Kasvukauden alku / Start of vegetation period

Creative Commons Nimeä 4.0 Kansainvälinen http://www.syke.fi/fi-FI/Avoin_tieto/Kayttolupa_ja_vastuut

no limitations

Moderate Resolution Imaging Spectrometer (MODIS) Terra Level 1B (1 km, 500 m), version 5 (V005) data were manually selected from the Level-1 and Atmosphere Archive and Distribution System (LAADS DAAC) for the period 2001-2008. For years 2009 - 2014 data were obtained from the satellite receiving station of the Finnish Meteorological Institute (FMI) in Sodankylä, Finland and gap-filled with data from LAADS DAAC. For years 2015 onwards version 6 (V006) data were obtained from LAADS DAAC. Terra Level 1B version 6.1 (V061) was used starting from 2022 onwards.

MODIS Level 1B data were calibrated to top-of-atmosphere reflectances and projected to a geographic latitude/longitude grid (datum WGS-84) using the software envimon by Technical Research Centre of Finland (VTT).

The Normalized Difference Water Index (NDWI, Gao et al.1996) was calculated from near-infrared (MODIS band 2) and mid-infrared (MODIS band 6) reflectance. Fractional Snow Cover (FSC) [0 1] was determined using the method by Metsämäki et al. (2012). FSC product specification are detailed in Böttcher et al. (2017). Cloud covered observations were removed using an automatic cloud masking algorithm by the Finnish Environment Institute.

Daily NDWI time series were averaged and smoothed for MODIS pixels with vegetation cover at a spatial grid resolution of 0.05 x 0.05 degrees. The day of the start of vegetation active period (VAP) was determined from NDWI time series applying the method by Delbart et al.(2005) and further described in Böttcher et al. (2016).

For the extraction of the start of vegetation active period (VAP), FSC was averaged at a spatial resolution of 0.05 x 0.05 degrees for MODIS pixels with dominant coverage of coniferous forest according to the CORINE Land Cover classification. A sigmoid function was fitted to the averaged FSC-time series and the start of the VAP was determined when FSC equals 0.99.

The start of the VAP in coniferous forest was compared to the start of VAP (the day when the Gross Primary Production (GPP) exceeded 15% of its summer maximum) from ground observations at 3 eddy covariance measurement sites in Finland. Satellite observations were highly correlated (R2=0.7) with the ground observations of VAP and the accuracy was 9 days for the period 2001-2016. The satellite product was in average 3 days late compared to the ground observations. The accuracy was higher (6 days, R2=0.84) and no bias was observed in pine forest compared to spruce forest that showed larger deviations to ground observations.

Since, the product is based on a proxy indicator for the start of VAP from optical satellite time series of FSC and it is therefore not a direct observation of the beginning of photosynthesis in trees. For years with very few snow, the direct effect of low temperature and high light are more relevant for the restriction of photosynthetic recovery in spring and therefore the method based on FSC may fail.

A main source of uncertainty in the detection method is the frequent cloud cover. The temporal resolution of the FSC time series is one day, but missing observations due to cloud cover led to gaps. Especially long temporal gaps (up to 1 month) were observed in northern Finland for year 2002, therefore the area of northern Finland was masked for this year. FSC time series with gaps longer than two weeks during the snow melt were discarded.

The vegetation active period in deciduous vegetation was compared to the visual observation of the bud break of birch from the phenological network of the Natural Resource Institute of Finland (Luke). Satellite observations were highly correlated with the ground observations of birch bud break. The accuracy was 7 days for the period 2001-2015 based on 84 site-years. The bias was negligible (0.4 days).

The product is based on the NDWI, which is sensitive to surface water and the water content of vegetation. Standing water e.g. in wetland areas can therefore influence the detection of the vegetation greening. The satellite product has not been validated over wetland areas.

A main source of uncertainty in the detection method is frequent cloud cover. The temporal resolution of the NDWI time series is one day, but missing observations due to cloud cover lead to gaps. For years 2001–2012 for the period April-July, depending on the location, in average 34%-73% of daily observations were missing. NDWI time series with gaps longer than two weeks during the spring vegetation development were discarded.

Reference:

Böttcher, K., Kervinen, M., Keto, V., Luojus, K., Manninen, T., Metsämäki, S., & Rautiainen, K. (2017). Report on EO products and comparison with in situ data, Monimet project deliverable, http://monimet.fmi.fi/index.php?style=warm&page=Deliverables&p=2 (accessed 14.10.2025).

Böttcher, K., Aurela, M., Kervinen, M., Markkanen, T., Mattila, O.-P., Kolari, P., Metsämäki, S., Aalto, T., Arslan, A.N., & Pulliainen, J. (2014). MODIS time-series-derived indicators for the beginning of the growing season in boreal coniferous forest — A comparison with CO2 flux measurements and phenological observations in Finland. Remote Sensing of Environment, 140, 625-638. http://dx.doi.org/10.1016/j.rse.2013.09.022

Böttcher, K., Markkanen, T., Thum, T., Aalto, T., Aurela, M., Reick, C., Kolari, P., Arslan, A., & Pulliainen, J. (2016). Evaluating biosphere model estimates of the start of the vegetation active season in boreal forests by satellite observations. Remote Sensing, 8, 580. http://dx.doi.org/10.3390/rs8070580

Delbart, N., Kergoat, L., Le Toan, T., L'Hermitte, J., Picard, G., 2005. Determination of phenological dates in boreal regions using normalized difference water index. Remote Sensing of Environment 97, 26-38.

Gao, B.-C., 1996. NDWI-A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sensing of Environment 58, 257-266.

Metsämäki, S., Mattila, O.-P., Pulliainen, J., Niemi, K., Luojus, K., & Böttcher, K. (2012). An optical reflectance model-based method for fractional snow cover mapping applicable to continental scale. Remote Sensing of Environment, 123, 508-521.