Ecological Engineering 21 (2003) 153–163 Hyperspectral reflectance of emergent macrophytes as an indicator of water column ammonia in an oligohaline, subtropical marsh David Rogers Tilley a,∗ , Muneer Ahmed b , Ji Ho Son c , Harish Badrinarayanan b a University of Maryland, Biological Resources Engineering, 1449 Animal Science/Agiculture Engineering. Building., College Park, MD 20742, USA b Department of Environmental Engineering, Texas A&M University, Kingsville, TX 78363, USA c Division of Environmental Systems Engineering, Pukyong National University, Pusan 608-737, South Korea Received 11 April 2003; received in revised form 26 September 2003; accepted 15 October 2003 Abstract Narrow spectral band reflectance indices of two emergent macrophytes (Typha latifolia and Borrichia frutescens) were found to correspond to changes in water column total ammonia (TA = NH 3 + NH + 4 ) of a constructed oligohaline (2–5 ppt) subtropical marsh (8ha) designed as a shrimp farm treatment wetland in south Texas, USA, near the western Gulf of Mexico. A handheld hyperspectral radiometer measured plant leaf reflectance across the visible (VIS, 300–700 nm) and near-infrared (NIR, 700–1100 nm) spectrum during the mid-growing season (July–August). Reflectance of the two species and water quality (TA; total phosphorus (TP); salinity, S) were measured along a dynamic nutrient gradient from wetland inlet to outlet. Mean TA concentration was 1.22 mg N l -1 ranging between 0.92 and 1.71 mg N l -1 . Salinity fluctuated between 2.5 and 4.5 ppt, while total phosphorus was between 0.01 and 0.86 mg P l -1 . T. latifolia, the more dominant wetland cover: (a) reflected 1–5% less in the VIS compared to B. frutescens, (b) generally reflected 7–8% more in the NIR, (c) exhibited an average red-edge (RE) (718 nm) 15 nm greater than B. frutescens, and (d) had an average leaf normalized difference vegetation index [NDVI = (R 774 - R 681 )/(R 774 + R 681 ), where Rλ is reflectance at wavelength λ] of 0.82, versus 0.65 for B. frutescens. A multivariate linear regression equation containing photochemical reflectance index (PRI) and RE for T. latifolia explained 93% of TA variation (P< 0.001) when 0.10 < TP < 0.20 mg P l -1 . For the entire range of water quality conditions, an equation containing PRI, RE and R 493 /R 678 for T. latifolia explained 72% of TA variation (P< 0.001). These results indicate that hyperspectral radiometry can quantify wetland ammonia concentrations in a treatment wetland, which could lead to development of remote techniques for continuously monitoring treatment effectiveness. © 2003 Elsevier B.V. All rights reserved. Keywords: Water quality; Remote sensing; Hyperspectral radiometry; Nitrogen; Treatment wetlands ∗ Corresponding author. Tel.: +1-301-405-8027; fax: +1-301-314-9023. E-mail address: dtilley@umd.edu (D.R. Tilley). 1. Introduction Assessing the effectiveness of constructed wetlands to reduce ammonia levels is one of several important criteria for judging their effectiveness in nutrient re- tention and overall water quality improvement (Kadlec 0925-8574/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ecoleng.2003.10.004