(1) 自然资源部第一海洋研究所

Oil spill is a common problem faced by marine countries,but there is still no reliable and practical method for oil slick accurate identification and quantity measurement via remote sensing technology.Based on the UAV hyperspectral imaging experiment,methods of oil spill detection and thickness estimation are studied.In the experiment, the UAV hyperspectral remote sensing and field spectral data of oil spill with different quantities are obtained in an oil spill experiment tank of large outdoor flume under the condition of simulating real marine environment. Then the feature spectral bands based oil spill detection and oil slick thickness estimation models are found. At last we get the following conclusions:(1)675nm and 699nm are the effective characteristic bands of oil spill detection,however,they have no detection capability for the very thin oil slick(thickness≤5mum),(2)3kinds of oil slick thickness estimation models witch are Normalized Difference Oil Spill Index(NDOSI)model,inverse proportion model and absorption line height model are proposed,in which the inverse ratio model is the only choice for thin and thick(thickness>50mum)oil slick.For the medium thickness oil slick,the NDOSI model is the best choice,and the inverse scale model and the oil spill absorption baseline height model have better inversion ability, and in cloudy weather,the inverse scale model is the best.