Quantification of irrigation and crop water productivity using soil moisture dynamics

Abstract

Estimating water budgets in agricultural fields is key to evaluating and improving water management. However, quantifying hydrological components remains a challenge, especially estimation of irrigation quantities and scheduling. Soil moisture dynamics is shown to hold key information about the water cycle and has been proven useful in retrieving rainfall for example. In this thesis, we hypothesize that the soil moisture dynamics can allow us to estimate irrigation and water budgets. We used a well-monitored groundwater-irrigated agricultural site in southern India for our study. The first step in this thesis was to retrieve soil moisture dynamics at different scales (point scale using HydraProbe, plot scale- 500 m using COSMOS and regional scale 9 km using remotely sensed datasets). The soil moisture dynamics were exploited to arrive at the soil hydraulic parameters using field and laboratory methods. These parameters were compared to the parameters available in the literature for different soil types in the catchment and it was concluded that field-derived parameters are more suitable for further study. We set up a synthetic experiment to evaluate the capacity of the model SM2RAIN to retrieve irrigation scheduling and quantity from soil moisture time series. In our soil-climatic condition, SM2RAIN was able to estimate accurately rainfall upto 40 mm but overestimated rainfall above 40 mm. As irrigation is usually less than 40 mm/d in our catchment SM2RAIN showed very good estimates. In field conditions, we found that the performance of SM2RAIN depended on the technology and scale used to produce soil moisture time series. The point scale soil moisture gave better irrigation as compared to larger scales oil moisture. However, the COSMOS scale was able to capture rainfall better than the HydraProbe scale. With soil hydraulic parameters and irrigation inputs, we used a coupled model to retrieve mass balance in the agricultural system. The crop dynamics were established using the crop model called STICS. Crop dynamics from STICS were fed toHydrus-1Dasaforcing variable to arrive at both crop and water dynamics. The reason for us to couple these two models is that the STICS model has good crop growth functions which are based on carbon distribution and it can also take information about agricultural processes as inputs but the water distribution in the model is done using the tipping-bucket model. On the other hand, the Hydrus-1D model solves Richard’s equation for water movement and the crop growth model built in this model is limited and cannot incorporate the agricultural processes in the field. Hence, we attempted to couple the two models to get the best part of both models and compensate for the difference using data from point scale soil moisture from HydraProbe. This coupled model was applied to both synthetic and real field datasets which proves their effectiveness of the methodology developed in the thesis. The results suggest that the soil moisture dynamics can be effectively used to retrieve soil hydraulic properties and quantify irrigation and crop water productivity (CWP). The CWP estimation was done under two cases, one under irrigation estimated using SM2RAIN and another case under irrigation estimated using the STICS model under no stress condition. This was done to understand if the irrigation scheduling was done based on crop water requirement, and by what amount the CWP could be improved. The results suggest that the CWP can be improved more in the rabi season up to 40% as compared to the kharif season which varied from negative value to 20%