Hydrological Modelling, Topographical Influence and Yield Mapping in Precision Agriculture

Abstract: The use of topographical data in hydrological and agricultural applications has increased with GPS data availability and the concept of precision agriculture. This thesis investigates the relationships between topography and in one case simulated soil moisture and in the other case yield. The yield was recorded with a new optical sensor, which uses a digital camera to detect objects and records the yield on tuber level. The accuracy of this yield monitor was determined to be better than that of alternative methods. The topographical influence on yield was determined with correlation to topographical indices. A spatial regression and a multiple step-wise regression method resulted in 20% of yield being explained by elevation, gradient and a new topographical index (horizontal position) in one field. In the other field studied, 18% of the yield was explained by topography, here the horizontal position was replaced by drainage area. Drainage area (a) calculation is a derivation of digital elevation models (DEMs) used within e.g. the TWI (topographical wetness index), ln (a / tan ƒÒ). The methods to quantify (a) are several and have an influence on the result of the TWI. Three drainage area algorithms were therefore tested together with TWI to map soil moisture for one German and one Swedish site. The methods tested included an eight-directions algorithm, a ¡¥stream tube¡¦ algorithm and a form-based algorithm. In the correlation with soil moisture the stream-tube and form-based methods performed best. Conventional drainage area calculations are applied to surfaces without any of the recurring features that may appear in cultivated fields, such as ridges. A new method for the calculation of drainage area in ridged fields was developed. Monte Carlo simulations showed a significant difference in estimated potential wetness between the new ¡¥ridged¡¦ and old methods. Simulation of soil moisture from three spatially distributed hydrological models, TOPMODEL, SMR and MIKE SHE, on the investigated fields ensued. The models all have a pronounced topographical driving force for the movement of water. The models were evaluated against observed soil moisture for two fields during two growing seasons. A correlation between simulated and observed soil moisture gave overall low correlation, 0.03 for SMR and MIKE SHE. It was concluded that the sole use of topography in explaining yield or estimating soil moisture is not enough. The spatial hydrological models tested need more tests on agricultural scale before they can be integrated into irrigation management systems.