Assimilating Remote Sensing-Based ET into SWAP Model for Improved Estimation of Hydrological Predictions
Summary (2 min read)
- Ccurate estimation of evapotranspiration (ET) plays an important role in quantification of the water balance at the pixel, watershed, basin and regional scale for better planning and managing water resources (Irmak et al., 2008a) .
- Furthermore, quantification of ET at multiple scales is spatially restricted because in-situ observations provide only point measurements.
- ET estimated from remote sensing observations can be used to calibrate hydrology model or estimate uncertain model parameters in the model via data assimilation.
- Similar works by Ines  , Srinuandee  , Chemin  , Kamble , Kulkarni  , Thapa  used remotely sensed information combined with a binary GA and SWAP model for optimizing soil hydraulic parameters.
- Additionally, the authors show here that such ET estimates may be used together with on-farm measurements of applied irrigation water to provide reliable estimates of soil moisture.
II. STUDY AREA
- The study was conducted at the South-central Nebraska.
- The long-term average rainfall in south central part of the state is about 680 mm although the annual total rainfall shows significant variations from year to year (e.g., 420 mm in 1988 to 1,040 mm in 1993).
- The dominant cropping system in south central Nebraska is corn-soybean rotation with increasing continuous corn production as the demand for ethanol production has been increasing.
- Most of the croplands in the region are irrigated with center pivots with the ground water pumped from the Ogallala aquifer being the dominant water supply for irrigation.
B. SWAP-GA Model Framework
- Genetic algorithm used in this research comprises of three components that are Remote sensing, SWAP model and Genetic algorithm.
- This process is termed as SWAP-GA method.
- The authors used SWAP-GA to estimate starting date of cropping, irrigation scheduling start time, time extent of cropping and the groundwater depths.
- The newly proposed parameters were fed to SWAP by GA according to the evaluation of the difference processes between SWAP output ETa values and the target ETa values.
- It is evident from the numerical figures in table that some crops are still developing on May and others are transpiring at higher rates.
- On June23, all the crops in the area are established.
- This indicates the variability of sowing dates and water management practices as influenced by water availability.
B. Remotely sensed Evapotranspiration Data Assimilation
- In simulations, hydraulic properties were based on measured values where possible; some values were altered slightly by optimizing the model to the local conditions until good agreement with measured ET was attained.
- Given these constraints, it was not possible to achieve perfect agreement with measurements for the wide range of ET conditions that occurred during the study.
- ET data from METRIC model were used as the "observed" RS data for the investigated pixel.
- Above results showed good fitness between the observed and simulated ET.
- Overall, the ET data assimilation results indicated that SWAP-GA performs well for the advective conditions of the study area with prediction errors of 10-20%.
C. Optimization of crop growth parameters from Data Assimilation
- The goal of the calibration process is to find optimal sets of configuration parameters for SWAP models.
- Optimal configuration parameters are determined by comparing the RMSE of the derived parameters, the convergence, the amount of a priori information used.
- The RMSE and Square of correlation coefficient then calculated from the observed and optimized values (Table 2 ).
- Good agreement was found between the optimized and observed ET.
- RMSE and r2 improvements occurred with the observed data when generation and population increased from 10 Generation 10 Population to 100 Generation and 100 Population.
D. Soil Moisture Estimation from Data Assimilation
- Soil-based measurements may be a far more practical and easy method for corn growers to use to schedule irrigations and assess current irrigation practices.
- At the start of the season the soil is moist from winter and spring rains; the readings are less than 0.3 cm 3 /cm 3 .
- Gradually the soil dries and the readings increase, beginning with the simulation at 0.5, 9.5, 27.5, 52.5 cm.
- The drying cycle resumed until a partial irrigation occurred in early May.
- The reason for the partial irrigation was that it was needed to replenish enough soil moisture to sow the crop through the germination process without excessive soil moisture depletion and crop stress.
- This study used remote sensing data to characterize system via a GA based hydrological data assimilation approach in Great Plain environment, and then the derived data were used as inputs to their water management optimization model.
- Reasonable parameters were successfully estimated and the ETa output from SWAP model matched with the actual ETa reasonably well.
- Soil moisture in the unsaturated zone in study area responded strongly to rainfall events because of the shallow water table in the great plain and additional net inputs from lateral saturated subsurface flows.
- If this evaluation is computationally expensive, the forward modeling approach can become impractical.
- These conclusions suggest that it is indeed necessary to couple a remotely sensed ET with a pixel-based hydrological model in order to study and explore the water management options.
Did you find this useful? Give us your feedback
Cites background or methods from "Assimilating Remote Sensing-Based E..."
...The eight-day composite images are treated as cloud-free products to monitor the Earth’s terrestrial activity [7,20]....
...Rather, we quantified the , NDVI, their interactions and the actual evapotranspiration values for the natural conditions of the fields, as reflected in the MODIS satellite images [7,20]....
...The declination extent primarily depends on the particular crop growth characteristics [6,7] and the irrigation management during the late season ....
Cites methods from "Assimilating Remote Sensing-Based E..."
...Kamble and Irmak (2008) attempted to assimilate ET, which was derived from Landsat TM/ETM data with a surface energy balance model, into a SWAP model....
Cites methods from "Assimilating Remote Sensing-Based E..."
...An Overview80 able method (Irmak and Kamble, 2009;, Kamble and Irmak, 2011;2008; Irmak et al., 2011)....
Related Papers (5)
Rolf H. Reichle, Dara Entekhabi +1 more
Rolf H. Reichle, Dennis McLaughlin +1 more
Jean-François Mahfouf, K. Bergaoui +4 more
Filipe Aires, C. Prigent +1 more
James G. Richman, Robert N. Miller +1 more