Aiming at the problem of low precision of abandoned land extraction caused by complex land cover and broken land， a method of abandoned land information extraction based on multi temporal collaborative change detection was proposed. Taking Luquan District， Shijiazhuang City， Hebei Province as the research area， the Normalized Difference Vegetation Index （NDVI） of various types of cultivated land cover was analyzed by using sentinel 2a and Landsat 7 multispectral images and supported by field samples Based on the classification system of seasonal abandonment， perennial abandonment， winter wheat and perennial garden， a multi temporal collaborative change detection model was constructed to carry out remote sensing monitoring of cultivated land abandonment in the study area. The results show that the classification accuracy of seasonal and perennial abandoned farmland based on Sentinel 2A image is 95.83% and 96.55% respectively； the classification accuracy of seasonal and perennial abandoned farmland based on Landsat 7 image is 91.67% and 93.10% respectively； the seasonal abandoned farmland accounts for 4.7% and perennial abandoned farmland accounts for 7.1% in Luquan District in 2019. This method can quickly and accurately obtain the spatial distribution and area information of cultivated land in the study area， and has good extraction accuracy for abandoned land in different resolution images.
Farmland is important for food production. It is thus of great importance to obtain timely and accurate information regarding non-agricultural farmlands for land resource management and policymaking. To investigate the changes of non-agricultural farmlands in Fuzhou over past 30 years， this study extracted the spatial information of farmlands using multi-temporal Landsat remote sensing images in 1989， 2000， 2010 and 2019 based on the Google Earth Engine （GEE） and random forest methods. We then used land transfer matrix， grid element method and geographic detector techniques to analyze the characteristics and driving factors of non-agricultural farmlands changes. The results show that： （1） The GEE platform integrating with random forest is suitable to extract farmlands in cloudy and rainy areas in southern part of China. The overall accuracy of the extracted farmlands is higher than 90%， and the Kappa coefficient is greater than 0.85. （2） The farmlands in Fuzhou has an imbalanced spatial distribution， where the area of farmlands deceases from east to west along time. From 1989 to 2019， the farmland changes mainly occurred at areas with an elevation of 100 m and a slope of less than 10°. The changed farmlands mainly consisted of forestlands and construction lands， in which the western region was mainly forestland， and the central and eastern region was construction land. （3） The natural factors are the prerequisite for the conversion of cultivated land， and the growth rate of urbanization and population data are the main driving factors. Moreover， urbanization rate and the proportion of primary industry growth rate were the factors forming the “fast-slow-stable” pattern of farmland non-agriculturalization.
In order to realize fast and accurate extraction of winter wheat planting spatial information by using high-resolution remote sensing image and deep learning semantic segmentation model， worldView-2 remote sensing image was used as the data source to produce the sample data sets with the scales of 128×128， 256×256 and 512×512， which trained the parameters of U-net and DeepLabv3+ semantic segmentation model to establish remote sensing classification model of winter wheat. The classification effects of deep learning was tested by comparing with maximum likelihood and random forest methods. The results showed that： （1） the overall accuracy and Kappa coefficient of the models obtained by training samples of different scales were more than 94% and 0.82， and the model accuracy was stable， which indicated that the sample sizes have little influence on the semantic segmentation model of winter wheat classification. （2） The overall classification accuracy and Kappa coefficient of the deep learning methods were above 94% and 0.89， while the maximum likelihood and random forest were below 92% and 0.85， respectively. This results suggested that the remote sensing classification model of winter wheat established in this study was superior to the traditional classification methods. The results can provide the references for the deep learning methods of crop planting information extraction with high resolution remote sensing image.
Wheat stripe rust is an air-borne disease that leads to large reduction in wheat production. The spread process is affected by many factors. Common crop diseases meteorological prediction models are difficult to simulate wheat stripe rust incidence accurately. In order to obtain accurate prediction of wheat stripe rust incidence， a Suscept-Exposed-Infectious-Removed StripeRust （SEIR-StripeRust） dynamic prediction model was constructed based on meteorological and remote sensing data. This paper chose the Longnan area of Gansu Province as a study area. First， meteorological factors and vegetation indexes were constructed based on meteorological data and MODIS data， respectively. Then， the above features were screened by correlation analysis to identify the sensitive factors. A new incidence prediction model named SEIR-StripeRust was constructed， coupled with the sensitive factors. Finally， compared the accuracy of SEIR-StripeRust model with used Back Propagation Neural Network （BPNN）， Support Vector Regression （SVR） and Multiple Linear Regression （MLR）. The results showed that the average temperature， relative humidity and normalized difference vegetation index were significantly correlated with the incidence of wheat stripe rust. The SEIR-StripeRust model constructed by the above three sensitive factors had the highest prediction accuracy， the coefficient of determination （R2 ） was 0.79， the Root Mean Square Error （RMSE） was 0.10， and the Mean Absolute Error （MAE） was 0.09， which were higher than the BPNN， SVR and MLR models under the same characteristic variables. The results showed that the SEIR-StripeRust model can effectively predict the incidence of wheat stripe rust and provide technical support for wheat stripe rust prediction and accurate prevention at county scale.
Drought is one of the main meteorological factors affecting peanut yield. Remote sensing assessment of peanut drought disaster is of great significance for yield estimation， disaster prevention and mitigation， and insurance claims. At present， the remote sensing evaluation of peanut drought mainly depends on the change information of spectral index， which is easily disturbed by the growth process in different regions， which limits the universality of spectral index method. Supported by multi temporal Sentinel-2 remote sensing images and field samples， this study analyzed the internal relationship between the daily average reflectance increment information of time-series bands and the drought disaster degree of peanut. Decision tree， random forest， logistic regression and other methods were used to classify the Drought Grades of peanut， and the overall accuracy and kappa coefficient were used to evaluate the accuracy of various methods. The results showed that the daily average increment of NIR reflectance in a single band was a strong indicator of peanut disaster. The results showed that the combination of multi spectral bands was better than single band in indicating drought degree of peanut， and the combination of red band， blue band and near infrared spectral band had the strongest indication， with the overall accuracy of 89.93% and Kappa coefficient of 0.847 1. Compared with Logistic regression and decision tree algorithm， random forest algorithm has the highest accuracy in drought assessment of peanut. In the analysis of the optimal time combination of drought grade， using the multi band daily average reflectance increment information of peanut growth peak period （July and August）， the overall accuracy of disaster grade remote sensing recognition can reach 88.62%， and the Kappa coefficient is 0.827 4. The results show that the drought disaster assessment method based on multi band reflectance daily increment in the growing period can effectively extract the disaster area and severity of peanut.
Forest plantations in plain area of China are highly fragmented and dispersed due to flexible planting system. Accurate and timely mapping plantation distribution in plain area plays an important role in plantation management and hydro-ecological functions. In this study， a hierarchical-based classification method， which optimizes variable combinations at each node， was proposed and applied in Linxi County， Anhui Province for mapping tree species in plain area. The object–based classification with combinations of various types of spectral and texture features derived from the fused image of ZY-3 multispectral and panchromatic band （ZY）， and the fused image of Sentinel-2 multispectral bands and ZY-3 panchromatic band （STZY） were conducted. The results showed that the proposed method provides higher classification accuracy than regression tree and random forest approaches for both datasets， especially for dominant poplar tree species. When only spectral features were used， STZY offered better results than ZY， overall accuracies increased by 2.49%~2.91%， indicating the importance of spectral information in classification. When textures were integrated with spectral features into classification procedure， overall accuracies increased by 10.19% and 4.99% for STZY and ZY respectively using the hierarchical-based classifier， implying that texture features are essential for classification. Thus， the proposed hierarchical-based classifier with optimized variable combinations is an effective method in tree species mapping in plain area using spectral and texture features derived from high spatial-resolution data.
Maize lodging caused by wind disaster may lead to a large reduction in maize production. Using remote sensing technology to accurately monitor maize lodging area and spatial distribution information is very important for disaster assessment.In this paper， Planet and Sentinel-2 images are combined with object-oriented and pixel-based methods to extract maize lodging in the study area， and different image features （spectral features， vegetation index and texture features） and different classification methods （support vector machine SVM， Random forest method RF and maximum likelihood method MLC） influence on the extraction accuracy of corn lodging.The results show that： ① The accuracy of corn lodging extraction using Planet images with high spatial resolution is generally higher than that of Sentinel-2 images；② From the perspective of classification accuracy and area accuracy， the spectral features， vegetation index， and mean feature of Planet image combined with object-oriented RF classification， the overall accuracy and Kappa coefficient are 93.77% and 0.87， respectively， and the average area error is the lowest 4.76%；③The accuracy of extracting maize lodging using Planet and Sentinel-2 images combined with object-oriented classification is higher than that of pixel-based classification. This research not only analyzes the advantages of object-oriented methods， but also evaluates the applicability of using different image data combined with object-oriented methods， which can provide a certain reference for remote sensing to extract crop lodging related research.
The quality and quantity of sample dataset from medium resolution remote sensing images is the key factor to contribute to the efficiency of low and medium resolution identification model. For winter wheat in this paper， we constructed a support vector regression model coupled with low and medium resolution images， to decomposed of mixed pixels， and exact winter wheat extent. Then analyzed the influences of sample quantity and quality of medium resolution remote sensing images respectively. The results states that only 10% quantity of samples are enough to achieve stable accuracy. Under this quantity， regional accuracy and pixel accuracy could reach higher than 98% and 92% respectively in typical winter wheat area. In terms of sample quality， the accuracy of result improved accompanying with the sample quality increment. We found that high accuracy could achieved when the sample quality is better than 60%. While in the area where medium resolution sample did not exist in area with medium samples， regional accuracy and pixel accuracy also increased accompanying with the sample amount and quality increment. In this area， 20% quantity of medium resolution sample was needed enough to achieve 97% of regional accuracy and 92% of pixel accuracy respectively. The above demonstrate the successful generalization of winter wheat identification by medium resolution sample to non-medium resolution area.
The monitoring of information such as irrigation area and drought conditions in irrigation districts is the basis of irrigation districts management， while the traditional way to get the information cannot meet the research needs. Satellite remote sensing is a powerful technical means for water resources management. Taking the Jiefangzha Irrigation district in Inner Mongolia Autonomous Region as a research area， the Landsat 8 satellite data were selected to calculate and analyze the distribution and change of the Temperature Vegetation Dryness Index（TVDI） and the Modified Perpendicular Drought Index（MPDI）. The paper built a remote sensing model of irrigation area based on drought index difference threshold to determine the threshold and extract the irrigation area. The results showed that the Jiefangzha Irrigation district received a large scale of irrigation in July to August in 2017. Through comparing the irrigation area extracted by using two drought index difference thresholds with the real irrigation area， the monitoring accuracy of TVDI and MPDI is 82.96% and 74.01%， respectively. And the high-resolution data of Google Earth is selected as the real data to calculate the confusion matrix. The results showed that the overall accuracy of MPDI extraction is 94.17%， which is higher than 58.90% of TVDI. The two results illustrate the feasibility of calculating drought index for irrigation drought monitoring and area extraction. However， in terms of spatial distribution， compared with TVDI， MPDI can better reflect the drought situation， and the spatial distribution of the irrigation district is more reasonable.
As an important food crop in China， winter wheat has accurate access to its spatial distribution and is of great significance for agricultural production management and agricultural monitoring. Taking Shangqiu City of Henan Province as an example， using the GF-1 data covering the whole growth period of winter wheat， the time series of Normalized Difference Vegetation Index （NDVI） and Enhanced Vegetation Index （EVI） was calculated. During the growth period， the data sets of different feature quantities were constructed， and the winter wheat was extracted by the support vector machine method. The results show that the EVI time series data can better describe the phenology of crops than NDVI， and the extraction accuracy is higher than NDVI. The EVI time series data and the key growth period image combination extraction precision is the highest， reaching 97.67%. At the same time， the principal component analysis method is used to reduce the dimensionality of the data， and try to improve the extraction efficiency of winter wheat by compressing the data volume of the feature set. The results show that the data after dimension reduction does not have a significant impact on the extraction accuracy， and the purpose of maintaining the accuracy of the compressed data is to provide reference value for large-area crop extraction.
Based on Google Earth Engine （GEE） cloud computing platform， we collaborate with Sentinel-2 images， WordClim bioclimatic data， SRTM topographic data， forest resources planning and design survey data and other data， and use Random Forest （RF）， Support Vector Machine （SVM） and Maximum Entropy （MaxEnt） machine learning algorithms were used as component classifiers to carry out the study of dominant tree species classification with multi-source features and multi-classifier decision fusion. Two serially integrated and three Bayesian parallel integrated models were constructed by the three component classifiers for determining the spatial distribution of 10 major dominant tree species in Shangri-La region of Yunnan. The classification results showed that the overall accuracy of the three component classifiers was lower than 67.17%， the overall accuracy of the three parallel integration methods was comparable， about 72%， the accuracy of the two serial integration methods was higher than 78.48%. Among them， the MaxEnt SVM serial integration method obtained the best accuracy （OA： 80.66%， Kappa： 0.78）， which improved the accuracy compared with the component classifiers by at least 13.49%. The study shows that the decision fusion method has higher accuracy than the component classifier in dominant tree species classification and effectively improves the classification accuracy of small sample tree species， which can be used for dominant tree species classification in large mountainous areas.
High-precision forest age can improve the estimation accuracy of forest biomass， stock and carbon storag， In order to improve the accuracy of forest age estimation in areas with frequent disturbances， taking Jiangle County， Fujian Province， where the forest disturbance intensity is high， as an example. the Landsat time series data of Jiangle County from 1987 to 2019 was constructed， and， the LandTrendr algorithm was used to obtain the node characteristics of the beginning of forest disturbance， through modeling with forest age information， realizing the mapping of the existing forest age in the disturbance area. Then， using the band， vegetation index， texture and topographic factor characteristics of the GF-1 image. through the recursive feature elimination of random forest algorithm， and the forest age modeling to achieve non- Estimation of the forest age in the disturbed area； In the end， the forest age of the two parts are combined to obtain the forest age in the study area in 2019. The results show that：①The total area of forest disturbance in Jiangle County from 1987 to 2019 is 346.37 km2， of which the - coniferous forest and broad-leaved forest account for 75.06% of the disturbance area；②The forest age error （RMSE=1.91 years） estimated by the LandTrendr algorithm’s disturbance start time node is small， and the model accuracy （R2=0.94） is high；③The R2 and RMSE of the age of coniferous forest and broad-leaved forest estimated by the random forest algorithm with the recursive feature elimination random forest algorithm are 0.64，0.48 and 4.71，12.71 years， respectively. Research shows that the disturbance algorithm combined with long time series can effectively improve the accuracy of forest age estimation in disturbance area， and provide a reference for forest age estimation at the regional scale in the subtropical mountainous area.
Strictly observing the ecological red line and protecting forest resources are important measures to build an ecological civilization and promote green development. It is of great significance for forest resource protection and management to grasp the changes of forest land in time and detect the forest destruction as soon as possible. Currently， the monitoring of deforestation and other destructive behaviors is passive and ex post. It is necessary to exploring the spatial distribution of deforest areas， evaluating and predict the risk of deforestation to prevent the occurrence of deforestation by Comprehensive supervise areas which are prone to deforestation. This article is based on deforest change maps of Baoting County， Hainan Province， which obtain by multi-source remote sensing change detection and field surveys during two consecutive years， to study the relationship between deforest areas and human activities， topography and landforms， and then to assess the risk of deforest， providing support for local forestry departments to carry out more targeted inspections， monitoring and protection measures， strengthen the effectiveness of forestland supervision， truly destructive behaviors such as deforest in the embryonic stage， strictly abiding by the ecological red line， and promoting the construction of ecological civilization.
We estimated forest stem volume using domestic active and passive remote sensing data GF-3 PolSAR and GF-6 PMS. And in order to find a way out of the redundancy problem of multi-source remote sensing data， feature combination is optimized. The research area is the natural forest land in ??Gongliu County， Xinjiang. We extracted spectral information， vegetation index， texture， vegetation coverage from GF-6 PMS data and then extracted backscattering coefficient and polarization decomposition parameters from GF-3 PolSAR data. Combining the extracted parameters， terrain factor and forest sample survey data， we estimated forest stem volume in the study area using K- Nearest Neighbor with Fast Iterative Features Selection （KNN-FIFS） method. Comparing and validating the estimation results when combined active and passive remote sensing data and a single remote sensing data source， we inverted the forest stem volume in the study area based on the optimal feature combination. The results show that the accuracy of combining GF-3 PolSAR data and GF-6 PMS data to estimate the forest stem volume in the study area is R2=0.72 and RMSE=92.48 m3/hm2， which is compared with the accuracy estimated using only GF-6 PMS data （R2=0.56， RMSE=118.8 m3/hm2）， R2 increased by 0.16 with an increase rate of 28.6% and RMSE decreased by 26.32 m3/hm2 with a decrease rate of 22.2%. It indicated that the cooperative inversion of active and passive remote sensing data can improve the estimation accuracy of forest stem volume， and the KNN-FIFS method can effectively estimate the forest stem volume of natural forests.
Based on UAV extraction of tree height characteristics of underground forests in graded karst Tiankeng， we explored the relationship between the growth strategy of tree height and the local enclosure habitat of graded Tiankeng， and studied the value of graded Tiankeng as a refuge for species. The graded Tiankeng was reconstructed in three dimensions by unmanned aerial remote sensing technology to extract tree height information inside and outside the graded Tiankeng.The results showed that the average tree height in the underground forest of degraded Tiankengs is about 5m higher than the surface. The average tree height in the underground forest is 10.47 m； the average tree height on the surface is 5.43 m； and the average tree height on the south slope of the surface is 5.75 m. The distribution characteristics of tree height in the Tiankeng are significantly influenced by elevation. Under the effect of karst Tiankeng microhabitats， the underground forest in the Tiankeng has a significant advantage in tree height compared with the surface outside the Tiankeng. Light is the main factor of intra- and interspecific competition among tree species in the underground forest， and vertical gradient is the primary feature of tree height distribution pattern of degraded Tiankeng vegetation. UAV remote sensing technology has the potential to be promoted as it can quickly obtain information on tree height in degraded Tiankeng underground forests.
Sun-Induced chlorophyll Fluorescence （SIF）， as a surrogate indicator of Gross Primary Productivity （GPP）， shows great potential in regional GPP estimation. The SIF and GPP have a good linear relationship， but the influence of different climate conditions on the SIF-GPP relationship is still unclear. In this study， we used MODIS GPP and GOME-2 SIF and environmental conditions （temperature， precipitation， radiation， etc.） in China during 2007~2018 to study the temporal and spatial patterns of GPP and SIF of terrestrial vegetation and the constraint of environmental factors. The results found that the spatial and temporal patterns of GPP and SIF of terrestrial vegetation are similar， but there are significant differences in the spatial distribution of GPP/SIF that act as a new indicator of light energy distribution. In addition， the yield of SIF （SIFYield） is controlled by the environmental factors （minimum temperature， saturated vapor pressure difference， soil moisture， and APAR） that restrict GPP， which indirectly confirms the close connection between SIF and GPP. Therefore， since the relationship between SIF and GPP in time and space is regulated by environmental conditions， the use of satellite-based SIF to accurately estimate regional GPP should consider the constraints of environmental factors.
Based on the GOME-2 satellite SIF dataset， we analyzed the spatial and temporal changes of SIF from 2007 to 2018 in China， and investigated the response of SIF to climate changes， such as temperature， precipitation， and radiation. The results showed that： （1） The SIF in China's vegetation region generally shows a decreasing distribution from southeast to northwest. The average annual SIF increases by 20.2% in last 12 years， with an amplitude of 0.034 mW/m2/sr/nm， and the increase area accounts for 80.3% of the whole China. The area with significant growth of SIF accounts for 25.7%， which were mainly distributed in eastern， southern and northeastern China. （2） The SIF increase in summer season during last twelve years is the largest with an amplitude of 0.065 mW/m2/sr/nm； the area with increased summer SIF accounts for 82.1% of the whole China， and the area with significant increase accounts for 19.4%. （3） The response of SIF to climate change was investigated using the partial correlation method. temperature is the main factor affecting the interannual variation of SIF； precipitation is the main driven factor for SIF in warm temperate and temperate vegetation regions； human activities are more likely to affect the growth of SIF in the green broad-leaved forest area； radiation is the driven factor for tropical monsoon rain forest areas located in low latitudes. The above results reveal the temporal and spatial changes of vegetation fluorescence in China from 2007 to 2018 and its response to climate change， which can provide important support for global carbon cycle research.
Cloud coverage hinders the effective range of earth observation by optical remote sensing satellite. Rapid and accurate cloud detection is an important step in the product generation process of remote sensing application. In view of the lack of suitable and high-quality cloud detection model in Google Earth Engine cloud platform， this study takes tropical cloudy Sri Lanka as the study area， constructs a Sentinel-2 image cloud detection model coupled with SVM and Cloud-Score algorithm. Through experiments， the cloud detection accuracy of this method is compared with that of QA60 method， Cloud-Score algorithm and Fmask from the point of view of visual interpretation and quantitative analysis. The results show that the cloud detection performance of Fmask model is the lowest， and the overall accuracy is only 63.45%. It has a serious phenomenon that water body is mistakenly divided into clouds， but its omission error is very low. The QA60 method has insufficient recognition of cirrus clouds， and the omission error is high. At the same time， it has a certain phenomenon of misclassification， and the low spatial resolution affects the details of cloud boundary extraction results. The cloud detection performance of the Cloud-Score algorithm is obviously better than that of the QA60 method， the overall accuracy is 89.83%， and the commission error is only 2.17%， but there is still a phenomenon that some cirrus clouds are missed. Compared with the other three cloud detection methods， the cloud detection model proposed in this study has the highest overall accuracy， reaching 98.21%， and has extremely low omission error and commission error. The model can accurately identify the boundary of the cloud， and can meet the cloud detection preprocessing requirements of Sentinel-2 remote sensing products.
Frequent forest fires have caused extensive vegetation destruction in the Amazon tropical rain forest. It’s of great importance to obtain the fire influence range and vegetation destruction in different years to understand the spatio-temporal evolution of fire in this area， study the interaction between fire and vegetation， and then explore the fire development mechanism， so as to provide a scientific basis for disaster forest and reduction. To this end， the MODIS vegetation index products and surface temperature products range from 2015 to 2019 were used in this paper to construct the MODIS Global Disturbance Model （MGDI）， combined with fire point data （hereinafter collectively referred to as MOD14A1） and Vegetation Continuous Field （VCF）to extract combustion scope and intensity at 1 000 m resolution， and the spatial and temporal law of burned area within 5 years of the study area was analyzed. The results revealed that ：（1） Burned area are mainly distributed in the central part of Brazil and the border between Brazil and Bolivia， accounting for about 67% of the total burning area；（2） The information of burned area and burned intensity comprehensively `indicated that the fire showing a “rise-drop-rise” trend；（3） The fire mainly occurred in shrub grassland（more than 50%） and broad-leaved forest（30%）， and most of them took place during the dry season； under the global warming circumstance， the fire frequency increased a lot；（4） The expansion of human activities， unreasonable agricultural reclamation and deforestation lead to serious grassland degradation in the study area， and agricultural land and construction land are increasing year by year， which provides good conditions for the occurrence and conduction of fire to a certain extent.
Water extraction is an essential step for rare earth monitoring of urban water environment. Extraction of small water bodies in the city has now become a hot depth study in the field of remote sensing images. However， deep learning requires a large number of sample datasets as input， and images with different spatial resolutions often need to construct different datasets. If the spatial resolution of the images is not much different， the sample transfer learning model can be used to ensure accuracy and save time. In this paper， the U-Net image segmentation model is selected to perform sample transfer learning for images with three different spatial resolutions—0.5 m， 0.8 m and 2 m respectively. It is found that after three migration learning of 2 meters to 0.8 meters， 2 meters to 0.5 meters， and 0.8 meters to 0.5 meters， the corresponding evaluation indexes F1-score， MioU and Kappa of the extracted water body are all above 0.80. Under the premise of little difference in resolution， this method of extracting urban water bodies from lower-resolution samples to higher-resolution images is basically feasible， and the accuracy of the results is better. It is suitable for water extraction in water-deficient cities.
As the central city of the lower reaches of the Qinhuai River Basin （QRB）， Nanjing has been suffering from ecocological envirnoment effects due to great changes in underlying surfaces under rapid urbanization processes. Impervious surfaces， one of the key indicator of regional urbanization， can bridge urban development and environment quality， which provide a new perspective of spatial goverance and coordination of urban and rural development. Under the sponge city concept，we extracted the impervious surfaces of multi-temporal Landsat images through the semi-automatic decision tree classification model in the QRB， China from 1988 to 2017. To extract the continuously changing impervious surfaces， we made full use of temporal， spatial and spectral characteristics through multi-filter methods to further improve the classification accuracy. We then explored the characteristics of impervious surface expansion， including the area， intensity， landscape expansion types of the QRB during the past 30 years. The results show that the QRB experienced rapid urbanization in the study period， and the impervious surface percentage increased from 3.09% in 1988 to 26.49% in 2017. Before 2006， the QRB kept extending mainly at a lower and medium speed in the urban cores of Nanjing city and built-up of Jiangning district. After that， it began to expand at a high speed， being located in the Jiangning district， Lishui District and Jurong city. The QRB was a constellation model in 1988， however， its shape of the impervious patches turned to be simpler with higher compactness based on the “multi-core expansion” around the urban core as well as the “point-axes expansion” along the main transportation lines.The impervious surface expansion in the basin showed an obvious spatial heterogeneity. Economic development， government behaviors， traffic infrastructure development and natural environmental conditions hadgreat impacts on the impervious surface expansion in the QRB. The urban-rural integration was constantly improving with higher patch infill growth.
Oil wells， wind turbines， and roads are the main artificial features in the Yellow River Delta （YRD）. It is of great significance to clarify their spatio-temporal characteristics to ensure the ecological security of the YRD. Based on medium and high-resolution satellite images， such as Landsat， SPOT， and GF2， the main artificial features in the YRD in 2000 and 2015 were extracted， and then their spatio-temporal characteristics were analyzed， using the methods of kernel density， average nearest neighbor， weighted analysis， Kriging interpolation， and spatial autocorrelation analysis. The results showed that： （1） The spatial distribution of wind turbines was uneven in 2015， showing a distribution pattern dominated by two core regions and assisted by three small agglomeration regions. The overall trend was northwest to southeast with an obvious spatial agglomeration characteristic. （2） The spatial distributions of oil wells in 2000 and 2015 were both uneven and characterized by time variation. In 2000， oil wells showed a spatial distribution pattern with two cores as a main part， supplemented by a patchy distribution of clusters. However， in 2015， oil wells showed a spatial pattern characteristic of three major core areas dominated and a saddle-shaped distribution of wells around the core areas. （3） From 2000 to 2015， the road network density had been continuously increasing， with obvious spatial agglomeration characteristics. The high road network density was mainly distributed in the central region of the YRD， and presented a spreading trend to the west and south. Overall， the wide distribution， fast growth， and the large number of artificial features in the YRD have posed a great threat to the ecological security of nature reserves. This study provides a scientific basis and decision-making reference for wetland ecosystem protection in this region by identifying and analyzing the spatio-temporal characteristics of artificial features in the YRD.
Ice clouds remote sensing is in an ongoing focus as ice clouds play an important role in atmospheric energy budgets due to their reflection of sunlight and their entrapment of infrared radiation. Submillimeter-wave radiometry is an important technique in detecting ice clouds because of its distinctive advantage over other remote sensors. Developing a complete radiative transfer model that links the ice cloud parameters and the brightness temperature observations is of critical importance. The paper builds up a forward model that rigorously handles the ice particle scattering based on the Atmospheric Radiative Transfer Simulator （ARTS）， and we conduct the simulations on a cloud cross section for the upcoming Ice Cloud Imager （ICI）. The results indicate that the ICI channels possess high sensitivity to the ice cloud microphysics， and only the low-frequency channels are sensitive to the liquid clouds. For the double sidebands with the same center frequency， the large frequency-offset channels show higher brightness temperature values in clear sky conditions， and they have larger BT depressions when encountering thick ice cloud layers. The forward model allows us to develop retrieval algorithms upon it in the future.
The accurate identification and expression of urban spatial structure is of great significance for mastering the current situation of urban spatial development and planning the future urban spatial. Using NPP/VIIRS annual average nighttime light remote sensing images and microblog check-in data， based on localized contour tree method and hierarchical structural Tupu， this study identified and expressed the urban center and their internal spatial structure in Zhengzhou city in 2014. In addition， combined with typical landmarks and urban functions， the identified urban centers were defined and classified. On this basis， 10 indicators were selected for quantitative analysis of identified urban centers from three aspects of social economy， urban morphology， and human activity. The results showed that： In 2014， three localized contour trees were identified in Zhengzhou， including 18 urban centers and 11 urban central composite areas； the contour tree level of the “main tree” in Zhengzhou city was 10. The urban centers of the old urban area were developed well and had a large spatial range. The spatial distribution of the urban centers in the northern area was unbalanced and relatively fragmented； According to the urban functions， the urban centers of Zhengzhou were divided into four categories （main center， comprehensive center， commercial center， and industrial center）， the number of which is 1， 10， 4 and 3， respectively. The spatial distribution of urban comprehensive centers is more uniform， followed by commercial center and industrial centers， and the results of dividing the urban centers into five grades are basically consistent with the levels of contour trees. In quantity， it is in the state of “less-more-less”， and in spatial distribution， it basically follows the trend of “from inside to outside” urban center grades decreasing in turn.
