LAPIG INDIRECT DETECTION OF LIQUID HYDROCARBON LEAKAGES ON CONTAMINED SOIL-VEGETATION SYSTEMS THROUGH REFLECTANCE AND IMAGING (PROSPECTIR-VS) SPECTROSCOPY: A POTENTIAL TOOL FOR EXTENSIVE PIPELINE MONITORING Carlos Roberto Souza Filho1, Lucíola Magalhães1; Giuliana Quitério1 Marcos Nopper1 , Teodoro Almeida2 ; Wilson Olveira3; Lis Rabaco3; Renato Rocha3 University of Campinas, Campinas, Brazil 1 University of São Paulo, São Paulo, Brazil2 Petrobras S.A./CENPES, Rio de Janeiro, Brazil3
Current methods of pipeline monitoring poses hindrances to the early detection of small hydrocarbon spills.
Assuming that liquid fuels (gasoline and diesel) are potential vegetation stressors, this study investigates the spectral characteristics of agricultural crops subjected to daily contamination of liquid hydrocarbons, using reflectance spectroscopy (portable FieldSpec Hi-Res sensor, with 2150 bands in the VNIR-SWIR range) and imaging spectroscopy (airborne ProSpecTIR-VS hyperspectral sensor, with 357 bands in the VNIR-SWIR range).
The study comprises both greenhouse and real scale experiments, where we will seek the probable impacts of hydrocarbon contamination through spectral changes in 5 plant species commonly present in the vicinity of pipelines, particularly in Brazil, with focus on Brachiaria brizantha (“grass”) and Neonotonia wightii (“soybean”)
THE EXPERIMENT 2 SCALES: small scale experiment (“lysimeter”) & macro scale (real, field scale). 3 types of spectral analysis Leaf (lysimeter) (FieldSpec Hi-Res – ASD) Canopy (macro scale experiment) (FieldSpec Hi-Res – ASD) Canopy (airborne hyperspectral survey) (357 bands at 5nm resolution -ProSpecTIR-VS)
2 contaminants: diesel (DSL) and gasoline (GSL)
THE EXPERIMENT Vegetation Species Resistance (i.e. plagues) and commonness Agronomic importance Extensive occurrence along pipelines Neonotonia wightii “soybean” Brachiaria brizantha “perenial grass”
LYSIMETERSNeonotonia wightii SHORTWAVE INFRARED (SWIR) REGION Spectral pattern at 2062 nm >> association with leaf biochemical analysis (rise in monosaccharides (“sugar”) content) monosaccharides Normalised average (% in relation to CTR) Wavelength (nm)
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) Two step algorithm (Almeida & Souza Filho, 2004; 2008): 1) Production of 15 spectral indices > enhancement of specific spectral signatures of vegetation properties 2) Principal Component Analysis applied to three sub-stes of spectral indices Indices applied to vegetation analysis Group Index Spectral Formula ProSpecTIR channels -Caroten Antocianin a-Chlorophyll b-Chlorophyll Carotenoids SIPI CSe NDVI VOG1 WBI MAC Leaf water Lignin Cellulose Nitrogen
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) GSL DSL CTR Degree of stress CSe Ratio Degree of stress based on the CSe (694nm/760nm) ratio
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) GSL DSL Degree of stress CTR PC1 - Group 1 (VNIR) a-Chlorophyll, b-Chlorophyll, Carotenoids Eigenvector matrix Eigenvector a-caroten antocian. a-Chlrop b-Chlorop Caroten. SIPI CSe
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) GSL Degree of stress DSL CTR PC1 - Group 2 (NIR) NDVI, VOG1 e WBI Eigenvector matrix Eigenvector
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) GSL Degree of stress DSL CTR PC2 – Group 3 (NIR/SWIR) Lignin EigenvectorLeaf Water Lignin Cellulose Nitrogen
INFORMATION EXTRACTION OF HYPERSPECTRAL DATA (PROSPECTIR VS) PC1 Group 1 GSL DSL PC1 Group 2 PC2 Group 3 CTR RGB Colour Composition R: PC1 - Group 1 (VNIR) G: PC1 - Group 2 (NIR) B: PC2 - Group 3 (NIR/SWIR)
The proposed methodology showed a high correlation between canopy spectral measurements taken at close range with the FieldSpec Hi-Res sensor and from the airborne ProspecTIR-VS sensor.
It was possible to characterize the reflectance of leaves grown in soils contaminated by low concentrations of gasoline and diesel and differentiate them from plants grown on soil without HCs.
The use of selected vegetation indices showed a high correlation with the behavior of vegetation stressed by the presence of HCs in all three scales of observations.
The spectral changes were similar among species but more prominent for gasoline (GSL) than diesel (DSL), occurring at different timeframes and under different doses of HCs
The results confirm the higher toxicity of gasoline for all selected crops.
The development of this work supports the possibility to preserve certain crops along pipelines that can be used as a bio-indicator of small leakages and the types of crops more susceptible to stress-induced leakage.
It also makes a first step on the establishment of the initial timing (i.e., exposure time and volume of injected hydrocarbons) when the contamination effects are more perceptible remotely.
LAPIG Thank you ! Geosciences Institute University of Campinas (UNICAMP) www.ige.unicamp.br www.ige.unicamp.br/sdm email@example.com