Combining ALOS-PALSAR imagery with field water level measurements for flood mapping of a tropical floodplain

Abstract

Radar imagery is potentially useful for the identification, mapping, and measurement of streams, lakes, and wetlands. Many studies showed that comparison of two consecutive radar images is useful for determining flood extent. However, the use of radar data series for flood mapping is still rarely reported. The purpose of this study is to explore the use of ALOS-PALSAR imagery for observing the dynamics of the Mahakam River floodplain in Kalimantan, Indonesia, by incorporating field water level measurements. Water level measurements were carried out along the river, lakes and at two peatland locations, using arrays of pressure transducers. The first peatland (P1) is part of the Mahakam floodplain, representing open peat area dominated by shrub and reed. The second peatland (P2) represents a forest covered peatland. A series of PALSAR imagery (polarity: HH; pixel spacing: 50 m) covering the middle and lower Mahakam area in the years 2007 and 2008 was collected. A land use/land cover map was available from a previous analysis of PALSAR imagery. To analyze Radar backscatter behavior for different land cover types, several regions of interest were selected, based on the land cover classes. A number of land cover classes (medium shrub, high shrub, fern/grass, and secondary forest) were found to be sensitive to flooding, whereas in some other classes (peat forest, riverine forest and tree plantation) backscatter signatures remained almost unchanged with flood inundation. Correlations between water level and radar backscatter of the regions of interest were used to distinguish between three types of flooding signal, viz. flooding of low vegetation, flooding of high vegetation, and the boundary shift of lakes. An analysis of the relationship between radar backscatter and water levels was carried out in each of the regions of interest. For lakes and shrub covered peatland in P1, where the range of water level variation was high, a good water level-backscatter correlation was obtained. In peat forest covered peatland in P2, subject to a small range of water level variation, water level-backscatter correlations were poor.