from 24.4% in 1990, to 20.1% in 2000 and 21.8% in 2010. Scenario results were of 22.4% (SQ), 43.2% (ND) and 28.4% (RRE) for 2050. Forest loss was identified as a product of anthropogenic drivers while regrowth was of physical & environmental drivers. When the area was segmented, regions with greater environmental condition resulted in improved values of landscape structure. SQ scenario was the most affected, losing small patches of forest that could function as structural connectors, and therefore potentially affect biodiversity and habitat. Mean annual water yield was reduced with forest regrowth by as much as 10.3% in ND. We concluded that the dynamics occurring in the landscape and the proposed scenarios affect mean annual water yield, regulation and landscape structure, allowing us to discuss differences between the scenarios and the relation between forest dynamics, landscape structure, hydrology and overtime potential effects over regulating and supporting ecosystem services. / Levantamentos do Instituto Florestal de São Paulo têm mostrado que no final do século XX a cobertura florestal nativa total do estado atingiu um patamar de perda e que se iniciou então um período de expansão. Rigidez de leis, fiscalização, benefícios econômicos, além de pressão social demonstrados nos últimos anos têm contribuído para essa expansão da vegetação nativa em certos locais. Este estudo propôs modelar a dinâmica da cobertura florestal nativa na bacia do Rio Piracicaba (12.500 km2), localizada no Estado de São Paulo, para averiguar os possíveis efeitos dessas mudanças nos serviços ecossistêmicos ligados à vazão e regulação de rios, além da própria estrutura da paisagem simulada, interligando-se com biodiversidade e habitat, promovidos pelos remanescentes florestais. Para atingir a proposta estabelecida nesta pesquisa, modelos de dinâmica da vegetação nativa foram desenvolvidos. Foram utilizados mapas temáticos de cobertura e uso do solo dos anos 1990, 2000 e 2010 originados a partir de imagens Landsat 5 TM. Com auxílio do software Dinamica EGO, especializado em modelagem da paisagem, criaram-se três modelos espaciais e temporais da dinâmica florestal, levando em consideração os cenários status quo (SQ), no deforestation (ND) e riparian restoration enforcement (RRE). Uma análise usando pesos de evidência foi utilizada para identificar as variáveis de transição florestal. As variáveis foram divididas em dois grupos, (1) físicas e ambientais, consistindo de tipos de solo, rede de drenagem, pluviosidade e presença de fragmentos florestais e (2) antrópicos, consistindo de densidade populacional, produto interno bruto, rede viária, zonas urbanas e predominância de atividade rural. Os cenários resultantes foram analisados por métricas de paisagem para fim de comparação e qualificação dos fragmentos em relação a sua estrutura, interligando-se aos serviços ecossistêmicos de suporte. Por último, foi realizada uma modelagem hidrológica usando o modelo Soil & Water Assessment Tool (SWAT) para averiguar a influência da mudança florestal na regulação de vazão de rios e portanto comparar os cenários em relação aos seus efeitos sobre serviços ecossistêmicos de regulação interligados à água. Resultados mostraram que transição florestal ocorreu, passando a cobertura florestal de 24,4% em 1990 para 20,1% em 2000 e então 21,8% em 2010. Cenários resultaram em uma cobertura florestal de 22,4% (SQ), 43,2% (LE) e 28,4% (RRE) para o ano de 2050. A perda de floresta foi identificada como produto de variáveis de natureza antrópica enquanto o ganho florestal foi de variáveis físicas e ambientais. Regiões com melhores condições ambientais resultaram em melhores valores de estrutura da paisagem. SQ foi afetado principalmente pela perda de pequenos fragmentos florestais que funcionam como conectores estruturais da paisagem, potencialmente afetando a biodiversidade e habitat. O deflúvio médio anual foi reduzido em até 10,3% com o incremento florestal observado em ND. Conclui-se que a cobertura florestal na paisagem e os cenários propostos afetam o deflúvio, regulação e a estrutura da paisagem, nos permitindo discutir nas diferenças entre cada cenário e a relação entre dinâmica florestal, estrutura da paisagem, hidrologia e potenciais efeitos nos serviços ecossistêmicos de suporte e regulação.

Estudos hidrológicos são extremamente importantes em projetos de obras lineares, nas quais o traçado deve minimizar o risco de instabilidade tanto na fase de construção como na fase de operação. O escoamento superficial, além de ser um parâmetro fundamental para a definição do traçado e dos projetos subsequentes, influencia diretamente a dinâmica dos processos geológico-geotécnicos na área do empreendimento. A presente pesquisa propõe um método para realizar uma estimativa das vazões máximas nas travessias de obras civis lineares, integrando modelos hidrológicos obtidos com Sistema de Informação Geográfica (SIG) e métodos usuais de cálculo de vazões de cheia. O método foi aplicado em um setor do oleoduto São Paulo – Brasília (OSBRA), entre os municípios de São Simão e Cravinhos (SP). O modelo hidrológico foi obtido no software ArcGIS 9.3, a partir da definição de uma área mínima de contribuição de 5 hectares, adequada para a escala da base cartográfica digital (1:10.000) e para as características da área de estudo. As 29 bacias de contribuição delimitadas foram posteriormente validadas na vistoria de campo. A compatibilidade entre as bacias geradas pelo modelo hidrológico e as bacias identificadas visualmente na área de estudo foi satisfatória. As vazões máximas foram calculadas pelo Método Racional (bacias com área inferior a 2 km²) e pelo Método Racional Modificado (bacias com área superior a 2 km²), que abrangeu o Método de McMath, o Método Racional com expoente redutor de área e o Método Racional com coeficiente de retardo, sendo o último o que apresentou os resultados mais razoáveis. O cálculo das vazões de cheia foi realizado para períodos de retorno de 10 e 50 anos, a partir dos quais foram gerados Mapas de Vazões Máximas, que foram utilizados para inferência de riscos de eventos perigosos de natureza geológico-geotécnica causados pela ação das águas pluviais na região do oleoduto. Como o traçado do oleoduto foi posicionado ao longo dos divisores principais, não foram identificados locais críticos. Posteriormente, o traçado do oleoduto foi utilizado como referência para uma aplicação do método proposto em um projeto de drenagem. As bacias com valores de vazões máximas classificados com muito baixos não foram incluídas devido aos fluxos pouco significativos. Para as bacias com valores de vazões de cheia classificados como baixos e médios, sugeriu-se a utilização de dispositivos de drenagem superficial. Para as bacias com vazões máximas classificadas como altas e muito altas, foi proposta a utilização de dispositivos de drenagem de transposição de talvegues. Os resultados obtidos na pesquisa permitiram a consolidação do método proposto para aplicação em outros tipos de obras lineares e em locais com diferentes características ambientais. / Hydrological studies are extremely important in linear constructions projects, in which the routing must minimize the risk of instability both during construction and in operation phase. The surface runoff, in addition of being a fundamental parameter to routing definition and subsequent projects, directly influences the dynamics of geological-geotechnical processes in the enterprise area. The present research proposes a method to estimate maximum flows in crossing locations in linear civil constructions, integrating hydrological models obtained with Geographic Information System (GIS) and usual methods of peak flows calculation. The method was applied in a sector of São Paulo – Brasília (OSBRA) oil pipeline, between the municipalities of São Simão and Cravinhos (SP). The hydrological model was obtained in ArcGIS 9.3 software, from the definition of a minimal area of contribution of 5 hectares, adequate for the scale of the digital cartographic base (1:10.000) and for the study area characteristics. The 29 delimited watersheds were subsequently validated in the field visit. The compatibility between watersheds generated in the hydrological model and watersheds visually identified in the study area was satisfactory. The maximum flows were calculated by Rational Method (watersheds with area less than 2 km²) e by Modified Rational Method (watersheds with area higher than 2 km²), which included McMath Method, Rational Method with reducing exponent area and Rational Method with retard coefficient, and the latter presented the most reasonable results. Peak flows calculation was performed for return periods of 10 and 50 years, from which were generated Maximum Flows Maps, which were used for inference of risks of hazardous events of geological-geotechnical nature caused by rainwater action in the pipeline region. The pipeline routing was positioned along the main dividers, therefore critical locations were not identified. Subsequently, the pipeline routing was used as a reference for application of the proposed method in a drainage project. Basins with maximum flows classified as very low were not included due to its low significance flows. For basins with peak flows values classified as low and medium, it was suggested the use of surface drainage devices. For basin with maximum flows classified as high and very high, it was proposed the use of thalwegs transposition drainage devices. The results obtained in the research allowed the consolidation of the proposed method to application in other types of linear constructions and in locals with different environmental characteristics.