The PPI time-series is affected by noise due to e.g. atmosphere and remaining cloud influence, resulting in some spikes and outlier values. Since large spikes and outliers might significantly affect the further function fitting, they first have to be removed from the data. This is done in an initial filtering process, further described in the TIMESAT software manual.

After the outlier removal the next step in the analysis is the determination of the number of growing seasons. This is based on a harmonic function fit (sine-cosine functions) to the data. The presence of a second season is established by evaluating the amplitudes of the first and second components of the harmonic fit. Presence of noise in the data complicates the decision on whether the given secondary maximum represents a true growing season or not. Therefore, an amplitude threshold is used to remove seasons that are smaller than the given threshold. A detailed description of the determination of the number of growing season is found in the TIMESAT software manual.

After the number of growing seasons have been determined double logistic functions are fitted to the data from each pixel. This is done to generate smooth continuous functions that well describe each individual growing season. It is assumed that most of the noise included in PPI (or any other vegetation index) results in negative bias of the values. Therefore, iterative adaptation of the logistic functions to the upper envelope of the data is applied in the following step. The function fit is performed on the PPI data. Values less than the first function fit are then considered as influenced by noise and thus less important, so their weights are decreased for the next iteration of the function fitting.

Phenological metrics (and other parameters describing character of the given growing season) are finally extracted from the fitted function data. The following parameters are extracted for each detected growing season to determine productivity:

Start-Of-Season (SOS): date of the start of the season defined as the date when the PPI has increased to the 20% level of the average annual PPI amplitude (Jin et al. 2017). The average annual PPI amplitude is the difference between the average peak level and the average base level for each pixel.

End-Of-Season (EOS): date of the end of the season defined as the date when the PPI drops under the 20 % level of the average annual PPI amplitude (Jin et al. 2017).

Large integral: integral of the fitted function between the start and end of the season.

Small integral: integral of the differences between the fitted function and the base level from start to end of the season.

Seasonal amplitude is calculated as a difference of the fitted curve maximum and the base level. The SOS and EOS points on the curve are then given as the fraction of the amplitude, i.e. the date when the fitted curve reaches/drops below the defined percent fraction of the seasonal amplitude. For this indicator 20% of the seasonal PPI amplitude was used as the SOS and EOS detection threshold.

The output of the process is a productivity metrics for each year of the time series 2000-2016 (17 years) covering the EEA39 territory. The spatial resolution of the productivity dataset is 500mx500m pixel size.

Detailed description of the methodology for calculating the productivity metric can be found in the TIMESAT software manual (publically available)

References:

Jönsson P., Eklundh L., 2004. TIMESAT—a program for analyzing time-series of satellite sensor data. Computers & Geosciences 30 (2004) 833–845.

Eklundh L., Jönsson P., 2015. TIMESAT: A Software Package for Time-Series Processing and Assessment of Vegetation Dynamics. In: Kuenzer C., Dech S., Wagner W. (eds) Remote Sensing Time Series. Remote Sensing and Digital Image Processing, vol 22. Springer, Chambridge.

Jin, H., Eklundh, L. 2014. A physically based vegetation index for improved monitoring of plant phenology, Remote Sensing of Environment, 152, 512 – 525.

Karkauskaite, P., Tagesson, T., Fensholt, R., 2017. Evaluation of the Plant Phenology Index (PPI), NDVI and EVI for Start-of-Season Trend Analysis of the Northern Hemisphere Boreal Zone, Remote Sensing, 9 (485), 21 pp.

Jin, H.X.; Jönsson, A.M.; Bolmgren, K.; Langvall, O.; Eklundh, L., 2017. Disentangling remotely-sensed plant phenology and snow seasonality at northern Europe using MODIS and the plant phenology index. Remote Sensing of Environment 2017,198, 203-212.

Abdi, A. M., N. Boke-Olén, H. Jin, L. Eklundh, T. Tagesson, V. Lehsten and J. Ardö (2019). First assessment of the plant phenology index (PPI) for estimating gross primary productivity in African semi-arid ecosystems. International Journal of Applied Earth Observation and Geoinformation 78: 249-260.

Jin, H., A. M. Jönsson, C. Olsson, J. Lindström, P. Jönsson and L. Eklundh (2019). New satellite-based estimates show significant trends in spring phenology and complex sensitivities to temperature and precipitation at northern European latitudes. International Journal of Biometeorology 63(6): 763-775.