Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application

Object-based image classification for land-cover mapping purposes using remote-sensing imagery has attracted significant attention in recent years. Numerous studies conducted over the past decade have investigated a broad array of sensors, feature selection, classifiers, and other factors of interest. However, these research results have not yet been synthesized to provide coherent guidance on the effect of different supervised object-based land-cover classification processes. In this study, we first construct a database with 28 fields using qualitative and quantitative information extracted from 254 experimental cases described in 173 scientific papers. Second, the results of the meta-analysis are reported, including general characteristics of the studies (e.g., the geographic range of relevant institutes, preferred journals) and the relationships between factors of interest (e.g., spatial resolution and study area or optimal segmentation scale, accuracy and number of targeted classes), especially with respect to the classification accuracy of different sensors, segmentation scale, training set size, supervised classifiers, and land-cover types. Third, useful data on supervised object-based image classification are determined from the meta-analysis. For example, we find that supervised object-based classification is currently experiencing rapid advances, while development of the fuzzy technique is limited in the object-based framework. Furthermore, spatial resolution correlates with the optimal segmentation scale and study area, and Random Forest (RF) shows the best performance in object-based classification. The area-based accuracy assessment method can obtain stable classification performance, and indicates a strong correlation between accuracy and training set size, while the accuracy of the point-based method is likely to be unstable due to mixed objects. In addition, the overall accuracy benefits from higher spatial resolution images (e.g., unmanned aerial vehicle) or agricultural sites where it also correlates with the number of targeted classes. More than 95.6% of studies involve an area less than 300 ha, and the spatial resolution of images is predominantly between 0 and 2 m. Furthermore, we identify some methods that may advance supervised object-based image classification. For example, deep learning and type-2 fuzzy techniques may further improve classification accuracy. Lastly, scientists are strongly encouraged to report results of uncertainty studies to further explore the effects of varied factors on supervised object-based image classification.

Inadequate water quality can mean that water is unsuitable for a variety of human uses, thus exacerbating freshwater scarcity. Previous large-scale water scarcity assessments mostly focused on the availability of sufficient freshwater quantity for providing supplies, but neglected the quality constraints on water usability. Here we report a comprehensive nationwide water scarcity assessment in China, which explicitly includes quality requirements for human water uses. We highlight the necessity of incorporating water scarcity assessment at multiple temporal and geographic scales. Our results show that inadequate water quality exacerbates China's water scarcity, which is unevenly distributed across the country. North China often suffers water scarcity throughout the year, whereas South China, despite sufficient quantities, experiences seasonal water scarcity due to inadequate quality. Over half of the population are affected by water scarcity, pointing to an urgent need for improving freshwater quantity and quality management to cope with water scarcity.

Abstract Satellite observations show that leaf area index (LAI) has increased globally since 1981, but the impact of this vegetation structural change on the global terrestrial carbon cycle has not been systematically evaluated. Through process-based diagnostic ecosystem modeling, we find that the increase in LAI alone was responsible for 12.4% of the accumulated terrestrial carbon sink (95 ± 5 Pg C) from 1981 to 2016, whereas other drivers of CO 2 fertilization, nitrogen deposition, and climate change (temperature, radiation, and precipitation) contributed to 47.0%, 1.1%, and −28.6% of the sink, respectively. The legacy effects of past changes in these drivers prior to 1981 are responsible for the remaining 65.5% of the accumulated sink from 1981 to 2016. These results refine the attribution of the land sink to the various drivers and would help constrain prognostic models that often have large uncertainties in simulating changes in vegetation and their impacts on the global carbon cycle.

The successful launch of the Landsat 8 satellite with two thermal infrared bands on February 11, 2013, for continuous Earth observation provided another opportunity for remote sensing of land surface temperature (LST). However, calibration notices issued by the United States Geological Survey (USGS) indicated that data from the Landsat 8 Thermal Infrared Sensor (TIRS) Band 11 have large uncertainty and suggested using TIRS Band 10 data as a single spectral band for LST estimation. In this study, we presented an improved mono-window (IMW) algorithm for LST retrieval from the Landsat 8 TIRS Band 10 data. Three essential parameters (ground emissivity, atmospheric transmittance and effective mean atmospheric temperature) were required for the IMW algorithm to retrieve LST. A new method was proposed to estimate the parameter of effective mean atmospheric temperature from local meteorological data. The other two essential parameters could be both estimated through the so-called land cover approach. Sensitivity analysis conducted for the IMW algorithm revealed that the possible error in estimating the required atmospheric water vapor content has the most significant impact on the probable LST estimation error. Under moderate errors in both water vapor content and ground emissivity, the algorithm had an accuracy of ~1.4 K for LST retrieval. Validation of the IMW algorithm using the simulated datasets for various situations indicated that the LST difference between the retrieved and the simulated ones was 0.67 K on average, with an RMSE of 0.43 K. Comparison of our IMW algorithm with the single-channel (SC) algorithm for three main atmosphere profiles indicated that the average error and RMSE of the IMW algorithm were −0.05 K and 0.84 K, respectively, which were less than the −2.86 K and 1.05 K of the SC algorithm. Application of the IMW algorithm to Nanjing and its vicinity in east China resulted in a reasonable LST estimation for the region. Spatial variation of the extremely hot weather, a frequently-occurring phenomenon of an abnormal heat flux process in summer along the Yangtze River Basin, had been thoroughly analyzed. This successful application suggested that the IMW algorithm presented in the study could be used as an efficient method for LST retrieval from the Landsat 8 TIRS Band 10 data.

Marine accumulations of terrigenous sediment are widely assumed to accurately record climatic- and tectonic-controlled mountain denudation and play an important role in understanding late Cenozoic mountain uplift and global cooling. Underpinning this is the assumption that the majority of sediment eroded from hinterland orogenic belts is transported to and ultimately stored in marine basins with little lag between erosion and deposition. Here we use a detailed and multi-technique sedimentary provenance dataset from the Yellow River to show that substantial amounts of sediment eroded from Northeast Tibet and carried by the river's upper reach are stored in the Chinese Loess Plateau and the western Mu Us desert. This finding revises our understanding of the origin of the Chinese Loess Plateau and provides a potential solution for mismatches between late Cenozoic terrestrial sedimentation and marine geochemistry records, as well as between global CO2 and erosion records.

Abstract. Satellite-based models have been widely used to simulate vegetation gross primary production (GPP) at the site, regional, or global scales in recent years. However, accurately reproducing the interannual variations in GPP remains a major challenge, and the long-term changes in GPP remain highly uncertain. In this study, we generated a long-term global GPP dataset at 0.05∘ latitude by 0.05∘ longitude and 8 d interval by revising a light use efficiency model (i.e., EC-LUE model). In the revised EC-LUE model, we integrated the regulations of several major environmental variables: atmospheric CO2 concentration, radiation components, and atmospheric vapor pressure deficit (VPD). These environmental variables showed substantial long-term changes, which could greatly impact the global vegetation productivity. Eddy covariance (EC) measurements at 95 towers from the FLUXNET2015 dataset, covering nine major ecosystem types around the globe, were used to calibrate and validate the model. In general, the revised EC-LUE model could effectively reproduce the spatial, seasonal, and annual variations in the tower-estimated GPP at most sites. The revised EC-LUE model could explain 71 % of the spatial variations in annual GPP over 95 sites. At more than 95 % of the sites, the correlation coefficients (R2) of seasonal changes between tower-estimated and model-simulated GPP are larger than 0.5. Particularly, the revised EC-LUE model improved the model performance in reproducing the interannual variations in GPP, and the averaged R2 between annual mean tower-estimated and model-simulated GPP is 0.44 over all 55 sites with observations longer than 5 years, which is significantly higher than those of the original EC-LUE model (R2=0.36) and other LUE models (R2 ranged from 0.06 to 0.30 with an average value of 0.16). At the global scale, GPP derived from light use efficiency models, machine learning models, and process-based biophysical models shows substantial differences in magnitude and interannual variations. The revised EC-LUE model quantified the mean global GPP from 1982 to 2017 as 106.2±2.9 Pg C yr−1 with the trend 0.15 Pg C yr−1. Sensitivity analysis indicated that GPP simulated by the revised EC-LUE model was sensitive to atmospheric CO2 concentration, VPD, and radiation. Over the period of 1982–2017, the CO2 fertilization effect on the global GPP (0.22±0.07 Pg C yr−1) could be partly offset by increased VPD (-0.17±0.06 Pg C yr−1). The long-term changes in the environmental variables could be well reflected in global GPP. Overall, the revised EC-LUE model is able to provide a reliable long-term estimate of global GPP. The GPP dataset is available at https://doi.org/10.6084/m9.figshare.8942336.v3 (Zheng et al., 2019).

This study provides a comprehensive evaluation of eight high spatial resolution gridded precipitation products in Adige Basin located in Italy within 45-47.1°N. The Adige Basin is characterized by a complex topography, and independent ground data are available from a network of 101 rain gauges during 2000-2010. The eight products include the Version 7 TRMM (Tropical Rainfall Measuring Mission) Multi-satellite Precipitation Analysis 3B42 product, three products from CMORPH (the Climate Prediction Center MORPHing technique), i.e., CMORPH_RAW, CMORPH_CRT and CMORPH_BLD, PCDR (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record), PGF (Global Meteorological Forcing Dataset for land surface modelling developed by Princeton University), CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data) and GSMaP_MVK (Global Satellite Mapping of Precipitation project Moving Vector with Kalman-filter product). All eight products are evaluated against interpolated rain gauge data at the common 0.25° spatial resolution, and additional evaluations at native finer spatial resolution are conducted for CHIRPS (0.05°) and GSMaP_MVK (0.10°). Evaluation is performed at multiple temporal (daily, monthly and annual) and spatial scales (grid and watershed). Evaluation results show that in terms of overall statistical metrics the CHIRPS, TRMM and CMORPH_BLD comparably rank as the top three best performing products, while the PGF performs worst. All eight products underestimate and overestimate the occurrence frequency of daily precipitation for some intensity ranges. All products tend to show higher error in the winter months (December-February) when precipitation is low. Very slight difference can be observed in the evaluation metrics and aspects between at the aggregated 0.25° spatial resolution and at the native finer resolutions (0.05°) for CHIRPS and (0.10°) for GSMaP_MVK products. This study has implications for precipitation product development and the global view of the performance of various precipitation products, and provides valuable guidance when choosing alternative precipitation data for local community.

The outbreak of Corona Virus Disease 2019 (COVID-19) is a grave global public health emergency. Nowadays, social media has become the main channel through which the public can obtain information and express their opinions and feelings. This study explored public opinion in the early stages of COVID-19 in China by analyzing Sina-Weibo (a Twitter-like microblogging system in China) texts in terms of space, time, and content. Temporal changes within one-hour intervals and the spatial distribution of COVID-19-related Weibo texts were analyzed. Based on the latent Dirichlet allocation model and the random forest algorithm, a topic extraction and classification model was developed to hierarchically identify seven COVID-19-relevant topics and 13 sub-topics from Weibo texts. The results indicate that the number of Weibo texts varied over time for different topics and sub-topics corresponding with the different developmental stages of the event. The spatial distribution of COVID-19-relevant Weibo was mainly concentrated in Wuhan, Beijing-Tianjin-Hebei, the Yangtze River Delta, the Pearl River Delta, and the Chengdu-Chongqing urban agglomeration. There is a synchronization between frequent daily discussions on Weibo and the trend of the COVID-19 outbreak in the real world. Public response is very sensitive to the epidemic and significant social events, especially in urban agglomerations with convenient transportation and a large population. The timely dissemination and updating of epidemic-related information and the popularization of such information by the government can contribute to stabilizing public sentiments. However, the surge of public demand and the hysteresis of social support demonstrated that the allocation of medical resources was under enormous pressure in the early stage of the epidemic. It is suggested that the government should strengthen the response in terms of public opinion and epidemic prevention and exert control in key epidemic areas, urban agglomerations, and transboundary areas at the province level. In controlling the crisis, accurate response countermeasures should be formulated following public help demands. The findings can help government and emergency agencies to better understand the public opinion and sentiments towards COVID-19, to accelerate emergency responses, and to support post-disaster management.

Food safety is a major concern for the Chinese public. This study collected 465 published papers on heavy metal pollution rates (the ratio of the samples exceeding the Grade II limits for Chinese soils, the Soil Environmental Quality Standard-1995) in farmland soil throughout China. The results showed that Cd had the highest pollution rate of 7.75%, followed by Hg, Cu, Ni and Zn, Pb and Cr had the lowest pollution rates at lower than 1%. The total pollution rate in Chinese farmland soil was 10.18%, mainly from Cd, Hg, Cu, and Ni. The human activities of mining and smelting, industry, irrigation by sewage, urban development, and fertilizer application released certain amounts of heavy metals into soil, which resulted in the farmland soil being polluted. Considering the spatial variations of grain production, about 13.86% of grain production was affected due to the heavy metal pollution in farmland soil. These results many provide valuable information for agricultural soil management and protection in China.

Lake size is sensitive to both climate change and human activities, and therefore serves as an excellent indicator to assess environmental changes. Using a large volume of various datasets, we provide a first complete picture of changes in China's lakes between 1960s–1980s and 2005–2006. Dramatic changes are found in both lake number and lake size; of these, 243 lakes vanished mainly in the northern provinces (and autonomous regions) and also in some southern provinces while 60 new lakes appeared mainly on the Tibetan Plateau and neighboring provinces. Limited evidence suggested that these geographically unbalanced changes might be associated primarily with climate change in North China and human activities in South China, yet targeted regional studies are required to confirm this preliminary observation.

El Niño's intensity change under anthropogenic warming is of great importance to society, yet current climate models' projections remain largely uncertain. The current classification of El Niño does not distinguish the strong from the moderate El Niño events, making it difficult to project future change of El Niño's intensity. Here we classify 33 El Niño events from 1901 to 2017 by cluster analysis of the onset and amplification processes, and the resultant 4 types of El Niño distinguish the strong from the moderate events and the onset from successive events. The 3 categories of El Niño onset exhibit distinct development mechanisms. We find El Niño onset regime has changed from eastern Pacific origin to western Pacific origin with more frequent occurrence of extreme events since the 1970s. This regime change is hypothesized to arise from a background warming in the western Pacific and the associated increased zonal and vertical sea-surface temperature (SST) gradients in the equatorial central Pacific, which reveals a controlling factor that could lead to increased extreme El Niño events in the future. The Coupled Model Intercomparison Project phase 5 (CMIP5) models' projections demonstrate that both the frequency and intensity of the strong El Niño events will increase significantly if the projected central Pacific zonal SST gradients become enhanced. If the currently observed background changes continue under future anthropogenic forcing, more frequent strong El Niño events are anticipated. The models' uncertainty in the projected equatorial zonal SST gradients, however, remains a major roadblock for faithful prediction of El Niño's future changes.

Scene classification from remote sensing images provides new possibilities for potential application of high spatial resolution imagery. How to efficiently implement scene recognition from high spatial resolution imagery remains a significant challenge in the remote sensing domain. Recently, convolutional neural networks (CNN) have attracted tremendous attention because of their excellent performance in different fields. However, most works focus on fully training a new deep CNN model for the target problems without considering the limited data and time-consuming issues. To alleviate the aforementioned drawbacks, some works have attempted to use the pretrained CNN models as feature extractors to build a feature representation of scene images for classification and achieved successful applications including remote sensing scene classification. However, existing works pay little attention to exploring the benefits of multilayer features for improving the scene classification in different aspects. As a matter of fact, the information hidden in different layers has great potential for improving feature discrimination capacity. Therefore, this paper presents a fusion strategy for integrating multilayer features of a pretrained CNN model for scene classification. Specifically, the pretrained CNN model is used as a feature extractor to extract deep features of different convolutional and fully connected layers; then, a multiscale improved Fisher kernel coding method is proposed to build a mid-level feature representation of convolutional deep features. Finally, the mid-level features extracted from convolutional layers and the features of fully connected layers are fused by a principal component analysis/spectral regression kernel discriminant analysis method for classification. For validation and comparison purposes, the proposed approach is evaluated via experiments with two challenging high-resolution remote sensing data sets, and shows the competitive performance compared with fully trained CNN models, fine-tuning CNN models, and other related works.

The aim of this research was to evaluate the predictive performances of frequency ratio (FR), logistic regression (LR) and weight of evidence (WoE), in flood susceptibility mapping in China. In addition, the ensemble WoE and LR and ensemble FR and LR techniques were applied and used in the evaluation. The flood inventory map, consisting of 196 flood locations, was extracted from a number of sources. The flood inventory data were randomly divided into a testing data-set, allocating 70% for training, and the remaining 30% for validation. The 15 flood conditioning factors included in the spatial database were altitude, slope, aspect, geology, distance from river, distance from road, distance from fault, soil type, land use/cover, rainfall, Normalized Difference Vegetation Index, Stream Power Index, Topographic Wetness Index, Sediment Transport Index and curvature. For validation, success and prediction rate curves were developed using area under the curve (AUC) method. The results indicated that the highest prediction rate of 90.36% was achieved using the ensemble technique of WoE and LR. The standalone WoE produced the highest prediction rate among the individual methods. It can be concluded that WoE offers a more advanced method of mapping prone areas, compared with the FR and LR methods.

Current methods of spatial prediction are based on either the First Law of Geography or the statistical principle or the combination of these two. The Second Law of Geography contributes to the revision of these methods so they are adaptive to local conditions but at the cost of increasing demand for samples. This paper presents a new thinking about spatial prediction based on the Third Law of Geography which focuses on the similarity of geographic configuration of locations. Under the Third Law of Geography, spatial prediction can be made on the basis of the similarity of geographic configurations between a sample and a prediction point. This allows the representativeness of a single sample to be used in prediction. A case study in predicting spatial variation of soil organic matter content was used to compare the spatial prediction based the Third Law of Geography with those based on the First Law and the statistical principle. It is concluded that spatial prediction based on the Third Law of Geography does not require samples to be over certain size nor to be of a particular spatial distribution to achieve a high quality prediction. The prediction uncertainty associated with spatial prediction based on the Third Law of Geography is more indicative to quality of the prediction, thus more effective in allocating error reduction efforts. These properties make spatial prediction based on the Third Law of Geography more suitable for prediction over large and complex geographic areas.

Great strides have been made in Geographic Information Systems (GIS) research over the past half-century. However, this progress has created both opportunities and challenges. From a geographic perspective, certain challenges remain, including the modelling of geographic-featured environments with GIS data model, the enhancement of GIS’s analysis functions for comprehensive geographic analysis and achieving human-oriented geographic information presentation. Several basic theoretical and technical ideas that follow the workflow and processes of geographic information induction, geographic scenario modelling, geographic process analysis and geographic environment representation are proposed to fill the gaps between GIS and geography. We also call for designing methods for big geographic data-oriented analysis, making best use of videos and developing virtual geographic scenario-based GIS for further evolution.

The integration of multi-platform, multi-angle, and multi-temporal LiDAR data has become important for geospatial data applications. This paper presents a comprehensive review of LiDAR data registration in the fields of photogrammetry and remote sensing. At present, a coarse-to-fine registration strategy is commonly used for LiDAR point clouds registration. The coarse registration method is first used to achieve a good initial position, based on which registration is then refined utilizing the fine registration method. According to the coarse-to-fine framework, this paper reviews current registration methods and their methodologies, and identifies important differences between them. The lack of standard data and unified evaluation systems is identified as a factor limiting objective comparison of different methods. The paper also describes the most commonly-used point cloud registration error analysis methods. Finally, avenues for future work on LiDAR data registration in terms of applications, data, and technology are discussed. In particular, there is a need to address registration of multi-angle and multi-scale data from various newly available types of LiDAR hardware, which will play an important role in diverse applications such as forest resource surveys, urban energy use, cultural heritage protection, and unmanned vehicles.

Lignin is the most abundant aromatic substrate on Earth and its valorization technologies are still under developed. Depolymerization and fragmentation are the predominant preparatory strategies for valorization of lignin to chemicals and fuels. However, due to the structural heterogeneity of lignin, depolymerization and fragmentation typically result in diverse product species, which require extensive separation and purification procedures to obtain target products. For lignin valorization, bacterial-based systems have attracted increasing attention because of their diverse metabolisms, which can be used to funnel multiple lignin-based compounds into specific target products. Here, recent advances in lignin valorization using bacteria are critically reviewed, including lignin-degrading bacteria that are able to degrade lignin and use lignin-associated aromatics, various associated metabolic pathways, and application of bacterial cultures for lignin valorization. This review will provide insight into the recent breakthroughs and future trends of lignin valorization based on bacterial systems.

Abstract Projecting future change of monsoon rainfall is essential for water resource management, food security, disaster mitigation, and infrastructure planning. Here we assess the future change and explore the causes of the changes using 15 models that participated in phase 6 of the Coupled Model Intercomparison Project (CMIP6). The multimodel ensemble projects that, under the shared socioeconomic pathway (SSP) 2–4.5, the total land monsoon rainfall will likely increase in the Northern Hemisphere (NH) by about 2.8% per one degree Celsius of global warming (2.8% °C −1 ) in contrast to little change in the Southern Hemisphere (SH; −0.3% °C −1 ). In addition, in the future the Asian–northern African monsoon likely becomes wetter while the North American monsoon becomes drier. Since the humidity increase is nearly uniform in all summer monsoon regions, the dynamic processes must play a fundamental role in shaping the spatial patterns of the global monsoon changes. Greenhouse gas (GHG) radiative forcing induces a “NH-warmer-than-SH” pattern, which favors increasing the NH monsoon rainfall and prolonging the NH monsoon rainy season while reducing the SH monsoon rainfall and shortening the SH monsoon rainy season. The GHG forcing induces a “land-warmer-than-ocean” pattern, which enhances Asian monsoon low pressure and increases Asian and northern African monsoon rainfall, and an El Niño–like warming, which reduces North American monsoon rainfall. The uncertainties in the projected monsoon precipitation changes are significantly related to the models’ projected hemispheric and land–ocean thermal contrasts as well as to the eastern Pacific Ocean warming. The CMIP6 models’ common biases and the processes by which convective heating drives monsoon circulation are also discussed.

Soil gross nitrogen (N) mineralization (GNM), a key microbial process in the global N cycle, is mainly controlled by climate and soil properties. This study provides for the first time a comprehensive analysis of the role of soil physicochemical properties and climate and their interactions with soil microbial biomass (MB) in controlling GNM globally. Through a meta-analysis of 970 observations from 337 published papers from various ecosystems, we found that GNM was positively correlated with MB, total carbon, total N and precipitation, and negatively correlated with bulk density (BD) and soil pH. Our multivariate analysis and structural equation modeling revealed that GNM is driven by MB and dominantly influenced by BD and precipitation. The higher total N accelerates GNM via increasing MB. The decrease in BD stimulates GNM via increasing total N and MB, whereas higher precipitation stimulates GNM via increasing total N. Moreover, the GNM varies with ecosystem type, being greater in forests and grasslands with high total carbon and MB contents and low BD and pH compared to croplands. The highest GNM was observed in tropical wet soils that receive high precipitation, which increases the supply of soil substrate (total N) to microbes. Our findings suggest that anthropogenic activities that affect soil microbial population size, BD, soil substrate availability, or soil pH may interact with changes in precipitation regime and land use to influence GNM, which may ultimately affect ecosystem productivity and N loss to the environment.

Prediction of Indian summer monsoon rainfall (ISMR) is at the heart of tropical climate prediction. Despite enormous progress having been made in predicting ISMR since 1886, the operational forecasts during recent decades (1989-2012) have little skill. Here we show, with both dynamical and physical-empirical models, that this recent failure is largely due to the models' inability to capture new predictability sources emerging during recent global warming, that is, the development of the central-Pacific El Nino-Southern Oscillation (CP-ENSO), the rapid deepening of the Asian Low and the strengthening of North and South Pacific Highs during boreal spring. A physical-empirical model that captures these new predictors can produce an independent forecast skill of 0.51 for 1989-2012 and a 92-year retrospective forecast skill of 0.64 for 1921-2012. The recent low skills of the dynamical models are attributed to deficiencies in capturing the developing CP-ENSO and anomalous Asian Low. The results reveal a considerable gap between ISMR prediction skill and predictability.