Daily NDVI relationship to clouds

1
Daily NDVI relationship to clouds

• TANG,Qiuhong
• The University of Tokyo
• IIS, OKIs Lab.

2
PUB

• Prediction in Ungauged Basins
• Relatively reliable Prediction in
  Un-ground-gauged Basins

Ungaused Basin B
Ungauged Basin A
3
PUB and observation

• If there isnt information of the basins, we
  cannot estimate hydrological response.
• UNGAUGED (Un-ground-gauged)
• Indirectly gauge (observation/information)
• Remote sensing Similar climate
  zone data
• Whats gauged? Whats ungauged? Whats PUB?
• PUB started with the geophysical science.

4
PUB model forcing data

• UNGAUGED
• Remote sensing Similar climate
  zone data

Available for RS Describe Land Surface DEM, Land
Cover,
Unavailable for RS Temperature, Wind speed Vapor
pressure
DEM geomorphologic factor to hydrological
response Land Cover vegetation phonology to
hydrological response Wind speed, vapor pressure
etc. regional climate forcing
RS
Interpolate
RS Unreliable, distributed data Ground
observation Reliable, point data
5
Land Cover NDVI data

• NDVI (NIR-Red) / (NIRRed)
• NDVI (normalized difference vegetation index)

Satellite NDVI
cloud
True NDVI
6
Cloud Index (NCI)
7
Cloud Index from Ground Observation

• Cloud Amount
• Sunshine Index n/N
• Where, n is duration of sunshine
• and N is maximum possible sunshine

8
Comparison (Satellite Ground Based Cloud Index)

• R-squared value between daily NCI values and the
  observed cloud amount (Left) and Sunshine time
  index (right). (1995-2000, 120 stations)

9
Comparison (Satellite Ground Based Cloud Index)

• RMSE value between daily NCI values and the
  observed cloud amount (Left) and Sunshine time
  index (right). (1995-2000, 120 stations)

10
Comparison (Satellite Ground Based Cloud Index)
Satellite data vs. Interpolate to all over the
River Basin
R-squared value between daily NCI values and the
observed cloud amount (Left) and Sunshine time
index (right). (1995-2000)
The poorest relationship in the Tibet Plateau
(red circle) might due to the distortion of
satellite scans.
11
Comparison (R2)
gt 80
gt 90
Distribution of R-squared values over the study
area. NCI and Cloud Amount Index (left), NCAI and
Sunshine Index (right).
12
Comparison (RMSE)

• The root-mean-square error (RMSE) associated with
  NCI and cloud amount ranges from 0.21 to 0.30
  with an averaged value of 0.25.
• For Sunshine time, RMSE ranges from 0.16 to 0.26
  with an averaged value of 0.20.

13
Comparison
USGS Land use
The daily NCI versus observed cloud amount and
SCAI relationships are strong and land cover
independent expect for water body
14
Conclusion

• The NCI and observed cloud index retain a high
  correlation coefficient suggesting that NCI may
  be useful for estimating clouds influence to
  solar radiation.

Satellite data (NDVI)
Ground based observation (cloud amount, sunshine)
Solar radiation
15
Thank you for your attention.
16
Solar and Scan Geometry
Sunshine NCI a
Sunshine NCI b