Validation Of Quantitative Remote Sensing Products Over Heterogeneous Land Surfaces

The satellite-derived products suffer from errors resulted from the satellite sensors' limitations,the algorithm instability and the surface heterogeneity.Thus,validation of satellite-derived products is an essential precursor to scientific research and subsequent environment monitoring.However,it is challenged by the large spatial scale difference between ground measurements and satellite product pixel.Heterogeneity,rather than homogeneity,is a common characteristic of most land surfaces.It complicates the validation process and is inevitable.One way to validate coarse-pixel products over heterogeneous land surfaces is to apply a multi-scale validation strategy that uses highresolution image,as an upscaling bridge to reduce the scale discrepancy(Liang et al.,2002;Peng et al.,2014).However,using the high-resolution image to calculate the reference value for validation may introduce additional errors from the inversion or geometric registration process,so the absolute accuracy of the remote sensing products could suffer from the effects of error propagation.Furthermore,the low update of highresolution maps restrict validation with continuity in time series.The wireless sensor network(WSN)technology provides access to continuously observe on the large pixel scale.Taking the albedo products as an example,this paper was dedicated to the validation of coarse-scale albedo products over heterogeneous surfaces based on the WSN observed data,which is aiming at narrowing down the uncertainty of results caused by the spatial scaling mismatch between satellite and ground measurements over heterogeneous surfaces.(1)Forward a spatio-temporal trend surface for long-term ground-measured albedo upscaling over heterogeneous land surface.The construction of the spatiotemporal trend surface integrated the spatial information provided by auxiliary remote sensing image and the temporal information provided by long time series ground observations.It used very limited high-resolution image and ground measurements time series for albedo spatial-temporal upscaling,and thus free from the effect of the spatial scale difference and the limited time frame of high-resolution ancillary image.The result showed that it can depict the spatial-temporal variation of surface albedo well and thereby enhancing the quality of validation over heterogeneous surfaces.(2)Validation strategy based on multiple measurements over heterogeneous surfaces.The performance of albedo validation depends on the quality of reference value at a corresponding coarse-pixel scale.However,the reference values based on high-resolution ancillary image suffer from uncertainties due to error propagation.By extracting the spatial-temporal characteristic of surface albedo from historical prior knowledge,a method of weighting is proposed to directly upscale WSN node albedo to a coarse-pixel scale.The results demonstrated the effectiveness of the upscaling method as well as the high accuracy of reference value at coarse pixel scale.The high-resolution albedo product is not a requirement to implement this albedo validation method,and therefore minimizes the uncertainty caused by the additional inversion errors and geometric registration processes.(3)The WSN observation constitutes a promising method for collecting land surface albedos for validating remote sensing albedo products,especially at coarsepixel scale.Due to the high cost of installation and maintenance of permanent networks,the method to minimize the number of measuring points without significant loss of information to characterize the behavior of surface albedo at coarse scale is required.An optimal sampling method based on spatial-temporal stability analysis and random combination was developed to use less WSN observations to provide the coarse scale ground albedo over heterogeneous land surfaces.This method has been proved effective and can guide the design of sparse validation networks in future.It is very complicated to perform validation over heterogeneous surfaces,which involves ground optimal sampling to capture the spatio-temporal variation of land surface parameters,upscaling transform of ground measurements from fine scale to coarse pixel scale and the assessment method.This paper makes some elementary study about the validation over heterogeneous surface.We will devote to further improve and develop new methods that are better able to account for the spatio-temporal characteristic of surface albedo in the future.Additionally,how to use the widely distributed single site measurements over the heterogeneous surfaces is also a question needed to be answered.