Earth Trends Modeler in TerrSet

The Earth Trends Modeler (ETM) is an integrated suite of tools within TerrSet for the analysis of image time series data associated with Earth Observation remotely sensed imagery. With Earth Trends Modeler, users can rapidly assess long term climate trends, measure seasonal trends in phenology, and decompose image time series to seek recurrent patterns in space and time.

TerrSet Brochure
  • Seasonal Trend

    The Earth Trends Modeler (ETM) is specially designed for the analysis of earth observation image time series. In this illustration, a series of 348 global images of monthly NDVI vegetation index data were analyzed for the presence of trends in seasonality. Pixels colored gray (which are almost absent) indicate a stable seasonality. All other colors represent trends. ETM provides an interactive legend (lower left) to interpret the trend for any area (the area in eastern Alabama and western Georgia in this case). The green curve represents the beginning of the series (1982) while the red one represents the end (2010). The X axis is time and the Y axis is NDVI. As can be seen, NDVI has generally increased with a growing bimodality. Spring is coming a bit earlier (11 days) and the autumn is extending longer (almost 30 days).

  • PCA Climate Change Analysis

    ETM can be used to analyze any kind of image time series. The most common formats are NetCDF and HDF, both of which are directly supported. However, many data are in more idiosyncratic formats. The analyses illustrated here use sea ice concentration data access for free from the U.S. National Snow and Ice Data Center (NSIDC). The documentation on the NSIDC web site indicates that the data are stored as unsigned byte values in flat binary files with a 300 byte header. This is easily imported using IDRISI’s flexible generic raster import routine. The images show part of a Principal Components Analysis as well as a non-linear trend analysis for the period from 1982-2008.

With ETM, users can address such questions as:

  • “How have temperatures changed over the past 30 years?”
  • “Are plants exhibiting a later senescence?”
  • “Are there recurrent spatial patterns in phytoplankton productivity?”
  • “What are the geographic impacts of climate events such as El Niño?”

No other software technology provides such a coordinated suite of data mining tools needed by the earth system science community for climate change analysis and impact assessment.

Tools in ETM include:

  • Parametric and non-parametric trend measures
  • Seasonal trend analysis
  • Principal Components / Empirical Orthogonal Function analysis (PCA/EOF)
  • Extended PCA/EOF for the co-analysis of multiple series (EPCA/EEOF)
  • Multichannel Singular Spectrum Analysis (MSSA)
  • Empirical Orthogonal Teleconnection (EOT) analysis and extended modes
  • Canonical Correlation Analysis (CCA)
  • Lagged Linear Modeling
  • Fourier PCA and Wavelet analysis

While ETM is intended for professional use, its simple and intuitive interface makes it an excellent tool for teaching and self-exploration. The system includes a full tutorial and sample data sets.

Earth Trends Modeler Key Features

  • Extract and analyze long-term global trends and their impacts
  • Examine the relationship between time series
  • Examine trends in seasonality
  • Isolate true change from normal environmental variability
  • Uncover and analyze patterns of variability across temporal scales
  • Preprocess image time series data including noise removal and deseasoning

Earth Trends Modeler Analytical Features

  • Animated 3-D display of space-time cubes
  • Dynamic lag correlation between index time series
  • Interactive Maximum Overlap Discrete Wavelet analysis
  • Trend analysis of index time series (linear trend, Theil-Sen median trend, polynomial (up to 9th order), moving average, Gaussian moving average, moving maximum)
  • Interactive temporal profiling
  • Image series trend analysis (linear trend, degree of linearity, Theil-Sen median trend, monotonic trend, Mann-Kendall trend significance)
  • Seasonal Trend Analysis (STA) including interactive interpretation and trend significance mapping
  • Principal Components (PCA) / Empirical Orthogonal Function (EOF) analysis (T-mode and S-mode, standardized/unstandardized and centered/uncentered)
  • Extended PCA/EOF (T-mode and S-mode, standardized/unstandardized and centered/uncentered)
  • Multichannel Singular Spectrum Analysis (T-mode and S-mode, standardized/unstandardized and centered/uncentered)
  • Empirical Orthogonal Teleconnection (EOT) analysis (S-mode, standardized/unstandardized and centered/uncentered)
  • Extended EOT (S-mode, standardized/unstandardized and centered/uncentered)
  • Cross-EOT (S-mode, standardized/unstandardized and centered/uncentered)
  • Multichannel EOT (MEOT) (S-mode, standardized/unstandardized and centered/uncentered)
  • Canonical Correlation Analysis (CCA) (T-mode and S-mode, standardized/unstandardized and centered/uncentered)
  • Fourier PCA (S-mode, unstandardized, centered)
  • Index to Image Series and Image to Image Series Multiple Linear Modeling (slope, intercept, R, R2, Adjusted R2, Partial R) at multiple lags plus residual series creation
  • Missing Data Interpolation (harmonic interpolation, spatial interpolation, linear temporal interpolation and climatology)
  • Denoising via temporal filtering (mean, Gaussian mean, maximum, cumulative sum, cumulative mean) with any arbitrary filter length
  • Denoising via Maximum Value Compositing (MVC).
  • Denoising via PCA (T-mode and S-mode)
  • Denoising via Inverse Fourier Analysis
  • Deseasoning (anomalies, standardized anomalies, temporal filter)
  • Series generation (linear, SIN, COS)
  • Series editing (lagging, truncation, extension, skip selection, temporal aggregation, spatial subsetting, automatic renaming)
  • Serial correlation analysis (Durbin-Watson)
  • Detrending (linear or difference series)
  • Cochrane-Orcutt transformation
  • Trend-preserving prewhitening (Wang and Swail procedure)

Learn more about Earth Trends Modeler