Study of the efficiency of monitoring the pollution of the earth's surface with oil products in the near IR range depending on the number of spectral channels

The paper presents a study of the efficiency of monitoring the pollution of the earth's surface with oil products in the near IR range. The paper presents the reflectance spectra of soil and sand samples contaminated with various oil products (commercial oil, diesel fuel, kerosene, gas condensate, gasoline, motor oil) in the spectral range of 900-2500 nm measured on a laboratory setup. The paper shows the results of the created neural network for 3 and 30 spectral channels for detecting pollution of the surface of samples with oil products. It has been established that the use of thirty spectral channels with a width of 10 nm in the spectral range of 1500-1800 nm allows for a probability of detecting oil products on the earth's surface of more than 99% in a wide range of weather conditions (dry soil, moderately wet soil, moderate rain after an oil spill) with a false alarm probability of less than 1%. It is shown that the use of three spectral channels with a width of 50 nm with central wavelengths of 1696, 1734 and 1775 nm makes it possible to ensure, in a wide range of weather conditions, a probability of detecting oil products on the earth's surface of more than 93% with a probability of false alarms of less than 3%.