Progress 01/01/08 to 12/31/08
Outputs
OUTPUTS: During the 5-year project period, our major emphasis has been placed on forest measurement, invasive weed mapping, wetland monitoring and environmental health studies in relation to schistosomiasis in China and plague in California all with remote sensing and geographic information systems technology. This has led to the graduation of 8 doctoral students in agricultural sciences; the organization of one international symposium on geoinformatics in 2007; the prototyping of 2 commercial software packages (Berkeleyimg Seg, an image segmentation software and Tiffs, a lidar data processing software for forest inventory); the world's first detailed comprehensive database on China's wetlands; and over 30 invited conference and on site seminar talks to academia and professionals. In addition, driven by the needs of this project, we established a new ecological research station on environment and health at Poyang Lake, China. PARTICIPANTS: Peng Gong, PI The following are project participants: Nick Clinton, Fire mapping and invasive cheatgrass monitoring Ashley Holt, Plague spread risk modeling with remote sensing and GIS Ruiliang Pu, LAI measurement and hyperspectral remote sensing Ryo Michishita, Thermal remote sensing and urban heat island Desheng Liu, High resolution data georeferencing and spatial temporal change detection Xin Miao, Hyperspectral monitoring of yellow starthistle Shaokui Ge, Salt cedar monitoring with field and aerial remote sensing Fernando Sedano, Land cover mapping in southern Africa with MISR and MODIS data Qi Chen, Forest measurement with airborne lidar data Qian Yu, High resolution detailed vegetation mapping Yongling Weng, Soil salination and hyperspectral remote sensing Huabing Huang, Lidar for forest inventory Fengming Hui, Wetland mapping Shaoqing Shen, WebGIS and wireless sensor network Weimin Li, Urban security and greenness interaction Formed alliances with the State Key Lab of Remote Sensing Science, jointly sponsored by Institute of Remote Sensing Applications, Chinese Academy of Sciences and Beijing Normal University; International Institute for Earth System Science, Nanjing University; School of Geography, Jiangxi Normal University; Sichun CDC and Jiangxi CDC in China. Maintained long-term collaboration with Greg Biging, Maggi Kelly, Ron Amundson, Denis Baldocchi in the College of Natural Resources, Robert Spear, Edmund Seto and Kirk Smith in the School of Public Health and Bin Yu in Statistics all at Berkeley. TARGET AUDIENCES: Forest inventory and invasive plant management sectors in USDA, resource management sectors in USGS, California Air Resource Board and Department of Water Resources. Private industries and relevant university researchers who undertake image segmentation and lidar data processing research. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.
Impacts
Our results along the line of developing the field of photo-ecometrics have improved the accuracy and efficiency in forest inventory and invasive weed monitoring. The algorithms developed in our research can potentially be widely commercialized. The development of spatial-temporal models for schistosomiasis transmission can be adapted to other health problems that are dependent on environmental change and spatial connectivity. Remote sensing proves to be effective in supplying accurate snapshot information into process-based models. Software developed out of research from this project has been acquired by researchers in the US, Canada, China and Europe. Two American companies were funded to provide services based on our publications of algorithm development.
Publications
- Huang HB, P. Gong, FM Hui, N. Clinton, 2008. Reduction of atmospheric and topographic effect on Landsat TM data for forest classification, International Journal of Remote Sensing. 29(19):5623-5642.
- Weng, YL, P. Gong, ZL Zhu, 2008, Reflectance spectroscopy for the assessment of salt content in soils of the Yellow River Delta of China, International Journal of Remote Sensing. 29(19):5511-5531.
- Weng YL., P. Gong, ZL Zhu, 2008, Soil salt content estimation on the Yellow River Delta with satellite hyperspectral data, Canadian Journal of Remote Sensing. 34(3):259-270.
- Xu, B., P. Gong, 2008, Noise estimation in a noise-adjusted principal component transformation and hyperspectral image restoration, Canadian Journal of Remote Sensing. 34(3):271-286.
- Yu Q., P. Gong, Y. Tian, R. Pu, 2008. Factors affecting spatial variation of classification uncertainties in an object-based vegetation mapping. Photogrammetric Engineering and Remote Sensing, 74(8):1007-1018.
- Pu R., P. Gong, Q. Yu, 2008. Comparative Analysis of EO-1 ALI and Hyperion, and Landsat ETM+ Data for Mapping Forest Crown Closure and Leaf Area Index, Sensors, 8, 3744-3766 DOI: 10.3390/s8063744
- Chen Q., D Baldocchi, P. Gong, T. Dowson, 2008. Modeling radiation and photosynthesis of a heterogeneous savanna woodland landscape with a hierarchy of model complexity, Agricultural and Forest Meteorology. 148:1005-1020.
- Sedano F., D. Gomez, P. Gong, G. Biging, 2008. Tree density estimation in a tropical woodland ecosystem with multiangular MISR and MODIS data. Remote Sensing of Environment. 112:2523-2537.
- Pu, R., P. Gong, Y. Tian, X. Miao, R. Carruthers, G. 2008. Anderson, Invasive species change detection using artificial neural networks and CASI hyperspectral imagery. Environmental Monitoring and Assessment, 140(1-3):15-32.
- Liu DS., K. Song, JRG Townshend, P. Gong, 2008. Using local transition probability models in Markov random fields in forest change detection, Remote Sensing of Environment, 112:2222-2231.
- Pu RL., P. Gong, R. Michishita, T. Sasagawa, 2008. Spectral mixture analysis for mapping abundance of urban surface components with Terra/ASTER data. Remote Sensing of Environment, 112:939-954.
- Pu, R.L., M. Kelly, G. Anderson, P. Gong, 2008. Using CASI hyperspectral imagery to detect mortality and vegetation stress associated with a new hardwood forest disease, Photogrammetric Engineering and Remote Sensing. 74(1):65-75.
- Pu R., P. Gong, Y. Tian, X. Miao, R. Carruthers, G. Anderson, 2008. Using classification and NDVI differencing methods for monitoring of sparse vegetation:a case study of saltcedar in Nevada, USA, International Journal of Remote Sensing. 29(14):3987-4011.
- Sedano F., T Lavergne, LM Ibanez, P. Gong, 2008. A neural network based scheme coupled with RPV model inversion package, Remote Sensing of Environment. 112:3271-3283.
- Shen SQ, X Cheng, P Gong, 2008. Sensor web oriented web-based GIS, in Proc. of Web and Wireless based Geographic Information Systems, collected in Lecture Notes in Computer Sciences, 5373:86-95.
- Hui FM, B. Xu, HB. Huang, Q. Yu, P. Gong, 2008. Modeling spatial-temporal change of Poyang Lake using multi-temporal Landsat imagery, International Journal of Remote Sensing. 29(20):5767-5784.
|
Progress 10/01/03 to 09/30/08
Outputs
OUTPUTS: During the 5-year project period, our major emphasis has been placed on forest measurement, invasive weed mapping, wetland monitoring and environmental health studies in relation to schistosomiasis in China and plague in California all with remote sensing and geographic information systems technology. This has led to the graduation of 8 doctoral students in agricultural sciences; the organization of one international symposium on geoinformatics in 2007; the prototyping of 2 commercial software packages (Berkeleyimg Seg, an image segmentation software and Tiffs, a lidar data processing software for forest inventory); the world's first detailed comprehensive database on China's wetlands; and over 30 invited conference and on site seminar talks to academia and professionals. In addition, driven by the needs of this project, we established a new ecological research station on environment and health at Poyang Lake, China. PARTICIPANTS: Nick Clinton, Fire mapping and invasive cheatgrass monitoring Ashley Holt, Plague spread risk modeling with remote sensing and GIS Ruiliang Pu, LAI measurement and hyperspectral remote sensing Ryo Michishita, Thermal remote sensing and urban heat island Desheng Liu, High resolution data georeferencing and spatial temporal change detection Xin Miao, Hyperspectral monitoring of yellow starthistle Shaokui Ge, Salt cedar monitoring with field and aerial remote sensing Fernando Sedano, Land cover mapping in southern Africa with MISR and MODIS data Qi Chen, Forest measurement with airborne lidar data Qian Yu, High resolution detailed vegetation mapping Yongling Weng, Soil salination and hyperspectral remote sensing Huabing Huang, Lidar for forest inventory Fengming Hui, Wetland mapping Shaoqing Shen, WebGIS and wireless sensor network Weimin Li, Urban security and greenness interaction Formed alliances with the State Key Lab of Remote Sensing Science, jointly sponsored by Institute of Remote Sensing Applications, Chinese Academy of Sciences and Beijing Normal University; International Institute for Earth System Science, Nanjing University; School of Geography, Jiangxi Normal University; Sichun CDC and Jiangxi CDC in China. Maintained long-term collaboration with Greg Biging, Maggi Kelly, Ron Amundson, Denis Baldocchi in the College of Natural Resources, Robert Spear, Edmund Seto and Kirk Smith in the School of Public Health and Bin Yu in Statistics all at Berkeley. TARGET AUDIENCES: Forest inventory and invasive plant management sectors in USDA, resource management sectors in USGS, California Air Resource Board and Department of Water Resources. Private industries and relevant university researchers who undertake image segmentation and lidar data processing research. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.
Impacts
Our results along the line of developing the field of photo-ecometrics have improved the accuracy and efficiency in forest inventory and invasive weed monitoring. The algorithms developed in our research can potentially be widely commercialized. The development of spatial-temporal models for schistosomiasis transmission can be adapted to other health problems that are dependent on environmental change and spatial connectivity. Remote sensing proves to be effective in supplying accurate snapshot information into process-based models. Software developed out of research from this project has been acquired by researchers in the US, Canada, China and Europe. Two American companies were funded to provide services based on our publications of algorithm development.
Publications
- No publications reported this period
|
Progress 01/01/07 to 12/31/07
Outputs
We have been continuing our work in three aspects: image segmentation; lidar data processing; and invasive weed mapping all with remotely sensed data. During the past year, we investigated and experimented a number of image segmentation algorithms including region growing algorithms and watershed segmentation algorithms. Through algorithm optimization, we improved the efficiency of the watershed segmentation and implemented a code for region growing and developed a software package out of it. The software is now being circulated for further evaluation to relevant researchers in the US, Canada and China. Our research activities on algorithm development for lidar data processing has been progressing well. Although lidar data has become more affordable for average users, how to effectively process the raw data and extract useful information remains a big challenge. The generation of digital elevation models (bare earth) is the largest and fastest growing application of
lidar data. However, the research on automating the production of bare earth is still in its infancy. Until recently, researchers tended not to publish their methods. Besides terrain mapping, there is an endless list of areas where lidar has a potential application but not adequately explored. We published three papers on lidar data processing. One details about the algorithm for tree and terrain separation. The second described the use of lidar data to determine canopy biomass. The third is a general application of lidar data for tree height estimation. The research results have been included in the software package (Tiffs: Toolbox for Lidar Data Filtering and Forest Studies) for processing lidar data and extracting bare earth and forest structure information. The software has the capability in tiling lidar data, filtering point cloud to separate earth and forest and shrub, isolating individual trees, extracting forest structural parameters. The third area of research efforts are on
the use of hyperspectral remote sensing and Landsat Thematic Mapper data in mapping salt cedar, yellow starthistle, cheatgrass and other invasive weeds in California, Utah and Nevada. With 2 m resolution airborne hyperspectral data with a spectral imaging capacity ranging from 400 nm to 1000 nm, we were able to map yellow starthistle at an accuracy of better than 85%. For the case of cheatgrass, we were able to use multiple Landsat senses to map the cheatgrass percentage in large tract of lands in Utah. Using an innovative set of algorithm that compares unknown multispectral data vector with reference data vector, we were able to map the distribution of non-native grasses in Southern California from multiple Landsat scenes. The spatial extent obtained agrees well with knowledge of field experts. We will continue our research on validating this new set of algorithms.
Impacts
Our results along the line of developing the field of photo-ecometrics have improved the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. The development of spatial-temporal models for schistosomiasis transmission can be adapted to other health problems that are dependent on environmental change and spatial connectivity. Remote sensing proves to be effective in supplying accurate snapshot information into process-based models.
Publications
- Kwak DA, WK Lee, JH Lee, GS Biging, P. Gong, 2007. Detection of individual trees and estimation of tree height using lidar data. Journal of Forest Research. 12(6):425-434.
- Chen Q., P. Gong, D. Baldocchi, Y. Tian, 2007. Estimating basal area and stem volume for individual trees from LIDAR data, Photogrammetric Engineering and Remote Sensing. 73(12):1355-1365.
- Liu D., M. Kelly, P. Gong, Q. Guo, 2007. Characterizing spatial-temporal tree mortality patterns associated with a new forest disease, Forest Ecology and Management. 253:220-231.
- Miao X., P. Gong, R. Pu, R. Carruthers, J. Heaton, 2007. Applying a class-based feature extraction approach for supervised classification of hyperspectral imagery. Canadian Journal of Remote Sensing. 33(3):162-175.
- Miao, X., P. Gong, R. Carruthers, 2007. Detection of yellow starthistle through band selection and feature extraction from hyperspectral imagery. Photogrammetric Engineering and Remote Sensing, 73(9):1005-1015.
- Xu, B., P. Gong, 2007. Land use/cover classification with multispectral and hyperspectral EO-1 data, Photogrammetric Engineering and Remote Sensing, 73(8): 955-965.
- Wu, YZ., J. Chen, JF. Ji, P. Gong, et al. 2007. The study of metal characterization and distribution in suburban soils combining geostatistical and chemical approach. Soil Science Society of America Journal. 71(3):918-926.
- Pu, R., Z. Li, P. Gong, I. Csiszar, R. Fraser, W.-M. Hao, S. Kondragunta, and F. Weng, 2007. Development and analysis of a 12-year daily 1-km forest fire data across the North America from NOAA/AVHRR data, Remote Sensing of Environment. 108:198-208.
- Tian, Y.Q., R. McDowell, Q. Yu, G.W. Sheath, W.T. Carlson, P. Gong, 2007. Modeling to analyze the impacts of animal treading effects on soil infiltration, Hydrological Processes, 21:1106-1114.
- Guo QH, Kelly M, Gong P, Liu DS, 2007. An object-based classification approach in mapping tree mortality using high spatial resolution imagery, GIScience & Remote Sensing. 44 (1): 24-47.
- Chen Q., P. Gong, D. Baldocchi, G. Xie, 2007. Filtering airborne laser scanning data with morphological methods. Photogrammetric Engineering and Remote Sensing. 73(2):175-185.
|
Progress 01/01/06 to 12/31/06
Outputs
We have been continuing our work on the development of photo-ecometrics. Our recent emphasis has been placed on the development of algorithms for lidar data processing. Although lidar data has become more affordable for average users, how to effectively process the raw data and extract useful information remains a big challenge. Although lidar is appealing in many aspects they have some characteristics that post new challenges. First, lidar is essentially a kind of vector data. Different from raster data, the spatial locations of laser points have to be explicitly stored, making the file size much larger than imagery given the same "nominal" spatial resolution. Second, how to extract useful information from these seemingly random points is a relatively new research topic. The generation of digital elevation models (bare earth) is the largest and fastest growing application of lidar data. However, the research on automating the production of bare earth is still in its
infancy. Until recently, researchers tended not to publish their methods. Besides terrain mapping, there is an endless list of areas where lidar has a potential application but not adequately explored. We have developed a software package (Tiffs: Toolbox for Lidar Data Filtering and Forest Studies) for processing lidar data and extracting bare earth and forest structure information. The software has the capability in tiling lidar data, filtering point cloud to separate earth and forest and shrub, isolating individual trees, extracting forest structural parameters. Another major progress achieved by our group is the development of a spatial-temporal model for schistosomiasis transmission with input from GIS and remote sensing. Schistosomiasis japonica is a disease caused by parasites that are transported via surface water and that live in both snail and human hosts. The model employed a spatial interaction matrix based on neighborhood relationships and hydrologic connectivity to assess
the effect of intervillage parasitic transport on disease transmission and control. Satellite remote-sensing data served as input to the model for predicting snail density within each village, and for deriving a digital elevation model that was used to quantify hydrologic connectivity. Simulations of the model with varying levels of connectivity and in the presence and absence of chemotherapy control were run for 227 villages near Xichang City, in southwest Sichuan province, China. Increasing connectivity resulted in a geographic clustering of parasites within particular villages that produced higher levels of worm burden than in low and no connectivity simulations. Worm burden within a village could either increase or decrease with connectivity, depending on the degree to which parasites were imported and exported. Simulations of mass chemotherapy in select villages can result in a beneficial reduction in worm burden that extends to downstream neighbors. These findings suggest that
better understanding of intervillage connectedness can be exploited in the design of cost-effective control strategies.
Impacts
Our results along the line of developing the field of photo-ecometrics have improved the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. The development of spatial-temporal models for schistosomiasis transmission can be adapted to other health problems that are dependent on environmental change and spatial connectivity. Remote sensing provides effective in supplying accurate snapshot information into process-based models.
Publications
- Liu D., Gong, P., Kelly, M. and Guo, Q. 2006. Automatic registration of airborne images by combining area-based methods with local transformation models. Photogrammetric Engineering and Remote Sensing, 72(9):1049-1059.
- Pu, R., Gong, P., Michishita, R. and Sasagawa, T. 2006. Assessment of multi-resolution and multi-sensor data for urban surface temperature retrieval, Remote Sensing of Environment. 104:211-255.
- Ge, SK, Carruthers, R. and Gong, P. 2006. Hyperspectral characteristics of canopy components and structure for phonological assessment of an invasive weed, Environmental Monitoring and Assessment, 120:109-126.
- Chen Q., Baldocchi, D. and Gong, P. 2006. Isolating individual trees in a savanna woodland using small-footprint LIDAR data, Photogrammetric Engineering and Remote Sensing, 72 (8): 923-932.
- Clinton, N., Gong, P. and Scott, K. 2006. Quantification of pollutants emitted from very large wildland fires in Southern California, U.S.A, Atmospheric Environment. 40 (20): 3686-3695
- Yu, Q., Gong, P., Clinton, N., Biging, G. and Schirokauer, D. 2006. Object-based detailed vegetation mapping using high spatial resolution imagery, Photogrammetric Engineering and Remote Sensing, 72(7):799-811.
- Ge S., Carruthers, R.I., Gong, P. and Herrera, A. 2006. Texture analysis for invasive tamarix paviflora mapping using aerial photography along Cache Creek, California. Environmental Monitoring and Assessment. 114 (1-3): 65-83.
- Gong, P., Xu, B. and Liang, S. 2006. Remote sensing and geographic information systems in the spatial temporal dynamics modeling of infectious diseases, Science in China, Series C, 36(2): 184-192. (in Chinese), 49(6):573-582.
- Guo Y., Gong, P., Amundson, R. and Yu, Q. 2006. Analysis of Factors Controlling Soil Carbon in the Conterminous United States. Soil Science Society of America Journal. 70(2):601-612.
- Guo Y., Amundson, R., Gong, P. and Yu, Q. 2006. Quantity and spatial variability of soil carbon in the conterminous United States. Soil Science Society of America Journal. 70 (2): 590-600.
- Liu D., Kelly, M. and Gong, P. 2006. A spatio-temporal approach to monitoring forest disease spread using multi-temporal high spatial resolution imagery. Remote Sensing of Environment. 101 (2): 167-180.
- Xu, B., Gong, P., Seto, E., Liang, S., Yang, C., Wen, S., Qiu, D., Gu, X.G. and Spear, R. 2006. A spatial-temporal model for assessing the effects of inter-village connectivity in schistosomiasis transmission, Annals of AAG. 96(1):31-46.
- Gong, P., Pu, R., Li, Z.Q., Scarborough, J., Clinton, N. and Levien, L. 2006. An integrated approach to burned area mapping in California with NOAA AVHRR data, Photogrammetric Engineering and Remote Sensing. 72(2):139-150.
|
Progress 01/01/05 to 12/31/05
Outputs
We have been continuing our work on the development of photo-ecometrics. A wide range of experiments have been carried out for extracting forest structural information from remotely sensed imagery. This is attempted in order to save cost of forest sampling. After undertaking atmospheric correction of airborne and satellite borne hyperspectral imagery, we compare their potential in estimating the canopy closure and leaf area index of a conifer forest. Our results indicate that higher spatial resolution at 4-5 m level such as that possessed by the airborne hyperspectral data is needed to achieve better accuracy in forest canopy closure and leaf area index. We also experimented with land cover classification with multi-temporal MODIS data from the Terra satellite. MODIS data has a coarser resolution of 250-1000 m. The classification has been performed over a Southern Africa Miombo ecosystem in Mozambique. After comparing a number of combination strategies of images
acquired from multiple dates we find it is important to synthesize the multitemporal data using principal component analysis applied to the band-wise differences of multitemporal imagery. By retaining the top two principal components in combination with the original data of one date, we can achieve the highest classification accuracy at the level of approximately 90%. This is substantially greater than that of the standard MODIS land cover product produced by the University of Maryland funded by NASA. We have made progress in developing new image processing algorithms for airborne imagery with a spatial resolution as high as 0.2 m. On high spatial resolution imagery over urbanized areas, shadow presents a serious problem in machine-based image analysis. We developed a shadow and shade reduction system. The system includes shadow simulation, ray tracing, integrated shadow detection, and shadow removal. The geodetic shadow is simulated from a digital surface model (DSM) and the sun
attitude. Using the camera model and collinearity equations, the corresponding scan line of the shadow is obtained. Ray tracing is used to determine the visibility of the shadow in the image. A building boundary driven height field ray tracing method is proposed to boost the tracing efficiency. Shadow segmentation is taken from the red-green-blue image at the base of the traced image shadows. The segmentation threshold is derived from the histogram of the traced image shadow area. Shadow removal includes the shadow region and companion region labelling, histogram processing, and intensity mapping. Our results show the effectiveness of the methods based on visual inspection.
Impacts
Our results along the line of developing the field of photo-ecometrics will considerably improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. The development of analysis methods for high resolution airborne image interpretation is valuable in achieving automatic information extraction from remotely sensed data. Our land cover mapping algorithms made innovative contribution to the use of texture with high spatial resolution images.
Publications
- Mao K, Z. Qin, J. Shi, P. Gong, 2005. A practical split-window algorithm for retrieving land-surface temperature from MODIS data, International Journal of Remote Sensing, 26(15):3181-3204.
- Sedano F., P. Gong, M. Ferrao, 2005. Land covers assessment with MODIS imagery in Southern African Miombo ecosystems, Remote Sensing of Environment. 98(4):429-441.
- Tian, Y., D.G. McCall, W. Dripps, Q. Yu, P.Gong, 2005. Using computer vision technology to evaluate the meat tenderness of grazing beef, Food Australia. 57(8):322-326.
- Anderson, G., R. Carruthers, S. Ge, P. Gong, 2005. Monitoring of invasive Tamarix distribution and effects of biological control with airborne hyperspectral remote sensing, International Journal of Remote Sensing. 26(12):2487-2489.
- Pu, R., Q. Yu, P. Gong, G.S. Biging, 2005. EO-1 Hyperion, ALI and Landsat 7 ETM+ data comparison for estimating forest crown closure and leaf area index, International Journal of Remote Sensing, 26(3):457-474.
- Gong, P., X. Miao, S. Ge, K. Tate, C. F. Battaglia, G.S. Biging, 2004. Water table level in relation to EO-1 ALI and Landsat ETM+ data over a mountainous meadow in California, Canadian Journal Remote Sensing. 32(5):691-696.
- Zhou SQ., X. Liang, Jing M. Chen, P. Gong, 2004. An assessment of VIC-3L hydrological model for the Yangtze River Basin based on remote sensing ? A case study of Baohe watershed, Canadian Journal Remote Sensing. 32(5): 840-853.
- Li. Y., P. Gong, 2005. An efficient texture image segmentation algorithm based on the GMRF model for classification of remotely sensed imagery, International Journal of Remote Sensing, 26(22):5149-5159.
- Li Y., P. Gong, T. Sasagawa, 2005, Integrated shadow removal based on photogrammetry and image analysis, International Journal of Remote Sensing, 26(18):3911-3929.
|
Progress 01/01/04 to 12/31/04
Outputs
We have been continuing our work on the development of photo-ecometrics. The emphasis has been placed on the detection of individual tree tops of conifer forest with high resolution remotely sensed imagery. This is attempted in order to save cost of forest sampling. We derive individual tree-crown boundaries and treetop locations under a unified framework involving a two-stage approach with edge detection followed by marker-controlled watershed segmentation. Our initial results show that the computer based method is comparable to manual identification but saves a significant amount of time. We also evaluated new ways to estimate leaf area index (LAI) with images from the only satellite hyperspectral sensor - EO-1 Hyperion. We compared the performance of three feature extraction methods for mapping forest crown closure (CC) and LAI. The methods are band selection (SB), principal component analysis (PCA) and wavelet transform (WT). A total of 38 field measurements of CC
and LAI were collected at Blodgett Forest Research Station, University of California at Berkeley. The experimental results indicate that the energy features extracted by the WT method are the most effective for mapping forest CC and LAI. We developed a wildfire-mapping algorithm and applied it to daily NOAA/AVHRR/HRPT data for wildland areas (scrub, chaparral, grassland, marsh, riparian forest, woodland, rangeland and forests) in California for September and October 1999. Daily AVHRR images acquired from two successive days are compared for active fire detection and burn scar mapping. We made some comparisons between the result mapped by the dynamic algorithm and the fire polygons collected by the California Department of Forestry and Fire Protection through ground survey. We found that our algorithm can track burn scars at different developmental stages at a daily level. We evaluated the potential of a frequency-based contextual classifier (FBC) for land-use classification with a
panchromatic Ikonos image (1 m resolution). To capture the spatial arrangement of image gray-level values and use such information in image classification, we applied texture spectrum (TS) directly in the FBC. The effects of several data preprocessing and reduction methods on the performance of the FBC were also evaluated. The methods include four gray-level reduction (GLR) techniques and modifications to the TS technique. The TS method performed considerably better than the other methods in identifying 9 land use types in a mountainous environment in China. In a different study, we used IKONOS 1-m panchromatic and 4-m multispectral images to map three species of mangroves in a study site along the Caribbean coast of Panama. Three different classification methods were investigated: maximum likelihood classification (MLC) at the pixel level, nearest neighbour (NN) classification at the object level, and a hybrid classification that integrates the pixel and object-based methods (MLCNN).
Among the three classification methods, MLCNN achieved the best average accuracy of 91.4 percent.
Impacts
Our results along the line of developing the field of photo-ecometrics will considerably improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. Our burned area mapping algorithm has improved mapping accuracy in comparison to other burnt scar mapping algorithms in California. This is helpful for the study of global carbon cycling. Our land cover mapping algorithms made innovative contribution to the use of texture with high spatial resolution images.
Publications
- Tian, Y., Radke, J.R., Gong, P. and Yu, Q. 2004. Spatially allocating PM2.5 emissions from residential wood burning in California, Atmospheric Environment. 38:833-843.
- Pu, R. and Gong, P. 2004. Prediction of burn scars using logistic regression and neural networks from a single date post fire Landsat 7 image, Photogrammetric Engineering and Remote Sensing. 70(7): 841-850.
- Wang L., Gong, P. and Biging, G.S. 2004. Individual tree crown delineation and treetop detection in high spatial resolution aerial imagery, Photogrammetric Engineering and Remote Sensing. 70(3):351-367.
- Wang L., Sousa, W. and Gong, P. 2004 Integration of object-based and pixel-based classification for mangrove mapping with IKONOS imagery. International Journal of Remote Sensing. 25(24):5655-5668.
- Gong P. and Chen, J.M. 2004. Special issue: remote sensing of water resources studies. Canadian Journal of Remote Sensing. 32(5):II-IV.
- Pu, R., Foschi, L. and Gong, P. 2004. Spectral feature analysis for assessment of water status and health level of coast live oak (quercus agrifolia) leaves, International Journal of Remote Sensing. 25 (20): 4267-4286.
- Wang L., Sousa, W., Gong, P. and Biging, G.S. 2004. Comparison of IKONOS and Quickbird images for mangrove mapping in Panama, Remote Sensing of Environment. 91(3-4):432-440.
- Xu B., Gong, P., Biging, G.S., Liang, S., Seto, E. and Spear, B. 2004. Snail density prediction for schistosomiasis control using IKONOS and ASTER images. Photogrammetric Engineering and Remote Sensing. 70(11): 1285-1294.
- Chen D., Stow, D. and Gong, P. 2004. Examining the effects of resolution on classification accuracy: an urban environmental case, International Journal of Remote Sensing. 25(11):2177-2192.
- Chen, Q. and Gong, P. 2004. Automatic variogram parameter extraction for textural classification of IKONOS imagery. IEEE Transactions on Geoscience and Remote Sensing. 42(5):1106-1115.
- Pu, R. and Gong, P. 2004. Wavelet transform applied to EO-1 Hyperion data for forest LAI and crown closure mapping, Remote Sensing of Environment. 91:212-224.
|
Progress 01/01/03 to 12/31/03
Outputs
GRADUATE STUDENTS = 6 We have been continuing our work on the development of photo-ecometrics. DSMs (digital surface models) automatically derived with digital photogrammetric systems are useful in land surface change monitoring including forest growth monitoring. Major efforts have been made to improve the automatic generation of orthophotographs of the forested canopy. Traditional orthophoto-making does not include the orthoprojection of forest canopy and therefore they are not truly orthophotos over forested and urban areas. Our procedure can produce high quality large scale orthophotos (upto 1:2,400 experimented). Such results have not been reported elsewhere. In addition to this, we have been working on tree species identification and individual tree mapping with satellite and airborne remote sensing data at a resolution between 0.5-1 m. We developed a new texture analysis algorithm to classify land cover and land use in a rural area with considerable success.
This part of works involved study areas from California, to Panama to mountainous China. We continued to make progress in estimating forest biophysical parameters such as crown closure and forest leaf area index. Good results are obtained for the range land of California and for plantation forest in New Zealand. With the internet based fire emission estimation system, we estimated the fire emissions for the 2003 southern California fire and for the entire North America during 1989-2000. We used burn scars determined from historical NOAA AVHRR imagery and validated them in northwest US. The validation indicates that the burn scars mapped for forested land are about 80% accurate. They are not as accurate in grassland and shrub areas. We are developing new methods to improve the fire mapping accuracy over non-forested areas. We have achieved some important progress in studying the soil diversity in the US through a GIS approach. Based on the State Soil Geographic Database, we produced an
inventory of all soil types down to the series level for the entire US with their area and spatial distribution. From this inventory, we can find what soils are rare in a state or in the US. Overlaying the soils map with the land use map of the US, we highlighted areas of soil endanger level by identifying areas of high level of land use such as urban and agricultural use. We are working on the survey of soil carbon both inorganic and organic for the entire US. We compared the quality of urban land cover derived from the night light satellite data with urban areas determined from daytime multispectral satellite data. This can improve our understanding of what could be misleading in urban areas derived from the night light data. As a result, the night light data must be used in an adaptive way in determining urban areas in the world. In the high energy consumption areas such as the US, it tended to over estimate the extent of urban areas.
Impacts
Our results along the line of developing the field of photo-ecometrics will considerably improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. Our North America burnt area map between 1989-2000 is highly demanded in the global ecosystem science community for the study of global carbon circulation. Our soil diversity work has caused a considerable amount of public attention on how to preserve our soil resources in the United States.
Publications
- Xu, B., P. Gong, S. Liang, E. Seto, B. Spear, 2003. Snail density estimation for schistosomiasis control by integrating field survey and multiscale satellite images, Geographic Information Sciences, 9(1-2):97-100.
- Guo, Y., R. Amundson, P. Gong, and R. Ahrens, 2003. Taxonomic structure, spatial distribution and relative abundance of the soils in the United States. Soil Science Society of America Journal. 67:1507-1516.
- Guo, Y., P. Gong, R. Amundson, 2003. Pedodiversity in the United States of America, Geoderma. 117:99-115.
- Pu, R. P. Gong and G. S. Biging, 2003. Simple calibration of AVIRIS data and LAI mapping of forest plantation in southern Argentina, International Journal of Remote Sensing. 24 (23): 4699-4714.
- Amundson, R., Y. Guo, P. Gong, 2003. Soil diversity and land use in the United States, Ecosystems. 6(5):470-482.
- Pu, R., P. Gong, G.S. Biging, M. Larrieu, 2003. Extraction of red edge optical parameters from Hyperion data for forest LAI estimation. IEEE Transactions on Geoscience and Remote Sensing 41(4): 916-921.
- Gong, P., R. Pu, G.S. Biging, M. Larrieu, 2003. Estimation of forest leaf area index using vegetation indices derived from Hyperion hyperspectral data. IEEE Transactions on Geoscience and Remote Sensing. 41(6): 1355-1362.
- Pu, R. B. Xu, P. Gong, 2003. Oakwood crown closure estimation by unmixing of Landsat TM data, International Journal of Remote Sensing. 24(22):4433-4445.
- Zhang, Q., J.Wang, P.Gong and P.Shi, 2003. Urban spatial pattern analysis from SPOT panchromatic imagery using textural analysis, International Journal of Remote Sensing. 24(21):4137-4160.
- Sun W, V. Heidt, P. Gong, G. Xu, 2003. Information fusion for land use classification from high resolution imagery, IEEE Trans. Geosci and Remote Sensing, 41(4): 883-890.
- Xu, B., P. Gong, R. Spear and E. Seto, 2003. Comparison of different gray level reduction schemes for a revised texture spectrum method for land-use classification using IKONOS imagery, Photogrammetric Engineering and Remote Sensing. 69(5):529-536.
- Pu, R., S. Ge, N.M. Kelly, P. Gong, 2003. Spectral absorption features as indicators of water status in Quercus Agrifolia leaves, International Journal of Remote Sensing. 24(9): 1799-1810.
- Xu, B., P. Gong, R. Pu, 2003. Crown closure estimation of oak savannah in a dry season with Landsat TM imagery: Comparison of Various Indices through Correlation Analysis, International Journal of Remote Sensing. 24(9):1811-1822.
- Pan, Y., Li, X., P. Gong, C. He, P. Shi, R. Pu, 2003. An integrative classification of vegetation in China with NOAA/AVHRR and vegetation-climate indices of Holdridge's life zone, International Journal of Remote Sensing. 24(5):1009-1027.
- Chen, J., P. Gong, C. He, R. Pu, P. Shi, 2003. Land use/cover change detection using improved change vector analysis, Photogrammetric Engineering and Remote Sensing. 69(4):369-379.
- Gong, P., S. Mahler, G.S. Biging, and D. Newburn, 2003. Vineyard identification in an oak woodland landscape with airborne digital camera imagery, International Journal of Remote Sensing. 24(6):1303-1315.
- Henderson M, E. Yeh, P. Gong, C. Elvidge, K. Baugh, 2003. Validation of urban boundaries derived from night-time satellite imagery, International Journal of Remote Sensing.24(3):595-609.
- Li Z., R. Fraser, J. Jin, A.A. Abuelgasim, I. Csiszar, P. Gong, R. Pu, W. Hao. 2003. Evaluation of the Algorithms Used for Developing a Long-term Fire Inventory across North America and a Close-look at 2000 US Western Fires. Journal of Geophysical Research, 108, D2, 4076 ACL20 1-22.
- Sheng Y., P. Gong, G.S. Biging, 2003. True orthoimage production from large scale aerial photographs. Photogrammetric Engineering and Remote Sensing. 69(3):259-266.
- Sheng, Y., P. Gong, G.S. Biging, 2003. Model-based conifer canopy surface reconstruction from photographic imagery: from single tree to a forest stand, Photogrammetric Engineering and Remote Sensing. 69(3):249-258.
|
Progress 01/01/02 to 12/31/02
Outputs
GSY = 6. We have been continuing our work on the development of photo-ecometrics. DSMs (digital surface models) automatically derived with digital photogrammetric systems are useful in land surface change monitoring including forest growth monitoring. Major efforts have been made to improve the automatic generation of orthophotographs of the forested canopy. Traditional orthophoto-making does not include the orthoprojection of forest canopy and therefore they are not truly orthophotos over forested and urban areas. Our procedure can produce high quality large scale orthophotos (up to 1:2,400 experimented). Such results have not been reported elsewhere. We developed an internet based fire emission estimation system. The system can estimate wildland fire, agricultural burning and residential wood burning based on fuel maps, burnt areas, and socio-economic and climate conditions for residential areas. The system is useful in prescribed burning and for postfire
estimation. It has been implemented for the state of California. We also mapped hotspots of North America for 1989-2000 using daily Advanced Very High Resolution Radiometer (AVHRR) on board the NOAA satellite series. Burn scar mapping and validation are underway. We continued to make progress in estimating forest biochemical constituents using in situ field spectral measurement data. At a Giant Sequoia plantation site, our results indicate that spectrally derived vegetation indices and linearly combined spectral data are correlated to total potassium, total phosphorus and total nitrogen of foliage. We also worked on spatio-temporal modeling of microbial contamination on grazed land in New Zealand. We built a grid-based and stream networked model to predict the fecal contaminants in a hilly land catchment extensively used for grazing. Experimental results indicate that our model captures the key features that control the population dynamics of the fecal contaminants. Through
international collaboration we expanded our forest leaf area index estimation work to Argentina and the carbon balance work to China. Our carbon balance work involved the use of NOAA AVHRR data from 1982-1998 for calculating the net primary productivity. For the first time, we estimated the carbon balance over the terrestrial ecosystems of China during 1982-1998.
Impacts
Our results along the line of developing the field of photo-ecometrics will considerably improve accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized. Our North America burnt area map between 1989-2000 is highly demanded in the global ecosystem science community for the study of global carbon circulation.
Publications
- Seto E., Xu, B., Liang, S., Spear, R., Gong, P., Wu, W., Davis, G., Qiu, D. and Gu, X. 2002. The use of remote sensing for predictive modelling of schistosomiasis in China, PE&RS. 68(2):167-174.
- Tian, Q., Xu, B., Gong, P., Wang, X., Tong, Q. and Guo, H. 2002. Reflectance, dielectric constant and chemical contents of sedimentary rocks, Interational Journal of Remote Sensing. 23(23):5123-5128.
- Tian, Y., Gong, P., Radke, J. and Scarborough, J. 2002. SAMC- a spatial and temporal system for analyzing microbial contaminants on grazing farmlands, Journal of Environmental Quality, 31:860-869.
- Zhang, Q., Wang, J., Peng, X., Gong, P. and Shi, P. 2002. Urban built-up land change detection with road density and spectral information from multi-temporal Landsat TM data, International Journal of Remote Sensing. 23(15):3057-3078.
- Gong, P., Sheng, Y. and Biging, G.S. 2002. 3D model-based tree measurement from high resolution aerial imagery, PE&RS, 68(11): 1203-1212.
- Gong, P., Xu, M., Chen, J., Chen, J.M., Qi, Y., Biging, G., Liu, J. and Wang, S. 2002. A preliminary study on the carbon dynamics of China's terrestrial ecosystems in the past 20 years, Earth Science Frontiers, 9(1):55-61.
- Gong, P., Mei, X., Biging, G.S. and Zhang, Z. 2002. Improvement of oak canopy model extracted from digital photogrammetry. PE&RS. 68(9): 919-924.
- Gong, P., Pu, R. and Heald, B. 2002. In situ hyperspectral data analysis for nutrient estimation of giant sequoia. IJRS. 23(9):1827-1850.
- Chen, PY, Srinivasan, R., Fedosejevs, G., Baez-Gonzalez, A. and Gong, P. 2002. Assessment of NDVI using merged NOAA-14 and NOAA-15 AVHRR data. Geographic Information Sciences, 8(1):31-38.
- Chen J., Gong, P., He, C., Luo, W., Tamura, M. and Shi, P. 2002. Assessment of urban development plan of Beijing by using CA-based urban growth model, PERS. 68(10): 1063-1071.
|
Progress 01/01/01 to 12/31/01
Outputs
GSY = 6. We have been continuing our work on the development of photo-ecometrics. DSMs (digital surface models) automatically derived with digital photogrammetric systems are useful in land surface change monitoring including forest growth monitoring. However, they cannot be applied directly to forest canopy change analysis with high accuracy due to the inevitable deficiencies of existing commercial digital photogrammetry packages. In a hardwood rangeland monitoring study, we found that the oak tree and woodland canopy boundaries were not well determined using several digital photogrammetry packages available to us. We developed a correction method for improvement at the erroneous canopy boundary locations in the DSM using shadow and boundary information extracted from imagery. Aerial photographs taken from oak woodland hills were tested. Using manual photogrammetric measurements as reference, we found that most of the points (88.3%) on the canopy boundaries were
displaced by greater than 1 m with a conventional digital photogrammetric package. After the proposed algorithms were applied, greater than 98.6% of the points on canopy boundaries were found to be within 1 m of their reference positions. In our efforts toward conifer tree species recognition we found that hyperspectral measurements taken from young trees in the 400 - 950 nm range can be transformed into derivative forms to improve the accuracy of tree species recognition from 60% to greater than 85%. Practically, we developed a 3D model-based tree interpreter, a semi-automatic method for tree measurement from high-resolution aerial images. It emphasizes on the extraction of the 3D geometric information such as tree location, tree height, crown depth (or crown height), crown radius and surface curvature. First, trees are modeled as 3D hemi-ellipsoids with the following parameters: tree-top coordinates, trunk base height, crown depth, crown radius, and crown surface curvature. This
model-based approach turns a tree interpretation task to a problem of optimal tree model determination. Multi-angular images are used to determine the optimal tree model for each tree. Tree-tops in each image of a stereo pair are identified interactively with an epipolar constraint, and the 3D geometry of trees can be determined automatically. With such a semi-automatic scheme, efficiency and reliability of 3D tree measurements are achieved by taking advantages of both the operator's interpretation skills and the machine's computation. The method was tested with a closed conifer stand on 1:2,400 photographs. An overall accuracy of 94% and 90% was obtained for tree height and crown radius measurements, respectively. At the continental scale, we have been continuing a fire history mapping of major forest fires over the US and Canada between 1985 and 2000. We adapted and developed new algorithms to detect hot spots and map fire scars using daily Advanced Very High Resolution Radiometer
(AVHRR) on board the NOAA satellite series. Validation is underway.
Impacts
Our results along the line of developing the field of photo-ecometrics will considerably improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized.
Publications
- Gong, P., Py, R., and Yu, B. 2001. Conifer species recognition: effects of data transformation, Int. J. Remote Sensing. 22(17):3471-3481.
- Tian, Y., Gong, P., Ibbitt, R, Elliott, A. 2001. Applications of GIS in analyzing spatial patterns of multiple runoff events, Geographic Information Sciences, 7(1):16-23. Sheng, Y., Gong, P., Biging, G.S. 2001. Model-based conifer crown surface reconstruction from high-resolution aerial images, PE&RS.67(8): 957-965.
- Tian, Y., Davies-Colley, R.J., and Gong P. 2001. Estimating solar radiation on slopes of arbitrary aspect, Agricultural and Forest Meteorology.109(1):67-74.
- Tian Q., Gong, P., Zhao, C., Guo, G. 2001. A feasibility study on diagnosing wheat water status using spectral reflectance, Chinese Science Bulletin, 40(8):669-671.
- Mei, X., Gong, P., and Biging, G.S. 2001. Image matching based on tracking matching paths in the similarity space, PE&RS, 67(4):453-460.
|
Progress 01/01/00 to 12/31/00
Outputs
GSY = 4. We further expanded our effort in the analysis of in situ hyperspectral data to differentiate oak species. While this work is still underway, some preliminary findings indicate that water contents in oak leaves collected in Marin County, California, are highly correlated with spectral data collected in the vicinity of 975 nm, 1200 nm and 1750 nm positions. However, our emphasis has primarily been on photo-ecometrics. We continued developing a number of new algorithms in digital photogrammetry and applied them to conifer and broadleaf crown reconstruction from a single tree level to the stand level from stereopairs of aerial photographes. Using 1:40,000 aerial photograph as ground truth, we evaluated the potential of Landsat Thematic Mapper (TM) imagery in crown closure estimation of California's hardwood rangeland. Using 1:2,400 and 1:12,000 aerial photographs, we developed new image matching algorithms to extract tree crown morphology and tree heights for
both broadleaf and conifer species in California. In addition, we experimented with methods for deriving digital surface models (DSM) and orthophotos generated with digital photogrammetry for the purpose of landscape characterization. Using 1:23,000 aerial stereopairs acquired over a coastal marshland area in Florida in 1951 and 1997, we produced a digital orthophoto for each year. Through measurement, we found the lower boundaries of the salty sandy zones on the marshland displaced toward the land for approximately 3-10 m from 1951 to 1997. Since it was unlikely that the sand type had changed over the 46-year period, the only possible explanation is that either the annual average sea level or the evaporation rate or both have undergone change in this area, leading to a conclusion that climate had changed over the 46-year period. At the continental scale, we started a fire history mapping of major forest fires over the US and Canada between 1985 and 2000. We developed algorithms to
detect hot spots and map fire scars using daily Advanced Very High Resolution Radiometer (AVHRR) on board the NOAA satellite series. Validation is underway.
Impacts
Our results along the line of developing the field of photo-ecometrics will dramatically improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized.
Publications
- Yi, C., Gong, P., Qi, Y., and Xu, M. 2000, The effect of buffer and temperature feedback on oceanic uptake of CO2, Geophysical Research Letters. 28(5):751-754.
- Pu, R., and Gong, P., 2000. Band selection from hyperspectral data for conifer species identification, Geographic Information Sciences, 6(2): 137-142.
- Gong, P., and Mu, L., 2000. Error detection in map databases: a consistency checking approach, Geographic Information Sciences, 6(2):188-193.
- Xu, M., Qi, Y., and Gong, P. 2000, China's new forest policy and its potential negative impacts, Science, Sept 22, 2000:2049.
- Gong, P. 2000. Digital surface model and topographic change monitoring, Quarternary Sciences, 20(3):247-251.
- Gong, P., Biging, G., and Standiford, R. 2000. The potential of digital surface model for hardwood rangeland monitoring, Journal of Range Management. 53:622-626.
- Zhou, Y., Maszle, D., Spear, R., Gong, P., Gu, X., Qiu, D., Liang, S. and Wen, S. 1999. GIS and spatial analysis in schistosomiasis transmission and control in moutainous areas, Chinese Journal of Schistosomiasis Control, 11(5):263-266.
|
Progress 01/01/99 to 12/31/99
Outputs
GRADUATE STUDENTS = 4. We expanded our effort in the analysis of in situ hyperspectral data collected in the field to differentiate vine species. This work is still underway. No conclusive findings are available. Our emphasis has been on photo-ecometrics. We employed digital photogrammetry techniques to measure oak woodland change. Instead of using only the radiometric values available in aerial photographs and digital camera imagery, we extracted 3D terrain and canopy surface morphological data. We used such data to determine changes in tree heights and crown closure. Automatic measurement results were compared with photo interpretation techniques. With 1:12,000 scale aerial photography obtained in 1970 and 1995 over Lucas Valley of Marin county, we found that the errors of crown closure and tree height estimation using the automated method were less than 0.7 percent and 1.5 m, respectively. We also developed new image matching algorithms that can improve accuracies
of 3D surface data extraction over broadleaf and conifer forestland. For conifer forest, it is almost impossible to reconstruct canopy surfaces with existing algorithms as implemented in any software package that is commercially available. With 20 cm resolution digital aerial photographs, our model-based image matching technique can produce tree height measurements that are within 10 percent.
Impacts
Our results along the line of developing the field of photo-ecometrics will dramatically improve the accuracy and efficiency in forest inventory. The algorithms developed in our research can potentially be widely commercialized.
Publications
- Gong, P., Wang, D., and Liang, S. 1999. Inverting a canopy reflectance model using an artificial neural network. International Journal of Remote Sensing. 20(1):111-122.
- Gong, P., Zhang, A., 1999. Noise effect on linear spectral unmixing, Geographic Information Sciences, 5(1): 52-57.
- Gong, P., Biging, G. S., Lee, S.M.,Mei, X., Sheng, Y., Pu, R., Xu, B., Schwarz, K-P. 1999. Photo-ecometrics for forest inventory, Geographic Information Sciences, 5(1): 9-14.
- Gong, P. 1999. Development of photo-ecometrics, Journal of Natural Resources, 14(4):313-317.
- Gong, P., Mei, X., Biging, G., Zhang, Z. 1999. Monitoring oak woodland change using digital photogrammetry, Journal of Remote Sensing. 3(4):285-289.
- Pu, R., Gong, P., Yang, R. 1999. Forest yield prediction with an artificial neural network and multiple regression, Chinese Journal of Applied Ecology, 10(2):129-134.
- Yu, B., Ostland, M., Gong, P., Pu, R. 1999. Penalized linear discriminant analysis for conifer species recognition, IEEE Transactions on Geoscience and Remote Sensing. 37(5):2569-2577.
- Spear, R., Gong, P., Seto, E., Xu, B., Zhou, Y., Maszle, D., Liang, S., Davis, G., Gu, X. 1998. GIS and remote sensing for schistosomiasis control in Sichuan, China, Geographic Information Sciences, 4(1-2): 14-22.
- Pu, R., and Gong, P. 1998. Modeling land-cover changes with gray systems theory and multitemporal aerial photographs, Geographic Information Sciences, 4(1-2): 74-79.
|
|