This document is also available in PDF: SWAT_parameter_sensitivity.pdf

Copyright reserved to Jaehak Jeong

1.      ALPHA_BF

 ALPHA_BF is a base flow recession constant which is a direct index of groundwater flow response to changes in recharge. Values vary from 0.1-0.3 for land with slow response to 0.9-1.0 for land with a rapid response. Calibration range is 0.001 ~ 1.0.

Model result changes as ALPHA_BF value changes as shown in Table 1. Figure 1 shows the difference in hydrograph at the outlet of LGA watershed with different alpha_bf values. As one can see, a greater alpha_bf value results in higher discharge amount in the hydrograph including peak flows.

  Table 1. Example of ALPHA_BF  calibration result

  Figure 1. Downstream hydrograph with different alpha_bf values

2.      CN2
CN2 is a moisture condition II curve number. SWAT reads this value from *.mgt files. An HRU with a small CN2 value have more infiltration than one with a large value. An impervious surface has CN2=100. CN2 is calibrated generally within the range of -4 ~ +4.

Model result changes CN2 value changes as shown in Table 2. Figure 2 shows the difference in hydrograph at the outlet of LGA watershed with different CN2 values. As expected, a greater CN2 value results in higher discharge amount in the hydrograph including peak flows.

  Table 2. Example of CN2 calibration result

  Figure 2. Downstream hydrograph with different CN2 values

3.      CH_K

CH_K is effective hydraulic conductivity of channel (mm/hr). This parameter is critical in computing transmission losses to channel bed. Therefore, larger ch_k value will trigger more loss of water to the river bed. CH_K value varies from 0.025 to larger than 127 depending on bed material group number.

Model result changes CH_K value changes as shown in Table 3. Figure 3 shows the difference in hydrograph at the outlet of LGA watershed with different CH_K values. As expected, a greater CH_K value results in draw down in the hydrograph including peak flows due to increased transmission losses.

  Table 3. Example of CH_K calibration result

  Figure 3. Downstream hydrograph with different CH_K values

4.      CH_N

CH_N is Manning’s roughness coefficient for the main channel. Chow (1959) suggests a range of 0.03 ~ 0.07 for mountain streams.

Since this parameter is related to channel flow, a change in CH_N does not affect WATYLD amount noticeably (see table 4), however, downstream hydrograph is influenced by this parameter: larger CH_N results in higher peak flow as shown in Figure 4.

    Table 4. Example of CH_N calibration result

  Figure 4. Downstream hydrograph with different CH_N values

5.      MSK_CO2

MSK_CO2 is a weighting factor for influence of low flow on storage time constant value paired with MSK_CO1. We found that SWAT is sensitive to MSK_CO2 value but to MSK_CO1. MSK_CO2 is  found in *.bsn file. Typical range for calibration is 0 ~ 10.

Since this parameter is related to channel flow (especially when using Muskingum routing method), a change in MSK_CO2 does not affect WATYLD amount noticeably (see table 5), however, downstream hydrograph is greatly influenced by this parameter: larger MSK_CO2 value results in loweer peak flow as shown in Figure 5.

  Table 5. Example of MUSK_CO2 calibration result

  Figure 5. Downstream hydrograph with different MUSK_CO2 values

6.      ESCO

ESCO is soil evaporation compensation coefficient. ESCO is negatively proportional to the maximum evaporation in a nonlinear fashion. Therefore, larger value results in less ET and thus, larger WTRYLD. A range of 0.001 ~ 1.0 is recommended.

As mentioned earlier, a large ESCO value results in low ET, but high WTRYLD value as shown in Table 6. This, however, barely affects downstream hydrograph as shown in Figure 7.

  Table 6. Example of ESCO calibration result

  Figure 6. Downstream hydrograph with different ESCO values

7.      GW_DELAY

GW_DELAY, delay time for aquifer recharge (days), accounts for the time delay in aquifer recharge once the water exits the soil profile. Literature recommends a range of 0.001 ~ 100 for calibration.

To this parameter, a marginal sensitivity is observed as shown in Table 7. A small GW_DELAY value results in slightly higher peak flow in the hydrograph as shown in Figure 7.

  Table 7. Example of GW_DELAY calibration result

  Figure 7. Downstream hydrograph with different GW_DELAY values

The calibration parameters listed above are the most sensitive ones in LGA watershed modeling. Since this watershed is small (A=2km²), result can be different when a large watershed model is calibrated.

Main channel capacity (flow rate, channel dimensions, etc..) may also affect the sensitivity of SWAT to parameters which are related to channel flow.