Integrating remote sensing and machine learning to detect turbidity anomalies in hydroelectric reservoirs.

Monitoring water quality in reservoirs is essential for the maintenance of aquatic ecosystems and socioeconomic services. In this scenario, the observation of abrupt elevations of physicochemical parameters, such as turbidity and other indicators, can signal anomalies associated with the occurrence of critical events, requiring operational actions and planning to mitigate negative environmental impacts on water resources. This work aims to integrate Machine Learning methods specialized in anomaly detection with data obtained from remote sensing images to identify with high turbidity events in the surface water of the Três Marias Hydroelectric Reservoir. Four distinct threshold-based scenarios were evaluated, in which the overall performance, based on F1-score, showed decreasing trends as the thresholds became more restrictive. In general, the anomaly identification maps generated through the models ratified the applicability of the methods in the diagnosis of surface water in reservoirs in distinct hydrological contexts (dry and wet), effectively identifying locations with anomalous turbidity values.

Modeling the effect of land use/land cover on nitrogen, phosphorous and dissolved oxygen loads in the Velhas River using the concept of exclusive contribution area.

Non-point source water pollution is a major problem in most parts of the world, but is also very difficult to quantify and control since it is not easily separated from point sources and can theoretically originate from the whole watershed. In this article, we evaluate the relationship between land use and land cover and four water pollution parameters in a watershed in Southeast Brazil. The four parameters are nitrate, total ammonia nitrogen, total phosphorous, and dissolved oxygen. To help concentrate on non-point source pollution, only data from the wet seasons of the time period (2001-2013) were analysed, based on the fact that precipitation causes runoff which is the main cause of diffuse pollution. The parameters measured were transformed into loads, which were in turn associated with an exclusive contribution area, so that every measuring station could be considered independent. Analyses were also performed on riparian zones of different widths to verify if the effect of the land cover on the water quality of the stream decreases with the increased distance. Pearson correlation coefficients indicate that urban areas and agriculture/pasture tend to worsen water quality (source). Conversely, forest and riparian areas have a reducing effect on pollution (sink). The best results were obtained for total ammonia nitrogen and dissolved oxygen using the whole exclusive contribution areas with determination coefficients better than R (2)≈0.8. Nitrate and total phosphorous did not produce valid models. We suspect that the transformation delay from total ammonia nitrogen to nitrate might be an important factor for the poor result for this parameter. For phosphorous, we think that the phosphorous sink in the bottom sediment might be the most limiting factor explaining the failure of our models.

A spatial-statistical approach for modeling the effect of non-point source pollution on different water quality parameters in the Velhas river watershed--Brazil.

In this article, a methodology for evaluating the effect of land use/land cover on the quality of nearby stream water in a semiarid environment is described and tested on a large watershed in Southeastern Brazil. The approach aims at identifying the width of the riparian area having the strongest effect on different water quality parameters. The land use/land cover data were generated from remotely sensed data while water quality point data were supplied by a government agency. Testing was conducted for both the rainy and dry seasons in an effort to understand the direct effect of surface runoff. The approach combines cartographic modelling using a geographical information system (GIS) and statistics to establish the strength of the relationship between water quality, land use and the distance from the stream. Results suggest a strong relationship between land use/land cover and turbidity, nitrogen and fecal coliforms. They also suggest that each of these parameters has a unique behavior when distance from the stream is considered. Finally, although it was expected that the models would apply better during the wet season, some parameters had the opposite behavior and displayed a better fit during the dry season.

An Evaluation of Radarsat-1 and ASTER Data for Mapping Veredas (Palm Swamps).

Veredas (palm swamps) are wetland complexes associated with the Brazilian savanna (cerrado) that often represent the only available source of water for the ecosystem during the dry months. Their extent and condition are mainly unknown and their cartography is an essential issue for their protection. This research article evaluates some of the fine resolution satellite data both in the radar (Radarsat-1) and optical domain (ASTER) for the delineation and characterization of veredas. Two separate approaches are evaluated. First, given the known potential of Radarsat-1 images for wetland inventories, the automatic delineation of veredas is tested using only Radarsat-1 data and a Markov random fields region-based segmentation. In this case, to increase performance, processing is limited to a buffer zone around the river network. Then, characterization of their type is attempted using traditional classification methods of ASTER optical data combined with Radarsat-1 data. The automatic classification of Radarsat data yielded results with an overall accuracy between 62 and 69%, that proved reliable enough for delineating wide and very humid veredas. Scenes from the wet season and with a smaller angle of incidence systematically yielded better results. For the classification of the main vegetation types, better results (overall success of 78.8%) were obtained by using only the visible and near infrared (VNIR) bands of the ASTER image. Radarsat data did not bring any improvement to these classification results. In fact, when using solely the Radarsat data from two different angle of incidence and two different dates, the classification results were low (50.8%) but remained powerful for delineating the permanently moist riparian forest portion of the veredas with an accuracy better than 75% in most cases. These results are considered good given the width of some types often less than 50 m wide compared with the resolution of the images (12.5 - 15 m). Comparing the classification results with the Radarsat-generated delineation allows an understanding of the relation between synthetic aperture radar (SAR) backscattering and vegetation types of the veredas.