An approach for evaluating the impact of gaps and measurement errors on satellite land surface phenology algorithms: Application to 20year NOAA AVHRR data over Canada

Kandasamy, Sivasathivel; Fernandes, Richard

doi:10.1016/j.rse.2015.04.014

Cited by 27 publications

(13 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cong, et al [33] argued that the reason for not finding a single best method could be different definitions of phenology parameters. Kandasamy and Fernandes [34] highlighted that the performance of different smoothing methods depends on both land surface condition and the clear sky identification approach adopted.…”

Section: Introductionmentioning

confidence: 99%

Performance of Smoothing Methods for Reconstructing NDVI Time-Series and Estimating Vegetation Phenology from MODIS Data

et al. 2017

View full text Add to dashboard Cite

Many time-series smoothing methods can be used for reducing noise and extracting plant phenological parameters from remotely-sensed data, but there is still no conclusive evidence in favor of one method over others. Here we use moderate-resolution imaging spectroradiometer (MODIS) derived normalized difference vegetation index (NDVI) to investigate five smoothing methods: Savitzky-Golay fitting (SG), locally weighted regression scatterplot smoothing (LO), spline smoothing (SP), asymmetric Gaussian function fitting (AG), and double logistic function fitting (DL). We use ground tower measured NDVI (10 sites) and gross primary productivity (GPP, 4 sites) to evaluate the smoothed satellite-derived NDVI time-series, and elevation data to evaluate phenology parameters derived from smoothed NDVI. The results indicate that all smoothing methods can reduce noise and improve signal quality, but that no single method always performs better than others. Overall, the local filtering methods (SG and LO) can generate very accurate results if smoothing parameters are optimally calibrated. If local calibration cannot be performed, cross validation is a way to automatically determine the smoothing parameter. However, this method may in some cases generate poor fits, and when calibration is not possible the function fitting methods (AG and DL) provide the most robust description of the seasonal dynamics.

show abstract

Section: Introductionmentioning

confidence: 99%

Performance of Smoothing Methods for Reconstructing NDVI Time-Series and Estimating Vegetation Phenology from MODIS Data

et al. 2017

View full text Add to dashboard Cite

show abstract

“…We adopt the strategy that uses high quality time series to generate validation dataset by artificially creating data gaps [19,20,28]. This is a widely used strategy to quantify the performance of time series reconstruction algorithms.…”

Section: Algorithm Validation and Comparisonmentioning

confidence: 99%

“…These algorithms have shown their efficiencies for smoothing time series in diverse ecosystems [17][18][19][20][21][22]. However, they may not perform well for time series with less high-quality observations.…”

Section: Introductionmentioning

confidence: 99%

Temporal Interpolation of Satellite-Derived Leaf Area Index Time Series by Introducing Spatial-Temporal Constraints for Heterogeneous Grasslands

2017

View full text Add to dashboard Cite

Continuous satellite-derived leaf area index (LAI) time series are critical for modeling land surface process. In this study, we present an interpolation algorithm to predict the missing data in LAI time series for ecosystems with high within-ecosystem heterogeneity, particularly heterogeneous grasslands. The algorithm is based on spatial-temporal constraints, i.e., the missing data in the LAI time series of a pixel are predicted by the phenological links with other pixels. To address the uncertainties in the construction and selection of reference curves in a heterogeneous landscape, the algorithm constructs a reference dataset for each missing data in the LAI time series from all pixels showing very strong linear phenological links with the target pixel within a region. We also use an iterative process to update the available spatial-temporal constraints. We tested the algorithm with an eight-day composite Moderate Resolution Imaging Spectroradiometer (MODIS) LAI product in the Songnen grasslands, Northeast China in 2010 and 2011. The validation dataset was generated based on high quality time series by artificially adding data gaps. The algorithm achieved high overall interpolation accuracies with high coefficient of determination R 2 (>0.9) and low root mean square error (RMSE) (<0.2) in both dry (2010) and wet (2011) years. The algorithm showed advantages in predicting missing data for different seasons and proportions of missing data versus the algorithm that uses regional average LAI curve as a reference. These results suggest that the proposed algorithm could more effectively characterize spatial-temporal constraint information in heterogeneous grasslands for temporal interpolation.

show abstract

“…Satellite time series are usually noisy and discontinuous, which can lead to large uncertainties in the estimation of phenological metrics [21]. To reduce such uncertainties, these data require smoothing and gap-filling before being used in phenological studies.…”

Section: The Savitzky-golay Filtermentioning

confidence: 99%

“…Different approaches have been identified to derive these metrics in a robust fashion [9][10][11][14][15][16][17][18]. Recent studies have also demonstrated that the uncertainty in the estimation of phenological metrics has a direct relation to the cloud contamination and the quality of satellite data [19][20][21]. However, the quality and research applicability of the Global Inventory Monitoring and Modeling Studies Leaf Area Index third-generation (GIMMS LAI3g) dataset were effectively evaluated and widely used in numerous research studies [22][23][24][25], which indicated that LAI3g data could be effectively used for deriving phenology.…”

Section: Introductionmentioning

confidence: 99%

Evaluating Vegetation Growing Season Changes in Northeastern China by Using GIMMS LAI3g Data

Xie

Zhou

et al. 2017

Climate

View full text Add to dashboard Cite

Accurate understanding and detecting of vegetation growth change is essential for providing suitable management strategies for ecosystems. Several studies using satellite based vegetation indices have demonstrated changes of vegetation growth and phenology. Temperature is considered a major determinant of vegetation phenology. To accurately detect the response of vegetation to climate variations, this study investigated the vegetation phenology in the northeast (NE) region of China by using in-situ temperature observations and satellite-based leaf area index estimates (LAI3g) for the period 1982-2011. Firstly, a spatial distribution of the averaged phenology over the 30 years was obtained. This distribution showed that a tendency for an early start of the growing season (SoS) and late end of the growing season (EoS) was observed towards of the southeastern part of NE China, with the late SoS and early EoS occurring at higher latitudes. Secondly, the temperature-based and satellite-based phenological trends were analyzed. Then the significant advanced trend (SAT), significant delayed trend (SDT), and nonsignificant trend (NT) of SOS and EOS in NE region of China were detected by using the Mann-Kendall trend test approach. Finally, changes in phenological trends were investigated by using the temperature-based and satellite-based phenology method. A comparison of the phenological trend shows that there are some significant advanced trends of SOS and significant delayed trends of EOS in the NE region of China over 30 years. The results of this study can provide important support of the view that a lengthening of growing season duration occurred at the northern high latitudes in recent decades.

show abstract

An approach for evaluating the impact of gaps and measurement errors on satellite land surface phenology algorithms: Application to 20year NOAA AVHRR data over Canada

Cited by 27 publications

References 59 publications

Performance of Smoothing Methods for Reconstructing NDVI Time-Series and Estimating Vegetation Phenology from MODIS Data

Performance of Smoothing Methods for Reconstructing NDVI Time-Series and Estimating Vegetation Phenology from MODIS Data

Temporal Interpolation of Satellite-Derived Leaf Area Index Time Series by Introducing Spatial-Temporal Constraints for Heterogeneous Grasslands

Evaluating Vegetation Growing Season Changes in Northeastern China by Using GIMMS LAI3g Data

Contact Info

Product

Resources

About