Accurate sea level pressure (SLP) data are critical for the forecasting and monitoring of tropical cyclones (TCs). Previous studies have explored SLP retrieval for TCs using passive microwave observations in a single oxygen band (60 or 118 GHz). Leveraging the Fengyun-3 (FY-3) satellite’s capability to simultaneously observe radiation in both 60 and 118 GHz bands, this study proposes a neural network-based algorithm to retrieve TC SLP fields from combined observations of brightness temperature (TB) and warm TB anomalies in both frequency bands. Application studies were carried out using the joint observations from the Microwave Temperature Sounder-2 (MWTS-2) and Microwave Humidity Sounder-2 (MWHS-2) onboard the FY-3D satellite. Optimal frequency channels for the SLP retrieval were selected based on an information content analysis. SLP retrievals within the TC core regions (within a 2° radius from the centers) were compared with SLP data from the Global Data Assimilation System Final Analysis (GDAS/FNL). The results showed root mean square errors (RMSEs) of 2.78 hPa for tropical depressions and storms, 4.23 hPa for hurricanes or typhoons, and 6.15 hPa for major hurricanes or severe typhoons. Retrievals were also compared within situobservations, and the corresponding RMSEs were 3.82, 4.86, and 5.14 hPa, respectively. Furthermore, comparative experiments between the new algorithm and the previous algorithm that used only observations from a single oxygen band demonstrated that the new algorithm provides more comprehensive SLP information and achieves higher accuracy.

The hostile environment and thick sulfuric acid clouds make it difficult to measure the Venusian lower and middle atmosphere. The Venus Volcano Imaging and Climate Explorer (VOICE), which is an orbiting mission, has planned to exploring the thermal structure and composition of the Venusian lower and middle atmosphere by the Microwave Radiometric Sounder (MWRS). This study presents a forward- and inversion- model for the Venusian lower and middle atmosphere at microwave and submillimeter wavelengths, which is developed to simulate the nadir and limb sounding from the MWRS. Important properties of the model with respect to the Venusian atmosphere are its inclusion of molecular thermal movement and collisions induced spectral lines broadening as well as the effects of refraction on the signal propagation in the atmosphere. The model also has potential applications for definition studies for future downward-looking and limb-looking sensors dedicated to Venus exploration.

As the successor to Sunway TaihuLight, the new Sunway supercomputer has ultra-high computing capacity, but the unique heterogeneous architecture presents performance optimization challenges for High Performance Computing (HPC) applications. Performance modeling is an effective way to discover the performance bottlenecks and then improve the performance of HPC applications. Existing performance modeling techniques do not work well on large-scale HPC applications due to high overhead and low accuracy, and are not suitable for the heterogeneous architecture due to a lack of support for multi-resource parameters. To address the above challenges, we propose an automatic multi-parameter performance modeling method for HPC applications on the new Sunway supercomputer. First, a lightweight performance profiling method is proposed to achieve low overhead performance profiling. Then, performance models with multiple resource parameters based on the Fourier neural operator are built, achieving high prediction accuracy and generalization ability. Finally, the Fourier neural operator is extended on the new Sunway supercomputer to realize the performance modeling automatically. Experimental results show that the average prediction error is less than 10% and the average overhead is less than 4%, and the results are superior to the baselines.

We propose a novel cost-efficient approach to threshold selection for binary web-page classification problems with imbalanced class distributions. In many binary-classification tasks the distribution of classes is highly skewed. In such problems, using uniform random sampling in constructing sample sets for threshold setting requires large sample sizes in order to include a statistically sufficient number of examples of the minority class. On the other hand, manually labeling examples is expensive and budgetary considerations require that the size of sample sets be limited. These conflicting requirements make threshold selection a challenging problem. Our method of sample-set construction is a novel approach based on stratified sampling, in which manually labeled examples are expanded to reflect the true class distribution of the web-page population. Our experimental results show that using false positive rate as the criterion for threshold setting results in lower-variance threshold estimates than using other widely used accuracy measures such as F1 and precision.