Following the progress of satellite data assimilation in the 1990s, the combination of meteorological satellites and numerical models has changed the way scientists understand the earth. With the evolution of numerica...Following the progress of satellite data assimilation in the 1990s, the combination of meteorological satellites and numerical models has changed the way scientists understand the earth. With the evolution of numerical weather prediction models and earth system models, meteorological satellites will play a more important role in earth sciences in the future. As part of the space-based infrastructure, the Fengyun (FY) meteorological satellites have contributed to earth science sustainability studies through an open data policy and stable data quality since the first launch of the FY-1A satellite in 1988. The capability of earth system monitoring was greatly enhanced after the second-generation polar orbiting FY-3 satellites and geostationary orbiting FY-4 satellites were developed. Meanwhile, the quality of the products generated from the FY-3 and FY-4 satellites is comparable to the well-known MODIS products. FY satellite data has been utilized broadly in weather forecasting, climate and climate change investigations, environmental disaster monitoring, etc. This article reviews the instruments mounted on the FY satellites. Sensor-dependent level 1 products (radiance data) and inversion algorithm-dependent level 2 products (geophysical parameters) are introduced. As an example, some typical geophysical parameters, such as wildfires, lightning, vegetation indices, aerosol products, soil moisture, and precipitation estimation have been demonstrated and validated by in-situ observations and other well-known satellite products. To help users access the FY products, a set of data sharing systems has been developed and operated. The newly developed data sharing system based on cloud technology has been illustrated to improve the efficiency of data delivery.展开更多
Model validation is the most important part of building a supervised model.For building a model with good generalization performance one must have a sensible data splitting strategy,and this is crucial for model valid...Model validation is the most important part of building a supervised model.For building a model with good generalization performance one must have a sensible data splitting strategy,and this is crucial for model validation.In this study,we con-ducted a comparative study on various reported data splitting methods.The MixSim model was employed to generate nine simulated datasets with different probabilities of mis-classification and variable sample sizes.Then partial least squares for discriminant analysis and support vector machines for classification were applied to these datasets.Data splitting methods tested included variants of cross-validation,bootstrapping,bootstrapped Latin partition,Kennard-Stone algorithm(K-S)and sample set partitioning based on joint X-Y distances algorithm(SPXY).These methods were employed to split the data into training and validation sets.The estimated generalization performances from the validation sets were then compared with the ones obtained from the blind test sets which were generated from the same distribution but were unseen by the train-ing/validation procedure used in model construction.The results showed that the size of the data is the deciding factor for the qualities of the generalization performance estimated from the validation set.We found that there was a significant gap between the performance estimated from the validation set and the one from the test set for the all the data splitting methods employed on small datasets.Such disparity decreased when more samples were available for training/validation,and this is because the models were then moving towards approximations of the central limit theory for the simulated datasets used.We also found that having too many or too few samples in the training set had a negative effect on the estimated model performance,suggesting that it is necessary to have a good balance between the sizes of training set and validation set to have a reliable estimation of model performance.We also found that systematic sampling method such a展开更多
基金This work was supported by the National Key Research and Development Program of China(2018YFB0504900,2018YFB0504905).
文摘Following the progress of satellite data assimilation in the 1990s, the combination of meteorological satellites and numerical models has changed the way scientists understand the earth. With the evolution of numerical weather prediction models and earth system models, meteorological satellites will play a more important role in earth sciences in the future. As part of the space-based infrastructure, the Fengyun (FY) meteorological satellites have contributed to earth science sustainability studies through an open data policy and stable data quality since the first launch of the FY-1A satellite in 1988. The capability of earth system monitoring was greatly enhanced after the second-generation polar orbiting FY-3 satellites and geostationary orbiting FY-4 satellites were developed. Meanwhile, the quality of the products generated from the FY-3 and FY-4 satellites is comparable to the well-known MODIS products. FY satellite data has been utilized broadly in weather forecasting, climate and climate change investigations, environmental disaster monitoring, etc. This article reviews the instruments mounted on the FY satellites. Sensor-dependent level 1 products (radiance data) and inversion algorithm-dependent level 2 products (geophysical parameters) are introduced. As an example, some typical geophysical parameters, such as wildfires, lightning, vegetation indices, aerosol products, soil moisture, and precipitation estimation have been demonstrated and validated by in-situ observations and other well-known satellite products. To help users access the FY products, a set of data sharing systems has been developed and operated. The newly developed data sharing system based on cloud technology has been illustrated to improve the efficiency of data delivery.
基金YX and RG thank Wellcome Trust for funding MetaboFlow(Grant 202952/Z/16/Z).
文摘Model validation is the most important part of building a supervised model.For building a model with good generalization performance one must have a sensible data splitting strategy,and this is crucial for model validation.In this study,we con-ducted a comparative study on various reported data splitting methods.The MixSim model was employed to generate nine simulated datasets with different probabilities of mis-classification and variable sample sizes.Then partial least squares for discriminant analysis and support vector machines for classification were applied to these datasets.Data splitting methods tested included variants of cross-validation,bootstrapping,bootstrapped Latin partition,Kennard-Stone algorithm(K-S)and sample set partitioning based on joint X-Y distances algorithm(SPXY).These methods were employed to split the data into training and validation sets.The estimated generalization performances from the validation sets were then compared with the ones obtained from the blind test sets which were generated from the same distribution but were unseen by the train-ing/validation procedure used in model construction.The results showed that the size of the data is the deciding factor for the qualities of the generalization performance estimated from the validation set.We found that there was a significant gap between the performance estimated from the validation set and the one from the test set for the all the data splitting methods employed on small datasets.Such disparity decreased when more samples were available for training/validation,and this is because the models were then moving towards approximations of the central limit theory for the simulated datasets used.We also found that having too many or too few samples in the training set had a negative effect on the estimated model performance,suggesting that it is necessary to have a good balance between the sizes of training set and validation set to have a reliable estimation of model performance.We also found that systematic sampling method such a
