Tim Bellerby

Satellite Oil Spill Detection Using Artificial Neural Networks

Oil spills represent a major threat to ocean ecosystems and their health. Illicit pollution requires continuous monitoring and satellite remote sensing technology represents an attractive option for operational oil spill detection. Previous studies have shown that active microwave satellite sensors, particularly Synthetic Aperture Radar (SAR) can be effectively used for the detection and classification of oil spills. Oil spills appear as dark spots in SAR images. However, similar dark spots may arise from a range of unrelated meteorological and oceanographic phenomena, resulting in misidentification. A major focus of research in this area is the development of algorithms to distinguish oil spills from `look-alikes. This paper describes the development of a new approach to SAR oil spill detection employing two different Artificial Neural Networks (ANN), used in sequence. The first ANN segments a SAR image to identify pixels belonging to candidate oil spill features. A set of statistical feature parameters are then extracted and used to drive a second ANN which classifies objects into oil spills or look-alikes. The proposed algorithm was trained using 97 ERS-2 SAR and ENVSAT ASAR images of individual verified oil spills or/and look-alikes. The algorithm was validated using a large dataset comprising full-swath images and correctly identified 91.6% of reported oil spills and 98.3% of look-alike phenomena. The segmentation stage of the new technique outperformed the established edge detection and adaptive thresholding approaches. An analysis of feature descriptors highlighted the importance of image gradient information in the classification stage.

LMODEL: A Satellite Precipitation Methodology Using Cloud Development Modeling. Part I: Algorithm Construction and Calibration

Abstract The Lagrangian Model (LMODEL) is a new multisensor satellite rainfall monitoring methodology based on the use of a conceptual cloud-development model that is driven by geostationary satellite imagery and is locally updated using microwave-based rainfall measurements from low earth-orbiting platforms. This paper describes the cloud development model and updating procedures; the companion paper presents model validation results. The model uses single-band thermal infrared geostationary satellite imagery to characterize cloud motion, growth, and dispersal at high spatial resolution (∼4 km). These inputs drive a simple, linear, semi-Lagrangian, conceptual cloud mass balance model, incorporating separate representations of convective and stratiform processes. The model is locally updated against microwave satellite data using a two-stage process that scales precipitable water fluxes into the model and then updates model states using a Kalman filter. Model calibration and updating employ an empirical r...

Detection and classification of oil spill and look-alike spots from SAR imagery using an Artificial Neural Network

Oil spills represent a major threat to ocean ecosystems and their health. The recent incident in the Gulf of Mexico demonstrates the potentially catastrophic nature of offshore oil spills. Illicit pollution requires continuous monitoring and satellite remote sensing technology represents an attractive option for operational oil spill detection. Previous studies have shown that active microwave satellite sensors, particularly Synthetic Aperture Radar (SAR) can be effectively used for the detection and classification of oil spills. Oil spills appear as dark spots in SAR images. However, similar dark spots may arise from a range of unrelated meteorological and oceanographic phenomena, resulting in misidentification. A major focus of research in this area is the development of algorithms to distinguish oil spills from `look-alikes. This paper describes the development of a new approach to SAR oil spill detection using two different Artificial Neural Networks (ANN). The first ANN segments a SAR image to identify pixels belonging to candidate oil spill features. A set of statistical feature parameters are then extracted and divided into subsets to facilitate sensitivity analyses. The second ANN classifies objects into oil spills or look-alikes according to their feature parameters. A pilot study employed sixty-two ERS-2 SAR and ENVSAT ASAR images of verified oil spills or look-alikes to train and evaluate the algorithm. Overall accuracies of 96.52 % were obtained for pixel segmentation and 95.2 % for feature classification. The segmentation approach outperformed established edge detection and adaptive thresholding techniques. An analysis of feature descriptors in the classification stage highlighted the importance of image gradient information.

WRF model sensitivity to choice of parameterization: a study of the ‘York Flood 1999’

Numerical weather modelling has gained considerable attention in the field of hydrology especially in un-gauged catchments and in conjunction with distributed models. As a consequence, the accuracy with which these models represent precipitation, sub-grid-scale processes and exceptional events has become of considerable concern to the hydrological community. This paper presents sensitivity analyses for the Weather Research Forecast (WRF) model with respect to the choice of physical parameterization schemes (both cumulus parameterisation (CPSs) and microphysics parameterization schemes (MPSs)) used to represent the ‘1999 York Flood’ event, which occurred over North Yorkshire, UK, 1st–14th March 1999. The study assessed four CPSs (Kain–Fritsch (KF2), Betts–Miller–Janjic (BMJ), Grell–Devenyi ensemble (GD) and the old Kain–Fritsch (KF1)) and four MPSs (Kessler, Lin et al., WRF single-moment 3-class (WSM3) and WRF single-moment 5-class (WSM5)] with respect to their influence on modelled rainfall. The study suggests that the BMJ scheme may be a better cumulus parameterization choice for the study region, giving a consistently better performance than other three CPSs, though there are suggestions of underestimation. The WSM3 was identified as the best MPSs and a combined WSM3/BMJ model setup produced realistic estimates of precipitation quantities for this exceptional flood event. This study analysed spatial variability in WRF performance through categorical indices, including POD, FBI, FAR and CSI during York Flood 1999 under various model settings. Moreover, the WRF model was good at predicting high-intensity rare events over the Yorkshire region, suggesting it has potential for operational use.

Passive microwave retrieval of ocean surface windspeeds for British coastal waters

Previous approaches to the estimation of ocean surface windspeeds from passive microwave data have been incapable of providing information within 100 km of a coastline due to land contamination in pixel footprints. This paper describes the application of a brightness-temperature separation method to the computation of Special Sensor Microwave Imager (SSM/I)-based estimates of ocean surface windspeed for the littoral zone surrounding the southern part of the British Isles over the three-month period of May-July 1996. When validated against in situ buoy data, the windspeed estimates for the coastal zone displayed broadly similar error characteristics to equivalent estimates made for the open ocean.