The influence of waterlogging on agricultural monitoring from space

Abstract

Waterlogging is the accumulation of excess water in the rootzone. When soils are waterlogged the rootzone is either saturated or submerged with water. As a result, waterlogging prevents aeration of the rootzone which has adverse effects on crop growth. Waterlogging is an issue in both irrigated and rain-fed agriculture and crop productivity issues related to waterlogging are reported around the globe. As most terrestrial crops are within days susceptible to the negative effects of waterlogging, timely information on waterlogging and its presence spatially is beneficial. In that regard, to understand the impact of waterlogging on agriculture and combat consequential negative effect, satellite remote sensing can play a role.

However, in the field of satellite Remote Sensing of agriculture, waterlogging has so far received little attention. Previous related research focused on remote sensing of inundation or monitoring surface water, but waterlogging in agriculture is an overlooked subject. Little is known about how waterlogging is present in (irrigated) agriculture and what the ability of different remote sensing techniques is to detect and monitor waterlogging. Therefore, this thesis aimed to extend knowledge on how waterlogging influences agricultural monitoring with satellite remote sensing. Ultimately, to set footsteps towards monitoring waterlogging with satellite remote sensing.

The results presented evolve around a sugarcane plantation in Xinavane, Mozambique. The plantation served as a case study to demonstrate different satellite remote sensing observations in the context of waterlogging. First, the case study is presented and a describption is provided on the ground data collected. By providing remote sensing evaporation estimates, the high demand of irrigation water is illustrated. Vast quantities of water is needed to sustain sugarcane crop growth in the semi-arid environment of the plantation.

To continue, with a thorough literature review and the case study it is demonstrated waterlogging is a major issue burdening crop productivity. By assessing different remote sensing evaporation algorithms the results showed currently available evaporation estimates interpret waterlogging stress as a need to irrigate. This implies, before evaporation estimates from satellite data can play a role in optimizing field-scale water use in irrigated areas, evaporation algorithms must be able to identify water stress only in the case of water deficit in the root-zone. Throughout the chapter the presence of waterlogging or crop response to waterlogging is illustrated in different satellite remote sensing observations. In sum, the results imply a need to integrate observations of multiple sensors and potentially ancillary data (e.g. DEMs) to unravel how to monitor waterlogging with satellite remote sensing.

In search for the influence of waterlogging on agricultural monitoring, the research continued by comparing optical vegetation indices, radar vegetation indices, and sugarcane yield over the growing season in the plantation. The analysis gave an interesting and unexpected result. Contrary to the expectation the results showed a negative correlation between the Cross Ratio (CR) and sugarcane yield over the growing season. A modeling study proved the negative correlation results from a change in the sugarcane's internal composition which affects the dielectric constant of sugarcane canopies observed. The chemical composition of plant water in sugarcane changes over the growing season. As a consequence of sucrose accumulation in the stalk, water is increasingly bound to sucrose and this process lowers the dielectric constant.
%The results predominantly show a decrease in observed vegetation water content, as a result of a change in chemical composition due to an increase in sucrose accumulation, lowers the backscattered signal.

To follow up, active and passive microwave observations, optical vegetation indices, and production data are evaluated in different seasons. In addition to a temporal change of sucrose and moisture, the results showed vertically the sucrose-moisture distribution changes as well over the growing season. Therefore, the vertical distribution of sucrose - plant moisture influences the dielectric constant and, hence, the backscattered signal. The results highlight the VV backscatter responds to the stalk biomass, which is also the reservoir of sucrose in the sugarcane crop.

Finally, the influence of waterlogging on Sentinel-1 backscatter was detected through benchmarking with passive microwave observations, optical vegetation indices, and production data in a period where waterlogging was reported. Despite a thick sugarcane canopy, an increase in VH and VV polarizations was observed as a result of waterlogging. The increase was present at all stages during the growing season. The difference in backscatter as a result of waterlogging was highest in the VH backscatter. Also, the effect of waterlogging is translated through to the CR, which proves CR can play a role in the discrimination of waterlogging.

The results presented in this thesis help to further understand the influence of waterlogging in agricultural monitoring. Also, this work shows to correctly interpret irrigation estimates and crop development, waterlogging and sucrose development need to be flagged or otherwise considered during the growing season. Especially radar observations from Sentinel-1 backscatter appeared to be useful in monitoring waterlogging and sucrose development.