Remote Sensing of Leaf, Canopy, and Vegetation Water Contents for Satellite Environmental Data Records - Satellite-based Applications on Climate Change - page 334

created to surround the data in order to determine the amount of water stress of one
pixel compared to the other pixels in the scatter plot (Ghulam et al. 2008 ). Another
line of research is the development of multiband indices, such as the ratio of WI/
NDVI (Pe˜uelas et al. 1997 ), and the ratio of MSI with the Normalized Multiband
Drought Index (Wang and Qu 2007 ).
The Soil Moisture Experiments 2004 and 2005 were conducted in Southeastern
Arizona and Central Iowa, respectively, to test microwave radiometer retrievals of
soil moisture. During these experiments, extensive data for CWC were measured
and related to Landsat NDII (Yilmaz et al. 2008a , b ) and MODIS and AVIRIS data
(Cheng et al. 2006 , 2008 ). A single linear relationship was found between NDII and
CWC (Fig. 20.3 ); RMSE of the regression equation was 0.091 kg m 2 (Hunt et al.
2011b ). F´ret et al. ( 2011 ) surveyed a large number of leaves from different species
and obtained a median LWC equal to 0.11 kg m 2 . The RMSE from Fig. 20.3 is
83% of the median LWC value from F´ret et al. ( 2011 ), so NDII can estimate LAI
to an accuracy of 0.83 for a land-cover type with known LWC. Asner et al. ( 2004 )

used NASA Earth-Observing-1 Hyperion data to show there was a significant
reduction in CWC during an imposed drought from rainfall exclusion in an Amazon
forest. However, Asner et al. ( 2004 ) also showed that LAI decreased, so the amount
of CWC decrease attributed to either water stress or LAI is not known.
The individual data sets are not significantly different from the overall regression
equation. Comparison of the RMSE in Fig. 20.3 to the median LWC value (F ´ ret
et al. 2011 ) indicates that there may be a limit to the accuracy of CWC achievable
from foliar water indices. Therefore, in the next three sections, we discuss three
applications for remotely sensed CWC in order to determine the accuracy
requirements for a CWC data product.

20.6 Estimation of Vegetation Water Content for Soil
Moisture Retrievals

VWC is one of the important parameters required for the retrieval of soil moisture
content using microwave sensors (Jackson et al. 2004 ; Entekhabi et al. 2010 ). VWC
may be estimated directly from the microwave data, but having independent data of
VWC from AVHRR/3, MODIS, or VIIRS would help to improve accuracy of the
retrieved soil moisture content. The central problem is that VWC is not estimated
directly from vegetation moisture indices such as NDII, because of the amount of
water in plant stems and other organs (Fig. 20.3 ).
Two of the Soil Moisture Experiments, 2002 and 2005, were conducted in
central Iowa, USA, where 70% of the area are agricultural fields planted in either
corn or soybean (Alan K. Stern, personal communication). Because stems support
the leaves, biophysical allometric relationships should exist between the amount of
water in stems and the amount of water in the foliage (Yilmaz et al. 2008b ). Unlike
the relationship between NDII and CWC (Fig. 20.3 ), corn and soybeans have very
different relationships between NDII and VWC, because proportionally more stem
mass is required to support the larger leaves of corn (Fig. 20.4a ).