202
Multiscale Hydrologic Remote Sensing: Perspectives and Applications
thus most of them are summarized in the validation section (Table 9.2). Some additional studies include analyses of the relationships between snow cover and terrain
and hydrometeorological characteristics (Poon and Valeo 2006; Tong et al. 2009a,b;
Jain et al. 2009; Xu and Li 2010), evaluation of the effects of cloud and forest masking on snow cover monitoring (Poon and Valeo 2006; Zhang et al. 2010), assessment
of different snow cover–related characteristics such as snow cover onset and melt
days or snow cover duration (Wang and Xie 2009; Gao et al. 2011), support for SWE
estimation (Drusch et al. 2004; Durand et al. 2008; Bavera and de Michele 2009;
Bocchiola and Groppelli 2010; Harshburger et al. 2010), and fractional and subpixel snow cover mapping (e.g., Kaufman et al. 2002; Salomonson and Appel 2006;
Dozier et al. 2008; Sirguey et al. 2008, 2009).
One of the main interests, from the hydrologic perspective, is the potential of
MODIS images for assisting in stream flow simulation and prediction. Related studies either implement MODIS SCA directly as a model input or assimilate MODIS
data into hydrologic model simulation, calibration, or validation. Rango et al. (2003,
2004) and Lee et al. (2005) used the daily MODIS snow cover product as an input
to simulate stream flow in the Rio Grande Basin using the snowmelt runoff model
(SRM). They found that snow depletion curves derived from MODIS enabled efficient stream flow simulations and forecasts, with the stream flow simulation accuracy
(coefficient of determination) ranging from 0.768 (Rango et al. 2003) to 0.89 (Lee
et al. 2005) in the Upper Rio Grande Basin and a somewhat lower accuracy of 0.57
in the smaller Rio Ojo Basin (Lee et al. 2005). An even lower accuracy (0.43) was
reported by Nitin (2004) in the Elaho Basin of British Columbia; however, the forecasts did not use direct observations of climate variables. The 8-day MODIS product
helped predict the general seasonal trend in snowmelt runoff but not the daily stream
flow variations. Wang et al. (2010) used the SRM to simulate the annual potential
snowmelt in the period 2000–2008. They reported a negative relationship between
annual air temperature and MODIS-derived SCA proportion and an increasing trend
of annual air temperatures and SCA since 2000.
Implementations of MODIS data for calibrating and validating watershed hydrologic models indicated that MODIS snow data generally improved the snow cover
simulations and did not change much the model performance with respect to runoff. For example, Rodell and Houser (2004) and Andreadis and Lettenmaier (2006)
assimilated MODIS snow cover observations into the snow water storage of a hydrologic model and assessed the assimilation efficiency against snow ground observations. They found that snow assimilation resulted in more accurate snow coverage
simulations and compared more favorably to ground snow measurements. Déry et al.
(2005) used the MODIS snow areal depletion curves to constrain the subgrid-scale
parameterization of the catchment-based land surface model (CLSM) and found
improvements in the timing and the amount of snow cover ablation and snowmelt
runoff. Udnaes et al. (2007), Parajka and Blöschl (2008b), and Şorman et al. (2009)
examined the potential of MODIS data for calibrating and validating a conceptual
hydrologic model. Their results indicated that the use of the MODIS snow cover
improved the snow model performance and also slightly improved the runoff model
efficiency. Parajka and Blöschl (2008b), for example, showed that, in a verification
mode, the median (Nash–Sutcliffe) model efficiency of runoff over 148 catchments
Multiscale Hydrologic Remote Sensing: Perspectives and Applications
thus most of them are summarized in the validation section (Table 9.2). Some additional studies include analyses of the relationships between snow cover and terrain
and hydrometeorological characteristics (Poon and Valeo 2006; Tong et al. 2009a,b;
Jain et al. 2009; Xu and Li 2010), evaluation of the effects of cloud and forest masking on snow cover monitoring (Poon and Valeo 2006; Zhang et al. 2010), assessment
of different snow cover–related characteristics such as snow cover onset and melt
days or snow cover duration (Wang and Xie 2009; Gao et al. 2011), support for SWE
estimation (Drusch et al. 2004; Durand et al. 2008; Bavera and de Michele 2009;
Bocchiola and Groppelli 2010; Harshburger et al. 2010), and fractional and subpixel snow cover mapping (e.g., Kaufman et al. 2002; Salomonson and Appel 2006;
Dozier et al. 2008; Sirguey et al. 2008, 2009).
One of the main interests, from the hydrologic perspective, is the potential of
MODIS images for assisting in stream flow simulation and prediction. Related studies either implement MODIS SCA directly as a model input or assimilate MODIS
data into hydrologic model simulation, calibration, or validation. Rango et al. (2003,
2004) and Lee et al. (2005) used the daily MODIS snow cover product as an input
to simulate stream flow in the Rio Grande Basin using the snowmelt runoff model
(SRM). They found that snow depletion curves derived from MODIS enabled efficient stream flow simulations and forecasts, with the stream flow simulation accuracy
(coefficient of determination) ranging from 0.768 (Rango et al. 2003) to 0.89 (Lee
et al. 2005) in the Upper Rio Grande Basin and a somewhat lower accuracy of 0.57
in the smaller Rio Ojo Basin (Lee et al. 2005). An even lower accuracy (0.43) was
reported by Nitin (2004) in the Elaho Basin of British Columbia; however, the forecasts did not use direct observations of climate variables. The 8-day MODIS product
helped predict the general seasonal trend in snowmelt runoff but not the daily stream
flow variations. Wang et al. (2010) used the SRM to simulate the annual potential
snowmelt in the period 2000–2008. They reported a negative relationship between
annual air temperature and MODIS-derived SCA proportion and an increasing trend
of annual air temperatures and SCA since 2000.
Implementations of MODIS data for calibrating and validating watershed hydrologic models indicated that MODIS snow data generally improved the snow cover
simulations and did not change much the model performance with respect to runoff. For example, Rodell and Houser (2004) and Andreadis and Lettenmaier (2006)
assimilated MODIS snow cover observations into the snow water storage of a hydrologic model and assessed the assimilation efficiency against snow ground observations. They found that snow assimilation resulted in more accurate snow coverage
simulations and compared more favorably to ground snow measurements. Déry et al.
(2005) used the MODIS snow areal depletion curves to constrain the subgrid-scale
parameterization of the catchment-based land surface model (CLSM) and found
improvements in the timing and the amount of snow cover ablation and snowmelt
runoff. Udnaes et al. (2007), Parajka and Blöschl (2008b), and Şorman et al. (2009)
examined the potential of MODIS data for calibrating and validating a conceptual
hydrologic model. Their results indicated that the use of the MODIS snow cover
improved the snow model performance and also slightly improved the runoff model
efficiency. Parajka and Blöschl (2008b), for example, showed that, in a verification
mode, the median (Nash–Sutcliffe) model efficiency of runoff over 148 catchments