ESTIMATION OF RIVER ESTIMATION OF RIVER DISCHARGE WITH MODIS DISCHARGE WITH MODIS

IMAGESIMAGES

The University of Tokyo, Institute of Industrial Science

(IIS)

Kohei Hashimoto and Kazuo Oki

BackgroundBackground The transport of sediment by water in rivers plays an

important role in hydrology and the ecological functioning of river floodplains and deltas.

River discharge estimation is useful for demonstrate River discharge estimation is useful for demonstrate these information.these information.

In addition, if low-cost and continuous river discharge estimation becomes available,

It is expected that communities with fewer financial resources such as those in developing countries could easily use this information for constructing effective flood control infrastructure such as dams and levees

BackgroundBackground One measure generally used for recognizing sediment

transport is suspended sediment concentration (SSC).

Satellite remote-sensing technology is useful in tracking spatial and temporal variations in SSC

Many researches attempting remote sensing of SSC have constructed empirical relationships between reflectance and in situ measurements monitored at the same time in the field.

In addition, suspended sediment concentrations (SSC) normally show a robust empirical relationship with such hydraulic flow parameters as discharge and velocity. (e.g. L-Q equation)

These facts raise the possibility of directly estimating river discharge or velocity from reflectance remotely sensed by satellite.

Objective of studyObjective of study In this study, MODIS images are used for estimating river

discharge. MODIS images are taken every day and can be used for free.

Problems: Although it is an advantage of observation by MODIS that it can

take images once a day, the narrow width of target rivers means that the spatial resolution of this sensor is too low (250m in band 1) to extract reflectance data directly from the pixel corresponding to the place in situ discharge observations have taken place.

To solve the problem: In this study, we extracted MODIS band 1 reflectance values from

a pixel near the river mouth.

Objects: To carry out the regression analysis between the reflectance

values and the in situ discharge data that were gathered the same days the satellite images were taken.

To evaluate monthly and annual average discharge estimated from the MODIS reflectance with the regression model.

Study Area and Used DataStudy Area and Used Data(a)Naka river and (b) Monobe river of MODIS images

Point A:All in situ measurements of river discharge were corrected by MLIT. In both rivers, discharge data were taken once a day throughout 2004.

MLIT: Ministry of Land, Infrastructure, Transport and Tourism

Point B:In this study, we extracted MODIS band 1 reflectance values from a pixel near the river mouth

Used MODIS data at Naka and Used MODIS data at Naka and Monobe RiversMonobe Rivers

*Level 1B images were used*Level 1B images were used (radiometrically and geometrically corrected ).(radiometrically and geometrically corrected ).

In this study, we used simple atmospheric correction for L1B data.

As reference data, the offshore data, which is assumed as no reflectance data, were used.

The reference from original data is subtracted.

Collected area of Offshore data

spatial resolution is too low spatial resolution is too low (250m in band 1)…(250m in band 1)…

Extract the reflectance values Extract the reflectance values from a pixel near the river from a pixel near the river mouth at point B.mouth at point B.

apply single regression apply single regression analysis to reflectance values analysis to reflectance values and natural logarithm of in and natural logarithm of in situ discharge.situ discharge.

RMS = 213.0381 RMS = 199.4644

Results <Naka River Results <Naka River (2004)>(2004)>

RMS = 73.62537 RMS = 76.16806

Results <Monobe River Results <Monobe River (2004)>(2004)>

RMS = 157.4799RMS = 157.4799

RMS = 35.38131

Relationship between observed and estimated Relationship between observed and estimated monthly dischargemonthly discharge

Estimated monthly discharge(m3/s)Estimated monthly discharge(m3/s)

Ob

serv

ed

mo

nth

ly d

isch

arg

e(m

3/s

)

Ob

serv

ed

mo

nth

ly d

isch

arg

e(m

3/s

)

In this study, we extracted MODIS band 1 reflectance values from a pixel near the river mouth (point B in Figure),then applied single regression analysis to reflectance values and the in situ discharge data that were gathered the same days the satellite images were taken.

Our method proved effective to estimate the discharge with MODIS data at narrow river.

ConclusionsConclusions

Thank you for your Thank you for your attention.attention.