# Table 1 Best-fit parameters of time functions

acτ (days)AICa
ΔVS
Path 126
f1(t)(7.4 ± 0.2) × 105 (m3)379 ± 31119.8
f2(t)b(6.1 ± 2.1) × 105 (m3)106 ± 183 (m3/day)313 ± 118121.2
Path 20
f1(t)(9.2 ± 1.2) × 105 (m3)641 ± 150191.5
f2(t)b(9.0 ± 25.3) × 105 (m3)10 ± 1437 (m3/day)632 ± 1300193.5
Qout
f1(t)(1.1 ± 0.03) × 1010 (kg)209 ± 16447.1
f2(t)(5.0 ± 0.1) × 109 (kg)(8.6 ± 0.1) × 106 (kg/day)47 ± 2274.2
ΔLd
Path 126
f1(t)− 1.90 ± 0.16 (cm)905 ± 115− 2350.7
f2(t)(− 2.63 ± 0.22) × 10−1 (cm)(− 1.04 ± 0.04) × 10−3 (cm/day)20 ± 11− 2445.1
Path 20
f1(t)− 1.34 ± 0.04 (cm)378 ± 26− 2345.2
f2(t)(− 4.62 ± 0.26) × 10−1 (cm)(− 0.88 ± 0.04) × 10−3 (cm/day)41 ± 10− 2465.9
1. a$${\text{AIC}} = n\log \left( {\frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \left( {O_{i} - C_{i} } \right)^{2} } \right) + 2k$$, where n is number of data, i is index, O is an observed value, C is a calculated value, k is number of model parameters (Akaike 1974)
2. bParameters included in f2(t) of ΔVS were not determined uniquely, which indicates that linear terms are completely unnecessary to characterize the time series of ΔVS