Rainfall-runoff quantification is one of the most important tasks in both engineering and watershed management as it allows to identify, forecast and explain watershed response. The division of the rainfall depth between infiltration and runoff has a high level of complexity due to the spatial heterogeneity in real catchments and the temporal precipitation variability, which provide scale effects on the overall runoff volumes. The Natural Resources Conservation Service Curve Number (NRCS CN) is the conceptual lumped model more recognized in the field of rainfall-runoff estimation. It has been applied worldwide for different purposes in hydrological models since it was formulated by the Soil Conservation Service (SCS). Its simplicity and ease of use has made this procedure attractive to be applied in any catchment by providing a reasonable response, especially when scarce information about the catchment response is available. In spite of its wide-spread use, some uncertainties must be examined for its proper application. Three uncertainties outstand among the rest: the procedure to select the most representative rainfall-runoff events, the best methodology to calibrate the parameters of the model (curve number CN and initial abstraction Ia) and the relationship between the mentioned parameters (λ=Ia/S).
This PhD Thesis exposes an advanced analysis to evaluate the influence of ratio λ for a set of the most representative watersheds of the Agricultural Research Service of the United Stated Department of Agriculture (ARS-USDA), which are characterized by different soil properties, land uses and climatic conditions. Moreover, to solve the inconveniences generated by the selection of the most representative rainfall-runoff events, a novel methodology based on the pattern of rainfall distribution is presented. Additionally, to improve the adaptation of NRCS CN model to different hypothesis about λ factor in ungauged watersheds, a new procedure based on the climate characteristics of each case is exposed. As a consequence, the use of NRCS CN model is updated and it is adapted to local conditions taking into account regional patterns of climate conditions.
On the other hand, the non-linearity process of runoff generation is conditioned by the watershed antecedent conditions which are commonly related to the initial soil moisture content. Despite the relevance of soil moisture measures when modelling flood is highlighted in several studies, much work is still needed to enhance its relation with runoff generation. The integration of those previous conditions in the NRCS CN model is frequently based on indicators linked to the Antecedent Precipitation Index (API). Alternatively, the integration of a Soil Moisture Accounting (SMA) procedure into the NRCSCN model achieved important improvements due to its relationship with a physically measurable variable in the watershed, instead of indirect estimations based on disconnected variables e.g. API. In order to overcome its recognized inconsistency in the examination of initial soil moisture store level, a new expression that correlates soil moisture and model parameters is provided in this work. The mentioned correlation establish a relationship between soil moisture content and model parameters generating a new rainfall runoff model called RSSa. Its suitability is revealed by the comparison against the original NRCS CN model and alternatives SMA procedures in 12 watersheds, located in six different countries, which are associated to different climatic conditions from Mediterranean to Semiarid zones. Based on statistical analysis, this new proposal, RSSa, exhibited the best fitting compared to previous proposed CN-based models. Furthermore, it was assessed the influence of soil moisture parameter for each watershed and the relative weight of scale effects over model parametrization.
Finally, due to the conceptual origin of the model, it is acknowledged that a certain error is implied on its use, being recognized among the likely error sources: the influence of the rainfall depth and its temporal distribution. However, the scarce literature concerning the rainfall intensity in the watersheds used as reference made the rainfall intensity parameter initially neglected. Even though the impact of the distribution and temporal sequences of storm intensity was not previously examined, its influence over runoff generation is known and, thus, the modeled outflow hydrographs derived from this process is well known it. Because of the high level of simplicity of the model, the CN parameter should integrate soil column rainfall-runoff response and a full list of heterogeneous and spatial distributed processes that may affect watershed response (i.e., slope, micro topography, overland flow filtration, etc...). To overcome that problem, the present PhD Thesis analyzes the performance of the NRCS CN model by evaluating the factors with larger influence on the model parameters, beyond the soil texture, surface conditions and vegetation, which are the ones usually considered. Then, in order to explicitly evaluate the impact of the initial soil moisture content in the NRCS CN model, a new version of the most recognizable methodology that integrates a Soil Moisture Account (SMA) procedure, named as SSab model was developed,. This new approach, which keeps its simplicity and conceptual hypothesis, is governed by three parameters: potential maximum retention S, a redefined runoff threshold Sa and a new parameter to ponder soil characteristics called b. Its accuracy was tested in different watersheds and in several synthetic rainfall-runoff events generated after solving the variably saturated flow equations for different combinations of soil textures, initial soil moisture content and climatic scenarios.
Subsequently, a factor analysis based on a Taguchi's orthogonal array and on the previous synthetic rainfall runoff dataset demonstrates the importance of differentiating the influence of variables over each parameter, and considering the current initial soil moisture conditions in the catchment. Furthermore, using that synthetic events, where no spatial variability is examined, the parameter Sa emerged as the most important parameter, showing its significant dependence with rainfall properties and soil texture, while the maximum storage parameter S and the new parameter b may be considered dependent on soil texture. Accordingly, the lack of generality of the original NRCS CN table is acknowledged, in spite of being widely used. Factors like the rainfall total depth and its temporal distribution should be considered when selecting model parameters, and the initial soil moisture content must be explicitly accounted in the evaluation of rainfall-runoff process. The proposed SSab model allows overcomes this issue and ease the integration of a NRCS CN-based model as a continuous soil moisture balance model.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados