With the advancement of agricultural modernization, agricultural machinery is widely used for crop harvesting. Traditionally, agricultural machines must be refueled at gas stations regularly, affecting the harvesting efficiency. A mobile refueling service has emerged in recent years, in which refueling tankers can move to serve the refueling request. However, the current mobile refueling system is still in an on-demand mode, which may not achieve timely response. Therefore, in this paper, we propose a new mobile refueling mode, i.e., predictive mobile refueling. To tackle the challenge of sparse rewards in predictive mobile refueling, we develop a two-stage reinforcement learning-based scheduling strategy MobRef, which decouples the scheduling process into a central request dispatcher and a distributed tanker reposition scheduler, and further introduces a potential energy-based reward shaping function to facilitate the training of the reposition scheduler. Extensive experiments on two real-world datasets demonstrate the effectiveness of MobRef, which outperforms the best baseline by 12.71% on average. We also present a deployed system based on MobRef, which is used internally in China National Petroleum Corporation.

In this research, we proposed a data management system for forestry remote sensing images. The methodological foundation of this research is the global subdivision theory. The storage and management of massive remote sensing data have become one of the key issues in the field of spatial information science, public service and commercial activities. The proposed management system for forestry remote sensing images is based on GeoSOT global subdivision model. This management system simplifies the process for forestry remote sensing data collection, storage, analysis and display procedures; besides, it demonstrates the potential to be applied to several forestry related applications.

To meet the demand of the real-time monitoring of urban land use, this work establishes a grid-coding index under the framework of a global subdivision grid and constructs a grid remote sensing intelligent interpretation model, based on the fine-tuning ResNet-18 model. The proposed model realizes real-time sharing and query statistics of grid learning results. Setting construction land identification as our objective, SPOT 6 satellite data of September 2018 and Gaofen-l satellite data of October 2018 were used in an experiment to train and validate the fine-tuning ResNet-18 model. The overall test accuracies of each data set were 97% and 91%, respectively, effectively realizing intelligent change monitoring and distribution statistics with regard to construction land.

On June 2, 2018, the Luojia 1-01 (LJ1-01) nighttime light satellite was launched from China with a spatial resolution of 130 m at nadir, which is a significant improvement over the resolutions of previous nighttime light satellites. However, few studies have focused on the applications of LJ1-01 imagery. This letter aimed to evaluate the potential of utilizing LJ1-01 data to extract built-up urban areas in comparison with Visible Infrared Imaging Radiometer Suite (VIIRS) day-night band (DNB) data by conducting a case study of 16 cities involved in the Belt and Road Initiative in China. The built-up urban areas of the 16 cities were extracted by thresholding segmentation in reference to administrative statistical data, and 30-m-resolution artificial impervious data were adopted as the benchmark (ground truth). Qualitative and quantitative assessments were implemented for an evaluation, revealing that the built-up urban areas extracted by LJ1-01 data outperformed the areas extracted by VIIRS DNB data. We argue that these improvements originate from the superior spatial resolution and image quality of LJ1-01 over its predecessors. The LJ1-01 data presented excellent suitability for the extraction of built-up urban areas at the city scale, and hence, these data may be further applied to other related investigations.

Nighttime light (NTL) remote sensing offers an efficient method to estimate electric power consumption (EPC). However, few research applied the estimation model of a certain year to other years' data for validation, and its transferability across years has not been investigated. Taking mainland China as an example, this paper aims to develop an EPC estimation model which is transferable across years. Partial least square regression (PLSR) was used to build the estimation model for EPC with entire DN series of NTL. Then, the model of each single year 2004-2013 was applied to the other years to validate its transferability across years. Results revealed that PLSR performed well both in model calibration (mean R2c=0.89) and model transfer (mean R2p=0.87). We concluded that PLSR models could effectively excavated reliable quantitative relationship between NTL and EPC, and held good transferability across years.

This research focuses on a booming sharing phenomenon- bicycle sharing in China. Although various bicycle sharing schemes have made residents' daily lives especially commuting behaviors more convenient and won praises from different aspects, they also give rise to several problems. Due to the high cost of data integration, low automation level and heterogeneity, we need a universal management system for the rapidly increasing sharing bicycles. In this research, we put forward GBSS (GeoSOT bicycle sharing system) which is an integrated management system for bicycle sharing in China. The system is based on the GeoSOT global subdivision grid. We acquired location data of Mobike in Suzhou, Jiangsu at two different timestamps on a single day to verify the reliability and efficiency of GBSS. The results proved that its application on the bicycle sharing system has the potential to improving traffic conditions, increasing service efficiency in the future.

Buildings are important elements in modern cities, and the identification of buildings provides the foundation to smart city construction. In this paper, we aimed to put forward a new identification code for buildings from the aspect of spatial fault tolerance. Taking the location information into consideration, buildings were identified with one-dimensional integer codes based on discrete global grid system (DGGS). Results illustrated that this identification code could express location information of buildings explicitly, and remained unchanged even when there were tiny spatial offsets. The percentage of unchanged codes was 90%, 81%, 65% 32% under different offsets of 0.5 m, 1 m, 2 m and 5 m, respectively. We concluded that this identification code was conducive to the more efficient management of buildings, and might additionally benefit the smart city construction.

With the rapid development of the internet technology and sensor technology, the global spatial data, along with its large-volume and multi-scale features, demonstrates higher requirements for data organization. The spatial grid has been widely used in many map index. However, when spatial objects have multi-scale, the spatial grid index with equivalent grids has been critized for its low-efficiencies. To solve this problem, we proposed a logical index as well as inverted index based on GeoSOT subdivision grids. A logical index is established for detailing the inside information of spatial objects and it is organized as inverted index which is an index data structure storing a mapping from content. By means of this index, we can achieve query operation efficiently regardless of the scales of spatial objects. A series of experiments are conducted to authenticate the advantages of the proposed method. In the experiment, we perform our model on MySQL database and randomly generate 50 thousand spatial object(polygons) as simulated data. Moreover, we compare our model to spatial grid index model with equivalent grids to verify its efficiency. Experimental results show that our approach performs 2 times faster than traditional method.

The entity's ID code is important for the identification information as well as the foundation of computer processing in spatial information organization and management system. Based on GeoSOT (Geographical coordinates Subdividing grid with One dimension integral coding on 2n-Tree) subdivision system, this paper puts forward a model for spatial entity's global unique ID code from three aspects of subdivision coding, transcoding, physical storage. Experiments on remote sensing image coding demonstrates the coding model's advantages in location identification and topological calculation, which indicates that, this model is efficient in identification and management for the global mass geo-spatial entities.

This documents put forward an advanced method to manage the rapid increasing spatial data. The core idea of the proposed method is the application of GeoSOT grid. In this geo-information management system, all the spatial data are marked by one or more codes for identification and the mentioned code is simply related to the geographical attributes regardless of the other characteristics. Further, we pay attention to the agricultural service applications and the dwelling service applications, and it turned out that our management method demonstrates the advantages of the traditional methods. Because of the high efficiency and convenience of this method, its application scenario is supposed to be quite broad in the future.

With the rapid development of remote sensing data, the instant and high-efficient analysis on remote sensing image become a challenge. In our research, we adopt the main idea of Merkle Tree and incorporate it with the regular quad-tree to generate an advanced Merkle Quad-Tree structure to assist the remote sensing image analysis. A series of experiments are conducted to authenticate the advantages of the proposed method. In the experiment, the remote sensing images from “Beijing 1” Micro-satellite Image in 2015 are modified on a different level to examine the comparison time consuming for various methods. It turns out that the Merkle Quad-Tree assisting method exhibits stable advantages with the different levels of modifications and which is around 1 percent of the time consumption of the traditional method. The considerable improvements seem to come primarily from the data-mapping process which results in a small and fixed data size.

In the application of geographic data inquiry, traditional search engine put up with queries using keyword matching method mostly, which, however, can hardly get the most accurate and expected answers for users, because spatial semantic queries described by human natural language are difficult for computers to understand. To solve this problem, a method of semantic translation in geographic space using global subdivision grid is put forward in this paper. By summing up the geographic spatial semantic expressing pattern, extracting spatial keywords from searching sentence, finally establish the relationship between spatial semantic meaning and spatial data results. Experiments are also designed and implemented, which results show that method in this paper could accomplish the geographic spatial semantic translation with fair accuracy.

Currently the UAV photographic data is unstable, turbulent and inaccurate, and these flaws make image stitching difficult. This paper proposed an innovative fast method based on global subdivision GeoSOT grid frame. And this method can apply to the high-speed image stitching of a certain area at different times on the UAV. The experiment result shows the advantages of the new method over the traditional one.

This document is a first exploration on the feasibility of the Non-sql database HBase and GeoSOT grids on the management of spatial data in big data time of spatial database. The traditional methods to face the challenge is mainly on the downward expanding; and the consumption of time and storage space is huge, also with a high price. We combine the distributed database HBase and global subdivision grid for data management in this research. The geocodes of the grid can present the spatial position of a spatial object and be regarded as the keyvalue in HBase. The design changes the traditional object-oriented database to geography-oriented database and more geographical characteristics are taken into account. We also make a comparison experiment on our theory and the result turns out that our new method has more advantages of geographical in query when the data volumn is huge.

This paper proposes a new method to express geographical objects. The method is based on GeoSOT global subdivision model. With a number of advantages of this model, our method is equipped with accuracy and concision. In addition, we focus on the geographical topology relation computation, the superiority of our method is also obvious.