Oil Spills or Look-Alikes? Classification Rank of Surface Ocean Slick Signatures in Satellite Data

Carvalho, Gustavo de Araújo; Minnett, Peter J.; Ebecken, Nelson F. F.; Landau, Luiz

doi:10.3390/rs13173466

Cited by 9 publications

(32 citation statements)

References 60 publications

(118 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The current research builds on the analyses of Carvalho et al [31,32], who exploited LDAs to classify oil spills from look-alikes. Their overall-accuracy classification results are the benchmark used here.…”

Section: Methodsmentioning

confidence: 99%

“…The database used in our experiment has two types of variables characterizing the ocean-slick targets: morphological information and MetOc parameters-Section 2.2 below. From the knowledge gained from past studies (e.g., [31,32]), we chose to analyze each of these variable types together and separately. As such, three combinations of variables were carried to the next stages:…”

Section: Stage 2: Combinations Of Variablesmentioning

confidence: 99%

“…Furthermore, these 36 algorithms were also compared with results from past research [31,32]. These authors also investigated oil spill versus look-alike classification algorithms applied to combinations of similar but different variables than used here; however, they used the simpler LDA technique (Section 2.4.1 below).…”

Section: Stage 4: Classification-algorithms Comparisonmentioning

confidence: 99%

“…Even though a number of oil slick databases exist and may be obtained in the relevant literature-for instance, from the Gulf of Mexico (e.g., [19,27]) and from the European Seas provided by European Maritime Safety Agency ( [59])-the main reason for our database choice is that it has been used in previous studies, also to discriminate spills from lookalikes but with a simpler ML technique (i.e., LDA in Carvalho et al [31,32]). Those past results are used here as a benchmark to evaluate our findings.…”

Section: Databasementioning

confidence: 99%

“…LDAs were explored by Carvalho et al [27][28][29] for seepspill classification: overall accuracies of ~70% were obtained using ~5000 samples imaged with RADARSAT-2 in Campeche Bay, Gulf of Mexico [30]. In a pair of follow-up papers, LDAs were used in a similar fashion, but to classify the signatures of oil spills and lookalikes off the coast of Brazil using RADARSAT-1 measurements: (i) exploiting ~770 samples, Carvalho et al [31] combined morphological information (e.g., area and perimeter) with Meteorological and Oceanographic contextual information ("MetOc parameters"), in a total of 39 combinations of variables, reaching overall accuracies of ~84%; and (ii) making use of ~550 samples, Carvalho et al [32] accounted for a total of 114 combinations of variables that included morphological, MetOc, and site-specific contextual geo-location information ("Geo-Loc attributes", e.g., latitude and longitude) to find overall accuracies up to ~85%. In the current paper, these oil spills versus look-alikes LDA overall-accuracy results serve as a benchmark to compare with the results of other ML methods.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Machine-Learning Classification of SAR Remotely-Sensed Sea-Surface Petroleum Signatures—Part 1: Training and Testing Cross Validation

et al. 2022

Self Cite

View full text Add to dashboard Cite

Sea-surface petroleum pollution is observed as “oil slicks” (i.e., “oil spills” or “oil seeps”) and can be confused with “look-alike slicks” (i.e., environmental phenomena, such as low-wind speed, upwelling conditions, chlorophyll, etc.) in synthetic aperture radar (SAR) measurements, the most proficient satellite sensor to detect mineral oil on the sea surface. Even though machine learning (ML) has become widely used to classify remotely-sensed petroleum signatures, few papers have been published comparing various ML methods to distinguish spills from look-alikes. Our research fills this gap by comparing and evaluating six traditional techniques: simple (naive Bayes (NB), K-nearest neighbor (KNN), decision trees (DT)) and advanced (random forest (RF), support vector machine (SVM), artificial neural network (ANN)) applied to different combinations of satellite-retrieved attributes. 36 ML algorithms were used to discriminate “ocean-slick signatures” (spills versus look-alikes) with ten-times repeated random subsampling cross validation (70-30 train-test partition). Our results found that the best algorithm (ANN: 90%) was >20% more effective than the least accurate one (DT: ~68%). Our empirical ML observations contribute to both scientific ocean remote-sensing research and to oil and gas industry activities, in that: (i) most techniques were superior when morphological information and Meteorological and Oceanographic (MetOc) parameters were included together, and less accurate when these variables were used separately; (ii) the algorithms with the better performance used more variables (without feature selection), while lower accuracy algorithms were those that used fewer variables (with feature selection); (iii) we created algorithms more effective than those of benchmark-past studies that used linear discriminant analysis (LDA: ~85%) on the same dataset; and (iv) accurate algorithms can assist in finding new offshore fossil fuel discoveries (i.e., misclassification reduction).

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Stage 2: Combinations Of Variablesmentioning

confidence: 99%

Section: Stage 4: Classification-algorithms Comparisonmentioning

confidence: 99%