The Department of Remote Sensing recently applied for a time series of RapidEye scenes to support a PhD thesis by Christian Bauer (Dep. of Remote Sensing, Würzburg) in the context of the project LaVaCCA. The proposal has recently been accepted, and the PhD work of Christian Bauer will be supported by a time series of RapidEye scenes acquired over two irrigated landscapes, Khorezm and Elliqkala, in Uzbekistan.
This study addressed the classification of multi-temporal satellite data from RapidEye by considering different classifier algorithms and decision fusion. Four non-parametric classifier algorithms, decision tree (DT), random forest (RF), support vector machine (SVM), and multilayer perceptron (MLP), were applied to map crop types in various irrigated landscapes in Central Asia. A novel decision fusion strategy to combine the outputs of the classifiers was proposed. This approach is based on randomly selecting subsets of the input data-set and aggregating the probabilistic outputs of the base classifiers with another meta-classifier. During the decision fusion, the reliability of each base classifier algorithm was considered to exclude less reliable inputs at the class-basis. The spatial and temporal transferability of the classifiers was evaluated using data sets from four different agricultural landscapes with different spatial extents and from different years. A detailed accuracy assessment showed that none of the stand-alone classifiers was the single best performing. Despite the very good performance of the base classifiers, there was still up to 50% disagreement in the maps produce by the two single best classifiers, RF and SVM. The proposed fusion strategy, however, increased overall accuracies up to 6%. In addition, it was less sensitive to reduced training set sizes and produced more realistic land use maps with less speckle. The proposed fusion approach was better transferable to data sets from other years, i.e. resulted in higher accuracies for the investigated classes. The fusion approach is computationally efficient and appears well suited for mapping diverse crop categories based on sensors with a similar high repetition rate and spatial resolution like RapidEye, for instance the upcoming Sentinel-2 mission.
“This field has been abandoned approximately 30 years ago” – such statements can be heard quite often when one talks with farmers in Khorezm, one of the largest irrigated agricultural landscapes in Central Asia. In this region, vast irrigation systems were installed during Soviet times, but after breakdown in the 1990s many fields were finally abandoned for many reasons (e.g. soil salinization, insufficient water supply). However, the spatial extent and timing of cropland abandonment in many regions in Central Asia remain unknown.
The LaVaCCA project aims at shedding more light on these issues, which could allow land managers to make better informed decisions about their land, or to propose alternative land uses such as pasture or afforestation on formerly abandoned fields – once these will have been identified. Staffers from different institutions in Central Asia (SIC-ICWC, KRASS, Al´Farabi University) recently visited the test sites in Khorezm together, where important information is collected during the project runtime (2015-2017) for calibration and validation of diverse remote sensing based methodologies (e.g. yield modelling, crop classification, salinity mapping).
Three PhD students from Usbekistan, Kazakhstan, and Germany investigate the recent and past status of agricultural areas in the lowlands of Central Asia, including Khorezm. They seek to back-trace the past development in order to better understand the various pathways that led to the widespread cropland abdandonment, which was already observed (see recent publication on this issue). Their overall aim is to make use of remote sensing for identifying abandoned fields and explaining their spatio-temporal pattern in order to gain more insight into the drivers of cropland abandonment in Central Asia.
From June 29th to July 2rd, experts from the Department of Remote Sensing, Würzburg University, met their colleagues of the LaVaCCA project from KRASS, SIC-ICWC, and Al´Farabi University at the office of KRASS in Urgench, Uzbekistan. During this meeting, the second R-course was done in the scope of the capacity building of the LaVaCCA Project. This course aimed at fostering the methodological skills of the participants in using R for geospatial data analysis, with a focus on agricultural monitoring. Participants have already gained first analytical skills in using R during the first training in Almaty, Kazakhstan in February 2015.
A new concept for this course was applied, in which the participants selected and suggested the course agenda, according to their specific project requirements. The course was then set up accordingly and held by Dr. Fabian Löw and Christian Bauer from Würzburg University.
Three major topics were focused on:
- Working with raster data: satellite image pre-processing in R (atmospherically calibration, stacking, sub-setting of large data sets, calculation of vegetation indices for assessing vegetation dynamics and within-field assessments of vegetation growth)
- Supervised image classification for crop type mapping, based on multi-temporal Landsat images and Random Forest algorithm (crop type distribution and acreage), accuracy assessments (confusion matrices)
- Regression analysis in R
After three intensive days of training, the participants togehther explored the subject of the LaVaCCA program “on-ground”: a field survey was organized and performed around Urgench city, where staffers from SIC-ICWC and KRASS maintain several test sites on cotton, wheat, and rice fields, which are the major crops in this region. On these selected fields, SIC-ICWC staffers assess biophysical parameters (fPAR, biomass), which will later be incorporated into yield models. The R course enables them to analyze important agricultural parameter (crop yield, crop acreage) with one of the most advanced programming languages, R. Staffers from KRASS also maintain soil salinity sampling test sites in this region, and the course enables them to analyse their field measurements, for example to perform regression analysis or supervised land use classifications, based on satellite images.
A new paper on cropland abandonment in Central Asia was published in Applied Geography in the context of the LaVaCCA Project.
In many regions worldwide, cropland abandonment is growing, which has strong and known environmental and socio-economic consequences. Yet, spatially explicit information on the spatial pattern of abandonment is sparse, particularly in post-Soviet countries of Central Asia. When thriving reaching for key Millennium Development Goals such as food security and poverty reduction, the issue of cropland abandonment is critical and therefore must be monitored and limited, or land use transformed into an alternative one. Central Asia experienced large changes of its agricultural system after the collapse of the Soviet Union in 1991. Land degradation, which started already before independence, and cropland abandonment is growing in extent, but their spatial pattern remains ill-understood.
The objective of this study was to map and analyse agricultural land use in the irrigated areas of Kyzyl-Orda, southern Kazakhstan, Central Asia. For mapping land use and identifying abandoned agricultural land, an object-based classification approach was applied. Random forest (RF) and support vector machines (SVM) algorithms permitted classifying Landsat and RapidEye data from 2009 to 2014. Overlaying these maps with information about irrigated land parcels, installed during the Soviet period, allowed indicating abandoned fields. Fusing the results of the two approaches, RF and SVM, resulted in classification accuracies of up to 97%. This was statistically significantly higher than with RF or SVM alone. Through the analysis of the land use trajectories, abandoned agricultural fields and a clear indication of abandoned land were identified on almost 50% of all fields in Kyzyl-Orda with an accuracy of approximately 80%. The outputs of this study may provide valuable information for planners, policy- and decision-makers to support better-informed decision-making like reducing possible environmental impacts of land abandonment, or identifying areas for sustainable intensification or re-cultivation.
Löw, F., Fliemann, E., Abdullaev, I., Conrad, C., & Lamers, J. P. A. (2015). Mapping abandoned agricultural land in Kyzyl-Orda, Kazakhstan using satellite remote sensing. Applied Geography, 8. doi:10.1016/j.apgeog.2015.05.009