Accurate crop classification is the basis of agricultural research, and remote sensing is the only effective measuring technique to classify crops over large areas. Optical remote sensing is effective in regions with good illumination; however, it usually fails to meet requirements for highly accurate crop classification in cloud-covered areas and rainy regions. Synthetic aperture radar (SAR) can achieve active data acquisition by transmitting signals; thus, it has strong resistance to cloud and rain interference. In this study, we designed an improved crop planting structure mapping framework for cloudy and rainy regions by combining optical data and SAR data, and we revealed the synchronous-response relationship of these two data types. First, we extracted geo-parcels from optical images with high spatial resolution. Second, we built a recurrent neural network (RNN)-based classifier suitable for remote sensing images on the geo-parcel scale. Third, we classified crops based on the two datasets and established the network. Fourth, we analyzed the synchronous response relationships of crops based on the results of the two classification schemes. This work is the basis for the application of remote sensing data for the fine mapping and growth monitoring of crop planting structures in cloudy and rainy areas in the future.

Accurate spatialization of socioeconomic data is conducive to understand the spatial and temporal distribution of human social development status and, thus, effectively support future scientific decision-making. This study focuses on population mapping, which is a classical spatialization of macroeconomic data of the social economy. Traditional population mapping based on rough grids or administrative divisions such as townships often has deficiencies in the accuracy of spatial pattern and prediction. In this article, hence, we employ residential geo-objects as basic mapping units and formalize the problem as a spatial prediction process using machine-learning (ML) methods with high-spatial-resolution (HSR) satellite remote sensing images and multisource geospatial data. The indicators of population spatial density, including residential geo-objects' area, building existence index, terrain slope, night light intensity, density of point of interest (POI) and road network from Internet electronic maps, and locational factors such as the distances from road and river, are jointly applied to establish the relationship between these multivariable factors and quantitative index of population density using ML algorithms such as Random Forests and XGBoost. The predicated values of population density from the mined nonlinear regression relation are further used to calculate the weights of disaggregation of each unit, and then the population quantity distribution at the scale of residential geo-objects is obtained under the control of the total amount of population statistics. Experiments with a county area show that the methodology has the ability to achieve better results than the traditional deterministic methods by reproducing a more accurate and finer geographic population distribution pattern. Meanwhile, it is found that the optimization of mapping results may benefit from the multisources geospatial data, and thus the methodological framework can be recommended to be extended to other spatialization areas of socioeconomic data.