Search

The Water Observation and Information System (WOIS) is an open source software tool for monitoring, assessing and inventorying water resources in a cost-effective manner using Earth Observation (EO) data. The WOIS has been developed by, among others, the authors of this paper under the TIGER-NET project, which is a major component of the TIGER initiative of the European Space Agency (ESA) and whose main goal is to support the African Earth Observation Capacity for Water Resource Monitoring. TIGER-NET aims to support the satellite-based assessment and monitoring of water resources from watershed to cross-border basin levels through the provision of a free and powerful software package, with associated capacity building, to African authorities. More than 28 EO data processing solutions for water resource management tasks have been developed, in correspondence with the requirements of the participating key African water authorities, and demonstrated with dedicated case studies utilizing the software in operational scenarios. They cover a wide range of themes and information products, including basin-wide characterization of land and water resources, lake water quality monitoring, hydrological modeling and flood forecasting and mapping. For each monitoring task, step-by-step workflows were developed, which can either be adjusted by the user or largely automatized to feed into existing data streams and reporting schemes. The WOIS enables African water authorities to fully exploit the increasing EO capacity offered by current and upcoming generations of satellites, including the Sentinel missions. [ABSTRACT FROM AUTHOR]

The type of algorithm employed to classify remote sensing imageries plays a great role in affecting the accuracy. In recent decades, machine learning (ML) has received great attention due to its robustness in remote sensing image classification. In this regard, random forest (RF) and support vector machine (SVM) are two of the most widely used ML algorithms to generate land cover (LC) maps from satellite imageries. Although several comparisons have been conducted between these two algorithms, the findings are contradicting. Moreover, the comparisons were made on local-scale LC map generation either from high or medium resolution images using various software, but not Python. In this paper, we compared the performance of these two algorithms for large area LC mapping of parts of Africa using coarse resolution imageries in the Python platform by the employing Scikit-Learn (sklearn) library. We employed a big dataset, 297 metrics, comprised of systematically selected 9-month composite FegnYun-3C (FY-3C) satellite images with 1 km resolution. Several experiments were performed using a range of values to determine the best values for the two most important parameters of each classifier, the number of trees and the number of variables, for RF, and penalty value and gamma for SVM, and to obtain the best model of each algorithm. Our results showed that RF outperformed SVM yielding 0.86 (OA) and 0.83 (k), which are 1–2% and 3% higher than the best SVM model, respectively. In addition, RF performed better in mixed class classification; however, it performed almost the same when classifying relatively pure classes with distinct spectral variation, i.e., consisting of less mixed pixels. Furthermore, RF is more efficient in handling large input datasets where the SVM fails. Hence, RF is a more robust ML algorithm especially for heterogeneous large area mapping using coarse resolution images. Finally, default parameter values in the sklearn library work well for satellite image classification with minor/or no adjustment for these algorithms. [ABSTRACT FROM AUTHOR]

Near real-time precipitation is essential to many applications. In Africa, the lack of dense rain-gauge networks and ground weather radars makes the use of satellite precipitation products unavoidable. Despite major progresses in estimating precipitation rate from remote sensing measurements over the past decades, satellite precipitation products still suffer from quantitative uncertainties and biases compared to ground data. Consequently, almost all precipitation products are provided in two modes: a real-time mode (also called early-run or raw product) and a corrected mode (also called final-run, adjusted or post-processed product) in which ground precipitation measurements are integrated in algorithms to correct for bias, generally at a monthly timescale. This paper describes a new methodology to provide a near-real-time precipitation product based on satellite precipitation and soil moisture measurements. Recent studies have shown that soil moisture intrinsically contains information on past precipitation and can be used to correct precipitation uncertainties. The PrISM (Precipitation inferred from Soil Moisture) methodology is presented and its performance is assessed for five in situ rainfall measurement networks located in Africa in semi-arid to wet areas: Niger, Benin, Burkina Faso, Central Africa, and East Africa. Results show that the use of SMOS (Soil Moisture and Ocean Salinity) satellite soil moisture measurements in the PrISM algorithm most often improves the real-time satellite precipitation products, and provides results comparable to existing adjusted products, such as TRMM (Tropical Rainfall Measuring Mission), GPCC (Global Precipitation Climatology Centre) and IMERG (Integrated Multi-satellitE Retrievals for GPM), which are available a few weeks or months after their detection. [ABSTRACT FROM AUTHOR]