| Confined by technical factors,a high spatial resolution may not be achieved with the high temporal resolution sustained while the sensors collecting the ground data,which leads to the incompatibility between spatial resolution and temporal resolution of image.And there will be the further negative effect on the remote sensing monitor of the ground data with wide coverage,high precision and quickly variation.The data spatial-temporal fusion technique that combines the distribution characteristics of spatial details of the data with high spatial resolution and the temporal variation data with high temporal resolution efficiently by the changing relation between data with high spatial resolution and data with high temporal resolution,is an effective method to solve the problems existed in data of single sensor.The performance of spatial-temporal fusion has improved gradually as the constant proposals and developments of the algorithms of spatial-temporal data fusion go on.And the analysis on the temporal variation of the data with high spatial resolution and high temporal resolution is the key to establish the changing relation between the data with high spatial and temporal resolution.Additionally,the high fusion of spatial and temporal resolution of data is hard to approach if the optical remote sensing data are collected in region with commonly cloudy and foggy weather,even supported by the effective spatial-temporal fusion algorithm of data.This research focuses on the pixel deviation of the same pixel locations from different sensors data,which has combined with the influence of ground features changes on the observed values from different sensors.In this research,the ground features has been separated into four types—cultivated fields,woodland,construction area and water,and the reflection deviation data of data with high spatial resolution and data with high temporal resolution from different ground features as well as the pixel deviation data of optical remote sensing and SAR da from different ground features has been collected.And the further discussion on the variation of reflection deviation of different sensors'data in different wave bands and ground features by the time series analysis.(1)The reflection deviation variation of data with high spatial resolution(Landsat)and the data with high temporal resolution(MCD34,VIIRS)in different wave bands(blue,green,red and near infrared)and different ground features(cultivated fields,woodland,construction area and water)are both white noise,which indicates that,among the all series of reflectance deviation,there is no correlation between each observed value and there is tendency and season changing information contained temporally in reflection deviation.So there is no necessity to carry out consequent analysis and modelling.Meanwhile the hypothesis that a linear variation of reflection deviation of different sensors is existed under the support of current spatial-temporal fusion algorithms,but that the establishment of the linear relation model to realize the spatial-temporal fusion of data with short interval is the best solution currently.(2)Subjected to the results of pixel deviation of optical remote sensing data and SAR data collected,the characteristics expressed in the temporal variation of pixel deviation of NDVI integrated with VH,VV and RVI from different ground features are more stable than the pixel deviation of single wave band and single polarization.Therefore,in this research,the pixel deviation analysis on optical remote sensing and SAR data only focuses on the pixel deviation of NDVI and VH,NDVI and VV,NDVI and RVI in different ground feature.Based on R language,the pixel deviation from different ground features of data each group will be analyzed respectively,and the pixel deviation series of each group will be fitted according to proper mathematical statistics models consequently in this research.The degree of fitting will be represented by the correlation coefficient R~2 to evaluate,and the results show that the correlation coefficient R~2 between the predicted value of tendency and practical value is up to 0.8+,which indicates that the predicted values of polynomial curve has fitted well with the practical values and been able to represent the practical values to some degree.(3)Based on the models of pixel deviation variation of NDVI and VH,NDVI and VV,NDVI and RVI,the improvement of Enhanced Spatial-Temporal Adaptive Reflectance Fusion Model(ESTARFM)has been completed and the Improved Spatial-Temporal Adaptive Reflectance Fusion Mode(ISTARFM)has consequently gained in this research.The same group of data has been fused respectively by ESTARFM and ISTARFM,and the correlation reviews of fusion results and observed data demonstrate that no matter whether the interval of data input is 120d or 324d,the correlation coefficients R~2 are both over the 0.7,and it has detected that fusion by ISTARFM usually gets a higher precision of fusion than the fusion by ESTARFM under a sustained time interval. |
PDF Full Text Request