Brazil is a country highly dependent on hydroelectricity. Because electricity generation is programmed monthly, one of the most critical information is the inflow of the main hydropower plants of the country. In this context, this study evaluates the predictability and accuracy of raw and bias-corrected ECMWF S2S forecasts as input of a hydrological model, focusing on high and low-flows within the Tocantins-Araguaia Basin in Brazil. Rainfall forecasts were also evaluated against observations across 22 sub-basins of the Tocantins- Araguaia River, considering lead times ranging from 1 to 28 days. Results indicated that the predictability of S2S forecasts vary with the sub-basin area and through seasons (wet and dry periods). Bias corrections were able to improve the accuracy of the forecasts for all lead-times and across sub-basins of different sizes. In the case of low-flows, using raw ECMWF S2S rainfall forecasts as input of the hydrological model resulted in streamflow forecast errors that increase with the scale of the sub-basin. For high-flows results are the opposite, errors are larger in small sub-basins and smaller at larger scales. This is due to the overestimation of dry season rainfall in the ECMWF S2S forecasts. When using bias-corrected rainfall forecasts as input, river flow forecasts are useful up to 28 days of lead-time. The results of this study revealed the potential of using subseasonal forecasts for decision-making in the Tocantins-Araguaia Basin, and encouraging further developments.

In global applications and data sparse regions, which comprise most of the earth, hydrologic model-based flood monitoring relies on precipitation data from satellite multisensor precipitation products or numerical weather forecasts. However, these products often exhibit substantial errors during the meteorological conditions that lead to flooding, including extreme rainfall. The propagation of precipitation forcing errors to predicted runoff and streamflow is scale-dependent and requires an understanding of the autocorrelation structure of precipitation errors, since error autocorrelation impacts the accumulation of precipitation errors over space and time in hydrologic models. Previous efforts to account for satellite precipitation uncertainty in hydrologic models have demonstrated the potential for improving streamflow estimates; however, these efforts use satellite precipitation error models that rely heavily on ground reference data such as rain gages or weather radar and do not characterize the nonstationarity of precipitation error autocorrelation structures. This work evaluates a new method, the Space-Time Rainfall Error and Autocorrelation Model (STREAM), which stochastically generates possible true precipitation fields, as input to the Hillslope Link Model to generate ensemble streamflow estimates. Unlike previous error models, STREAM represents the nonstationary and anisotropic autocorrelation structure of satellite precipitation error and does not use any ground reference to do so. Ensemble streamflow predictions are compared with streamflow generated using satellite precipitation fields as well as a radar-gage precipitation dataset during peak flow events. Results demonstrate that this approach to accounting for precipitation uncertainty effectively characterizes the uncertainty in streamflow estimates and reduces the error of predicted streamflow. Streamflow ensembles forced by STREAM improve streamflow prediction nearly to the level obtained using ground-reference forcing data across basin sizes.

The availability of land, favorable edaphoclimatic conditions, and water have consolidated the role of Brazil as a global player in the agricultural commodities market, which has driven intensive land use conversion in recent decades. In this context, the extensive use of biofuels to mitigate fossil fuel CO2 emissions has facilitated their replacement as pasture by sugarcane monoculture, particularly in Sao Paulo State that is responsible for more than 52% of the country's sugarcane production. This expansion has occurred without evaluating its impact on water availability for other uses. Therefore, this study investigates the effects of the sugarcane expansion on the hydrological cycle of the Aguapei River basin in Sao Paulo State, during 1985-2017. To achieve this goal, we first validated the hydrological model simulation by comparing the time trajectories within the Budyko framework for the sub-basin with the longest observation records. We then assessed the impact of the expansion of the area under sugarcane by comparing the streamflow over the study period against a baseline scenario, which assumed that no land use changes occurred after 1985 in six sub-basins of the Aguapei River basin. The results revealed a reduction in streamflow with sugarcane expansion. Although these changes corresponded to only 2-5% of the mean streamflow of the baseline scenario, these can increase to 13-27% when the effect of sugarcane expansion is isolated. Our results suggested that the impacts of sugarcane on streamflow are gradually diluted at larger scales. We also estimated that the extra water consumption due to the sugarcane expansion in Sao Paulo State was 2150 hm(3) yr(-1) in 2017 compared to 1985.

This work investigates the use of a stochastic error model (the 2-Dimensional Satellite Rainfall Error Model-SREM2D) to generate an ensemble of rainfall fields, based on the forecasts from the Eta regional weather forecast model. To evaluate the usefulness of this approach against traditional techniques, streamflow probabilistic forecasts from a distributed hydrological model forced with two sources of rainfall data are compared in the Tocantins-Araguaia basin in Brazil. The first dataset is an empirical rainfall ensemble produced by the SREM2D model applied to the Eta model, and the second is a state-of-the-art rainfall ensemble produced by the ECMWF model. Results show the potential of the stochastic error model to generate precipitation ensemble fields from a regional numerical weather forecasting model removing around 60% and 12% of the systematic and random error, respectively. Moreover, SREM2D is proven to be an efficient technique that involves a low computational cost when compared to the more sophisticated ensemble techniques used by the ECMWF model.

This study investigates the potential of observations with improved frequency and latency time of upcoming altimetry missions on the accuracy of flood forecasting and early warnings. To achieve this, we assessed the skill of the forecasts of a distributed hydrological model by assimilating different historical discharge time frequencies and latencies in a framework that mimics an operational forecast system, using the European Ensemble Forecasting system as the forcing. Numerical experiments were performed in 22 sub-basins of the Tocantins-Araguaia Basin. Forecast skills were evaluated in terms of the Relative Operational Characteristics (ROC) as a function of the drainage area and the forecasts’ lead time. The results showed that increasing the frequency of data collection and reducing the latency time (especially 1 d update and low latency) had a significant impact on steep headwater sub-basins, where floods are usually more destructive. In larger basins, although the increased frequency of data collection improved the accuracy of the forecasts, the potential benefits were limited to the earlier lead times.

We critically examined the performance of probabilistic streamflow forecasting in the prediction of flood events in 19 subbasins of the Doce River in Brazil using the Eta (4 members, 5 km spatial resolution) and European Centre for Medium‐Range Weather Forecasts (ECMWF; 51 members, 32 km resolution) weather forecast models as inputs for the MHD‐INPE hydrological model. We observed that the shapes and orientations of subbasins influenced the predictability of floods due to the orientation of rainfall events. Streamflow forecasts that use the ECMWF data as input showed higher skill scores than those that used the Eta model for subbasins with drainage areas larger than 20,000 km2. Since the skill scores were similar for both models in smaller subbasins, we concluded that the grid size of the weather model could be important for smaller catchments, while the number of members was crucial for larger scales. We also evaluated the performance of probabilistic streamflow forecasting for the severe flood event of late 2013 through a comparison of observations and streamflow estimations derived from interpolated rainfall fields. In many cases, the mean of the ensemble outperformed the streamflow estimations from the interpolated rainfall because the spatial structure of a rainfall event is better captured by weather forecast models.

This study investigates the efficiency of correcting radar rainfall estimates using a stochastic error model in the upper Iguacu river basin in Southern Brazil for improving streamflow simulations. The 2-Dimensional Satellite Rainfall Error Model (SREM2D) is adopted here and modified to account for topographic complexity, seasonality, and distance from the radar. SREM2D was used to correct the radar rainfall estimates and produce an ensemble of equally probable rainfall fields, that were then used to force a distributed hydrological model. Systematic and random errors in simulated streamflow were evaluated for a cascade of sub-basins of the Iguacu catchment, with drainage area ranging from 1,808 to 21,536 km(2)). Results showed an improvement in the statistical metrics when the SREM2D ensemble was used as input to the hydrological model in place of the radar rainfall estimates in most sub-basins. Specifically, SREM2D was able to remove the relative bias (up to 50%) in the radar rainfall dataset regardless of the basin dimension, whereas the random error was reduced more prominently in the larger basins (up to 100 m(3) s(-1)). An event scale evaluation was also performed for nine selected flood events in three sub-basins. SREM2D reduced the overestimation in the cumulative rainfall and streamflow volumes during these events.

Resumo Uma das principais aplicações das estimativas de precipitação por satélite é a modelagem hidrológica em bacias onde a rede convencional e em tempo real de pluviômetros são precárias no que se refere à resolução espacial e temporal de dados. Neste trabalho discute-se o desempenho do modelo de erro de precipitação por satélite estocástico multidimensional - SREM2D (do inglês, Two-Dimensional Satellite Rainfall Error Model), o qual simula conjuntos de campos diários de precipitação com os mesmos padrões estatísticos (dispersão) que a diferença dos campos de chuva estimados por satélite e pluviômetro de uma série maior. A maioria dos modelos tratam o erro como uma medida uni-dimensional sem o reconhecimento que a precipitação é um processo intermitente no tempo e no espaço. O modelo SREM2D caracteriza a estrutura espacial, a dinâmica temporal e a variabilidade espacial do erro de estimativa das taxas de precipitação. Este trabalho avalia os resultados das simulações do SREM2D para diversos algoritmos de estimativa de precipitação por satélite na bacia dos rios Tocantins-Araguaia. Resultados mostram que o conjunto obtido através das realizações do modelo SREM2D reduziram o viés dos algoritmos de estimativa de precipitação por satélite principalmente para bacias com área de drenagem superior a 12.000 km2.

This study investigates the applicability of error corrections to satellite-based precipitation products in streamflow simulations. A three-year time series (2008-2011) is considered across 19 sub-basins of the Tocantins-Araguaia basin (764,000 km(2)), located in the center-north region of Brazil. A raingauge network (24 h accumulation) of approximately 300 collection points (similar to 1 gauge every 2500 km(2)) is used as reference for evaluating the following four satellite rainfall products: the Tropical Rainfall Measuring Mission real-time 3B42 product (3B42RT), the Climate Prediction Center morphing technique (CMORPH), the Global Satellite Mapping of Precipitation (GSMaP), and the NOAA Hydroestimator (HYDRO-E). Ensemble streamflow simulations, for both dry and rainy seasons, are obtained by forcing the Distributed Hydrological Model developed by the Brazilian National Institute for Space Research (MHD-INPE) with the satellite rainfall products, corrected using a two-dimensional stochastic satellite rainfall error model (SREM2D). The ensemble simulations are evaluated using streamflow output derived by forcing the model with reference rainfall gauge data. SREM2D is able to correct for errors in the satellite precipitation data by pushing the modeled streamflow ensemble closer to the reference river discharge, when compared to the simulations forced with uncorrected rainfall input. Ensemble streamflow error statistics (MAE and RMSE) show a decreasing trend as a function of the catchment area for all satellite products, but the rainfall-to-streamflow error propagation does not show any dependence on the basin size. (C) 2015 Elsevier B.V. All rights reserved.