Comparison of RetinaNet-Based Single-Target Cascading and Multi-Target Detection Models for Administrative Regions in Network Map Pictures.

There is a critical need for detection of administrative regions through network map pictures in map censorship tasks, which can be implemented by target detection technology. However, on map images there tend to be numerous administrative regions overlaying map annotations and symbols, thus making it difficult to accurately detect each region. Using a RetinaNet-based target detection model integrating ResNet50 and a feature pyramid network (FPN), this study built a multi-target model and a single-target cascading model from three single-target models by taking Taiwan, Tibet, and the Chinese mainland as target examples. Two models were evaluated both in classification and localization accuracy to investigate their administrative region detection performance. The results show that the single-target cascading model was able to detect more administrative regions, with a higher f1_score of 0.86 and mAP of 0.85 compared to the multi-target model (0.56 and 0.52, respectively). Furthermore, location box size distribution from the single-target cascading model looks more similar to that of manually annotated box sizes, which signifies that the proposed cascading model is superior to the multi-target model. This study is promising in providing support for computer map reading and intelligent map censorship.

Heat stress may cause a significant reduction of rice yield in China under future climate scenarios.

Extreme heat events have become more frequent and severe under climate change and seriously threaten rice growth. Most existing crop models tend to underestimate the impacts of heat stress on rice yields. Heat stress modules in crop models have not been extensively explored, particularly on a large scale. This study modeled rice growth under heat stress at the flowering and filling stages through two heat stress models which coupled into the CERES-Rice model. We evaluated the advanced model with provincial statistics and Gridded Observed Rice Yield. Our improved CERES-Rice model produced more accurate estimates on rice yield than the original model evidenced by an increased correlation coefficient (R) of 12.72% and d-index of 0.02%. The RMSE and MAE decreased by 5.94% and 6.01%, respectively. Most pseudo positive correlations between rice yield and the number of heat days were corrected to the negative ones by the improved model. The future projections from the improved model signifies multi-model ensemble yield projection without CO2 effect (MME-I-NOCO2) has an apparent fall from 2020 to 2099 under RCP4.5, RCP6.0 and RCP8.5 with the decreasing percentages of 6%, 14%, and 37%, respectively, whereas the decreasing trend (12%) only occurs under RCP8.5 with CO2 effect (MME-I-CO2). The apparently decreasing trends of yield projection from MME-I-NOCO2 will occur in most rice-planted regions of China with the decreasing rate < 50 kg/ha/a especially in the central-south and southern cropping regions, and this decreasing trend will be slowed down for MME-I-CO2. Relative to rice yield of historical period, rice yield variations of MME-I-NOCO2 for different growing seasons show a downward trend with the decrease of approximately 54%, 60%, and 43%, respectively. Our study highlights the importance of modeling crop yields under heat stress to food security, agricultural adaptation and mitigation to climate change.

Landslide Susceptibility Mapping by Fusing Convolutional Neural Networks and Vision Transformer.

Landslide susceptibility mapping (LSM) is an important decision basis for regional landslide hazard risk management, territorial spatial planning and landslide decision making. The current convolutional neural network (CNN)-based landslide susceptibility mapping models do not adequately take into account the spatial nature of texture features, and vision transformer (ViT)-based LSM models have high requirements for the amount of training data. In this study, we overcome the shortcomings of CNN and ViT by fusing these two deep learning models (bottleneck transformer network (BoTNet) and convolutional vision transformer network (ConViT)), and the fused model was used to predict the probability of landslide occurrence. First, we integrated historical landslide data and landslide evaluation factors and analysed whether there was covariance in the landslide evaluation factors. Then, the testing accuracy and generalisation ability of the CNN, ViT, BoTNet and ConViT models were compared and analysed. Finally, four landslide susceptibility mapping models were used to predict the probability of landslide occurrence in Pingwu County, Sichuan Province, China. Among them, BoTNet and ConViT had the highest accuracy, both at 87.78%, an improvement of 1.11% compared to a single model, while ConViT had the highest F1-socre at 87.64%, an improvement of 1.28% compared to a single model. The results indicate that the fusion model of CNN and ViT has better LSM performance than the single model. Meanwhile, the evaluation results of this study can be used as one of the basic tools for landslide hazard risk quantification and disaster prevention in Pingwu County.