Walter Collischonn

O desastre climático de abril e maio de 2024 no Rio Grande do Sul ocorreu em um contexto de mudanças climáticas, que vêm intensificando eventos extremos na região. A crise causada por chuvas extraordinárias expôs falhas estruturais e de gestão de riscos,evidenciando a urgência de comunicação eficaz. Este artigo apresenta uma cronologia do desastre e um relato das ações voluntárias de pesquisadores na produção e disseminação de informações essenciais, incluindo previsão do nível do Guaíba, mapas de risco, suporte técnico no Vale do Taquari, monitoramento de abrigos, informações sobre saneamento e o Repositório de Mapas. Análises pioneiras sobre os impactos das inundações foram cruciais para a compreensão rápida do evento. O relato destaca desafios como limitações tecnológicas, dificuldades de articulação com a gestão pública e a ausência de um sistema institucional robusto. A inexistência de agências de bacias hidrográficas compromete a gestão integrada dos recursos hídricos e a resposta a desastres no longo prazo. O evento reforça a necessidade de investimentos substanciais em governança hídrica, infraestrutura de monitoramento eestratégias de adaptação para fortalecer a resiliência da sociedade.

In 2023, an unprecedented drought and heat wave severely affected Amazon waters, leading to high mortality of fishes and river dolphins. Five of 10 lakes monitored had exceptionally high daytime water temperatures (over 37°C), with one large lake reaching up to 41°C in the entire approximately 2-meter-deep water column and up to 13°C of diel variation. Modeling showed that high solar radiation, reduced water depth and wind speed, and turbid waters were the main drivers of the high temperatures. This extreme heating of Amazon waters follows a long-term increase of 0.6°C/decade revealed by satellite estimates across the region’s lakes between 1990 and 2023. With ongoing climate change, temperatures that approach or exceed thermal tolerances for aquatic life are likely to become more common in tropical aquatic systems.

Knowing river flows in space and time is fundamental for several hydrological and environmental applications. One of the greatest challenges in hydrology, however, is having this information at every river reach, as we can only focus our resources in obtaining measurements at particular sites. In recent years, regression techniques based on Machine Learning (ML) approaches have been extensively developed, presenting great results in all areas of knowledge. In hydrological sciences, the potential of using these techniques is enormous and still to be explored. In this context, we have built a ML regression modeling pipeline to estimate reference unitary mean flows (qm) and low flows (q95). We trained the models using most up-to-date state-of-the-art environmental data and evaluated results against consistent streamflow data from 1069 gauges spread across Brazil, using a nested 10-fold cross validation approach. In the evaluation, R2 scores of >0.8 for qm and 0.7 for q95 were obtained. Key findings indicate that while simpler models like multiple linear regression performed well for qm, more complex models such as Gradient Boosting Machines and Random Forest offered improved predictions for q95. Average precipitation was the most significant predictor across models, with drainage density and soil organic content playing a crucial role in low flow estimations.

The increase in atmospheric CO2 concentration elevates atmospheric temperature and enhances water storage capacity. This could lead to more extreme precipitation events, while simultaneously, higher temperatures may cause land and soil to dry out. Such alterations create ambiguous expectations regarding the direction of hydrological changes in the following decades. This work quantifies streamflow changes on flood discharges in South America based on the MGB‐SA hydrodynamic‐hydrological model forced with the Climate Model Intercomparison Project (CMIP6) ensemble of climate projections. Future projections indicate that floods are expected to increase in magnitude and become up to five times more frequent in Southern Brazil, a region recently impacted by unprecedented flooding. In contrast, ecosystems such as the Amazon and the Pantanal will likely experience less frequent floods in the coming decades. These projections have relevant implications not only for flood risk in populated areas but also for ecological dynamics.

Climate change is expected to increase the frequency and intensity of flooding, emphasizing the need to better understand these events. Satellite observations provide essential information, and the recent launch of SWOT offers new opportunities to investigate flood dynamics. Here, we use SWOT observations to detect significant hydrologic changes during an unprecedented 2024 flood in southern Brazil. In the Jacuí and Guaíba Rivers, water slopes increased dramatically–up to 11 times (from 0.82 to 9.59 cm/km) and 21 times (0.31–6.58 cm/km) compared to stable conditions. SWOT‐derived surface water elevations captured 99% of water level variability compared to in situ observations. Additionally, flooding in Patos Lagoon was intensified by NE–SW wind forcing, causing water levels to rise by up to 40 cm, worsening the flood event. SWOT proves a unique opportunity to understand extreme flooding events, providing new support for flood risk management in the context of climate change.

We present an overview of the unprecedent hydrological event that occurred in Rio Grande do Sul state in late April and early May 2024, causing floods and landslides in a scale that was not seen before in Brazil. The rainfall that caused the disaster is incomparable to any recent event in Brazil at large spatial and temporal scale. The intense and long-lasting precipitation led to record floods from small rivers with drainage areas below 1000 km² to the Patos Lagoon, which drains more than 180,000 km². The rainfall also led to landslides that rank first in Brazil in terms of spatial scale and number of occurrences. Mountain rivers experienced flash floods with extreme water levels, while in the lowlands the large rivers and the Patos Lagoon experienced long duration flooding, with stage remaining high for about 30 days. The flood event is probably ranked first in terms of displaced people and damage to property and infrastructure in Brazil. Almost 200 fatalities occurred due to the disaster and protection structure failures. The April-May hydrological event show a glimpse of what can be expected along the coming decades due to climate change in a region that covers South Brazil, and parts of the neighboring countries.

Esta nota técnica apresenta uma análise comparativa entre as chuvas intensas ocorridas no Rio Grande do Sul, entre abril e maio de 2024, com os maiores eventos de chuva já observados no Brasil, entre 1961 e 2022. A análise comparativa foi realizada utilizando a técnica Depth Area Duration, que é adequada para a análise de chuvas intensas em grande escala espacial e temporal. Concluímos que as chuvas ocorridas no Rio Grande do Sul superaram amplamente os eventos de precipitação de larga escala mais intensos historicamente observados no Brasil, de 1961 a 2022, pelo menos para áreas de 2.000 a 100.000 km2 e durações de chuva de 3 a 14 dias. Em alguns casos (chuva de 3 dias em 10.000 km2), a precipitação de 2024 foi 47% maior que o recorde anterior de precipitação. Considerando que não podemos descartar que eventos semelhantes ocorram em nenhuma outra bacia, pelo menos no Sul do Brasil, pode ser necessário rever as regras originais de projeto e operação de estruturas hidráulicas de alto risco e sistemas de defesa contra inundações em toda a região Sul do país.

Digital elevation models (DEMs) have a wide range of applications and play a crucial role in many studies. Numerous public DEMs, frequently acquired using radar and optical satellite imagery, are currently available; however, DEM datasets tend to exhibit elevation values influenced by vegetation height and coverage, compromising the accuracy of models in representing terrain elevation. In this study, we developed a digital terrain model for South America using a novel methodology to remove vegetation bias in the Copernicus DEM GLO-30 (COPDEM) model using machine learning, Global Ecosystem Dynamics Investigation (GEDI) elevation data, and multispectral remote sensing products. Our results indicate considerable improvements compared to COPDEM in representing terrain elevation, reducing average errors (BIAS) from 9.6 m to 1.5 m. Furthermore, we evaluated our product (ANADEM) by comparison with other global DEMs, obtaining more accurate results for different conditions of vegetation fraction cover and land use. As a publicly available and open-source dataset, ANADEM will play a crucial role in advancing studies that demand accurate terrain elevation representations at large scales.

Digital elevation models (DEMs) have a wide range of applications and play a crucial role in many studies. Numerous public DEMs, frequently acquired using radar and optical satellite imagery, are currently available; however, DEM datasets tend to exhibit elevation values influenced by vegetation height and coverage, which can compromise the accuracy of models in representing terrain elevation. In this study, we developed a digital terrain model for South America using a novel methodology to remove vegetation bias in the Copernicus DEM GLO-30 (COPDEM) model using machine learning, Global Ecosystem Dynamics Investigation (GEDI) elevation data, and multispectral remote sensing products. Our results indicate considerable improvements compared to COPDEM in representing terrain elevation, reducing average errors (BIAS) from 9.6 m to 1.5 m. Furthermore, we evaluated our product (ANADEM) by comparison with other global DEMs, obtaining more accurate results for different conditions of vegetation fraction cover and land use. As a publicly available and open-source dataset, ANADEM will play a crucial role in advancing studies that demand accurate terrain elevation representations at large scales.

This technical note presents a comparative analysis between the heavy rainfall that occurred in Rio Grande do Sul, between April and May 2024, with the heaviest rainfall events ever observed in Brazil, between 1961 and 2022. The comparative analysis was performed using the Depth Area Duration technique, suitable for the analysis of heavy rainfall on a large spatial and temporal scale. We concluded that following our analysis, the rainfall that occurred in Rio Grande do Sul largely exceeded the most intense large-scale precipitation events historically observed in Brazil, from 1961 to 2022, at least for areas from 2,000 to 100,000 km² and rainfall durations of 3 to 14 days. In some cases (3-day rainfall in 10,000 km²) the 2024 rainfall was 47% higher than the previous record rainfall. Considering that we cannot rule out that similar events occur in any other basin, at least in South Brazil, it may be necessary to review the original design and operation rules of high-risk hydraulic structures and flood defense systems throughout the Southern region of the country.