Nowcast flood predictions in the Amazon watershed based on the remotely sensed rainfall product PDIRnow and artificial neural networks.

The aim of this study was to develop artificial neural network (ANN) models to predict floods in the Branco River, Amazon basin. The input data for the models included the river levels and the average rainfall within the drainage area of the basin, which was estimated from the remotely sensed rainfall product PDIRnow. The hourly water level data used in the study were recorded by fluviometric telemetric stations belonging to the National Agency of Water. The multilayer perceptron was used as the neural framework of the ANNs, and the number of neurons in each layer of the model was determined via optimization with the SCE-UA algorithm. Most of the fitted ANN models showed Nash-Sutcliffe efficiency index values greater than 0.9. It is possible to conclude that the ANNs are effective for predicting the flood levels of the Branco River, with horizons of 6, 12 and 24 h; thus, constituting a viable option for use in river-flood warning systems in the Amazon basin. For the forecast with a 24-h horizon, it is essential to include the average rainfall of the basin that accumulated over the last 48 h as input data into the ANNs, along with the levels measured by the streamflow stations. The indirect rainfall estimates provided by PDIRnow are an excellent alternative as input data for ANN models used to predict floods and constitute a viable solution for regions where the density of rain gauge stations is low, as is the case in the Amazon basin.

Environmental vulnerability assessment of the Doce River basin, southeastern Brazil.

Environmental vulnerability is an important tool to understand the natural and anthropogenic impacts associated with the susceptibility to environmental damage. This study aims to assess the environmental vulnerability of the Doce River basin in Brazil through Multicriteria Decision Analysis based on Geographic Information Systems (GIS-MCDA). Natural factors (slope, elevation, relief dissection, rainfall, pedology, and geology) and anthropogenic factors (distance from urban centers, roads, mining dams, and land use) were used to determine the environmental vulnerability index (EVI). The EVI was classified into five classes, identifying associated land uses. Vulnerability was verified in water resource management units (UGRHs) and municipalities using hot spot analysis. The study employed the water quality index (WQI) to assess the EVI and global sensitivity analysis (GSA) to evaluate the model input parameters that most influence the basin's environmental vulnerability. The results showed that the regions near the middle Doce River were considered environmentally more vulnerable, especially the UGRHs Guandu, Manhuaçu, and Caratinga; and 35.9% of the basin has high and very high vulnerabilities. Hot spot analysis identified regions with low EVI values (cold spot) in the north and northwest, while areas with high values (hot spot) were concentrated mainly in the middle Doce region. Water monitoring stations with the worst WQI values were found in the most environmentally vulnerable areas. The GSA determined that land use and slope were the primary factors influencing the model's response. The results of this study provide valuable information for supporting environmental planning in the Doce River basin.

Effect of environmental covariable selection in the hydrological modeling using machine learning models to predict daily streamflow.

There are different methods for predicting streamflow, and, recently machine learning has been widely used for this purpose. This technique uses a wide set of covariables in the prediction process that must undergo a selection to increase the precision and stability of the models. Thus, this work aimed to analyze the effect of covariable selection with Recursive Feature Elimination (RFE) and Forward Feature Selection (FFS) in the performance of machine learning models to predict daily streamflow. The study was carried out in the Piranga river basin, located in the State of Minas Gerais, Brazil. The database consisted of an 18-year-old historical series (2000-2017) of streamflow data at the outlet of the basin and the covariables derived from the streamflow of affluent rivers, precipitation, land use and land cover, products from the MODIS sensors, and time. The highly correlated covariables were eliminated and the selection of covariables by the level of importance was carried out by the RFE and FFS methods for the Multivariate Adaptive Regression (EARTH), Multiple Linear Regression (MLR), and Random Forest (RF) models. The data were partitioned into two groups: 75% for training and 25% for validation. The models were run 50 times and had their performance evaluated by the Nash Sutcliffe efficiency coefficient (NSE), Determination coefficient (R2), and Root of Mean Square Error (RMSE). The three models tested showed satisfactory performance with both covariable selection methods, however, all of them proved to be inaccurate for predicting values associated with maximum streamflow events. The use of FFS, in most cases, improved the performance of the models and reduced the number of selected covariables. The use of machine learning to predict daily streamflow proved to be efficient and the use of FFS in the selection of covariables enhanced this efficiency.

Multivariate statistics for spatial and seasonal quality assessment of water in the Doce River basin, Southeastern Brazil.

This study employed multivariate statistical techniques in one of the main river basins in Brazil, the Doce River basin, to select and evaluate the most representative parameters of the current water quality aspects, and to group the stations according to the similarity of the selected parameters, for both dry and rainy seasons. Data from 63 qualitative monitoring stations, belonging to the Minas Gerais Water Management Institute network were used, considering 38 parameters for the hydrological year 2017/2018. Principal component analysis (PCA) and hierarchical cluster analysis (HCA) were used to reduce the total number of variables and to group stations with similar characteristics, respectively. Using PCA, four principal components were selected as indicators of water quality, explaining the cumulative variance of 68% in the rainy season and 65% in the dry season. The HCA grouped the stations into four groups in the rainy season and three groups in the dry season, showing the influence of seasonality on the grouping of stations. Moreover, the HCA made it possible to differentiate water quality stations located in the headwaters of the basin, in the main river channel, and near urban centers. The results obtained through multivariate statistics proved to be important in understanding the current water quality situation in the basin and can be used to improve the management of water resources because the collection and analysis of all parameters in all monitoring stations require greater availability of financial resources.

Influence of changes in land use and land cover and rainfall on the streamflow regime of a watershed located in the transitioning region of the Brazilian Biomes Atlantic Forest and Cerrado.

Climate change and the intensification of anthropogenic activities in watersheds have been substantially changing the streamflow regime, which is a problem for water resource managers. This study assesses the influence of the changes in land use and land cover and rainfall on the streamflow regime. This study also models the pattern of these streamflows according to the rainfall and land use and land cover in the Santo Antônio River watershed, located in the transitioning region of the Brazilian Biomes Atlantic Forest and Cerrado. To assess the dynamic relationship between land use and land cover and the streamflow regime, five classes of land use and land cover were used. To characterize the hydrological pattern, data from six streamflow gauges and 24 rainfall gauges that influence the study area were used. Multiple regression models were adjusted to estimate streamflow using the explanatory variables rainfall and land use and land cover. As result, a direct relationship was found, as the decrease in streamflow in some drainage areas was influenced by the decrease in rainfall over the base period. The relationship between land use and land cover and streamflow was not significant. The reductions in the streamflow regimes over the years in the watershed were influenced by reductions in annual rainfall, which reduced about 19% while the water withdrawals from 2003 to 2014 increased 2350%. The results found in this study are useful to the water managers since they can estimate streamflow in any part of the studied river through rainfall and land use and land cover data. This helps to reduce the risks associated with the water allocation process.

Use of multivariate statistical methods to analyze the monitoring of surface water quality in the Doce River basin, Minas Gerais, Brazil.

The objective of the present study was to evaluate the water quality data in the Minas Gerais portion of the Doce River basin in order to analyze the current monitoring network by identifying the main variables to be maintained in the network, their possible sources of pollution, and the best sampling frequency. Multivariate statistical techniques (factor analysis/principal components analysis, FA/PCA and cluster analysis, CA) complemented by the analysis of violation of the framing classes were used for this purpose. Water quality variables common to 64 monitoring sites were analyzed for the base period from 2010 to 2017. The water quality variables were analyzed considering the different monitoring campaigns: (a) partial campaigns; (b) total campaigns; and (c) monthly campaigns. It was identified from the FA/PCA results, that, when the partial campaign data were analyzed, the variables selected represent the high susceptibility that the basin presents to erosion and the release of domestic effluents in its water bodies. When the data of total campaigns were evaluated, representative variables of the contamination by heavy metals from industrial and mining activities were included. Therefore, the analysis of violation of the framing classes made possible to identify five critical variables: thermotolerant coliforms, dissolved iron, total phosphorus, and total manganese, which reinforced the results obtained in FA/PCA. Based on the results of the analyses, it was recommended to include variables associated with heavy metal contamination in the partial campaigns, prioritizing the dissolved iron and total manganese, as well as total chloride sampling only for the total campaigns. The evaluated data from the monthly campaigns, the CA showed that although the quarterly monitoring frequency is satisfactory, the monthly monitoring is more appropriate for the monitoring of water quality in the Minas Gerais portion of the Doce River basin.

Methodological proposal for the allocation of water quality monitoring stations using strategic decision analysis.

In order to fill a gap in the monitoring of water quality in Brazil, the objective of this study was to propose a methodology to support the allocation of water quality monitoring stations in river basins. To achieve this goal, eight criteria were selected and weighted according to their degree of importance. It was taken into account the opinion of water resources management experts. In addition, a decision support system was designed so that the methodology could be used in the allocation of water quality monitoring stations by researchers and management bodies of water resources, to be fully implemented in geographic information system environment. In order to demonstrate the potential of the proposed methodology, which can be used in places that have or not existing monitoring networks, it has been applied in the Minas Gerais portion of the Doce river basin. Because the area already has a monitoring network with 65 stations in operation under the responsibility of the Minas Gerais Water Management Institute (IGAM), an expansion of the network was suggested and a simulation of a scenario was performed considering that the study area did not have an established network. The results of the analyses consisted of maps of suitability, indicating the locations with greater and lesser suitability for the establishment of the stations. With the application of the methodology, seven new sites were proposed so that the study area had the density recommended by the National Water Agency (ANA), and it was verified that the Caratinga River Water Resources Management Unit (UGRH5 Caratinga) has the most deficiency of stations among the six units evaluated in the Minas Gerais portion of the Doce river basin. In the simulated scenario considering the non-existence of a network, the adequacy map obtained was compared with the existing monitoring network and it was possible to classify the stations according to the purpose for which they were established, such as monitoring environments under anthropic activities or establishing benchmarks for the water bodies. Overall, the proposed methodology proved itself robust, and although the results were specific to one basin, the criteria and decision support system used are fully applicable to other areas of study.

Environmental flow in the River Ondas basin in Bahia, Brazilian Cerrado.

This paper aimed to estimate the environmental flow of a water basin located in the Brazilian Cerrado using the bidimensional model River2D. The study was carried out in a stretch of the lower portion of the River Ondas in the western part of the state of Bahia, Brazil. To carry out the ecohydrological modeling, the following were used: topobathymetry, hydraulic characterization, the streamflows with the probability of non-exceedances (Q50, Q60, Q70, Q80, Q90, and Q95), and the Habitat Suitability Index for species of the genus Hypostomus. In the River2D, the weighted usable areas (WUAs) pertaining to the streamflows associated with different non-exceedances were simulated for the later construction of optimization and identification matrices of the streamflows that maximize the habitat area throughout the year. For juvenile Hypostomus, WUA increased as streamflow increased, with higher values associated with Q50. For adult specimens, lower WUA values were observed associated with Q50, while higher values were associated with Q95, which shows a preference for lower streamflows. The environmental flows found for the stretch of the River Ondas varied over the course of the year. The lowest environmental flows were observed in September (30.31 m(3) s(-1)) and October (29.98 m(3) s(-1)), while the highest were observed in February (44.22 m(3) s(-1)) and March (43.16 m(3) s(-1)). The environmental flow regime obtained restricts the water availability in the basin, for the purpose of water capture, which shows the importance of ecohydrological studies in forming a basis for water resource management actions.