Quantum-mimic Optical Coherence Tomography (Qm-OCT) images are cluttered with artefacts - parasitic peaks which emerge as a by-product of the algorithm used in this method. However, the shape and ...behaviour of an artefact are uniquely related to Group Velocity Dispersion (GVD) of the layer this artefact corresponds to and consequently, the GVD values can be inferred by carefully analysing them. Since for multi-layered objects the number of artefacts is too high to enable layer-specific analysis, we employ a solution based on Machine Learning. We train a neural network with Qm-OCT data as an input and dispersion profiles, i.e. depth distribution of GVD within an A-scan, as an output. By accounting for noise during training, we process experimental data and estimate the GVD values of BK7 and sapphire as well as provide a qualitative GVD value distribution in a grape and cucumber. Compared to other GVD-retrieving methods, our solution does not require user input, automatically provides dispersion values for all the visualised layers and is scalable. We analyse the factors affecting the accuracy of determining GVD: noise in the experimental data as well as general physical limitations of the detection of GVD-induced changes, and suggest possible solutions.
•Extreme value modelling is proposed based on reciprocals of positive block maxima•S-shaped Gumbel plots are predicted, reflecting bounded environmental variables
There has long been interest in ...making inferences about future low-probability natural events that have magnitudes greater than any in the past record. Given a stationary time series, the unbounded Type 1 and Type 2 asymptotic extreme value distributions are often invoked as giving theoretical justification for extrapolating to large magnitudes and long return periods for hydrological variables such as rainfall and river discharge. However, there is a problem in that environmental extremes are bounded above by the bounded nature of their causal variables. Extrapolation using unbounded asymptotic models therefore cannot be justified from extreme value theory and at some point there will be over-prediction of future magnitudes. This creates the apparent contradiction, for example, of annual rainfall maxima being well approximated by Type 2 extreme value distributions despite the bounded nature of rainfall magnitudes. An alternative asymptotic extreme value approach is suggested for further investigation, with the model being the asymptotic distribution of minima (Weibull distribution) applied to block maxima reciprocals. Two examples are presented where data that are well matched by Type 1 or Type 2 extreme value distributions give reciprocals suggestive of lower bounds (upper bound γ to the original data). The asymptotic model here is a 3-parameter Weibull distribution for the reciprocals, with positive location parameter γ−1. When this situation is demonstrated from data, parameter estimation can be carried out with respect to the distribution of reciprocals of 3-parameter Weibull random variables. This distribution is referenced here as the bounded inverse Weibull distribution. A maximum likelihood parameter estimation methodology is presented, together with a parametric bootstrap approach for obtaining one-sided upper confidence limits to γ. When data permits estimation of γ, the bounded inverse Weibull distribution is suggested as an improved alternative to Type 1 or Type 2 extreme value distributions because the upper bound reality is recognised. However, extensive application to many data sets is required to evaluate the practical utility of the bounded approach for extrapolating beyond the largest recorded event.
Creating new windbreaks may reduce lee evaporation by reducing surface wind speed (u). However, applying theoretical models to anticipate the extent of evaporation reduction may be liable to error ...because the physical processes of windbreak evaporative impacts are not fully understood.
An alternative, statistical approach is proposed for sites where eddy correlation time series are available. For low wind speeds the Penman-Monteith equation indicates that recorded evaporation should be approximated as a linear function of Rn, neglecting ground heat flux. That is, air temperature and saturation vapour pressure become of lesser importance. Linear regression with Rn could then be applied as a simple means to anticipate the extent of site evaporation reduction that would occur if u could be sufficiently lowered.
Evaporation linearity with Rn may not always be achievable in the lee of actual constructed windbreaks. However, a linear model could still be useful to obtain an upper bound to evaporation reduction, aiding decisions as to whether to construct a windbreak. Preliminary analysis for an eddy correlation site in Canterbury (South Island, New Zealand) indicates that summer evaporation for wind speed around 1 ms-1 is well approximated as a linear function of Rn, in this case indicating up to 20% possible evaporation reduction. This assumes that consequential changes in other environmental variables can be neglected when wind speed is reduced. If confirmed by further work, the regression approach may find general application as a simple means to anticipate the maximum extent of evaporation reduction from a new windbreak.
The variability of sea surface temperatures (SSTs) is crucial in climate dynamics, influencing marine ecosystems and human activities. This study leverages graph neural networks (GNNs), specifically ...a GraphSAGE model, to forecast SSTs and their anomalies (SSTAs), focusing on the global scale structure of climatological data. We introduce an improved graph construction technique for SST teleconnection representation. Our results highlight the GraphSAGE model's capability in 1‐month‐ahead global SST and SSTA forecasting, with SST predictions spanning up to 2 years with a recursive approach. Notably, regions with persistent currents exhibited enhanced SSTA predictability, contrasting with equatorial and Antarctic areas. Our GNN outperformed both the persistence model and traditional methods. Additionally, we offer an SST and SSTA graph data set based on ERA5 and a graph generation tool. This GNN case study has shown how the GraphSAGE can be used in SST and SSTA forecasting, and will provide a foundation for further refinements such as adjusting graph construction, optimizing imbalanced regression techniques for extreme SSTAs, and integrating GNNs with other temporal pattern learning methods to improve long‐term predictions.
Plain Language Summary
Have you ever wondered how scientists predict climate changes like warming oceans or unusual weather patterns? Our team is improving these predictions by focusing on sea surface temperatures (SSTs) and SST anomalies (SSTAs)—changes in ocean temperatures that greatly impact our environment and society. Traditionally, scientists use grid‐based machine learning models for these forecasts. We have taken a new approach using graph neural networks (GNNs), which are excellent at learning spatial patterns. This method does not just predict temperatures; it understands how different ocean areas are interconnected. Our technique successfully forecasts SSTs up to 2 years ahead and SSTAs 1 month in advance, particularly in regions with strong ocean currents. This accuracy is crucial for anticipating and preparing for climate impacts on marine and human activities. We have developed tools and data sets for researchers, which will foster advancements in graph‐based climate forecasting. Our findings demonstrate that using appropriate graph re‐sampling and GNNs can help understand the complex climate system. In summary, we are pioneering new ways to predict ocean temperature changes through machine learning, contributing to our understanding of and adaptation to climate change. This work benefits not just scientists, but anyone interested in climate and ocean studies.
Key Points
Explores the power of graph representation for global sea surface temperature (SST) and SST anomaly (SSTA) forecasting
Addresses the issues with traditional grid‐based models and provides an SST graph data set and a tool for converting SST grid data to graphs
Shows the efficacy of the GraphSAGE model for global SST and SSTA forecasting and identifies the regions where SSTAs are more predictable
Invasive plants can have both positive and negative impacts on ecosystem services (ES), with decisions on control often being characterised by conflicts over loss of their perceived positive impacts ...on individual ES following removal. We present an analytical framework to aid in reducing such conflicts by allocating control effort to both minimise negative impacts and to maximise positive impacts on multiple ES. We used spatial models to map the negative impacts of invasive conifers on biodiversity, perceived landscape quality, and water yield and their positive impacts on erosion protection and carbon storage across a major catchment in the South Island of New Zealand. We tested the effect of distribution type (i.e. hotspot vs Gaussian) on trade-offs among these ES. We also tested whether using a non-linear function optimisation algorithm to fit variable weights to individual ES significantly reduced trade-offs. We show that an optimised multiple-ES approach could considerably reduce conflicts around invasive tree management arising from their contrasting impacts on different ecosystem services (i.e. by reducing trade-offs between ES), but cannot remove such conflict altogether. Our results are consistent with studies showing that ES with a hot-spot type distribution are the most vulnerable to trade-offs in multi-ES prioritisation, and hence will be the most likely to cause conflict in invasive tree control decisions. Our approach also shows that giving higher priority to ES with hot-spot distributions could reduce conflicts (by reducing tradeoffs). We argue that, when ES data are available, including estimates of ES impacts should be among the “due diligence” requirements for developing invasive tree control strategies.
Expressed as 11-year running means (1945-2008), winter headwater flows in the Waitaki River (New Zealand) and the IPO index show similar patterns in the way they vary over time. However, this effect ...is not evident in the other seasons. A possible explanation is the IPO in its positive phase is associated with warmer air temperatures in winter, but not in other seasons. Analysis of Mt Cook (Hermitage station) daily rainfalls suggests that warmer winter air temperatures on precipitation days are associated with higher mean daily precipitation, so the positive IPO may result in greater amounts of winter precipitation falling as rain rather than snow. This could provide the link between the winter IPO and headwater discharge. On the other hand, temperature is only weakly associated with precipitation magnitudes in other seasons, when higher average temperatures do not provide the same restricting effect on precipitation as in winter.
Artefacts in quantum-mimic optical coherence tomography are considered detrimental because they scramble the images even for the simplest objects. They are a side effect of autocorrelation, which is ...used in the quantum entanglement mimicking algorithm behind this method. Interestingly, the autocorrelation imprints certain characteristics onto an artefact - it makes its shape and characteristics depend on the amount of dispersion exhibited by the layer that artefact corresponds to. In our method, a neural network learns the unique relationship between the artefacts' shape and GVD, and consequently, it is able to provide a good qualitative representation of object's dispersion profile for never-seen-before data: computer-generated single dispersive layers and experimental pieces of glass. We show that the autocorrelation peaks - additional peaks in the A-scan appearing due to the interference of light reflected from the object - affect the GVD profiles. Through relevant calculations, simulations and experimental testing, the mechanism leading to the observed GVD changes is identified and explained. Finally, the network performance is tested in the presence of noise in the data and with the experimental data representing single layers of quartz, sapphire and BK7.
Copulas and other multivariate models can give joint exceedance probabilities for multivariate events in the natural environment. However, the choice of the most appropriate multivariate model may ...not always be evident in the absence of knowledge of dependence structures. A simple nonparametric alternative is to approximate multivariate dependencies using “line mesh distributions”, introduced here as a data-based finite mixture of univariate distributions defined on a mesh of
L
= C(
m
, 2) lines extending through Euclidean
n
-space. That is,
m
data points in
n
-space define a total of
L
lines, where C() denotes the binomial coefficient. The utilitarian simplicity of the method has attraction for joint exceedance probabilities because just the data and a single bandwidth parameter within the 0, 1 interval are needed to define a line mesh distribution. All bivariate planes in these distributions have the same Pearson correlation coefficients as the corresponding data. Marginal means and variances are similarly preserved. Using an example from the literature, a 5-parameter bivariate Gumbel model is replaced with a 1-parameter line mesh distribution. A second illustration for three dimensions applies line mesh distributions to data simulated from a trivariate copula.
Optimised wavelength selection is important to the development of new types of inexpensive and portable near infrared instruments that might be used on fruit in orchards. The use of discrete ...bandwidth devices, such as light-emitting diodes, requires preselection of a small number of discrete wavelengths. In this work, a kiwifruit data set consisting of 834 absorbance spectra and corresponding fruit dry-matter measurements, an important maturity indicator for kiwifruit, has been subjected to an exhaustive wavelength search to build optimal multiple linear regression models of up to seven wavelengths. Using a standard partial least-squares model as a benchmark, a six-wavelength model has been identified as an optimum, predicting kiwifruit dry matter with r2 of 0.88 and root mean square error of prediction (RMSEP) of 1.22%. The sensitivity of the model to shifts in the key wavelengths was also evaluated, revealing that a 1 nm offset or a 0.25 nm random noise component would be enough to increase the RMSEP by around 0.04% in actual dry matter value or 3% in relative percentage terms.
PDF