Lars Eklundh

TIMESAT - a program for analyzing time-series of satellite sensor data

Three different least-squares methods for processing time-series of satellite sensor data are presented. The first method uses local polynomial functions and can be classified as an adaptive Savitzky-Golay filter. The other two methods are more clear cut least-squares methods, where data are fit to a basis of harmonic functions and asymmetric Gaussian functions, respectively. The methods incorporate qualitative information on cloud contamination from ancillary datasets. The resulting smooth curves are used for extracting seasonal parameters related to the growing seasons. The methods are implemented in a computer program, TIMESAT, and applied to NASA/NOAA Pathfinder AVHRR Land Normalized Difference Vegetation Index data over Africa, giving spatially coherent images of seasonal parameters such as beginnings and ends of growing seasons, seasonally integrated NDVI and seasonal amplitudes. Based on general principles, the TIMESAT program can be used also for other types of satellite-derived time-series data.

Seasonality extraction by function fitting to time-series of satellite sensor data

A new method for extracting seasonality information from time-series of satellite sensor data is presented. The method is based on nonlinear least squares fits of asymmetric Gaussian model functions to the time-series. The smooth model functions are then used for defining key seasonality parameters, such as the number of growing seasons, the beginning and end of the seasons, and the rates of growth and decline. The method is implemented in a computer program TIMESAT and tested on Advanced Very High Resolution Radiometer (AVHRR) normalized difference vegetation index (NDVI) data over Africa. Ancillary cloud data [clouds from AVHRR (CLAVR)] are used as estimates of the uncertainty levels of the data values. Being general in nature, the proposed method can be applied also to new types of satellite-derived time-series data.

Investigating relationships between Landsat ETM+ sensor data and leaf area index in a boreal conifer forest

The aim of this paper is to investigate the feasibility of using Landsat ETM+ data for the determination of leaf area index (LAI). The investigation is prompted by the need for obtaining spatially distributed data on LAI to be used as input for carbon modelling of northern boreal forests. Detailed field data have been collected in a coniferous forest area in Uppland, central Sweden, dominated by Norway spruce and Scots pine. A forest canopy reflectance model (Kuusk and Nilson, 2000) has been used to simulate stand reflectances in the Landsat ETM+ wavelength bands as a means of investigating the theoretical reflectance response to LAI changes. The analysis shows that the response to changes in LAI is strongest in the visible wavelength bands, particularly Channel 3, whereas only weak response is noted in the NIR band and for some vegetation indices [simple ratio (SR) and NDVI]. Modelled reflectances are influenced by various other factors, particularly ground reflectance and leaf biochemical properties. Observed reflectances from the Landsat ETM+ sensor have been compared with reflectance modelling results and with field-based LAI estimates. The results indicate that LAI estimation using inverse canopy reflectance modelling may be difficult, given the large number of input parameters required and the current level of uncertainty in these parameters. Statistical relationships between LAI and observed ETM+ reflectances are strongest in ETM+ Channel 7.

Estimating relations between AVHRR NDVI and rainfall in East Africa at 10-day and monthly time scales

Temporal relations between AVHRR NDVI and rainfall data over East Africa at 10-day and monthly time scales have been analysed using distributed lag models. On average, only 10 per cent of the variation in 10-day NDVI values could be explained by concurrent and preceding rainfall. Corresponding values for monthly data was 36 per cent. If it is assumed that rainfall data can be used as an indicator of vegetation development the study indicates that AVHRR NDVI may have limitations for temporal vegetation monitoring in these environments.

Fourier Series for analysis of temporal sequences of satellite sensor imagery

Fourier Series and the derivative were used in this study for analysing time series of remotely-sensed data. The technique allows fundamental characteristics of time series data to be quantified. In Fourier analysis a function in space or time is broken down into sinusoidal components, or harmonics. The first and second harmonics are a function of the mono or bi-modality of the curve, demonstrated in the study on Global Vegetation Index data classified into typical mono and bi-modal vegetation index zones. The last harmonic explains close to 100 per cent of the variance in the curve. Other important parameters of the time series, such as extreme points and rate of change, can be extracted from the derivative of the Fourier Series. Fourier Series may form a basis for a quantitative approach to the problem of handling temporal sequences of remotely-sensed data.

A Comparative analysis of standardised and unstandardised Principal Component Analysis in remote sensing

Abstract In this study Principal Components have been calculated using covariance and correlation matrices for Tour data sets: Monthly NOAA-NDVI maximum-value composites, NOAA-LAC data, Landsat-TM data, and SPOT multi-spectral data. An analysis of the results shows consistent improvements in the signal to noise ratio (SNR) using the correlation matrix in comparison to the covariance matrix in the principal components analysis for all the data sets

A ground-validated NDVI dataset for monitoring vegetation dynamics and mapping phenology in Fennoscandia and the Kola peninsula

An NDVI dataset covering Fennoscandia and the Kola peninsula was created for vegetation and climate studies, using Moderate Resolution Imaging Spectroradiometer 16‐day maximum value composite data from 2000 to 2005. To create the dataset, (1) the influence of the polar night and snow on the NDVI values was removed by replacing NDVI values in winter with a pixel‐specific NDVI value representing the NDVI outside the growing season when the pixel is free of snow; and (2) yearly NDVI time series were modelled for each pixel using a double logistic function defined by six parameters. Estimates of the onset of spring and the end of autumn were then mapped using the modelled dataset and compared with ground observations of the onset of leafing and the end of leaf fall in birch, respectively. Missing and poor‐quality data prevented estimates from being produced for all pixels in the study area. Applying a 5 km×5 km mean filter increased the number of modelled pixels without decreasing the accuracy of the predictions. The comparison shows good agreement between the modelled and observed dates (root mean square error = 12 days, n = 108 for spring; root mean square error = 10 days, n = 26, for autumn). Fennoscandia shows a range in the onset of spring of more than 2 months within a single year and locally the onset of spring varies with up to one month between years. The end of autumn varies by one and a half months across the region. While continued validation with ground data is needed, this new dataset facilitates the detailed monitoring of vegetation activity in Fennoscandia and the Kola peninsula.

Ground-Based Optical Measurements at European Flux Sites: A Review of Methods, Instruments and Current Controversies

This paper reviews the currently available optical sensors, their limitations and opportunities for deployment at Eddy Covariance (EC) sites in Europe. This review is based on the results obtained from an online survey designed and disseminated by the Co-cooperation in Science and Technology (COST) Action ESO903—“Spectral Sampling Tools for Vegetation Biophysical Parameters and Flux Measurements in Europe” that provided a complete view on spectral sampling activities carried out within the different research teams in European countries. The results have highlighted that a wide variety of optical sensors are in use at flux sites across Europe, and responses further demonstrated that users were not always fully aware of the key issues underpinning repeatability and the reproducibility of their spectral measurements. The key findings of this survey point towards the need for greater awareness of the need for standardisation and development of a common protocol of optical sampling at the European EC sites.

Investigating the use of Landsat thematic mapper data for estimation of forest leaf area index in southern Sweden

The study aims at investigating the use of Landsat thematic mapper (TM) for mapping leaf area index (LAI) in coniferous and deciduous forests in southern Sweden. LAI has been estimated in the field with optical measurements, allometric equations, and litter-trap data, and empirical relationships between LAI estimates and satellite-measured reflectances have been analysed. Several common vegetation indices and multiple regressions where estimated LAI is predicted as a function of various spectral bands are tested. The results indicate significant relationships between Landsat TM reflectances and parameters related to LAI, and the relationships are improved when separating coniferous and deciduous stands. The best relationships occur between Landsat TM data and the product of effective LAI as estimated with the LAI-2000 instrument and a needle clumping factor (LG), which explains about 80% of the variation in coniferous stands and about 50% of the variation in deciduous stands. The best single bands in coniferous stands are the middle-infrared bands (TM5 and TM7), and the best vegetation index is the moisture stress index (TM5/TM4). The best single band in deciduous stands is TM4, and the best vegetation index is the simple ratio (SR).

An Optical Sensor Network for Vegetation Phenology Monitoring and Satellite Data Calibration

We present a network of sites across Fennoscandia for optical sampling of vegetation properties relevant for phenology monitoring and satellite data calibration. The network currently consists of five sites, distributed along an N-S gradient through Sweden and Finland. Two sites are located in coniferous forests, one in a deciduous forest, and two on peatland. The instrumentation consists of dual-beam sensors measuring incoming and reflected red, green, NIR, and PAR fluxes at 10-min intervals, year-round. The sensors are mounted on separate masts or in flux towers in order to capture radiation reflected from within the flux footprint of current eddy covariance measurements. Our computations and model simulations demonstrate the validity of using off-nadir sampling, and we show the results from the first year of measurement. NDVI is computed and compared to that of the MODIS instrument on-board Aqua and Terra satellite platforms. PAR fluxes are partitioned into reflected and absorbed components for the ground and canopy. The measurements demonstrate that the instrumentation provides detailed information about the vegetation phenology and variations in reflectance due to snow cover variations and vegetation development. Valuable information about PAR absorption of ground and canopy is obtained that may be linked to vegetation productivity.