Introduction
Worldwide, 324 million hectares of land, which contribute 40% of the total food production, are equipped for irrigation (FAO, 2016). At a global scale, average annual water consumption by irrigation is 7,700 m3 per hectare. In Brazil, irrigation’s water uptake is approximately 1000 m3/s, characterizing the major consumption in the territory. That quantity is distributed through irrigations systems covering 7 million ha with potential to expand by 43.3% (ANA, 2017). In the countryside, rivers and small dams are the major sources of irrigation, despite the substantial volumetric impact on water resources during the dry season. Considering that both urban and rural water supplies are provided by superficial water, scarcity has been a critical concern. For example, in the biennium 2014–2015, the reservoirs of São Paulo metropolitan region were reduced to single-digit percentages, leading to a profound water crisis (Millington, 2018).
Almost 20% of the Brazilian territory is covered by wetlands, and their distribution has been dramatically reduced by land conversion, thus affecting their natural water storage and ecological functions (Junk et al., 2013; Rosolen, Oliveira & Bueno, 2015). Around the world, uncontrolled exploitation is damaging ecosystem services that depend on wetlands (Verones, Bartl, Pfister, Vilchez & Hellweg, 2012).
In the Cerrado region—the world’s most threatened savanna environment and Brazil’s last remaining agricultural frontier areas under native vegetation—the pronounced seasonality necessitates the use of extensive irrigation for expanding agricultural land use (Rodrigues et al., 2018). This biome is characterized by long dry periods and by the presence of herbaceous plants and scattered woody plants and trees that cover well-drained lateritic soils of low fertility. The Cerrado’s wetlands represent islands of natural water storage. In the southeast portion of the ecosystem, our study area in this research, the water dynamic and volume are closely related to rainfall seasonality and the fluctuation of shallow groundwater, comprising a freshwater system integrated with rivers in the catchments. Surrounding the study area, large farms with center-pivot irrigation systems have been using the wetland’s water for grain cultivation. Although the allocation of water is legally required by some farmers, the resource’s conservation is jeopardized by a lack of knowledge about the balance between input and output of water in wetlands, and about the hydrological dynamic between aquifers and soils with hydromorphic properties. Despite closed wetlands store water and promote groundwater infiltration, influencing the hydrologic cycle at the watershed scale, few studies have relied on field data to further our understanding of the impacts, either individual or cumulative, of natural landscape characteristics and wetland drainage at the local scale (Haque & Badiou, 2018).
Hydrodynamic models typically partition rainfall into: soil moisture storage, groundwater recharge and surface runoff. This partitioning largely depends not only on the hydrological conditions, but also on the pedogeological conditions. Invasive data acquisition techniques such as drilling or trench excavations are labor-intensive, costly and involve extensive fieldwork sampling and laboratory analysis. They can be efficiently complemented by non-invasive geophysical and remote sensing methods which are time and cost effective, i.e. they succeed in covering large areas of land cheaper and quicker than the soil sampling techniques.
Unmanned Aerial Vehicles (UAVs), and in particular micro-UAVs (payload less than 1.5–2 kg), represent the latest frontier in land and water monitoring because of low-altitude flight and flexible payload design (Anderson and Gaston, 2013). In recent years, miniaturized components (GNSS receivers, inertial measurement units, autopilots) have advanced (Watts et al., 2012), and UAVs have been used also for a wide range of hydrological applications such as fluvial monitoring; river bathymetry and photogrammetric DEM generation using very high resolution (VHR) imagery (Lejot et al., 2007).
Innovative quantification techniques that use remote sensing, and especially those using high spatio-temporal resolution imagery, have been applied in non-invasive mapping of soil surface furthering our understanding of the relationship between surface soil water and aquifer in wetlands (Lin, 2012).
In this field, UAVs equipped with digital cameras had the capability to acquire VHR and describe the distribution of soil moisture in wetlands.
Traditional techniques emphasized in pedological studies (e.g., grain size analysis, mineralogy, chemical etc) carried out in a few meters thick do not provide adequate information about soil-water relationships as well the possibility to quantify the volume of water in soil. The use of images generated by UAV allow obtain information not only from surface features but also to provide excellent estimated values of, e.g., the water volume that the wetland can storage.
Another technique is electrical resistivity tomography (ERT), which can be successfully applied in waterlogged areas, since it is highly sensitive to the presence of water and can provide valuable information regarding the structure and volume of storage in rocks and soils (Carrara, Pece & Robertini, 1994; Zhang et al, 2016; Luo et al., 2019). These techniques have been widely applied in resource exploration and hydrogeology surveys, having the advantage of being rapid and cost-effective means of generating information (Kearey & Brooks, 1991; Helaly, 2017).
Besides, geophysical ERT technique allow to assess soil architecture, from surface to saprolite, highlighting zones with distinctive presence of water in pore space. That information is a key to understand the soil itself; soil-water transfers and to provide a model of water flow and hydrological process.
Aerial imaging is a widely applied technique in the study of wetlands, but digital aerial photogrammetry, i.e., the calculation trough UAV images of the area and the potential volume of water that can be stocked in the wetlands is a lack of knowledge, so that is one of our propositions in this work. Still, the second aim of this work is to propose an empirical model for estimating the spatial distribution of water in a specific Cerrado wetland of Minas Gerais state, Brazil, using UAV and ERT techniques to assess groundwater flow and aquifer recharge.
Combining both, UAV and ERT techniques, we employed noninvasively tools to directly quantify soil architecture and hydrodynamic. The methodological choice aimed to improve the needed linkage between digital architecture soil mapping with hydropedology (Lin, 2012).