Estimation of urban vegetation fraction from HJ-1B CCD data: Methods and comparison

Urban vegetation is an important contributor to the quality of urban environment and habitat situation. Researches on estimating urban vegetation fraction (UVF) accurately from remotely sensed imageries have been conducted massively and widely. However, in view of obtaining UVF and its spatio-temporal dynamics, shortcomings are still there in these widely used data, which are acquired by sensors, including Terra/Acqua MODIS, NOAA AVHRR, SPOT and Landsat series (e.g.1–5, 7). Therefore, it is desirable to exploit the usage of other kind of data, observed by the newly launched China satellites (e.g. HJ-1A/B, CBERS) for those issues. These data also are provided freely to the common users. In this paper, the possibility was evaluated that using HJ-1B CCD multispectral data for estimating UVF with optimal and practical methods. Two methodologies were discussed and compared, which specifically were based on vegetation index (VI) and spectral mixture analysis (SMA). Findings in this paper demonstrate that UVF was able to be obtained accurately through both the VI method (e.g. based on NDVI, SAVI, and EVI) and the SMA. Although the SMA performed superiorly to the VI method, results modeled from both methods were parallel with those obtained from Landsat TM/ETM+ (in previous practices). Our research suggests that these new satellites, including HJ-1A/B, can provide the useful data sources for monitoring urban environment, with both high spatial and temporal resolution.