چكيده به لاتين
The significance of increasing the accuracy of various hydrological models estimation among scientists has ever been a source of interest over the last two decades. In this regard, data assimilation is always considered as one of the promising methods for joint estimation of state variables and parameters of hydrological models. On the other hand, a wide range of observations that represent the real values of hydrological state variables has provided researchers with significant potential in this field. In this study, in order to estimate the joint states and parameter values of SWAT, surface soil moisture observations from the AMSR-E sensor have been assimilated.
These observations have been modeled alongside the measured values of surface runoff by taking the advantages of the particle filtering algorithm. For this purpose, changes in SWAT source codes have been conducted to introduce soil moisture observation values in the SWAT execution process. In order to evaluate the quality of soil moisture and runoff estimation by using the particle filter and specifically PF-SIR version, first the catchment area limited to the upper part of Bitas watershed metering the so-called station from the sub-basins of Urmia Lake catchment area was modeled. Then, three separate scenarios were defined, including: 1) assimilation of surface runoff observations and evaluating its effect on soil moisture and runoff estimation quality. 2) assimilation of satellite soil moisture observations and evaluating its effect on the quality of runoff and soil moisture estimation and 3) Simultaneous assimilation of runoff and surface soil moisture observations and evaluating the results of estimating these two variables under the combined assimilation approach.
To optimally evaluate the particle filtering algorithm before any action, the criterion problem was analyzed using a different number of particles. Then the capabilities of the particle filtering algorithm to estimate the parameters and runoff for a synthetic case study were tested. In order to better evaluate the potential of the particle filtering algorithm, a real case study was taken into account. The estimation of both parameters and runoff of the basin in was examined in comparison with the results of the SUFI2 calibration algorithm. In this study, which used four measures of fitness and two accuracy indices for evaluating the quality of batch estimation, The results showed a remarkable improvement in the accuracy of point and batch estimation of state variables using a particle filter. Finally, the performance of the particle filter alongside the impact of executable file of the SWAT modified version in assimilation of runoff and soil moisture observations has been evaluated. The results show that the assimilation of these variables observations, despite increasing the accuracy of estimating of assimilated state, will not help the improvement of other state estimation accuracy. This point is while the assimilation of both variables observations has improved the accuracy of the prediction values of both runoff and soil moisture.
