Observation of western Siberian Wetlands by using remote sensing techniques: Estimation of methane emission
M. Tamura and Y. Yasuoka
National Institute for Environmental Studies,
16-2 Onogawa, Tsukuba, Ibaraki, 305 Japan
Nakanihon Service, Air 2-3-15 Kyobashi, Chuou, Tokyo, 104 Japan
Phone : +81-298-50-2479; Fax: +81-298-51-4732;
E-mail: [email protected]
A SPOT/HRV image was obtained at Plotnikovo test site in western Siberian wetlands. The image area was classified into eight categories of ecosystems (birch forest, conifer forest, bog_1, bog_2, bog_3, water, grass and bare soil) using ground truth data and aerial photographs. Methane emission from the image area was estimated by combining the result of ecosystem classification and the methane flux data measured on the ground. This estimate of methane emission was in good agreement with the average estimate obtained from airborne methane measurements. The agreement shows the usefulness of remote sensing techniques in extrapolating ground methane flux measurements to a regional or global scale estimate of methane emissions.
Western Siberian wetlands are presumed to be large sources of atmospheric methane, which is one of the most significant greenhouse gases, Recent Japan-Russia joint research Project conducted by our research institute and Russian co-workers is yielding the results supporting this resumption (Panikov, 1994). To evaluate the role of the western Siberian wetlands as sources of atmospheric methane, it is necessary to classify wetland ecosystems and to measure mean methane flux for each ecosystem type. In this research we investigate the vegetation in western Siberian wetlands by using remote sensing techniques and estimate regional methane emissions by combining the results of satellite observations with ground methane measurements. We use two categories of satellite sensors: a wide-coverage coarse-spatial-resolution sensor like NOAA/AVHRR and high-spatial-resolution sensors like SPOT/HRV and JERS-I/OPS & SAR. The former is used to estimate overall wetland distributions and to monitor the seasonal change of vegetation in the whole western Siberian wetlands. The latter is used to classify wetland ecosystems is selected test sites.
In this paper we analyze a SPO/HRV image at Plotnikovo test site in west Siberia to classify wetland ecosystems. Methane emission from the image area is estimated by combining the result of emission estimate with those obtained from airborne methane measurements.
II. SPOT/HRV image at Plotnikovo in western Siberian wetlands
A SPOT/HRV image was obtained at Plotnikovo test site on 8 July 1995. Plotnikovo settlement is located at the map coordinates (85o05’E, 56o51’N) in the basin of Ob’ river and belong to the south east part of the western Siberian wetlands (see Figure I). The SPOT/HRV is a high resolution imaging system with a ground resolution of 20 m and has three spectral bands of green, red and near infrared wavelengths (0.50-0.59, 0.61-0.68 and 0.79-0.89 um). Figure 2 shows the SPOT/HRV image in the near infrared band. We selected this place as one of our test sites because ground observations of vegetation and atmospheric gasses have been made since 1993 by the Moscow Institute of Microbiology. Circle symbols in Figure 2 shows the locations of methods flux measurements.
Figure 1: Location of Platnikovo
Figure 2: SPOT/HRV image at Plotnikovo. Circle symbols show ground measurement points of methane fluxes.
Based on the ground truth data and aerial photographs obtained in 1994 and 1995, we classified the land cover types in the image area into eight categories (birch forest, conifer forest, bog_1, bog_2, bog_3, water, grass and bare soil)using the supervised maximum likelihood classification technique. Figure 3 shows the result of land cover classification. Birch trees are dominant in forested areas and coniferous
forests are found along rivers because soils have richer nutrients there. Bog_1 is identified as peat land with low pine trees and shrubs, bog_2 as peat land with sparse dwarf trees and shrubs, and bog_3 as peat land mainly with grasses such as sedges and cotton-grasses.
Figure 3: Result of land cover classification