Optimizing and comparing gap-filling techniques using simulated NDVI time series from remotely sensed global data

Abstract NDVI (Normalized Difference Vegetation Index) time series usually suffer from remaining cloud presence, even after data pre-processing. To address this issue, numerous gap-filling (or reconstruction) techniques have been developed in the literature, although their comparison has mainly been local to regional, with only two global studies to date, and has led to sometimes contradictory results. This study builds on these different comparisons, by testing different parameterizations for five NDVI temporal profile reconstruction techniques, namely HANTS (Harmonic Analysis of Time Series), IDR (iterative Interpolation for Data Reconstruction), Savitzky-Golay, Asymmetric Gaussian and Double Logistic, and then comparing them as generally parameterized, and then with the best of the tested parameterizations. These comparisons show that the HANTS reconstruction technique provides lower errors in cloud prone areas, while the IDR method works best with shorter cloud covers. However, the remaining errors in cloud prone areas are still high, and there is room for new reconstruction techniques. Although these results are only applicable to the range of the tested parameterizations, these latter have been chosen within widely used configurations, and should provide interested users with a better understanding of the different methods and the best parameterization for their needs, as well as an estimate of the expected error in the reconstruction of NDVI time series.

[1]  Jin Chen,et al.  A simple method for reconstructing a high-quality NDVI time-series data set based on the Savitzky-Golay filter , 2004 .

[2]  Gang Yang,et al.  Missing Information Reconstruction of Remote Sensing Data: A Technical Review , 2015, IEEE Geoscience and Remote Sensing Magazine.

[3]  Frank Veroustraete,et al.  Extending the SPOT-VEGETATION NDVI Time Series (1998–2006) Back in Time With NOAA-AVHRR Data (1985–1998) for Southern Africa , 2008, IEEE Transactions on Geoscience and Remote Sensing.

[4]  José A. Sobrino,et al.  Comparison of cloud-reconstruction methods for time series of composite NDVI data , 2010 .

[5]  Zhenyu Jin,et al.  Empirical comparison of noise reduction techniques for NDVI time-series based on a new measure , 2014 .

[6]  Jing M. Chen,et al.  Locally adjusted cubic-spline capping for reconstructing seasonal trajectories of a satellite-derived surface parameter , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Mingguo Ma,et al.  Comparison of Eight Techniques for Reconstructing Multi-Satellite Sensor Time-Series NDVI Data Sets in the Heihe River Basin, China , 2014, Remote. Sens..

[8]  Chao-Hung Lin,et al.  Patch-Based Information Reconstruction of Cloud-Contaminated Multitemporal Images , 2014, IEEE Transactions on Geoscience and Remote Sensing.

[9]  Qiang Liu,et al.  Reconstruction of Satellite-Retrieved Land-Surface Reflectance Based on Temporally-Continuous Vegetation Indices , 2015, Remote. Sens..

[10]  Edwin W. Pak,et al.  An extended AVHRR 8‐km NDVI dataset compatible with MODIS and SPOT vegetation NDVI data , 2005 .

[11]  F. Veroustraete,et al.  Reconstructing pathfinder AVHRR land NDVI time-series data for the Northwest of China , 2006 .

[12]  Ben Gorte,et al.  On the performance of remote sensing time series reconstruction methods – A spatial comparison , 2016 .

[13]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[14]  G. Schwarz.  Estimating the Dimension of a Model , 1978 .

[15]  Frédéric Baret,et al.  Evaluation of the representativeness of networks of sites for the global validation and intercomparison of land biophysical products: proposition of the CEOS-BELMANIP , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[16]  Damien Garcia,et al.  Robust smoothing of gridded data in one and higher dimensions with missing values , 2010, Comput. Stat. Data Anal..

[17]  Tao Zhang,et al.  A Temporal-Spatial Iteration Method to Reconstruct NDVI Time Series Datasets , 2015, Remote. Sens..

[18]  David P. Roy,et al.  Generating a long-term land data record from the AVHRR and MODIS Instruments , 2007, 2007 IEEE International Geoscience and Remote Sensing Symposium.

[19]  Iain Brown,et al.  Spatio-temporal MODIS EVI gap filling under cloud cover: An example in Scotland , 2012 .

[20]  Mathew R. Schwaller,et al.  On the blending of the Landsat and MODIS surface reflectance: predicting daily Landsat surface reflectance , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[21]  Myoung-Seok Suh,et al.  Detection of contaminated pixels based on the short-term continuity of NDVI and correction using spatio-temporal continuity , 2013, Asia-Pacific Journal of Atmospheric Sciences.

[22]  B. Holben.  Characteristics of maximum-value composite images from temporal AVHRR data , 1986 .

[23]  Feng Gao,et al.  A simple and effective method for filling gaps in Landsat ETM+ SLC-off images , 2011 .

[24]  Jennifer N. Hird,et al.  Noise reduction of NDVI time series: An empirical comparison of selected techniques , 2009 .

[25]  A. Belward,et al.  The Best Index Slope Extraction ( BISE): A method for reducing noise in NDVI time-series , 1992 .

[26]  Rasmus Fensholt,et al.  An ESTARFM Fusion Framework for the Generation of Large-Scale Time Series in Cloud-Prone and Heterogeneous Landscapes , 2016, Remote. Sens..

[27]  José A. Sobrino,et al.  TISSBERT: A benchmark for the validation and comparison of NDVI time series reconstruction methods , 2018, Revista de Teledetección.

[28]  María Amparo Gilabert,et al.  Noise Reduction and Gap Filling of fAPAR Time Series Using an Adapted Local Regression Filter , 2014, Remote. Sens..

[29]  P. Atkinson,et al.  Inter-comparison of four models for smoothing satellite sensor time-series data to estimate vegetation phenology , 2012 .

[30]  Per Jönsson,et al.  TIMESAT - a program for analyzing time-series of satellite sensor data , 2004, Comput. Geosci..

[31]  P. Eilers.  A perfect smoother. , 2003, Analytical chemistry.

[32]  A. Dijk,et al.  Smoothing vegetation index profiles: an alternative method for reducing radiometric disturbance in NOAA/AVHRR data , 1987 .

[33]  W. Verhoef,et al.  Reconstructing cloudfree NDVI composites using Fourier analysis of time series , 2000 .

[34]  M. Schaepman,et al.  Intercomparison, interpretation, and assessment of spring phenology in North America estimated from remote sensing for 1982–2006 , 2009 .

[35]  Yang Liu,et al.  Global evaluation of gap-filling approaches for seasonal NDVI with considering vegetation growth trajectory, protection of key point, noise resistance and curve stability , 2017 .

[36]  Gang Yang,et al.  A Moving Weighted Harmonic Analysis Method for Reconstructing High-Quality SPOT VEGETATION NDVI Time-Series Data , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[37]  P. Beck,et al.  Improved monitoring of vegetation dynamics at very high latitudes: A new method using MODIS NDVI , 2006 .

[38]  P. Eilers,et al.  Evaluating the effectiveness of smoothing algorithms in the absence of ground reference measurements , 2011 .

[39]  Peter M. Atkinson,et al.  An effective approach for gap-filling continental scale remotely sensed time-series , 2014, ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing.

[40]  Massimo Menenti,et al.  Reconstruction of global MODIS NDVI time series: performance of harmonic analysis of time series (HANTS). , 2015 .