Including land cover change in analysis of greenness trends using all available Landsat 5, 7, and 8 images: A case study from Guangzhou, China (2000–2014)

Abstract

Yingchun Fu); zhezhu@usgs.gov (Zhe Zhu). ABSTRACT:Remote sensing has proven a useful way of evaluating long-term trends in vegetation -greenness‖ through the use of vegetation indices like Normalized Differences Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI). In particular, analyses of greenness trends have been performed for large areas (continents, for example) in an attempt to understand vegetation response to climate. These studies have been most often used coarse resolution sensors like Moderate Resolution Image Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR). However, trends in greenness are also important at more local scales, particularly in and around cities as vegetation offers a variety of valuable ecosystem services ranging from minimizing air pollution to mitigating urban heat island effects.To explore the ability to monitor greenness trends in and around cities, this paper presents a new way for analyzing greenness trends based on all available Landsat 5, 7, and 8 images and applies it to Guangzhou, China. This method is capable of including the effects of land cover change in the evaluation of greenness trends by separating the effects of abrupt and gradual changes, and providing information on the timing of greenness trends.An assessment of the consistency of surface reflectance from Landsat 8 with past Landsat sensors indicates biases in the visible bands of Landsat 8, especially the blue band. Landsat 8 NDVI values were found to have a larger bias than the EVI values; therefore, EVI was used in the analysis of greenness trends for Guangzhou. In spite of massive amounts of development in Guangzhou from 2000 to 2014, greenness was found to increase, mostly as a result of gradual change. Comparison of the greening magnitudes estimated from the approach presented here and a Simple Linear Trend (SLT) method indicated large differences for certain time intervals as the SLT method does not include consideration for abrupt land cover changes. Overall, this analysis demonstrates the importance of considering land cover change when analyzing trends in greenness from satellite time series in areas where land cover change is common.

Human Activity Influences on Vegetation Cover Changes in Beijing, China, from 2000 to 2015

“…These images were processed as level 1T (precision and terrain corrected) products, so we completed pre-processing, including radiometric calibration, FLAASH atmospheric correction, and ortho-rectification. Previous studies have suggested that Landsat-8 OLI data show biased values, especially in bands RED and NIR, in comparison with TM data [29,30]; therefore, we completed histogram matching between OLI and TM to make the radiation values of the multi-source images comparable. We also used the same atmospheric correction method for all Landsat data, as well as using the same selectable calibration parameters, which might help alleviate the bias [30].…”

Leveraging Multi-Sensor Time Series Datasets to Map Short- and Long-Term Tropical Forest Disturbances in the Colombian Andes

“…Although other vegetation indices have shown less sensitivity to saturation issues in tropical forests, recent studies have demonstrated that NDVI is less affected by sun-sensor geometry variation [20,52] as well as variation in the frequency distribution of observations [29], both of which are relevant for time series analysis. The cross calibration between Thematic Mapper (TM), and Enhanced Thematic Mapper Plus (ETM+) sensors supports a continuity of cross-sensor NDVI trends during the study period [50,53,54]. However, TM/ETM+ and Landsat 8 Operational Land Imager (OLI) do not offer such consistency.…”

“…In total, this amounted to 374 Landsat 5, 7 and 8 images from 1996-2015; images with over 50% cloud cover were removed, leaving 149 Landsat images for the analysis (Table S1, Figure S1). Cloud masks produced using Landsat Ecosystem Disturbance Adaptive Processing System (LEDAPS), [48], Function of Mask (FMask) [49], and Landsat 8 Surface Reflectance (L8SR) algorithms [50,51] were applied, and NDVI was calculated for all images. Although other vegetation indices have shown less sensitivity to saturation issues in tropical forests, recent studies have demonstrated that NDVI is less affected by sun-sensor geometry variation [20,52] as well as variation in the frequency distribution of observations [29], both of which are relevant for time series analysis.…”

“…However, TM/ETM+ and Landsat 8 Operational Land Imager (OLI) do not offer such consistency. Zhu et al [50] found that OLI-derived NDVI values were positively biased with respect to ETM+ by about 0.04. Consequently, we transformed NDVI surface reflectance-corrected data using the methods described by Roy et al [55] to minimize OLI's positive bias and ensure NDVI consistency across Landsat sensors.…”

Evaluating the Sensitivity of Ecosystems to Soil Salinization in the Manas River Basin

“…The surface reflectance in OLI has a resolution of 30 × 30 m in seven multispectral bands. The image, acquired for 10 Sept. 2015 (a nearly cloudfree day) covered the entire study area [24]. The image was subjected to geometric correction and radiometric correction [25][26], and spliced and cut as required to represent the chosen study area.…”