|TSM/TDM Science Team Meeting 2016|
|Science Team Meetings|
|How to Submit a Proposal|
|COFUR Price List (scientific use)|
|TanDEM-X Science Service System|
|Title||Multi-Frequency SAR for Crop Acreage Estimation|
|Investigator||SHANG, JIALI - Agriculture and Agri-Food Canada, Research Branch|
Early-season crop inventory provides valuable spatial information on changing land use and agricultural production. Agriculture land use monitoring is critical for forecasting, early warning, program response, policy development, and performance measurement. Timely geospatial information on crop inventory can be used to track a number of sustainable production issues and crop resources. Crop type information is also needed as input by various crop growth and yield models.This proposed research aims at developing a methodology to provide crop acreage estimate early in the growingseason. Recent research conducted at Agriculture and Agri-Food Canada (AAFC) under TerraSAR-X science LAN_0337 has shown that satisfactory classification accuracies can be achieved using SAR data alone at the end of the growing season. This proposed research will examine the potential of multi-frequency SAR data to derive crop acreage information early in the growing season. On-going research activities will continue collecting L-Band (ALOS PALSAR) and C-Band (RADARSAT-2 polarimetric) SAR data to identify crops in diverse agro-ecosystems. With the addition of the X-band SAR data, this project will permit the unique opportunity to demonstrate the delivery of in-season a crop inventory using multi-frequency (X- C- and L-Band) SAR. The project objectives include:
1. Determine to what level of accuracy TerraSAR-X can classify crops in Canada’s main agricultural regions.
2. Determine whether multi-frequency (X- C- and L-Band) SAR data alone can produce classification accuracies targeted by AAFC (overall and individual accuracies of 85%) at the early stages of the growing season.
3. Determine whether incorporation of coherence change detection can assist in delivering crop type information.
The team will use a random forest classifier for crop classification. In addition advanced non-parametric classification methods and classifiers designed specifically for polarimetric data will be explored using existing data over Canada. Optimal pre-processing methodologies will be examined such as speckle filter selection. The inclusion of polarimetric-derived inputs into the classifier, such as decomposition parameters, will be explored. Classification of TerraSAR alone, and in combination with C- (RADARSAT-2) and L-Band (PALSAR) data, is planned. The project team is requesting a total of 72 images over a three-year period (12 images per site per year). Data required are mainly dual polarization stripmaps. This project falls within the scope of existing internally funded research projects at AAFC. RADARSAT-2 data are available through the Canadian government allocation. ALOS PALSAR will be acquired through an existing JAXA project. This research will produce a preliminary and final report on the contribution of TerraSAR-X for in-season crop mapping, as well as scientific publications. Results will demonstrate the applicability of TerraSAR-X for operational mapping, promoting the use of these data in both public and private agencies.
Back to list of proposals
|© DLR 2004-2016||