T. J. Peterson

Seasonal and event dynamics of spatial soil moisture patterns at the small catchment scale

[1] Our understanding of short- and long-term dynamics of spatial soil moisture patterns is limited due to measurement constraints. Using new highly detailed data, this research aims to examine seasonal and event-scale spatial soil moisture dynamics in the topsoil and subsoil of the small spruce-covered Wustebach catchment, Germany. To accomplish this, univariate and geo-statistical analyses were performed for a 1 year long 4-D data set obtained with the wireless sensor network SoilNet. We found large variations in spatial soil moisture patterns in the topsoil, mostly related to meteorological forcing. In the subsoil, temporal dynamics were diminished due to soil water redistribution processes and root water uptake. Topsoil range generally increased with decreasing soil moisture. The relationship between the spatial standard deviation of the topsoil soil moisture (SD� ) and mean water content (� ) showed a convex shape, as has often been found in humid temperate climate conditions. Observed scatter in topsoil SD� (� ) was explained by seasonal and event-scale SD� (� ) dynamics, possibly involving hysteresis at both time scales. Clockwise hysteretic SD� (� ) dynamics at the event scale were generated under moderate soil moisture conditions only for intense precipitation that rapidly wetted the topsoil and increased soil moisture variability controlled by spruce throughfall patterns. This hysteretic effect increased with increasing precipitation, reduced root water uptake, and high groundwater

Simulating runoff under changing climatic conditions: Revisiting an apparent deficiency of conceptual rainfall-runoff models

Hydrologic models have potential to be useful tools in planning for future climate variability. However, recent literature suggests that the current generation of conceptual rainfall runoff models tend to underestimate the sensitivity of runoff to a given change in rainfall, leading to poor performance when evaluated over multiyear droughts. This research revisited this conclusion, investigating whether the observed poor performance could be due to insufficient model calibration and evaluation techniques. We applied an approach based on Pareto optimality to explore trade-offs between model performance in different climatic conditions. Five conceptual rainfall runoff model structures were tested in 86 catchments in Australia, for a total of 430 Pareto analyses. The Pareto results were then compared with results from a commonly used model calibration and evaluation method, the Differential Split Sample Test. We found that the latter often missed potentially promising parameter sets within a given model structure, giving a false negative impression of the capabilities of the model. This suggests that models may be more capable under changing climatic conditions than previously thought. Of the 282[347] cases of apparent model failure under the split sample test using the lower [higher] of two model performance criteria trialed, 155[120] were false negatives. We discuss potential causes of remaining model failures, including the role of data errors. Although the Pareto approach proved useful, our aim was not to suggest an alternative calibration strategy, but to critically assess existing methods of model calibration and evaluation. We recommend caution when interpreting split sample results.

Nonlinear time‐series modeling of unconfined groundwater head

This paper presents a nonlinear transfer function noise model for time-series modeling of unconfined groundwater hydrographs. The motivation for its development was that existing groundwater time-series models were unable to simulate large recharge events and multiyear droughts. This was because existing methods do not partition rainfall to runoff and do not account for nonlinear soil water drainage. To account for these nonlinear processes, a vertically integrated soil moisture module was added to an existing transfer function noise model. The soil moisture module has a highly flexible structure that allowed 84 different forms to be built. Application of the time-series model requires numerical calibration of parameters for the transfer functions, noise model and, for the nonlinear models, the soil moisture module. This was undertaken using the Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES) global calibration scheme. However, reproducible calibration to the global optima was challenging and a number of modifications were required to the transfer function noise model. In trialing the 84 nonlinear models and 2 linear models, each was applied to eleven observation bores within a paired catchment study area in Great Western, Victoria, Australia. In comparison with existing groundwater hydrograph time-series models, the proposed nonlinear time-series model performed significantly better at all observation bores during calibration and evaluation periods. Both the linear and nonlinear models were also used to quantify the impact of revegetation within the paired catchment; however, results were inconclusive, which is likely due to time-series data for the state of the revegetation being unavailable. By analyzing the application of 84 nonlinear models to each bore, an optimal structure for the soil moisture module was identified. It is unlikely, however, that this model structure would be appropriate for all climates and geologies. To encourage further investigations, open-source code for the highly flexible groundwater time-series modeling framework is available and we invite others to develop new models.

Top-down groundwater hydrograph time-series modeling for climate-pumping decomposition

Groundwater time-series modeling has emerged as an efficient approach for simulating the impacts of multiple drivers of groundwater-head variation such as rainfall, evaporation and groundwater pumping. However, a bottom-up approach has generally been adopted whereby the input drivers have been assumed without statistical evidence for their inclusion. In this study, a parsimonious time-series model was adopted which accounts for various drivers and is able to simulate the overall groundwater-head variation. It can also separate the effects of pumping and climate drivers on multi-annual time series of groundwater-level variation. The time-series model consists of a soil-moisture layer to account for non-linearity between rainfall and recharge, as well as different pumping response functions to account for pumping from a single well, lake-induced recharge and the effects of multiple pumping bores. The method was applied to a groundwater-pumping region in south-eastern Australia. The results showed that the model is able to separate the effects of pumping from the effects of climate on groundwater-head variation. However, improved estimation of those influences requires a flexible model structure that can account for spatially varying physical processes within the study region such as the relative influence of single or multiple pumping bores and induced recharge from surface-water bodies.RésuméLa modélisation des séries temporelles de piézométrie a émergé en tant qu’approche efficace pour simuler les impacts de multiples causes de variation de la piézométrie des eaux souterraines, telles que la pluie, l’évaporation et les pompages d’eau souterraine. Cependant, une approche ascendante a généralement été adoptée en vertu de laquelle les causes d’entrée ont été supposées sans preuve statistique pour leur implication. Dans cette étude, un modèle parcimonieux de séries chronologiques a été adopté prenant en considération différentes causes et étant capable de simuler la variation piézométrique dans sa globalité. Il peut aussi séparer les effets des pompages des causes climatiques pour des séries chronologiques pluriannuelles de variation du niveau piézométrique. Le modèle de séries chronologiques consiste en une couche sol-humidité afin d’intégrer la non linéarité entre la pluie et la recharge, ainsi que différentes fonctions de réponse aux pompages pour prendre en compte le pompage dans un puits unique, la recharge induite par un lac et les effets de multiples forages d’exploitation. La méthode a été appliquée à une région d’exploitation des eaux souterraines du Sud-Est de l’Australie. Les résultats montrent que le modèle est capable de séparer les effets des pompages des effets du climat sur les variations du niveau piézométrique. Cependant, l’amélioration de l’estimation de ces influences requiert une structure de modèle flexible qui peut prendre en considération des processus physiques variant spatialement dans la région d’étude, telles que l’influence relative d’un unique ou de multiples forages d’exploitation et la recharge induite par des cours d’eau ou plans d’eau de surface.ResumenEl modelado de series de tiempo de agua subterránea se ha convertido en un enfoque eficiente para la simulación de los impactos de las múltiples causas de la variación de la carga hidráulica del agua subterránea, tales como precipitación, evaporación y bombeo de agua subterránea. Sin embargo, por lo general se ha adoptado un enfoque “de abajo a arriba” por el cual las causas de ingreso se han asumido sin evidencias estadísticas para su inclusión. En este estudio, se adoptó un modelado de series de tiempo parsimoniosas que representa a varias causas y es capaz de simular la variación global de la carga hidráulica del agua subterránea. También puede separar los efectos del bombeo y las causas climáticas en series de tiempo multianuales de variación de los niveles de agua subterránea. El modelo de series de tiempo consiste de una capa de humedad del suelo para tener cuenta la no linealidad entre la precipitación y la recarga, así como diferentes funciones de respuesta al bombeo para dar cuenta del bombeo desde un pozo único, recarga inducida por un lago y los efectos de múltiples pozos de bombeo. Se aplicó el método a una región de bombeo de agua subterránea en el sudeste de Australia. Los resultados mostraron que el modelo es capaz de separar los efectos del bombeo de los efectos de las variaciones climáticas sobre la variación de la carga hidráulica del agua subterránea. Sin embargo, una mejor estimación de esas influencias requiere una estructura de modelo flexible que puede dar cuenta de procesos físicos espacialmente variables dentro de la región de estudio, tal como la influencia relativa de simples o múltiples pozos de bombeo y recarga inducida desde cuerpos de agua superficial.摘要作为模拟地下水头变化的多重驱动因素影响一个有效的方法,地下水时序模拟应运而生,这些因素包括降雨、蒸发及地下水的抽取。然而,一种自下而上的方法普遍被采用,在这种方法中,假定输入驱动因素无其入选统计论据。在本研究中,采用了一种特别简单的时序模型,这个模型解释了各种驱动因素,能够模拟全部的地下水头变化。模型还可以区分抽水因素和气候因素对地下水位变化多个年度时序的影响。时序模型由一个解释降雨和补给之间的非线性误差的土壤水分层以及解释单井中抽水、湖泊诱发的补给及多个抽水井影响的不同抽水响应函数组成。此方法应用于澳大利亚东南部的地下水抽水区。结果显示,模型能够区分抽水和气候对地下水头变化的影响。然而,提高这些影响的估测水平需要一个切实可行的模型结构,这个模型结构要能够解释研究区内空间上变化的物理过程,诸如单个或多个钻井的相对影响及地表水体诱发的补给。ResumoA modelação de séries temporais de níveis piezométricos surgiu como uma abordagem eficiente para simular os impactes de múltiplos controladores da variação do potencial hidráulico, tais como a precipitação, a evaporação e o bombeamento da água subterrânea. No entanto, tem sido geralmente adotada uma abordagem ascendente, pela qual os controladores de entrada são assumidos sem evidência estatística da sua inclusão. Neste estudo, foi adotado um modelo parcimonioso de séries temporais que responde por vários controladores e é capaz de simular a variação global do potencial hidráulico. Também pode separar os efeitos de controladores de bombeamento e de clima em séries temporais multianuais de variação do nível piezométrico. O modelo de séries temporais é composto por uma camada de humidade do solo para explicar a não linearidade entre a precipitação e a recarga, assim como diferentes funções de resposta ao bombeamento num único furo, recarga induzida a partir de um lago e o efeito de vários furos de bombeamento. O método foi aplicado numa região de extração de água subterrânea no sudeste da Austrália. Os resultados mostraram que o modelo é capaz de separar os efeitos do bombeamento dos efeitos do clima na variação do potencial hidráulico. No entanto, a melhoria da estimativa dessas influências requer uma estrutura de modelo flexível que possa responder pela variação espacial dos processos físicos dentro da região de estudo, tais como a influência relativa de furos de bombeamento singulares ou múltiplos e a recarga induzida a partir de corpos de água superficiais.

Decomposing groundwater head variations into meteorological and pumping components: a synthetic study

Time-series modeling is often used to decompose groundwater hydrographs into individual drivers such as pumping and meteorological factors. To date, there has been an assumption that a simulation fitting the total hydrograph produces reliable estimates of the impact from each driver. That is, assessment of the decomposition has not used an independent estimate of each decomposition result. To begin to address this, a synthetic study is undertaken so that the impact of each driver is known. In this study, 500 MODFLOW groundwater models of a one-layer unconfined aquifer were constructed. For each model, three hydrogeological properties (saturated hydraulic conductivity, storativity and depth to aquifer basement), the distance between observation and pumping bores, and extraction rate were set randomly and synthetic groundwater hydrographs were derived. For each hydrograph, the influence of individual drivers was estimated using six different time-series models. These estimates were then compared to the known meteorological and pumping influences derived from the MODFLOW models. The results demonstrate that hydrograph separations obtained from time-series models do not always result in reliable estimation of pumping and meteorological influences even when the overall hydrograph fit is good. However, when the time-series model represents the important processes (e.g. phreatic evaporation is included for shallow water tables) and the (head) variance of the pumping signal to the meteorological signal is between 0.1 and 10, the time-series model has the potential to adequately separate the influence of pumping and climate.RésuméLa modélisation des séries chronologiques est souvent utilisée pour décomposer les hydrogrammes des eaux souterraines en leurs déterminants pris individuellement, par exemple le pompage et les facteurs météorologiques. Jusqu’à ce jour, il y avait une hypothèse selon laquelle une simulation qui concorde avec l’hydrogramme dans sa totalité fournit une estimation fiable de l’impact imputable à chaque déterminant. Autrement dit, l’évaluation de la décomposition n’utilisait pas une estimation indépendante de chaque résultat de décomposition. Pour avancer dans la résolution de ce problème, une étude de synthèse destinée à connaître l’impact de chaque déterminant a été entreprise. Dans cette étude, 500 modèles MODFLOW d’aquifère libre mono-couche ont été établis. Pour chaque modèle, trois propriétés hydrogéologiques (conductivité hydraulique de la zone saturée, coefficient d’emmagasinement et profondeur du mur de l’aquifère), la distance entre piézomètre et puits de pompage et le débit de pompage ont été fixés de manière aléatoire et des hydrogrammes synthétiques des eaux souterraines ont été déduits. Pour chaque hydrogramme, l’influence de chaque facteur a été estimée d’après la modélisation de six chroniques différentes. Ces évaluations ont été ensuite comparées aux influences connues de la météorologie et du pompage telles que déduites des modèles MODFLOW. Les résultats montrent que les séparations d’hydrogramme obtenues par la modélisation des séries temporelles ne se traduisent pas toujours par une estimation fiable des influences du pompage et de la météorologie, même quand la correspondance avec l’hydrogramme global est bonne. Cependant, quand le modèle des séries chronologiques représente les processus importants (par exemple l’évaporation phréatique est comptabilisée pour une surface de nappe libre peu profonde) et que la variance du signal de pompage par rapport au signal météorologique est comprise entre 0.1 et 10, le modèle de séries chronologiques est capable de séparer correctement l’influence du pompage de celle du climat.ResumenEl modelado de series de tiempo se usa a menudo para descomponer hidrogramas de agua subterránea en componentes, tales como el bombeo y los factores meteorológicos. Hasta la fecha, ha existido el supuesto que una simulación adecuada del hidrograma total produce estimaciones fiables de los efectos de cada componente. Es decir, la evaluación de la descomposición no ha utilizado una estimación independiente de cada resultado de la descomposición. Para comenzar a abordar esto, se llevó a cabo un estudio sintético de modo de conocer el impacto de cada componente. En este estudio, se construyeron 500 modelos MODFLOW de aguas subterráneas de un acuífero no confinado de una sola capa. Para cada modelo, se fijaron al azar las propiedades hidrogeológicas (conductividad hidráulica saturada, almacenamiento y profundidad al basamento acuífero), la distancia entre pozos de observación y de bombeo y la tasa de extracción y a partir de ello fueron derivados los hidrogramas sintéticos de agua subterránea. Para cada hidrograma, se estimó la influencia de los componentes individuales usando seis diferentes modelos de series de tiempo. Estas estimaciones se compararon con las influencias meteorológicas y de bombeos conocidas, derivadas a partir de los modelos MODFLOW. Los resultados demuestran que las separaciones de hidrogramas obtenidas a partir de los modelos de series de tiempo no siempre resultan en estimaciones seguras de las influencias meteorológicas y del bombeo aún cuando el ajuste general del hidrograma es bueno. Sin embargo, cuando el modelo de series de tiempo representa los procesos importantes (por ejemplo, la evaporación desde la freática es incluida para niveles freáticos someros) y la (carga hidráulica) la varianza entre la señal de bombeo y la señal meteorológica es entre 0.1 y 10, el modelo de series de tiempo tiene el potencial para separar adecuadamente la influencia de bombeo y el clima.摘要时间序列模拟常常用于分解地下水水位曲线到单个的驱动因素中,如抽水和气象因素。迄今为止,有一个假定就是,拟合整个水文曲线的模拟从每个驱动因素中可得出可靠的影响估算结果。这就是说,分解评价没有使用每个分解结果的独立估算值。为了首先强调这点,进行了综合研究,以便获知每个驱动因素的影响。在本项研究中,建立了一个单层非承压含水层500个MODFLOW地下水模型。每个模型,随机设定了三个水文地质特性(饱和水力传导系数、储存系数和含水层底部的深度)、观测井和抽水井的距离及抽水速度,得到了综合地下水水位曲线图。针对每个水位曲线图,利用六个不同的时间序列模型估算了每个驱动因素的影响。然后,把这些估算值与由MODFLOW模型得到的已知气象和抽水影响进行了对比。结果显示,即使是整体水位曲线图拟合非常好,时间序列模型得到的水位曲线图也并不总能得出抽水和气象影响的可靠估算结果。然而,当时间序列模型展示重要过程(例如,浅层水位中包括潜水蒸发)时及抽水信号对气象信号的(水头)变化在0.1和10之间时,时间序列模型具有充分分离抽水和气候影响的潜力。ResumoModelagem de séries temporais é comumente usada para decompor hidrogramas de água subterrânea em componentes forçantes individuais, como bombeamento e fatores meteorológicos. Até o momento, tem existido uma hipótese de que uma simulação que ajusta o hidrograma total produz uma estimativa confiável do impacto de cada componente. Isto é, uma avaliação de decomposição não utiliza uma estimativa independente de cada resultado da decomposição. Para começar a lidar com o problema, um estudo sintético foi feito de forma que o impacto de cada componente seja conhecido. Neste estudo, foram construídos 500 modelos de água subterrânea MODFLOW de um aquífero não confinado de uma camada. Para cada modelo, três propriedades hidrológicas (condutividade hidráulica saturada, coeficiente de armazenamento e profundidade da base do aquífero), a distância entre os poços de observação e de bombeamento e a taxa de extração foram definidas de forma aleatória, tendo seus hidrogramas de água subterrânea derivados. Para cada hidrograma, a influência das componentes forçantes individuais foi estimada usando seis modelos de séries temporais distintos. Estas estimativas foram então comparadas com influências meteorológicas e de bombeamento conhecidas, derivadas dos modelos MODFLOW. Os resultados demonstram que a separação dos hidrogramas obtidos através dos modelos de séries temporais nem sempre resultam em estimativas confiáveis da influência de bombeamento e de condições meteorológicas, mesmo quando o ajuste do hidrograma é bom. Entretanto, quando um modelo de séries temporais representa o processo importante (p. ex. evaporação freática é incluída em aquíferos rasos) e a variância (de carga) entre o sinal de bombeamento com o sinal meteorológico está entre 0.1 e 10, o modelo de séries temporais tem o potencial de separar adequadamente a influencia de bombeamento e clima.

A synthetic study to evaluate the utility of hydrological signatures for calibrating a base flow separation filter

Estimation of baseflow from streamflow hydrographs has been a major challenge in hydrology for decades, leading to developments of baseflow separation filters. When without tracer or groundwater data to calibrate the filters, the standard approach to apply these filters in practice involves some degrees of subjectivity in choosing the filter parameters. This paper investigates the use of signature-based calibration in implementing baseflow filtering by testing seven possible hydrological signatures of baseflow against modelled daily baseflow produced by Li et al. [2014] for a range of synthetic catchments simulated with HydroGeoSphere. Our evaluation demonstrates that such a calibration method with few selected signatures as objectives is capable of calibrating a filter - Eckhardt filter - to yield satisfactory baseflow estimates at daily, monthly and long-term time scales, outperforming the standard approach. The best performing signatures can be readily derived from streamflow timeseries. While their performance depends on the catchment characteristics, the catchments where the signature method performs can be distinguished using commonly-used descriptors of flow dynamics. This article is protected by copyright. All rights reserved.

Can we manage groundwater? A method to determine the quantitative testability of groundwater management plans

Groundwater is the worlds largest freshwater resource and due to overextraction, levels have declined in many regions causing extensive social and environmental impacts. Groundwater management seeks to balance and mitigate the detrimental impacts of development, with plans commonly used to outline management pathways. Thus, plan efficiency is crucial, but seldom are plans systematically and quantitatively assessed for effectiveness. This study frames groundwater management as a system control problem in order to develop a novel testability assessment rubric to determine if plans meet the requirements of a control loop, and subsequently, whether they can be quantitatively tested. Seven components of a management plan equivalent to basic components of a control loop were determined, and requirements of each component necessary to enable testability were defined. Each component was weighted based upon proposed relative importance, then segmented into rated categories depending on the degree the requirements were met. Component importance varied but, a defined objective or acceptable impact was necessary for plans to be testable. The rubric was developed within the context of the Australian groundwater management industry, and while use of the rubric is not limited to Australia, it was applied to 15 Australian groundwater management plans and approximately 47% were found to be testable. Considering the importance of effective groundwater management, and the central role of plans, our lack of ability to test many plans is concerning.

The good, the bad and the outliers: automated detection of errors and outliers from groundwater hydrographs

Suspicious groundwater-level observations are common and can arise for many reasons ranging from an unforeseen biophysical process to bore failure and data management errors. Unforeseen observations may provide valuable insights that challenge existing expectations and can be deemed outliers, while monitoring and data handling failures can be deemed errors, and, if ignored, may compromise trend analysis and groundwater model calibration. Ideally, outliers and errors should be identified but to date this has been a subjective process that is not reproducible and is inefficient. This paper presents an approach to objectively and efficiently identify multiple types of errors and outliers. The approach requires only the observed groundwater hydrograph, requires no particular consideration of the hydrogeology, the drivers (e.g. pumping) or the monitoring frequency, and is freely available in the HydroSight toolbox. Herein, the algorithms and time-series model are detailed and applied to four observation bores with varying dynamics. The detection of outliers was most reliable when the observation data were acquired quarterly or more frequently. Outlier detection where the groundwater-level variance is nonstationary or the absolute trend increases rapidly was more challenging, with the former likely to result in an under-estimation of the number of outliers and the latter an overestimation in the number of outliers.RésuméLes observations suspectes du niveau des eaux souterraines sont fréquentes et peuvent survenir pour de nombreuses raisons, allant d’un processus biophysique imprévu aux défauts de forage et aux erreurs de gestion de données. Les observations imprévues peuvent fournir de précieuses informations qui remettent en questions les prévisions existantes et peuvent être considérées comme des valeurs aberrantes, tandis que les défauts de suivi et de traitement des données peuvent être considérés comme des erreurs, et, si ignorées, peuvent compromettre l’analyse des tendances et la calibration des modèles hydrogéologiques. Idéalement, des valeurs aberrantes et des erreurs doivent être identifiées, mais à ce jour il s’agit d’un processus subjectif qui n’est. pas reproductible et qui est. inefficace. Cet article présente une approche permettant d’identifier de manière objective et efficaces de multiples types d’erreurs et de valeurs aberrantes. L’approche ne nécessite que l’hydrogramme des niveaux d’eaux souterraines observés, ne requiert aucune attention particulière concernant l’hydrogéologie, des paramètres d’influence (par exemple les pompages) ou la fréquence du suivi, et est. disponible gratuitement dans la boîte à outils HydroSight. Dans ce cas, les algorithmes et les modèles de séries chronologiques sont détaillés et appliqués à quatre piézomètres possédant des dynamiques variées. La détection des valeurs aberrantes était la plus fiable lorsque les données d’observation étaient acquises trimestriellement ou plus fréquemment. La détection des valeurs aberrantes où la variance du niveau d’eaux souterraines est. non stationnaire ou la tendance absolue augmente rapidement était plus difficile, la première pouvant entraîner une sous-estimation du nombre de valeurs aberrantes et la dernière une surestimation du nombre de valeurs aberrantes.ResumenLas observaciones sospechosas del nivel de agua subterránea son comunes y pueden surgir por muchas razones que van desde un proceso biofísico imprevisto hasta errores por fallas en la perforación o en el manejo de los datos. Las observaciones imprevistas pueden aportar valiosas ideas que desafían las expectativas existentes y pueden considerarse valores atípicos, mientras que las fallas en el monitoreo y en el manejo de datos pueden considerarse errores y, si se ignoran, pueden comprometer el análisis de tendencias y la calibración del modelo de agua subterránea. Idealmente, se deben identificar los valores atípicos y los errores, pero hasta la fecha esto ha sido un proceso subjetivo que no es reproducible y es ineficiente. Este artículo presenta un enfoque para identificar objetiva y eficientemente múltiples tipos de errores y valores atípicos. El enfoque sólo requiere el hidrograma de agua subterránea observado, no requiere consideración especial de la hidrogeología, de los impulsos (por ejemplo, el bombeo) o de la frecuencia de monitoreo, y está libremente disponible en la caja de herramientas de HydroSight. Aquí, los algoritmos y el modelo de serie temporal se detallan y se aplican a cuatro pozos de observación con variables dinámicas. La detección de valores atípicos fue más confiable cuando los datos de observación se adquirieron trimestralmente o con mayor frecuencia. La detección de valores atípicos en que la varianza del nivel del agua subterránea no es estacionaria o la tendencia absoluta aumenta rápidamente era más difícil, ya que la primera probablemente daría lugar a una subestimación del número de valores atípicos y la última a una sobreestimación del número de valores atípicos.摘要可疑的地下水位观测结果很常见,有多种原因可造成意料之外的生物物理过程、钻孔故障及资料管理误差。意料之外的观测结果可提供宝贵的启示,这些启示挑战已有的期望值,可被认为是异常值,而监测和数据处理故障可被认为是误差,如果忽略不计,这些异常值和误差可危害趋势分析和地下水模型校正。理想的是,异常值和误差应当辨别出来,但到目前为止,这一直是个凭经验的过程,这个过程是不可复制的,也是低效的。本文介绍了一种客观、有效地辨别多种类型的误差和异常值的方法。该方法只需要观测的地下水位图,不需要特别考虑水文地质条件、驱动因素(例如抽水)或者监测频率,在HydroSight工具箱免费获得。在此,详述了算法和时间序列模型,并应用到四个具有不同动力学的观测孔中。当每个季度或者更频繁地需要观测数据时,异常值的检测最可靠。地下水位变化非稳定或者绝对趋势快速增加的地方,异常值检测更具挑战,前者可能导致异常值数量的低估,后者可能导致异常值数量的高估。ResumoObservações suspeitas sobre o nível das águas subterrâneas são comuns e podem surgir por diversas razões que vão desde um processo biofísico imprevisto até a falha no furo e erros de gerenciamento de dados. As observações imprevistas podem fornecer informações valiosas que desafiam as expectativas existentes e podem ser considerados dados discrepantes, enquanto o monitoramento e as falhas no tratamento de dados podem ser considerados erros e, se ignorados, podem comprometer a análise de tendências e a calibração de modelo de águas subterrâneas. Idealmente, dados discrepantes e erros devem ser identificados, mas até agora este tem sido um processo subjetivo que não é reprodutível e é ineficiente. Este artigo apresenta uma abordagem objetiva e eficiente para identificar múltiplos tipos de erros e dados discrepantes. A abordagem requer apenas o hidrograma de águas subterrâneas observado, não requer nenhuma consideração particular da hidrogeologia, as forçantes (p.ex. bombeamento) ou a frequência de monitoramento, e é disponível gratuitamente na caixa de ferramentas HydroSight. Aqui, os algoritmos e modelos de séries temporais são detalhados e aplicados em quatro furos de observação com diferentes dinâmicas. A detecção de dados discrepantes foi mais confiável quando os dados de observação foram adquiridos trimestralmente ou mais frequentemente. A detecção de dados discrepantes em que a variância do nível da água subterrânea não é estacionária ou a tendência absoluta aumenta rapidamente foi mais desafiadora, com a primeira provavelmente resultando em uma subestimação do número de dados discrepantes e a última uma sobre-estimava no número de dados discrepantes.

A One Stage Damage Detection Technique Using Spectral Density Analysis and Parallel Genetic Algorithms

This paper describes a new global damage identification framework for the continuous/periodic monitoring of civil structures. In order to localize and estimate the severity of damage regions, a one-stage model-based Bayesian probabilistic damage detection approach is proposed. This method, which is based on the response power spectral density of the structure, enjoys the advantage of broadband frequency information and can be implemented on input-output as well as output-only damage identification studies. A parallel genetic algorithm is subsequently used to evolve the optimal model parameters introduced for different damage conditions. Given the complex search space and the need to perform multiple time-consuming objective function evaluations, a parallel meta-heuristic provides a robust optimization tool in this domain. It is shown that this approach is capable of detecting structural damage in both noisy and noise-free environments.

Error propagation in computer models: analytic approaches, advantages, disadvantages and constraints

Uncertainty and its propagation in computer models has relevance in many disciplines, including hydrology, environmental engineering, ecology and climate change. Error propagation in a model results in uncertainty in prediction due to uncertainties in model inputs and parameters. Common methods for quantifying error propagation are reviewed, namely Differential Error Analysis and Monte Carlo Simulation, including underlying principles, together with a discussion on their differences, advantages and disadvantages. The separate case of uncertainty in the model calibration process is different to error propagation in a fixed model in that it is associated with a dynamic process of iterative parameter adjustment, and is compared in the context of non-linear regression and Bayesian approaches, such as Markov Chain Monte Carlo Simulation. Error propagation is investigated for a soil model representing the organic carbon depth profile and also a streamflow model using probabilistic simulation. Different sources of error are compared, including uncertainty in inputs, parameters and geometry. The results provided insights into error propagation and its computation in systems and models in general.