Uncertainty of Remote Sensing Data in Monitoring Vegetation Phenology: A Comparison of MODIS C5 and C6 Vegetation Index Products on the Tibetan Plateau

Zhoutao Zheng,Wenquan Zhu

doi:10.3390/rs9121288

Abstract

Vegetation phenology is considered a sensitive indicator of climate change, which controls carbon, nitrogen, and water cycles within terrestrial ecosystems. The Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) is an important moderate resolution remote sensing data for monitoring vegetation phenology. However, Terra MODIS Collection 5 (C5) vegetation index products were identified to be affected by sensor degradation, which has been addressed in the recently released MODIS Collection 6 (C6) vegetation index products. In order to compare the difference between MODIS C5 and C6 NDVI in monitoring vegetation phenology, the start and end of growing season (SOS and EOS) of the alpine grassland on the Tibetan Plateau (TP) were extracted using four common methods. Then, the C5 and C6 NDVI-derived SOS (SOSC5 and SOSC6) and EOS (EOSC5 and EOSC6) were compared with ground-observed phenology data. Results showed that the multi-year average growing season NDVIs of C6 were lower than those of C5 in most areas, while the inter-annual variation patterns of regional average SOSC5 and SOSC6 (EOSC5 and EOSC6) were consistent. However, large spatial differences in phenological trends were found between C5 and C6 NDVI products. From C5 to C6, pixels with a SOS (EOS) trend shifting from significant to insignificant or from insignificant to significant accounted for at least 14.58% (9.07%) of the total pixels. SOSC5 was more consistent than SOSC6 with the ground-observed green-up dates. C5 NDVI may be more appropriate for monitoring SOS than C6 NDVI in the study region, but more ground-observed phenology records are needed to confirm it due to only four observational sites in this study. However, large differences and poor correlations existed between EOSC5 (EOSC6) and the ground-observed beginning of leaf coloring. To further evaluate the uncertainty of MODIS C5 and C6 NDVI in monitoring vegetation phenology, higher resolution near-surface remote sensing data and corresponding validation methods should be applied.

Highlights

Vegetation phenology dynamics can reflect the response of terrestrial ecosystem to climate change and play an important role in adjusting the cycling of carbon, nitrogen, and water [1,2,3]
Due to the fact that Terra data is more affected by the sensor degradation than Aqua data [13,17], this study focused on the Terra Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) products
Collection 5 (C5) NDVI may be more appropriate for monitoring SOS than Collection 6 (C6) NDVI in this area, but more ground-observed phenology records are needed to confirm it due to only four available sites in our study

Summary

Introduction

Vegetation phenology dynamics can reflect the response of terrestrial ecosystem to climate change and play an important role in adjusting the cycling of carbon, nitrogen, and water [1,2,3]. Moderate Resolution Imaging Spectroradiometer (MODIS) remote sensing data have been increasingly used for monitoring vegetation phenology. Severe data problems were found to exist in MODIS Collection 5 (C5) ocean colors [10], aerosols [11], and NDVI products [12,13,14], mainly resulting from sensor degradation. When compared with C5, the C6 Level 1B data, including the top-of-atmosphere reflectance in the near-infrared, red and blue bands were calibrated [13]. Unlike C5 NDVI, which uses daily reflectance data, C6 NDVI uses pre-composed (8-day) surface reflectance data that are atmospherically corrected with a modified compositing algorithm that aims to reduce the aerosol issues (minimizing the blue band) [17,19]

Objectives

Methods

Findings

Discussion

Conclusion