Main content area

A biophysically based and objective satellite seasonality observation method for applications over the Arctic

Chen, Wenjun, Foy, Norah, Olthof, Ian, Zhang, Yu, Fraser, Robert, Latifovic, Rasim, Poitevin, Jean, Zorn, Paul, McLennan, Donald
International journal of remote sensing 2014 v.35 no.18 pp. 6742-6763
aerosols, algorithms, climate, growing season, monitoring, national parks, phenology, remote sensing, snowpack, vegetation index, Arctic region, Canada
Despite wide applications of remote-sensing data with high temporal resolution for monitoring phenology, two persistent problems have prevented the realization of their full potential. The first is the subjectivity in defining thresholds for a phenological event (e.g. the start or end of growing season − SOS or EOS). The second is the use of various arbitrarily selected filtering and smoothing algorithms for constructing vegetation index seasonal profiles in order to reduce the noise caused by residue cloud contamination and aerosol variations. In this study, we addressed both problems by developing a biophysically based and objective satellite seasonality observation method (BLOSSOM) for application over Canada’s Arctic. Application of the BLOSSOM method to three northern Canadian national parks (Ivvavik, Wapusk, and Sirmilik) proved that the method is operational. Using the uncertainties in the vegetation index and its threshold, we estimated the overall mean uncertainties as being −5.3 to 3.4 days, −4.2 to 5.2 days, and −6.2 to 8.4 days, respectively, for SOS, EOS, and growing season length (GSL). Further independent tests against SOS, determined using records of snow cover at nearby climate stations (as ‘truth’), indicate that the mean absolute error is less than 3.6 ± 0.2 days.