Predicting runoff series in ungauged basins of the Brazilian Cerrado biome

Highlights

• A regional-sample data set is made available with catchments from the Cerrado biome.

• Regionalization of a hydrological model is assessed by spatial or physical proximity.

• Streamflow series are obtained for 4531 hydrologic units in the Cerrado biome.

Abstract

Information concerning water availability in a basin can be key to trustworthy and robust decisions, and reduce disputes over water among its multiple users. The Cerrado region, however, lacks basic hydrological information concerning water availability in many of its basins. In this study, two regionalization frameworks based on the donation of parameter sets from a hydrological model calibrated in gauged catchments were assessed. These approaches were evaluated using a leave-one-out cross-validation for the gauged catchments. Parameters donation by spatial proximity led to KGE and rNSE of 0.58 and 0.56, while by attributes proximity led to KGE and rNSE of 0.56 and 0.45, respectively. A regional-sample data set for the Cerrado (HydroCerrado) was made available with information compiled for the 411 gauged catchments used in this study and runoff time series simulated for 4531 level 5 ottobasins by donating parameter sets by spatial proximity.

Introduction

With an extension of approximately 200 million hectares and occupying 24% of the Brazilian territory, the Cerrado is Brazil's second largest biome. Despite its acidic soils, which are poor in organic matter and nutrients (Klink, 2014; Rada, 2013), the Cerrado was fundamental for the ascension of the Brazilian agriculture in recent decades (Cremaq, 2010; Rada, 2013). The region presents highly technified farms with potential for high productivity, as seen in its latest agriculture frontier, the MATOPIBA (an acronym for the states Maranhão, Tocantins, Piauí, and Bahia).

One of the reasons for the agriculture success in the Cerrado is its long wet season, which lasts from October to April, enabling double-cropping within the same season (Spangler et al., 2017). In addition to climate, water availability plays a key factor in the region. The Cerrado has a high water yield, contributing to the availability of surface water in many of the main hydrographic regions of Brazil it encompasses (Lima, 2011) and with a high potential for the expansion of irrigated areas (Althoff and Rodrigues, 2019; FEALQ, 2014).

However, over the last few years, there has been an increasing number of conflicts over water use in parts of the Cerrado where irrigated agriculture expanded with poor planning and lack of hydrological information (Pousa et al., 2019). Irrigation represents alone ∼67% of all water consumption in the country (ANA, 2017). Besides, studies considering different projections of climate change have highlighted several risks for sustainable socio-economic development in the region. For instance, Chou et al. (2014) predicted an increase in temperature and decrease in rainfall, while others predict an increase in the length of the dry season (Pires et al., 2016) and a decrease in river discharge (Oliveira et al., 2017).

To guarantee social well-being and sustainable development in the region it is crucial to have enough technical information to support the planning and management of water resources. Water allocation among multiple users usually comes with conflicting interests (Rodrigues et al., 2014), requiring knowledge of the historic and current water availability and water demand within a basin. Pousa et al. (2019) suggested that the knowledge of water availability in a basin, which serves as a management unit, can be key to trustworthy and robust decisions and can contribute to the reduction of conflicts during the dry season. The Cerrado region, however, lacks basic hydrological information that can be used to develop reliable estimates of water availability in specific regions of its basins.

There are many ways to determine water availability in hydrographic basins where streamflow measurements or other hydrometric data are lacking. For example, the complex interaction between physiographic factors such as climate, land use and land cover, topography, and geology, make the application of mathematical and computational models the best alternative to estimate target variables in ungauged basins. Such models can be valuable tools in the planning and management of water resources because they can simulate the impact of different factors on the basin's water availability. A large number of models have already been developed to assess water availability (Beck et al., 2017; Devia et al., 2015; Douglas-Mankin et al., 2010; Perrin et al., 2003; Steenhuis et al., 2009). These models should preferably be simple and easy to implement, making use of recent advances of different data sources such as satellite or remote sensing-derived products including evapotranspiration, net radiation, snow cover, or soil moisture data (Chen and Wang, 2018; Jiang and Wang, 2019; Tran et al., 2018).

The use of gridded products has excelled in hydroclimatic modeling in the past 20 years for both conceptual (lumped/bucket-style) and physically-based (distributed) models, especially in regions with data scarcity (Jiang and Wang, 2019; Tran et al., 2018). However, runoff simulations and predictions in ungauged basins (PUB) remain a crucial challenge in water resource planning and management (Guo et al., 2021; Qi et al., 2020). PUB has been addressed mainly by regionalization approaches. These approaches usually consist of “donating” hydrological model parameters calibrated at gauged catchments to predict runoff time series or hydrological signatures in ungauged catchments (Guo et al., 2021). In a large sample study, Qi et al. (2020) assessed several regionalization approaches for 2277 gauged catchments worldwide and found that donating hydrological parameters based on spatial proximity generally outperformed donation based on physical similarity (“attributes proximity”), whereas both outperformed regression techniques. Considering that previous efforts to regionalize hydrologic information in parts of the Cerrado have mainly been addressed by regression techniques and for hydrologic signatures (Lopes et al., 2017; Morais et al., 2020; Pruski et al., 2013), exploring the donation of hydrological parameters can be considered an advance since obtaining runoff series allows for more dynamic analyses to be carried out.

Based on the current state of hydrological data availability in the Cerrado and its importance for Brazil's agriculture, there is a need to improve information on water availability in the many unmonitored parts of the biome to help stakeholders make decisions based on scientific knowledge (Pousa et al., 2019). Thus, the objectives of the present study were to (i) assess two frameworks for regionalization of hydrological model's parameters and (ii) to develop a runoff time series data set for all (4531) of the Cerrado level 5 ottobasins. Ottobasins are basins and interbasins hierarchically classified according to the topological system proposed by Otto Pfafstetter (see Furnans and Olivera, 2001). The development of this data set enables the analysis of different hydrological signatures and water availability assessment across the Cerrado biome.