Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Global estimation of the
1-m soil water content using microwave measurements from
satellite
R. Lindau & C. Simmer
University of Bonn
Soil Moisture Measurements
50 stations in the former SU provide:
• Soil moisture in the uppermost meter
• Total number: 17748
• Frequency: 10-daily
• Period: 1952 - 1985
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
ANOVA
Illustration of decomposition
Total variance External variance Internal variance
= Variance between + Mean variance the means of the within the
subsamples subsamples
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
ANOVA Soil Moisture
Soil moisture within the uppermost meter measured at 48 stations in the former Soviet Union
Total number of observations: 7009
Total variance: 10682 mm
2Varian ce in mm
2Numbe r of bins
Error of the
total mean
Seemin g externa
l varianc
e
Error of external
means
Internal variance
True externa
l varianc
e
Relative external variance
Annual
Cycle 36 2 388 51 1034
3 338 3.16
%
Interstati
on 48 2 913
3 10 1558 912
3 85.40
%
Interannu
al 8 2 39 12 1065
4 26 0.25
%
Local longtime means
single cumulative
Climatolog. rain 58.6 58.6
Soil texture 0.5 69.0
Vegetation 37.7 72.8
Terrain slope 2.8 73.0
73% of the soil moisture variance
is explained by four parameters :
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Temporal Anomalies
In a second step 10 Ghz measurements are used to retrieve the remaining
temporal part of the variance.
A correlation of 0.609 is
attained .
Radio Frequency Interference
Time series of 6 GHz
brightness temperature from SMMR in France
Until 1981 the normal annual cycle is found.
After 1981 the 6 Ghz signal is
completely unusable due to
noise.
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Westward intensification
The scatter is low in Sibiria, increases westward and
reaches maximum values near St.
Petersburg
Two-step Retrieval
Longterm local mean of soil moisture
• Longterm mean of precipitation
• Soil texture
• Vegetation density
• Terrain slope
Anomaly against the longterm local mean
• Brightness temperatures at 10 Ghz
• Anomalies of rain and air temperature
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Soil Moisture
C lim at o lo g ic a l m ea n Temporal
anomalies
+
Algorithm test Illinois
Verification by independent measurements from Illinois
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Algorithm test China
•
Underestimation of the mean soil moisture by
74mm. (200 mm / 274 mm)
•
Correlation low with 0.523
•
Mainly due to four desert
stations.
10-years mean soil moisture
REMO Algorithm
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Oder Catchment
REMO
• Stronger interannual variability
• Stronger annual cycle
• Annual cycle
delayed
AMSR Data
August
• Humid winter in SH
• Constant moisture in tropics
• Summer monsoon in NH
• Dry-out in NH mid-latitudes
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Annual Cycle 2003
• Tropics:
constant moisture
• India:
Monsoon increase
• Europe:
Summer dry-out
But spatial differences dominate anyhow
Brazzaville Bombay
Budapest
Conclusions
Pure spatial variance dominates the soil moisture variability
Two step algorithm for soil moisture
• Local longterm mean
• Anomalie (temporal variation at each site)
10 GHz channel is used, because 6 GHz is disturbed by RFI
Verification of the algorithm by Illinois and Chinese data
Application
• Validation of REMO (1979-1988) with SMMR
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005
Chinese Soil Moisture Data
40 Stations
Measurements of 1m soil
moisture for the period 1981 -
1999
Comparison of monthly maps
REMO Dez 1979 Algo
REMO Sep 1985 Algo
Measuring Soil Water Contents at Different Scales – FZ Jülich – 17th November 2005