Ben Ingram

A comparison of multiple indicator kriging and area-to-point Poisson kriging for mapping patterns of herbivore species abundance in Kruger National Park, South Africa

Kruger National Park (KNP), South Africa, provides protected habitats for the unique animals of the African savannah. For the past 40 years, annual aerial surveys of herbivores have been conducted to aid management decisions based on (1) the spatial distribution of species throughout the park and (2) total species populations in a year. The surveys are extremely time consuming and costly. For many years, the whole park was surveyed, but in 1998 a transect survey approach was adopted. This is cheaper and less time consuming but leaves gaps in the data spatially. Also the distance method currently employed by the park only gives estimates of total species populations but not their spatial distribution. We compare the ability of multiple indicator kriging and area-to-point Poisson kriging to accurately map species distribution in the park. A leave-one-out cross-validation approach indicates that multiple indicator kriging makes poor estimates of the number of animals, particularly the few large counts, as the indicator variograms for such high thresholds are pure nugget. Poisson kriging was applied to the prediction of two types of abundance data: spatial density and proportion of a given species. Both Poisson approaches had standardized mean absolute errors (St. MAEs) of animal counts at least an order of magnitude lower than multiple indicator kriging. The spatial density, Poisson approach (1), gave the lowest St. MAEs for the most abundant species and the proportion, Poisson approach (2), did for the least abundant species. Incorporating environmental data into Poisson approach (2) further reduced St. MAEs.

Detection and Segmentation of Vine Canopy in Ultra-High Spatial Resolution RGB Imagery Obtained from Unmanned Aerial Vehicle (UAV): A Case Study in a Commercial Vineyard

The use of Unmanned Aerial Vehicles (UAVs) in viticulture permits the capture of aerial Red-Green-Blue (RGB) images with an ultra-high spatial resolution. Recent studies have demonstrated that RGB images can be used to monitor spatial variability of vine biophysical parameters. However, for estimating these parameters, accurate and automated segmentation methods are required to extract relevant information from RGB images. Manual segmentation of aerial images is a laborious and time-consuming process. Traditional classification methods have shown satisfactory results in the segmentation of RGB images for diverse applications and surfaces, however, in the case of commercial vineyards, it is necessary to consider some particularities inherent to canopy size in the vertical trellis systems (VSP) such as shadow effect and different soil conditions in inter-rows (mixed information of soil and weeds). Therefore, the objective of this study was to compare the performance of four classification methods (K-means, Artificial Neural Networks (ANN), Random Forest (RForest) and Spectral Indices (SI)) to detect canopy in a vineyard trained on VSP. Six flights were carried out from post-flowering to harvest in a commercial vineyard cv. Carmenere using a low-cost UAV equipped with a conventional RGB camera. The results show that the ANN and the simple SI method complemented with the Otsu method for thresholding presented the best performance for the detection of the vine canopy with high overall accuracy values for all study days. Spectral indices presented the best performance in the detection of Plant class (Vine canopy) with an overall accuracy of around 0.99. However, considering the performance pixel by pixel, the Spectral indices are not able to discriminate between Soil and Shadow class. The best performance in the classification of three classes (Plant, Soil, and Shadow) of vineyard RGB images, was obtained when the SI values were used as input data in trained methods (ANN and RForest), reaching overall accuracy values around 0.98 with high sensitivity values for the three classes.

Projected sequential Gaussian processes: A C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed.

Spatial analysis of drug poisoning deaths in the American West, particularly Utah.

BACKGROUND Most states in the Western US have high rates of drug poisoning death (DPD), especially New Mexico, Nevada, Arizona and Utah (UT). This seems paradoxical in UT where illicit drug use, smoking and drinking rates are low. To investigate this, spatial analysis of county level DPD data and other relevant factors in the Western US and UT was undertaken. METHODS Poisson kriging was used to smooth the DPD data, populate data gaps and improve the reliability of rates recorded in sparsely populated counties. Links between DPD and economic, environmental, health, lifestyle, and demographic factors were investigated at four scales using multiple linear regression. LDS church membership and altitude, factors not previously considered, were included. Spatial change in the strength and sign of relationships was investigated using geographically weighted regression and significant DPD clusters were identified using the Local Morans I. RESULTS Economic factors, like the sharp social gradient between rural and urban areas were important to DPD throughout the west. Higher DPD rates were also found in areas of higher elevation and the desert rural areas in the south. The unique characteristics of DPD in UT in terms of health and lifestyle factors, as well as the demographic structure of DPD in the most LDS populous states (UT, Idaho, Wyoming), suggest that high DPD in heavily LDS areas are predominantly prescription opioid related whereas in other Western states a larger proportion of DPD might come from illicit drugs. CONCLUSION Drug policies need to be adapted to the geographical differences in the dominant type of drug causing death. Educational materials need to be marketed to the demographic groups at greatest risk and take into account differences in population characteristics between and within States. Some suggestions about how such adaptations can be made are given and future research needs outlined.

A Web-based GIS Decision Support Tool for Determining Corn Aflatoxin Risk: A Case Study Data from Southern Georgia, USA

Aflatoxin is a fungal toxin contaminating corn and causing liver cancer in humans and animals. Contamination is driven by high temperatures and drought. Aflatoxin assessment is expensive so extension services need to identify high risk areas so irrigation, planting strategies and corn varieties can be adapted. This research presents a web-based decision support tool for risk illustrated with a case study from southern Georgia. The tool employs the approach, developed by Kerry et al. (2017b) where exceedance of key thresholds in temperatures, rainfall, soil type and corn production are used to determine risk. The tool also includes NDVI to indicate drought stress and could be further expanded to include new risk factors and adapted to other crops.

Modeling the spatial distribution of African buffalo (Syncerus caffer) in the Kruger National Park, South Africa

The population density of wildlife reservoirs contributes to disease transmission risk for domestic animals. The objective of this study was to model the African buffalo distribution of the Kruger National Park. A secondary objective was to collect field data to evaluate models and determine environmental predictors of buffalo detection. Spatial distribution models were created using buffalo census information and archived data from previous research. Field data were collected during the dry (August 2012) and wet (January 2013) seasons using a random walk design. The fit of the prediction models were assessed descriptively and formally by calculating the root mean square error (rMSE) of deviations from field observations. Logistic regression was used to estimate the effects of environmental variables on the detection of buffalo herds and linear regression was used to identify predictors of larger herd sizes. A zero-inflated Poisson model produced distributions that were most consistent with expected buffalo behavior. Field data confirmed that environmental factors including season (P = 0.008), vegetation type (P = 0.002), and vegetation density (P = 0.010) were significant predictors of buffalo detection. Bachelor herds were more likely to be detected in dense vegetation (P = 0.005) and during the wet season (P = 0.022) compared to the larger mixed-sex herds. Static distribution models for African buffalo can produce biologically reasonable results but environmental factors have significant effects and therefore could be used to improve model performance. Accurate distribution models are critical for the evaluation of disease risk and to model disease transmission.