A critical grain size for Landsat ETM+ investigations into intertidal sediments: a case study of the Gomso tidal flats, Korea

Abstract

Remote sensing combined with a field survey is an effective method of monitoring intertidal sediments. The 0.0625 mm grain size has long been used as a critical one, and the percentage of grains larger than the critical grain size has been widely used to classify tidal flat deposits. The remote sensing community has applied this principle to interpret optical remote sensing images without a detailed understanding of the correlation between grain size and the remotely sensed data. This paper aims to establish a new critical grain size effective for Landsat enhanced thematic mapper plus (ETM+) data. The influence of water content and topography on the spectral reflectance is also investigated. Tests were undertaken in the Gomso Bay tidal flats, Korea. Critical grain size tests were conducted for particles measuring 0.0625 mm, 0.125 mm, and 0.25 mm, which revealed that a critical grain size of 0.25 mm rather than 0.0625 mm was a more effective standard. Surface water and interstitial water content were considered for their influence on optical reflectance, as Landsat ETM+ imaging is more responsive of the former. The 5, 7, and 4 ETM+ bands responded to surface water cover with a correlation coefficient of 0.90, 0.82, and 0.63, respectively. The interstitial water content related well to grain size with correlation coefficients of −0.59 for 0.0625 mm, −0.62 for 0.125 mm, and −0.81 for 0.25 mm. A linear model was derived for the ETM+ band 4 data based on a 0.25 mm critical grain size whist employing the principal component analysis method. The R 2 value obtained was 0.85. Sediments with grain sizes less than 20% of 0.25 mm, or larger grains, did not fit the model. Anomalies were interpreted as the effect of local topographic features. Employing Landsat data for intertidal sediment studies is indicated using the ETM+ band 4 and a target critical grain size of 0.25 mm. Surface sediment classification based on remotely sensed data must circumspectly consider an effective critical grain size, water content, local topography, and intertidal structures.