Faculty of Mathematics and Natural Sciences - Earth Observation Lab

News archive (before 2014)

published 21st Nov 2013

Two new publications by Patrick Griffiths and colleagues. The first paper, published in RSE,attempts to characterise forest dynamics over the entire Carpathians using a series of composited datasets. The paper addressed changes in forest types (looking into the fate of spruce monocultures), forest disturbances and post-disturbance recovery. The most important finding here, was that disturbance levels throughout the Carpathians dropped considerably after the early transition period and that forest composition shifted towards a more mixed-deciduous forests while conifer forests experienced increasing disturbance levels.

The most recent paper was published in ERL. Here the authors addressed changes in agricultural land use throughout the Carpathians. Compositing algorithms were adapted to create specific seasonal best observation composites along with seasonal metrics. The methodology however is only briefly described here... The key findings were that cropland abandonment until 2000 accounted for a third of socialist farmland, and that over the past decade this process has been reversed in many regions (i.e. cropland recultivation and also grassland conversion). Relating these processes to agricultural suitability (Global Agro-ecological Zones) confirmed assumptions that abandonment occurred on less productive land, while recultivation targeted highly productive lands.

published 26th Sep 2013

A new paper is published by Magdlalena Main-Knorn et.al. in Remote Sensing of Environment on  "Monitoring coniferous forest biomass change using a Landsat trajectory-based approach."

We used satellite data (Landsat, SPOT, and IRS LISS) to estimate changes in conifer aboveground biomass in the Western Carpathian Mountains. First, Random Forests (RF) were used to create near annual biomass maps over 25 years. Then, we applied the LandTrendr algorithm to detect and describe biomass trends between 1985 and 2010.

In this study, we showed the effectiveness of the trajectory-based change detection approach (LandTrendr) to map abrupt and gradual biomass changes. A near-annual time series provided detailed information on stand-replacing disturbances, long-term degradation, and forest regeneration; processes that strongly affect forest biomass.

Magdalena Main-Knorn has finished her PhD at the Geomatics Lab 2012 and is now working at the DLR Earth Observation Center
Remote Sensing Technology Institute (IMF)
Photogrammetry and Image Analysis

published 17th Sep 2013

New paper on "Analyzing the drivers of tree planting in Yunnan, China, with Baysian networks", by our guest scientist Dr. Daniel Müller et.al. in Land Use Policy. Read...

We use a Bayesian network to analyze the factors that drive the decision to plant trees in Yunnan, China, with both quantitative household survey data and qualitative insights from extensive field work. We could show how state forest policies were the main underlying driver to the forest transitionin the past, but private afforestation activities increasingly dominate the expansion of tree cover. Farmers use tree planting to cash in on the growing economic opportunities of tree crops and to adapt to growing labor scarcities induced by massive rural emigration.

published 16th Sep 2013

Patrick Griffiths published the second paper out of his PhD research at the Geomatics Lab. It describes the development of a compositing algorithm for Landsat data and assesses the output composites regarding seasonal/annual and radiometric consistency as well as the suitability for land cover mapping. A more detailed description of the paper can be found here.

published 5th Aug 2013

Pedro Leitão, postdoctoral researcher at the Geomatics Lab, published as co-author a commentary paper on "Modelling species distributions with remote sensing data: bridging disciplinary perspectives" in the Journal of Biogeography. Interested - Read the online early version.

published 23th Aug 2013

New paper published in Remote Sensing of Environment by Akpona Okujeni et al.: “Support vector regression and synthetically mixed training data for quantifying urban land cover”.

In this study, we used HyMap data and SVR to quantify four spectrally complex urban land cover types. To overcome the problem of deriving reliable training data for land cover mapping with regression techniques, we trained SVR models with synthetically mixed data generated from a spectral library. We evaluated our approach by comparing land cover fraction estimates to umixing results obtained from traditional MESMA. The high accuracies achieved demonstrate (i) the suitability of the proposed workflow for utilizing SVR for land cover mapping, and (ii) the strength of kernel-methods when analyzing heterogeneous environments using imaging spectrometer data.

published 16th Jul 2013

New paper on Import Vector Machines for Quantitative Analysis of Hyperspectral Data in IEEE Geoscience and Remote Sensing Letters published by Stefan Suess, a doctoral researcher of the Geomatics Lab of the Land System Science Cluster.

published 6th Mar 2013

New paper published in Remote sensing (Open Access): "Mapping Rubber Plantations and Natural Forests in Xishuangbanna (Southwest China) Using Multi-Spectral Phenological Metrics from MODIS Time Series" by Cornelius Senf et.al., a doctoral researcher of our lab.

Basically, we test whether phenological metrics derived from MODIS EVI and SWIR time series can be used to differentiate between rubber plantations and natural forests, and which metrics work out best. Results are promising, with only low differences in accuracy between phenological metrics and full time series data. Furthermore, we could highlight the important information captured in the SWIR reflectance and relate the phenological metrics to natural phenomena.

published 6th Mar 2013

Pedro Leitão, postdoctoral researcher at our lab, is co-author of a fully published paper on "Collinearity: a review of methods to deal with it and a simulation study evaluating their performance".

Appendices are published here.

published 4th Mar 2013

New paper on Effectiveness of protected areas in European Russia in Remote Sensing of Environment published by Anika Sieber, a doctocal researcher of the Land System Science Cluster.

published 12th Feb 2013

Satellite NASA Landsat Data Continuity Mission

Image courtesy of dream designs / FreeDigitalPhotos.net

Patrick Hostert is a member of the Landsat Science Team. He was on site, when the eighth Landsat Satellite was successfully launched on February 11th 2013 from Vandenberg Air Force Base, California (videos).

published 11th Feb 2013

New paper in Agricultural Systems (in press) by Daniel Müller and Pedro J. Leitão et.al., both researchers at our lab.

In essence, we find that market principles became increasingly critical in shaping agricultural landscapes. But the underlying causes of land-use change showed distinct differences between the two countries. In Albania, variables that adversely affected subsistence agricultural production often resulted in fundamental changes in livelihood strategies, frequently away from agriculture and toward migration. In Romania, the factors that impaired the profitability of agriculture were the crucial determinants of cropland abandonment. The BRTs proved useful for such analysis, because they combine high predictive performance with appealing options for interpreting results.

published 5th Feb 2013

Daniel Müller has successfully presented his habilitation. He is a guest scientist from the IAMO in our lab. On February 4th 2013 he successfully delivered his habilitation lecture on "Conceptualizing, Measuring and Mapping Land-Use Intensity".

published 1st Feb 2013

deadline for session proposals has expired on the 31st Jan 2013


Global Land Project 2nd Open Science Meeting, March 2014 Berlin, Germany

published 21st Nov 2012

Patrick Griffiths doctoral researcher of our Lab published a paper on: "A pixel-based Landsat compositing algorithm for large area land cover mapping" ►►►

published 7th Nov 2012

New paper in Remote Sensing by He Yin, doctoral researcher of the Lab:

In this paper we highlighted differences in trends from different NDVI archives in arid to sub-arid environments. We cautioned that AVHRR GIMMS and NDVI products from other sensors cannot be combined into homogeneous, long-term time series across the globe without regional sensitivity analyses. A more thorough understanding of the factors introducing uncertainties in AVHRR GIMMS time series is needed.

published 1st Nov 2012

IRI THESys project office initiated

published 26th Oct 2012

Patrick Hostert, Chair of the Geomatics Lab, was recently appointed as a member of the Landsat Science Team ►►►

published 19th Oct 2012

Future Earth 2012 - discussion Science Year 2012 "Project Earth: Our Future":  Presentation and discussion on the "Future of Land Use" at the Berlin Brandenburg Academy of Sciences