Journal Article

Assimilation of Sentinel‐Based Leaf Area Index for Modeling Surface‐Ground Water Interactions in Irrigation Districts

Zafarmomen, Nima, Bayat, Mehrad, Ehtiat, Majid, Moradkhani, Hamid, Alizadeh, Hosein

2024

Overview

Vegetation‐related processes, such as evapotranspiration (ET), irrigation water withdrawal, and groundwater recharge, are influencing surface water (SW)—groundwater (GW) interaction in irrigation districts. Meanwhile, conventional numerical models of SW‐GW interaction are not developed based on satellite‐based observations of vegetation indices. In this paper, we propose a novel methodology for multivariate assimilation of Sentinel‐based leaf area index (LAI) as well as in‐situ records of streamflow. Moreover, the GW model is initially calibrated based on water table observations. These observations are assimilated into the SWAT‐MODFLOW model to accurately analyze the advantage of considering high‐resolution LAI data for SW‐GW modeling. We develop a data assimilation (DA) framework for SWAT‐MODFLOW model using the particle filter based on the sampling importance resampling (PF‐SIR). Parameters of MODFLOW are calibrated using the parameter estimation (PEST) algorithm and based on in‐situ observation of the GW table. The methodology is implemented over the Mahabad Irrigation Plain, located in the Urmia Lake Basin in Iran. Some DA scenarios are closely examined, including univariate LAI assimilation (L‐DA), univariate streamflow assimilation (S‐DA), and multivariate streamflow‐LAI assimilation (SL‐DA). Results show that the SL‐DA scenario results in the best estimations of streamflow, LAI, and GW level, compared to other DA scenarios. The streamflow DA does not improve the accuracy of LAI estimation, while the LAI assimilation scenario results in significant improvements in streamflow simulation, where, compared to the open loop run, the (absolute) bias decreases from 75% to 6%. Moreover, S‐DA, compared to L‐DA, underestimates irrigation water use and demand as well as potential and actual crop yield. Key Points Using source code modification, SWAT‐MODFLOW is connected to sequential DA Multivariate assimilation of streamflow, GW‐level and leaf area index (LAI) shows the best results Streamflow data assimilation does not improve LAI simulation, while LAI data assimilation improves streamflow simulation

• Publisher: John Wiley & Sons, Inc,Wiley
• Subject: Algorithms / Crop yield / Data assimilation / Data collection / Evapotranspiration / Groundwater / Groundwater irrigation / Groundwater recharge / Groundwater table / hydrologic models / intensive irrigation / Iran / Irrigation / Irrigation districts / Irrigation water / Lake basins / Leaf area / Leaf area index / Leaves / Mathematical models / Modelling / Multivariate analysis / Numerical models / Parameter estimation / remote sensing / Resampling / Satellite observation / satellites / Stream discharge / Stream flow / Surface water / surface water‐groundwater interaction / Surface-groundwater relations / SWAT‐MODFLOW / Vegetation / Vegetation index / Water table / Water use / watersheds