The model domain (Figure 2.1) was divided into 81 x 66 cells with 4 km resolution using the WPS program. All weather forcing data and data for the initial conditions were interpolated spatially into 4 km and temporally into 1 hour in the HRLDAS platform. Interpolation of the parametric fields was carried out by WPS
program. Resolution on the order of 2 km was recommended by a previous study (Sridhar et al., 2003) in order to best represent the highly heterogeneous regions. However, since the resolution of input weather data is coarse (32 km for NARR), going to a resolution higher than 1 km may not be of much help. The resolution of 4 km was selected considering the computational difficulties arising due to long-term model runs, data processing, and storage.
The total depth of the soil column was 2 m with the root zone depths read from the vegetation look-up table. HRLDAS v3.1 was used to run Noah LSM in an uncoupled mode with the changes explained in Section 3.1.2. Four experiments were performed. Long-term simulations were done for 30 years with 4 km spatial resolution and with the irrigation scheme in it. The Noah LSM was initialized at 00 hour in 1 January 1979. The first 24 months were treated as the spin-up period. Model simulations spanned through 31 December 2010 in 1 hour time steps. Vegetation information is static for the whole simulation period. Short-term simulations were done from 1 January 2008 through 31 December 2010 in two ways, leaving 2008 simulation year as the spin-up period. The first experimental simulation was with the irrigation scheme and the spatial resolution of
4 km. The second experiment was without irrigation but with a 4 km resolution in order to access the effects of irrigation. The last experimental set up was with an irrigation scheme, but from 1 January 2009 through 31 December 2010. Only 2009 simulation year was excluded as the spin-up period in order to evaluate suitable spin-up time for this region.
Figure 3.9. Outlines of the 3 WRF Domains (Outermost Domain Is the Figure Box)
3.4.2 WRF
The output from WRF V3.2.1 was used in this study with three nested domains. WRF output was obtained from a parallel project carried out at Boise State University. Three model domains were created with the innermost domain being the same as the
domain used in Noah LSM (Figure 3.9). The outermost domain is 36 km in resolution with 98 rows and 89 columns of grid cells. The middle domain has 102 x 114 grid cells of 12 km and it helps to resolve orographic effects. The innermost domain is 4 km and consists of 81 x 61 grid cells. All three domains have 38 vertical grid cells of varying heights, with the lowest having an average height of 36 m. Due to the extensive
computational power required, WRF results were obtained only for a selected time period of 1 March 2010 to 30 September 2010.
3.5 Validation
Both inputs and outputs of the model were subjected to validation. Model forcing data extracted from NARR was verified using the data collected from field observation sites at HL and RR and AgriMet weather stations. This kind of verification is necessary since the NARR data used to force the models are also modeled data. It is important to know if the errors in the input data are significant enough to propagate in the model output. When Chen et al. (2007) verified their input data against IHOP_2002 field observations, they found that NCEP stage-IV rainfall was slightly overestimated in dry regions and underestimated over the transitional and wet regions. While the largest error was found in satellite-derived solar radiation, other atmospheric forcing data obtained from NCEP EDAS were accurate with low errors.
Model output was validated in two ways, as spatial data and point-scale data. For point-scale output, data for the station locations were extracted. Spatial validation was
done by comparing the model-simulated ET estimates with METRIC images for the years 1996, 2000, 2002, and 2006.
3.5.1 Field Observations
The surface energy balance components for the model, latent heat, sensible heat, ground heat, and net radiation, were validated against fluxes measured at HL and RR sites in 2010. The input variables of air temperature and precipitation were validated by comparing them with the field observations. These variables are important drivers of land-surface processes and have been measured at all observation sites.
3.5.2 METRIC Images
METRIC (Satellite-Based Energy Balance for Mapping Evapotranspiration with Internalized Calibration) is a satellite-based, surface energy balance model to predict evapotranspiration. It uses an image processing tool, such as ERDAS, to link various modules (Allen et al., 2007a). It calculates actual ET as a residual of energy balance using satellite images containing both shortwave and thermal information. It is based on another satellite-based model, SEBAL (Bastiaanssen et al., 1998). The intention was to produce higher resolution spatial maps of ET with more accuracy compared to other models. The resolution of the ET map depends on the resolution of the satellite images. METRIC computes net radiation, soil heat flux and sensible heat flux, and ET, then is calculated as the residual according to the energy balance equation. It has been designed to be calibrated internally using ground-based alfalfa reference ET, calculated from
hourly weather data. The instantaneous ET computed for the time of the satellite image can be converted to daily ET or seasonal ET using a reference ET fraction (ETrF). It assumes the ETrF at the image time is the same as the average ETrF over the day (24 hour) in calculating the daily ET. METRIC provides useful information of ET, which can be used in the field for various purposes. Monitoring water-rights compliance and aquifer depletion in Idaho, mapping ET from agricultural and riparian vegetation in New Mexico, and assessing irrigation adequacy and salinity management in California are examples of usages of METRIC in the field (Allen et al., 2007b).
CHAPTER FOUR: VALIDATION AND RESULTS OF IRRIGATION IN THE MODEL
4.1 Evaluation of NARR Data with Field Observations