linking modis imagery with turbidity and tss time series data generation over lake tana, ethiopia
DESCRIPTION
LINKING MODIS IMAGERY WITH TURBIDITY AND TSS TIME SERIES DATA GENERATION OVER LAKE TANA, ETHIOPIA. Essayas K. Ayana* 1,3 , William D. Philpot 2 and Tammo S. Steenhuis 1,3 1 Department of Biological and Environmental Engineering, Cornell University, Ithaca, NY, USA - PowerPoint PPT PresentationTRANSCRIPT
•Moderate Resolution Imaging Spectroradiometer (MODIS) Moderate Resolution Imaging Spectroradiometer (MODIS)
250m resolution images used (250m resolution images used (LPDAAC 2010))
•Bulk water samples and GPS coordinates of sampling locations Bulk water samples and GPS coordinates of sampling locations
are collected during the satellite overpass time over are collected during the satellite overpass time over Lake TanaLake Tana
•Total suspended solids and turbidity measuredTotal suspended solids and turbidity measured
•Reflectance values of the sampling locations are extractedReflectance values of the sampling locations are extracted
•Multiple regression analysis was performed on two Multiple regression analysis was performed on two set of set of
measurements (taken 27 November 2010 and 13 May 2011)measurements (taken 27 November 2010 and 13 May 2011)
•Third data set (collected November 7, 2011) is used for Third data set (collected November 7, 2011) is used for
validationvalidation
Various conservation programs are being implemented in Various conservation programs are being implemented in
developing countries with the potential benefit of reduced developing countries with the potential benefit of reduced
sediment inflow into fresh water lakes. These claims are hard to sediment inflow into fresh water lakes. These claims are hard to
verify due to prohibitive costs of continuous stream sediment verify due to prohibitive costs of continuous stream sediment
load sampling and analysis. load sampling and analysis. RRemote sensing can potentially aid emote sensing can potentially aid
in monitoring sediment concentration. in monitoring sediment concentration. Various regression models Various regression models
have been developed using remotely sensed images (have been developed using remotely sensed images (
Hu et al., 2004Hu et al., 2004, Chen et al., 2007, Wang and Lu, 2010). However . However
these regression models are not universal and hence cannot be these regression models are not universal and hence cannot be
applied to estimate the same parameters elsewhere. The applied to estimate the same parameters elsewhere. The
relationship between in situ water quality parameters and their relationship between in situ water quality parameters and their
corresponding reflectance measurement are almost always site-corresponding reflectance measurement are almost always site-
specific specific ((Liu et al. 2003))..
LINKING MODIS IMAGERY WITH TURBIDITY AND TSS TIME SERIES DATA GENERATION OVER LAKE TANA, ETHIOPIA
INTRODUCTION
Essayas K. Ayana*Essayas K. Ayana*1,31,3, , William D. PhilpotWilliam D. Philpot22 and Tammo S. Steenhuis and Tammo S. Steenhuis1,31,3
11Department of Biological and Environmental Engineering, Cornell University, Ithaca, NY, USADepartment of Biological and Environmental Engineering, Cornell University, Ithaca, NY, USA22Department of Civil and Environmental Engineering, Cornell University, Ithaca, NY, USADepartment of Civil and Environmental Engineering, Cornell University, Ithaca, NY, USA
33School of Civil and Water Resources Engineering, Bahir Dar University, Bahir Dar, EthiopiaSchool of Civil and Water Resources Engineering, Bahir Dar University, Bahir Dar, Ethiopia
1. CHEN, Z., HU, C. & MULLER-KARGER, F. 2007. Monitoring turbidity in Tampa Bay using MODIS/Aqua 250-m imagery. Remote Sensing of Environment, 109, 207-220.
2. GETIS, A., and ORD, J. K., 2010, The analysis of spatial association by use of distance statistics. Perspectives on Spatial Data Analysis, 127-145.
3. HU, C., CHEN, Z., CLAYTON, T. D., SWARZENSKI, P., BROCK, J. C. & MULLER–KARGER, F. E. 2004. Assessment of estuarine water-quality indicators using MODIS medium-resolution bands: Initial results from Tampa Bay, FL. Remote Sensing of
Environment, 93, 423-441.
4. LPDAAC, 2010, Surface Reflectance Daily L2G Global 250m.
5. LIU, Y., ISLAM, M. A., and GAO, J., 2003, Quantification of shallow water quality parameters by means of remote sensing. Progress in Physical Geography, 27, 24-43.
6. WANG, J. J. & LU, X. 2010. Estimation of suspended sediment concentrations using Terra MODIS: An example from the Lower Yangtze River, China. Science of the Total Environment, 408, 1131-1138. *[email protected]
REFERENCES
•water reflectance in the NIR band fits best resulting in a linear water reflectance in the NIR band fits best resulting in a linear correlation with measured TSS (R2=0.95) and turbidity correlation with measured TSS (R2=0.95) and turbidity (R2=0.89)(R2=0.89) •Secchi depth correlate with NIR reflectance exponentially Secchi depth correlate with NIR reflectance exponentially (R2=0.74) (R2=0.74)
•The root mean square error (RMSE) in using these equations The root mean square error (RMSE) in using these equations was 16.5 mg lwas 16.5 mg l-1-1 ,15.6 NTU, and 0.11 m to measured TSS, ,15.6 NTU, and 0.11 m to measured TSS, turbidity and Secchi depth respectivelyturbidity and Secchi depth respectively
•TThe Getis–Ord Gi* statistic data mining technique (he Getis–Ord Gi* statistic data mining technique (Getis and Ord 2010) is used to ) is used to
select the representative “muddiest pixel”select the representative “muddiest pixel”
•Regression equation established for TSS is applied on the “muddiest pixel” to Regression equation established for TSS is applied on the “muddiest pixel” to
estimate the maximum TSS for the given dayestimate the maximum TSS for the given day
•A 10 years (2000-2009) TSS time series data is generated A 10 years (2000-2009) TSS time series data is generated
•A SWAT model is calibrated and validated using the MODIS image generated 10 A SWAT model is calibrated and validated using the MODIS image generated 10 years TSS time series.years TSS time series. •Nash–Sutcliffe efficiencies of 0.39 for calibration period and 0.32 for validation. Nash–Sutcliffe efficiencies of 0.39 for calibration period and 0.32 for validation.
TSS TIME SERIES GENERATION
•Single band relations are found to be most accurate to measure turbidity, TSS and Single band relations are found to be most accurate to measure turbidity, TSS and
Secchi depthSecchi depth•MODIS images MODIS images are a potential cost effective tool to monitor suspended sediment to monitor suspended sediment
concentration and to obtain past history of concentrations which will help to evaluate concentration and to obtain past history of concentrations which will help to evaluate
the effect of best management practices the effect of best management practices •Results showed that MODIS images are not sensitive enough to detect turbidity Results showed that MODIS images are not sensitive enough to detect turbidity
variations below 60 NTU. variations below 60 NTU.
APPLICATION FOR WATER QUALITY MODELING
CONCLUSION
METHODS
RESULTS
Acknowledgement
USDA-SRE and the HED provided partial support for this study.
NSE=0.39NSE=0.32