29.10.2014 Views

Recharge systems for protecting and enhancing groundwate

Recharge systems for protecting and enhancing groundwate

Recharge systems for protecting and enhancing groundwate

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

424<br />

TOPIC 3<br />

Modelling aspects <strong>and</strong> <strong>groundwate</strong>r hydraulics<br />

as unacceptable (Ghulam et al. 2004). The root mean square error is useful to quantify the differences between<br />

observed data <strong>and</strong> simulated data with the optimized parameters:<br />

O<br />

=<br />

n<br />

∑<br />

i=<br />

1<br />

n<br />

Oi<br />

⎛<br />

N<br />

1<br />

⎞<br />

2<br />

(7) RMSE = ⎜ ∑ [ Mi<br />

− Si()<br />

b ] (87)<br />

⎝ N<br />

⎠<br />

where Mi <strong>and</strong> Si (b) are measured <strong>and</strong> simulated values <strong>for</strong> an output variable.<br />

i=<br />

1<br />

During recharge there was decrease in RMSE with the time (Table 3). RMSE decreased from 1.24 m after 1 hr of<br />

recharge to 0.83 m after 87 hrs of recharge. Similar trends were shown by ME also. ME increased from 0.97 to 0.98<br />

from 1 hr of recharge to 87 hrs of recharge. Initially high RMSE <strong>and</strong> low ME revealed that during the start of<br />

recharge there was much deviation between the experimental head <strong>and</strong> simulated head. Similar trends were<br />

observed during the recovery also i.e. from 1 hr to 34 hrs recovery, RMSE decresed from 3.57 m to 3.05 m <strong>and</strong> ME<br />

increased from 0.85 to 0.87 (Table 4). Overall matching between observed <strong>and</strong> simulated data points was quite<br />

satisfactory during calibration except during start of recharge <strong>and</strong> recovery in each cycle (Fig. 3). It might have<br />

been due to the facts: 1) that hydrus 2D model does not take into account the storage coefficient in the aquifer<br />

<strong>and</strong> /or 2) the entry points resistances of piezometers might have delayed their response to drawup <strong>and</strong> drawdown<br />

pressure heads under actual field conditions.<br />

Table 3. Number of observations N, RMSE <strong>and</strong> ME in drawup during calibration <strong>and</strong> validation in pressure heads<br />

Time (h)<br />

Calibration<br />

N RMSE* (m) ME** N<br />

* Root mean square error ; ** Modelling efficiency.<br />

Validation<br />

RMSE (m)<br />

IInd IIIrd IVth<br />

Mean ME<br />

1 31 1.24 0.97 31 1.21 1.22 1.23 0.97<br />

15 35 1.13 0.97 35 1.17 1.17 1.19 0.96<br />

45 39 0.98 0.98 39 1.15 1.15 1.21 0.98<br />

87 46 0.83 0.98 46 1.00 1.00 1.15 0.99<br />

Table 4. Number of observations N, RMSE <strong>and</strong> ME in drawdown during calibration <strong>and</strong> validation in pressure heads<br />

Time (h)<br />

Calibration<br />

N RMSE* (m) ME** N<br />

Validation<br />

RMSE (m)<br />

IInd IIIrd IVth<br />

Mean ME<br />

1 34 3.57 0.87 300 3.50 3.80 3.82 0.85<br />

16 44 3.20 0.88 300 3.10 3.37 3.51 0.86<br />

34 47 3.05 0.89 300 2.95 3.10 3.21 0.87<br />

Figure 3. Simulated <strong>and</strong> experimental<br />

drawup/drawdown of piezometric head<br />

as given by model during successive recovery<br />

<strong>and</strong> recovery cycles<br />

ISMAR 2005 ■ AQUIFER RECHARGE ■ 5th International Symposium ■ 10 –16 June 2005, Berlin

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!