The impact of spatial resolution on inland water quality monitoring from space

Abstract Remote sensing of inland waters can provide timely and global water quality information to a wide variety of stakeholders. One of the parameters that determines the feasibility of using optical space-based instruments for monitoring inland waters is the ground sampling distance (GSD), defined as the width of a pixel projected on the Earth’s surface. We assume that to analyze a body of water with optical imagery, its characteristic width must be larger than 3 times the GSD to obtain an ‘unmixed’ pixel that doesn’t contain signal from the adjacent land. Here we obtain the size distribution of river lengths, river areas, and lake areas—as a function of width—for rivers and lakes in the Western United States (US) and in Australia. We base this analysis on the Surface Water and Ocean Topography River Database (SWORD) and HydroLAKES databases, extrapolated to 5 m-wide features. We show that the fraction of river length and river area larger than a certain width increases sharply as the width decreases, indicating that even small decreases in the GSD result in significant increases in the number of bodies that can be surveyed. On the other hand, the distribution of lake areas shows a ‘knee’ at around 400 m, indicating that gains from GSDs smaller than 130 m will be modest. We found that a satellite instrument with a GSD capability of 18 m can provide coverage of 4.4% of total river lengths, 38% of total river area, and 94% of total lake area within the study areas. We argue that decreasing the GSD incurs penalties associated with loss of signal-to-noise, larger instrument, smaller swath, and longer revisit times.