|Sudduth, Kenneth - Ken|
Submitted to: Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 3/1/2006
Publication Date: 3/1/2006
Citation: Sudduth, K.A., Jang, G., Lerch, R.N., Sadler, E.J. 2006. Hyperspectral reflectance estimation of reservoir water quality. In: Yang, C., Everitt, J.H. (editors), Proceedings 20th Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment. October 4-6, 2005, Weslaco, Texas. 2006 CDROM.
Interpretive Summary: Impairment of the environment by agricultural activities is an ongoing concern of agriculturalists, environmentalists, and the general public. To assess the effects of existing and new agricultural practices over large watersheds, we need to find more efficient ways to measure various water quality variables such as chlorophyll content, turbidity (cloudiness of the water, caused mainly by sediment), and nutrients such as nitrogen and phosphorus. One method that can collect data quickly over large areas and has been used successfully for estimating water quality is remote sensing. In this study, we evaluated the use of remote sensing to estimate chlorophyll, turbidity, and nutrients in Mark Twain Lake, a large manmade reservoir in northeast Missouri. Using data collected on a single date in 2004, we obtained good relationships between remote sensing data in the visible and near-infrared wavelength ranges and the water quality variables. Using data collected over three dates in 2004 and 2005, the relationships were not as good, but could still be useful. To estimate chlorophyll and nutrients, it was important to use hyperspectral remote sensing data, which consists of measurements at many (in this case more than 100) wavelengths. We were able to estimate turbidity with data at only 4 wavelengths (so-called multispectral data). This research demonstrates that our approach to using remote sensing to estimate water quality could provide good results. The best estimates required relationships for each measurement date, but using a single relationship across multiple dates might be possible if slightly less accuracy was acceptable. These findings will help us and other researchers develop ways to more efficiently estimate differences in water quality across large watersheds and within reservoirs.
Technical Abstract: Remotely sensed estimates of water quality parameters would facilitate efforts in spatial and temporal monitoring. In 2004, we collected hyperspectral water reflectance data with airborne and ground-based sensing systems for multiple arms of Mark Twain Lake, a large man-made reservoir in northeast Missouri. In 2005, we obtained ground-based data at two additional dates. Water samples were also collected and analyzed in the laboratory for chlorophyll, nutrients, and turbidity. Previously reported reflectance indices and wavelength-selection (i.e., stepwise multiple regression) methods were used to develop relationships between spectral and water quality data. Within a single measurement date, all measured water quality parameters were strongly related (R2 > 0.6) to reflectance data. When ground-based data were pooled across dates, relationships were generally less predictive (R2 > 0.5). Stepwise regression estimates were generally better than those obtained with previously developed narrow-band indices, particularly across multiple dates. Wide-band, multispectral reflectance was strongly related only to turbidity and those other parameters highly correlated to turbidity. Regression estimates developed with calibration data from a single date were of variable accuracy when applied to data obtained on other sampling dates. For best accuracy, laboratory water quality data used to calibrate the remote sensing relationships should span the range of variability in the parameters of interest. Thus, collection of calibration samples at each sensing date would be a reasonable way to insure the accuracy of remote sensing estimates of water quality.