Michael James Friedel

University of Colorado - University of Hawaii

The availability of freshwater and low-cost electricity are limiting factors for sustainable living in Hawaii. This raises the question: Can technology be developed to locate and characterize freshwater and geothermal resources simultaneously? We present a multimodal machine learning (MML) workflow to characterize the 3D distribution of features (p...

Knowing the location where groundwater denitrification occurs, or by proxy the groundwater redox status (oxic, mixed, and anoxic), is valuable information for assessing and managing potential agricultural land-use impacts on freshwater quality. We compare the efficacy of supervised (Linear Discriminant Analysis LDA; Boosted Regression Trees, BRT; a...

Unravelling sub-catchment scale nitrogen delivery to waterways  In large and heterogeneous catchments (100s-1000s of km 2), it is very difficult to link a nitrogen flux observed at the catchment outlet to the many past and present activities that collectively are responsible for it.  We therefore aim to establish cause-effect relationships at the...

To be effective and efficient, decision making on land use, land management, mitigation measures, as well as policy, need to be based on a clear understanding of cause-effect relationships. Present practice is to link activities on the land and water quality outcomes at spatial scales of 100s to 1000s of km 2. However, such large catchments are ine...

The aim of this work is to understand the exchange of water between the Serra Geral aquifer system (SGAS) and Guarani aquifer system (GAS). The objectives are twofold. First, introduce the capability of the modified self-organizing maps (MSOM) as an unbiased nonlinear approach to estimate missing values of hydrochemistry and hydraulic transmissivit...

The Australian machine-learning workflows apply fusion, clustering, and estimation operations to hydrogeophysical data for deriving hydrostratigraphic units (HSUs). Data fusion is performed by training a self-organizing map (SOM) with these data. The application of Davies-Bouldin criteria to K-means clustering of SOM nodes determines the number and...

An unsupervised machine-learning workflow is proposed for estimating fractional landscape soils and vegetation components from remotely-sensed hyperspectral imagery. The workflow is applied to EO-1 Hyperion satellite imagery collected near Ibirací, Minas Gerais, Brazil. The proposed workflow includes subset feature selection, learning and estimatio...

The aim of this work is twofold: to present the capability of the Self-organizing maps (SOM) as an alternative approach to imputate missing values of hydrochemistry and hydraulic transmissivity for the Serra Geral aquifer (SGAS), a transboundary fractured system covering four countries across South America; and to identify areas with potential conn...

: Development of successful surface-groundwater management strategies depend on quality of data provided for analysis. This study evaluates statistical robustness when using a modified self-organizing map (MSOM) technique to estimate missing values for various hypersurface models: (1) synoptic groundwater-surface water hydrochemistry; (2) time-seri...

Aim Numerical models provide a way to evaluate groundwater systems, but determining the hydrostratigraphy used in their construction is often challenged by spatially-limited field data. The aim of this study is to develop a machine-learning based approach to assist in the groundwater model calibration process. Method We propose a two-step approach...

Joint inversions facilitate the integration of different data sets for improved subsurface interpretations. We present a framework for the implicit joint inversion of data sets using available numerical models and model-independent parameter estimation (PEST) software package. Matlab scripts are used to prepare PEST files and build appropriate join...

The objectives of this research were to determine the role of preferential flow paths in the transport of atrazine (2-chloro-4-(ethylamino)-6-(isopropylamino)-s-triazine) and dicamba (3-6-dichloro-2-methoxybenzoic acid) through silt and loam soils overlying the High Plains aquifer in Nebraska. In a previous study, 3 of 6 study areas demonstrated hi...

The Geoscience Australia commissioned GNS Science to define a collaborative project in which machine learning (ML) is used for characterization and mapping of hydrostratigraphic units (HSUs). This report proposes a ML-based approach to reduce processing time and enhance reliability in modeling of the hydrostratigraphy at the Broken Hill Managed Aqu...

Numerical models provide a way to evaluate groundwater systems, but determining the hydrostratigraphic units (HSUs) used in constructing these models remains subjective, nonunique, and uncertain. A three-step machine-learning approach is proposed in which fusion, estimation, and clustering operations are performed on different data sets to arrive a...

The objective of this research was to determine the process of atrazine transport compared to bromide and δO18 transport in sands near Denver. Three 1.5 × 2 × 1.5-m plots were installed and allowed to equilibrate for 2 years before research initiation and were instrumented with 1.5 × 2-m zero-tension pan lysimeters installed at 1.5-m depths. Additi...

Save Money and Reduce Time (SMART) in characterization and mapping of aquifers

Aquatic ecosystem models can potentially be used to predict the influence of stresses on catchment resource quality. Given that catchment responses are functions of natural and anthropogenic stresses reflected in sparse and spatiotemporal biological, physical, and chemical measurements, an ecosystem is difficult to model using statistical or numeri...

Artificial adaptive systems are evaluated for their usefulness in modeling earthquake hydrology of the Canterbury region, NZ. For example, an unsupervised machine-learning technique, self-organizing map, is used to fuse about 200 disparate and sparse data variables (such as, well pressure response, ground acceleration, intensity, shaking, stress an...

Since the development of personal computers, the modeling of groundwater systems shifted from analytical equations to numerical models. Given the ill-posed nature and non-uniqueness of numerical groundwater models, the use of alternate data fusion and knowledge extraction paradigms is being explored to reduce uncertainty through improvements in the...

A hybrid modeling approach is proposed for near real-time three-dimensional (3D) mapping of surficial aquifers. First, airborne frequency-domain electromagnetic (FDEM) measurements are numerically inverted to obtain subsurface resistivities. Second, a machine-learning (ML) algorithm is trained using the FDEM measurements and inverted resistivity pr...

Identify optimal subset of variables to improve ecosystem models

We adopt a hybrid approach to map the 3D hydrostratigraphy of an alluvial aquifer using big data collected in the Ettrick basin, Otago New Zealand. First, a subset (1%) of the 18 million regional helicopter electromagnetic (HEM) sounding measurements (300 Hz, Horizontal co-planar; 3300 Hz, vertical co-planar; 8200 Hz, horizontal co-planar; 40 kHz,...

Agriculture is related with land-use/cover changes (LUCC) over large areas and, in recent years, increase in demand of ethanol fuel has been influence in expansion of areas occupied with corn and sugar cane, raw material for ethanol production. Nevertheless, there´s a concern regarding the impacts on food security, such as, decrease in areas plante...

systems (AAS) to assist in conceptualizing groundwater systems. First, vector-quantization and visualization techniques are used to explore conceptual model connectivity among ground-water-quality variables in in deltaic, basin-fill, and crystalline bedrock aquifers. Second, an evolutionary input selection is combined with machine-learning to impro...

Understand influence of natural and anthropogenic stresses on catchment resource quality

The joint inversion of receiver function (RF) and surface‐wave dispersion data is popular because it reduces the nonuniqueness of the modeled subsurface‐seismic velocities. Whereas various inverse procedures have been used in joint subsurface imaging, the evaluation of uncertainties in the estimated parameter distribution is usually overlooked or c...

A novel two-step modeling approach is presented to obtain optimal starting values and geostatistical constraints for numerical inverse problems otherwise characterized by spatially-limited field data. First, a type of unsupervised neural network, called the self-organizing map (SOM), is trained to recognize nonlinear relations among environmental v...

The sensor Hyperion was launched in late 2000 and represents an advance to hyperspectral remote sensing. However, due from technical aspects has always presented a large amount of nonlinear noise among the bands. Our goal was apply self-organizing map (SOM) technic on Hyperion data without noise correction for mapping tropical soil characteristics....

Few studies attempt to model the economic feasibility of mining undiscovered mineral resources given the sparseness of data; and the coupled, nonlinear, spatial, and temporal relationships among variables. In this study, a type of unsupervised artificial neural network, called a self-organized map (SOM), is trained using data from 203 porphyry copp...

A hydrogeologic study was conducted to support mineral-resource assessment activities in Mauritania, Africa. Airborne magnetic depth estimates reveal two primary groundwater basins: the porous coastal Continental Terminal Basin (fill deposits); and the interior, fractured interior Taoudeni Basin. In the Continental Terminal Basin, there is uniform...

It is not possible, using numerical methods, to model groundwater flow and transport in the fractured crystalline rock of northeastern Brazil. As an alternative, the usefulness of self-organizing map (SOM), k-means clustering, and Davies-Bouldin techniques to conceptualize the hydrogeology was evaluated. Also estimated was the well yield and ground...

The remediation of land containing munitions and explosives of concern, otherwise known as unexploded ordnance, is an ongoing problem facing the U.S. Department of Defense and similar agencies worldwide that have used or are transferring training ranges or munitions disposal areas to civilian control. The expense associated with cleanup of land pre...

The joint inversion of surface-wave dispersion curves and derivative information from receiver function data reduces the non-uniqueness of modeled sub-surface seismic velocities. The evaluation of uncertainties in the estimated velocity distributions however are usually overlooked or considered qualitatively. In this study, we present and use an op...

Few studies attempt to model the range of possible post-fire hydrologic and geomorphic hazards because of the sparseness of data and the coupled, nonlinear, spatial, and temporal relationships among landscape variables. In this study, a type of unsupervised artificial neural network, called a self-organized map (SOM), is trained using data from 540...

This study demonstrates the novel application of genetic programming to evolve nonlinear post-fire debris-flow volume equations from variables associated with a data-driven conceptual model of the western United States. The search space is constrained using a multi-component objective function that simultaneously minimizes root-mean squared and uni...

We demonstrate the efficacy of an unsupervised artificial neural network, called a self-organizing map (SOM), to facilitate modeling of undiscovered porphyry copper deposits at the global scale. Specifically, the SOM can provide relevant model input for quantifying the amounts of undiscovered metals, and predicting the economic feasibility of minin...

The Self-organizing map (SOM) technique is used to estimate missing hydrogeologic (hydraulic and hydrochemical) properties and evaluate potential connectivity between the Serra Geral and Guarani aquifer system. K-means clustering of SOM neurons is useful for identifying hydrogeologic units (conceptual models) in which the Serra Geral waters are car...

We demonstrate the efficacy of an unsupervised artificial neural network, called a self-organizing map (SOM), to facilitate modeling and classifying targets based on images from the EO1-Hyperion hyperspectral sensor. Preprocessing of these images included their conversion to reflectance values with atmospheric corrections based on a radiative trans...

Climate change is one possible external driver of ecosystem services. In the tropical Pacific, short-term climate change is influenced by oceanic Kelvin waves that induce remote temperatures to rise (El Niño event) or decrease (La Niña event). This teleconnection is not globally uniform; in the United States (U.S.) drought conditions induced by El...

Self-organizing map (SOM) and geographic information system (GIS) models were used to investigate the nonlinear relationships associated with geochemical weathering processes at local (~100 km 2) and regional (~50.000 km 2) scales. The dataset consisted of 304 samples, 19 B-horizon soil variables (P, C, pH, Al, total acidity, Ca, Mg, K, total catio...

We have previously shown that climate variables are self-similar in nature, and thus represent long-memory processes. Therefore, these processes need to account for fractional differences in the time series. This requires that we make use of fractal analysis in time series and use fractional differencing. To begin, it is possible to make short term...

Every year thousands of people are affected by floods and landslide hazards caused by rainstorms. The problem is more serious in tropical developing countries because of the susceptibility as a result of the high amount of available energy to form storms, and the high vulnerability due to poor economic and social conditions. Predictive models of ha...

Fractal modeling of reconstructed climate variables revealed that the models coincided with historical anecdotes of global weather patterns. By using the fractal information dimension, it was possible to identify respective cold and warm periods as negative and positive system shocks. For example, the medieval warm period was revealed by a positive...

Studies of climate variability often rely on surface temperature change anomalies. Here regional Pacific sea and land surface temperature data were extended from a century to millennial scale using a type of unsupervised artificial neural network. In this approach, the imputation of annual climate fields was done based on the nonlinear and self-org...

Species distribution models (SDM) use presence field data and environmental variables to locate suitable regions for species occurrence. The majority of presence data, especially in tropical countries, are from herbariums and museum collections, which usually contain geographical positioning errors. Despite the use of global positioning systems (GP...

The Development of Mineral Environmental Assessment Methodologies Project (MEAP) is a new project within the U.S. Geological Survey’s Mineral Resources Program. One objective of the MEAP is to predict potential environmental impacts at a watershed scale, with associated uncertainty, from future mining activities. One initial task is to determine th...

In this paper, we present a joint inverse algorithm for improved subsurface imaging using helicopter frequency-domain electromagnetic (HEM) data. In this algorithm, we introduce a regularization strategy that incorporates soft prior information in the form of difference equations between and across profiles to constrain the simultaneous solution of...

A regularized joint inverse procedure is presented and used to estimate the magnitude of extreme rainfall events in ungauged coastal river basins of El Salvador: Paz, Jiboa, Grande de San Miguel, and Goascoran. Since streamflow measurements reflect temporal and spatial rainfall information, peak-flow discharge is hypothesized to represent a similar...

A regionalization procedure is presented and used to predict probable flooding in four ungauged coastal river basins of El Salvador: Paz, Jiboa, Grande de San Miguel, and Goascoran. The flood-prediction problem is sequentially solved for two regions: upstream mountains and downstream alluvial plains. In the upstream mountains, a set of rainfall-run...

Hydrogeologic maps were constructed for the Islamic Republic of Mauritania. The ground-water flow system in the country can best be described as two interconnected regional systems: the porous Continental Terminal coastal system and the interior, fractured sedimentary Taoudeni Basin system. In these systems, ground-water flow occurs in fill deposit...

The Romanian government is interested in understanding the effects of existing and future mining activities on long-term dispersal, storage, and remobilization of sediment-associated metals. An initial Soil and Water Assessment Tool (SWAT) model was prepared using available data to evaluate hypothetical failure of the Valea Sesei tailings dam at th...

The USGS entered into an agreement with the Mauritania Ministry of Mines and Industry to inventory and review the quality of information collected as part of the Project for Strengthening of the Institutions in the Mining Sector (PRISM). Whereas the PRISM program collected geophysical, geochemical, geological, satellite, and hydrogeologic informati...

Postmining restoration is one of the greatest concerns for uranium in-situ leach (ISL) mining operations. The ISL-affected aquifer needs to be returned to conditions specified in the mining permit (either premining or other specified conditions). When uranium ISL operations are completed, postmining restoration is usually achieved by injecting redu...

The representativeness of soil pore water extracted by suction lysimeters in ground-water monitoring studies is a problem that often confounds interpretation of measured data. Current soil water sampling techniques cannot identify the soil volume from which a pore water sample is extracted, neither macroscopic, microscopic, or preferential flowpath...