Time series data from environmental monitoring stations are often analysed with machine learning methods on an individual basis, however recent advances in the machine learning field point to the advantages of incorporating multiple related time series from the same monitoring network within a ‘global’ model. This approach provides the opportunity for larger training data sets, allows information to be shared across the network, leading to greater generalisability, and can overcome issues encountered in the individual time series, such as small datasets or missing data. We present a case study involving the analysis of 165 time series from groundwater monitoring wells in the Namoi region of Australia. Analyses of the multiple time series using a variety of different aggregations are compared and contrasted (with single time series, subsets, and all of the time series together), using variations of the multilayer perceptron (MLP), self-organizing map (SOM), long short-term memory (LSTM), and a recently developed LSTM extension (DeepAR) that incorporates autoregressive terms and handles multiple time series. The benefits, in terms of prediction performance, of these various approaches are investigated, and challenges such as differing measurement frequencies and variations in temporal patterns between the time series are discussed. We conclude with some discussion regarding recommendations and opportunities associated with using networks of environmental data to help inform future resource-related decision making.
In this study, we developed a workflow that applies a complex groundwater model for purpose-driven groundwater monitoring network design and uses linear uncertainty analysis to explore the predictive dependencies and provide insights into the veracity of the monitoring design. A numerical groundwater flow model was used in a probabilistic modelling framework for obtaining the spatial distribution of predicted drawdown for a wide range of plausible combination of uncertain parameters pertaining to the deep sedimentary basin and groundwater flow processes. Reduced rank spatial prediction was used to characterize dominant trends in these spatial drawdown patterns using empirical orthogonal functions (EOF). A differential evolution algorithm was used to identify optimal locations for multi-level piezometers for collecting groundwater pressure data to minimize predictive uncertainty in groundwater drawdown. Data-worth analysis helps to explore the veracity of the design by using only the sensitivities of the observations to predictions independent of the absolute values of predictions. A 10-bore monitoring network that collects drawdown data from multiple depths at each location was designed. The data-worth analysis revealed that the design honours sensitivities of the predictions of interest to parameters. The designed network provided relatively high data-worth for minimizing uncertainty in the drawdown prediction at locations of interest.
Monitoring groundwater quality in economically important and other aquifers is carried out regularly as part of regulatory processes for water and other resource development. Many water quality parameters are measured as part of baseline monitoring around mining and onshore gas resource development regions to develop improved understanding of the hydrogeological system as well as to inform managerial decisions to assess and manage contamination risks and health hazards. Water quality distribution in an aquifer is most often inferred from point measurements from limited number of bores drilled at arbitrary locations. Estimating the distribution of water quality parameters in the aquifer based on these point measurements is often a challenging task and results in high uncertainty in the estimates due to limited data availability. Minimizing uncertainty can be achieved by drilling more bores to collect water quality data and several approaches are available to identify optimal bore hole locations to minimize estimation uncertainty. However, optimization of borehole locations is difficult when multiple water quality parameters are of interest and have different spatial distributions in the aquifer. In this study we use geostatistical kriging to interpolate a large number of groundwater quality parameters. Then we integrate these predicted values and use the Differential Evolution algorithm to determine optimal locations for bores that would simultaneously reduce spatial prediction uncertainty of all parameters. The method is applied for designing a groundwater monitoring network in the Namoi region of Australia for monitoring groundwater quality in an economically important aquifer of the Great Artesian Basin. Optimal locations for 10 new monitoring bores are identified using this approach.
We used remote sensing data, field observations and numerical groundwater modelling to investigate long-term groundwater storage losses in the regional aquifer of the Ganga Basin in India. This comprised trend analysis for groundwater level observations from 2851 monitoring bores, groundwater storage anomaly estimation using GRACE and Global Land Data Assimilation System (GLDAS) data sets and numerical modelling of long-term groundwater storage changes underpinned by over 50,000 groundwater level observations and uncertainty analysis. Three analyses based on different methods consistently informed that groundwater storage in the aquifer is declining at a significant rate. Groundwater level trend indicated storage loss in the range - 1.1 to - 3.3 cm year-1 (median - 2.6 cm year-1) while the modelling and GRACE storage anomaly methods indicated the storage loss in the range of - 2.1 to - 4.5 cm year-1 (median - 3.2 cm year-1) and - 1.0 to - 4.2 cm year-1 (median - 1.7 cm year-1) respectively. Probabilistic modelling analysis also indicated that the average groundwater storage is declining in all the major basin states, the highest declining trend being in the western states of Rajasthan, Haryana and Delhi. While smaller compared to the western states, average groundwater storage in states further towards east-Uttar Pradesh, Bihar and West Bengal within the basin are also declining. Time series of storage anomalies obtained from the three methods showed similar trends. Probabilistic storage analysis using the numerical model vetted by observed trend analysis and GRACE data provides the opportunity for predictive analysis of storage changes for future climate and other scenarios.
Efficient and reliable diagnostic tools for the routine indexing and certification of clean propagating material are essential for the management of pospiviroid diseases in horticultural crops. This study describes the development of a true multiplexed diagnostic method for the detection and identification of all nine currently recognized pospiviroid species in one assay using Luminex bead-based suspension array technology. In addition, a new data-driven, statistical method is presented for establishing thresholds for positivity for individual assays within multiplexed arrays. When applied to the multiplexed array data generated in this study, the new method was shown to have better control of false positives and false negative results than two other commonly used approaches for setting thresholds. The 11-plex Luminex MagPlex-TAG pospiviroid array described here has a unique hierarchical assay design, incorporating a near-universal assay in addition to nine species-specific assays, and a co-amplified plant internal control assay for quality assurance purposes. All assays of the multiplexed array were shown to be 100% specific, sensitive and reproducible. The multiplexed array described herein is robust, easy to use, displays unambiguous results and has strong potential for use in routine pospiviroid indexing to improve disease management strategies.
Abstract Recent advances in aerial drones offer new insights into the biology, ecology and behaviour of marine wildlife found on or near the ocean’s surface. While opening up new opportunities for enhanced wildlife monitoring, the impacts of drone sampling and how it might influence interpretations of animal behaviour are only just beginning to be understood. The capacity of drones to record bottlenose dolphin ( Tursiops spp.) behaviour was investigated, along with how the presence of a small drone at varying altitudes influences dolphin behaviour. Over 3 years and eight locations, 361 drone flights were completed between altitudes of 5 and 60 m above the ocean. Analyses showed that dolphins were increasingly likely to change behaviour with decreasing drone altitude. A positive correlation was also found between time spent hovering above a group of dolphins and the probability of recording a behavioural response. Dolphin group size also influenced the frequency of an observed behavioural change, displaying a positive correlation between behaviour change and group size. Overall, although drones have the potential to impact coastal dolphins when flown at low altitudes, they represent a useful tool for collecting ecological information on coastal dolphins owing to their convenience, low cost and capacity to observe behaviours underwater. To maximize benefits and minimize impacts, this study suggests that drones should be flown 30 m above coastal bottlenose dolphins.
'/%3e%3c/defs%3e%3cpath fill='%23012169' d='M0 0h10080v5040H0z'/%3e%3cpath stroke='%23fff' stroke-width='.6' d='m0 0 6 3m0-3L0 3' clip-path='url(%23b)' transform='scale(840)'/%3e%3cpath stroke='%23e4002b' stroke-width='.4' d='m0 0 6 3m0-3L0 3' clip-path='url(%23c)' transform='scale(840)'/%3e%3cpath stroke='%23fff' stroke-width='840' d='M2520 0v2520M0 1260h5040'/%3e%3cpath stroke='%23e4002b' stroke-width='504' d='M2520 0v2520M0 1260h5040'/%3e%3cg fill='%23fff'%3e%3cuse xlink:href='%23d' x='2520' y='3780'/%3e%3cuse xlink:href='%23a' x='7560' y='4200'/%3e%3cuse xlink:href='%23a' x='6300' y='2205'/%3e%3cuse xlink:href='%23a' x='7560' y='840'/%3e%3cuse xlink:href='%23a' x='8680' y='1869'/%3e%3cuse xlink:href='%23e' x='8064' y='2730'/%3e%3c/g%3e%3c/svg%3e)
