Scheduled maintenance: 8-10 am ACT/NSW each Wednesday. Expect shutdowns and restarts.

Contact us: data at auscover.org.au  |  Disclaimer: Please read
Product pages » Crop type mapping in the Western Queensland cropping region

Crop type mapping in the Western Queensland cropping region

Last modified by Michael Schmidt on 2018/06/27 12:23

Active crop type mapping in the Western Queensland cropping region

Summer_2017.png

Link to the data

DescriptorData linkLayer name
Dataset digital object identifier (DOI) 
GeoNetwork recordNot Available Yet

QLD Tiff mosaics - THREDDShttp://qld.auscover.org.au/thredds/catalog/auscover/landsat/crop_frequency/qld/crop_type/catalog.html

QLD Tiff mosaics - FTP accessftp://qld.auscover.org.au/landsat/crop_frequency/qld/crop_type
Forage reporthttps://www.longpaddock.qld.gov.au/forage/cropfrequency.phpcrop frequency report 

Data licence and Access rights

ItemDetail
RightsCopyright 2010-2015. Rights owned by the Queensland Department of Environment and Science (DES).
LicenceCreative Common Attribution (CC-BY) 3.0
AccessWhile every care is taken to ensure the accuracy of this information, DES makes no representations or warranties about its accuracy, reliability, completeness or suitability for any particular purpose and disclaims all responsibility and all liability (including without limitation, liability in negligence) for all expenses, losses, damages (including indirect or consequential damage) and costs which might be incurred as a result of the information being inaccurate or incomplete in any way and for any reason.
 CitationPringle, M., Schmidt, M., and Tindall, D. (2018, inpress): Multi-decade, multi-sensor time-series modelling based on geostatistical concepts to predict broad groups of crops. Remote Sensing of Environment.

Alternate title

Crop type mapping - Western Queensland cropping zone 1987-2017 derived from USGS Landsat images, Sentinel-2 images and MODIS imagery .

This dataset shows broad classes of crops in areas that are actively cropped in the western Strategic Cropping Lands (SCL) in Queensland, for the winter and summer growing seasons from 1987 to the current year. Predefined growing seasons were used for winter (June to October) and summer (November to May), and actively growing crops with a green vigour were classified for each season separately. Temporal (within-season) and spectral information were used in a regression-tree-based classifier (‘Random Forest’; Breiman, 2001), using field observations as training data. All available Landsat 5 TM, Landsat 7 ETM+, Landsat 8 OLI, and Sentinel-2 data were used MODIS imagery was used as a backup in case of large, temporal data gaps. All imagery was cloud masked and NBAR-corrected (Flood et al. 2013, Flood 2014), to generate a cloud-free composite at the date of maximum greenness for each growing season. These composite images were used to produce spatially homogeneous image segments (i.e. field boundaries), on the basis of which further analyses are performed (Bunting, et al. 2013). Seasonal phenological characteristics are derived per segment and used in the classification model. The modal prediction of the multi-class classifier is shown in the crop-type maps (in GeoTIFF format).

The classification differentiates these classes in winter:

  • Cereal crop (e.g. wheat, barley, oats), with a raster value of 1;
  • Pulse crop (e.g. chickpea), with a raster value of 2; and
  • Bare soil with a raster value of 3.

The classes differentiated in summer are:

  • Coarse-grain & Pulse (e.g. sorghum, maize, mungbean), with a raster value of 1;
  • Cotton crop, with a raster value of 2; and
  • Bare soil with a raster value of 3.

The class Bare soil is to be interpreted as bare for a "substantial" time during the respective season. This is not an exclusive classification and may also include fields with a certain degree of litter or stubble.

This dataset is an improvement over a former active crop mapping approach (Schmidt et al, 2016).

The modal crop type for the per-season image classification is reported. Zero values indicate null data.

A number of ancillary datasets have been used to mask out certain features from the dataset which are not related to the detection and mapping of actively growing crops: - Nature conservation, Mining, Transport infrastructure, Waste and Disposal, Manufacturing and industrial infrastructure, Stock routes, Townships, Forestry and plantation areas have been masked using the nearest contemporary Queensland Land Use Mapping Program (QLUMP) dataset. A Landsat-derived water index was used to mask water bodies from each individual image (Danaher and Collett 2006). The SRTM data consisted of a DEM and slope layer as processed by Geosciences Australia. 

Validation status

The overall classification accuracy (tau statistic) varies with season and region where separate models are validated:

-  The summer classification for the crop groups is with a kappa statistic of 85.4%

- The winter classification for the crop groups is with a kappa statistic of 88.7%

Limitations:

The largest sources of error were that ‘Bare soil’ is likely to be under-represented, and that, during winter, ‘Pulse' can be misclassified as ‘Cereal’. Other errors may occur in the data which are due to unknown effects.

Pastures: pastures may be incorrectly classified as cropping, particularly in wet years or seasons when pastures green-up rapidly and at similar times to actively growing crops.

Tree Crops: horticulture is excluded from the crop mapping based on the QLUMP mapping.

Land-in-transition: formerly cropped land may still appear as cropped if the vegetation greens-up during the growing period, e.g. weeds or redundant or abandoned crops and crop residue are dominating the ground cover.

Failed crop: a failed crop can only be detected if it reached a certain level of greenness, which may vary between region, seasons and years.

Clearing: some recent vegetation clearing and conversion to agriculture may not be included in the mapping but are expected to be mapped in future updates. Water: vegetated areas around watercourses and dams may be classified as crop due to the strong greening-up phase after high rainfall events. Some areas of shallow water or water with emergent or floating vegetation may be incorrectly classified as cropped where they have not been removed by the water mask and have seasonal patterns in greenness which may be similar to crops.

Topographic Effects: areas with steep slopes (i.e. a slope of greater than 10%) are excluded. Inaccuracies in the DEM used for identifying these areas may result in some areas being included.

Spatial and Temporal extents

ItemDetail
Spatial resolution (metres)30
Spatial coverage (degrees)north:-26; south:-29; west:145.5 east:152.5
Temporal resolutionSeasonally (winter and summer)
Temporal coverage1987 ongoing
Sensor & platformLandsat 5TM, Landsat 7 ETM+, Landsat 8 OLI, Sentinel-2 MSI, MODIS Terra and Aqua
ItemDetail
Spatial representation typegrid
Spatial reference systemAustralian Albers. EPSG:3577

Point of contact

ItemDetail
NameMichael Schmidt
OrganisationQLD Department of Environmnet and Science
PositionSenior Scientist (Remote Sensing)
Emailmichael.schmidt@des.qld.gov.au
Rolepoint of contact
AddressRemote Sensing Centre, DES, EcoSciences Precinct
Telephone+61 7 3170 5675
URLhttps://www.des.qld.gov.au/

Credit

Landsat 5 TM, Landsat 7 ETM+, Landsat 8 QLI and MODIS images were acquired from United States Geologic Survey. Sentinel-2 imagery from the Copernicus program were downloaded from the Amazon Web Service.

Filenaming convention

Filenames follow a simple convention:

ActiveCropType<version>_<season><year>_modal.tif

Example: 

ActiveCropTypeV2_winter1987_modal.tif

Keywords

ThesauriKeyword
GCMDAgriculture
CFvegetation_area_fraction
FoRDIVISION 07 AGRICULTURAL AND VETERINARY SCIENCES >> GROUP 0703 CROP AND PASTURE PRODUCTION

There are three main thesauri that AusCover recommends:

  1. Global Change Master Directory (http://gcmd.nasa.gov)
  2. Climate and Forecast (CF) convention standard names (http://cfconventions.org/standard-names.html).
  3. Fields of Research codes (http://www.abs.gov.au/ausstats/abs@.nsf/0/6BB427AB9696C225CA2574180004463E?opendocument).

Data quality

Horizontal Positional Accuracy

All the data described here has been generated on the basis of Landsat TM, ETM+ and OLI data, which has a spatial resolution of approximately 30 m. Sentinel-2 and MODIS imagery were re-sampled to the same grid. The imagery is rectified using control points measured with a differential GPS ensuring a maximum root mean square (RMS) error of 20 m at these control points. However, it is possible that errors up to ±50 m occur between these control points. The imagery has been corrected for height displacement using a 3" digital elevation model (DEM) based on the National Aeronautics and Space Administration (NASA), Shuttle Radar Topography Mission (SRTM). It is not recommended that these data sets be used at scales more detailed than 1:100,000.

Vertical Positional Accuracy

All the data described here has been generated from the analysis of Landsat TM, ETM+ and OLI data, which has a spatial resolution of approximately 30 m. The imagery is rectified using control points measured with a differential GPS ensuring a maximum root mean square (RMS) error of 20 m at these control points. However, it is possible that errors up to ±50 m occur between these control points. The imagery has been corrected for height displacement using a 3" digital elevation model (DEM) based on the National Aeronautics and Space Administration (NASA), Shuttle Radar Topography Mission (SRTM). It is not recommended that these data sets be used at scales more detailed than 1:100,000.

Related products

Active crop mapping: http://data.auscover.org.au/xwiki/bin/view/Product+pages/Cropping

Crop Frequency: http://data.auscover.org.au/xwiki/bin/view/Product+pages/Cropping+Frequency

References and further reading:

ItemDetail or link
 PublicationBreiman, L. (2001). Random Forests. Machine Learning, 45, 5-32. http://dx.doi.org/10.1023/A:1010933404324
PublicationBunting, P., Clewley, D. Lucas, R.M. and Gillingham, S. (2014). The Remote Sensing and GIS Software Library (RSGISLib). Computers and Geosciences, 62, 216-226. http://dx.doi.org/10.1016/j.cageo.2013.08.007.
PublicationDanaher, T. and Collett, L (2006). Development, Optimisation and Multi-temporal Application of a Simple Landsat Based Water Index. Proceedings of the 13th Australasian Remote Sensing and Photogrammetry Conference, Canberra, Australia, November 2006.
PublicationFlood, N. (2014). Continuity of Reflectance Data between Landsat-7 ETM+ and Landsat-8 OLI, for Both Top-of-Atmosphere and Surface Reflectance: A Study in the Australian Landscape. Remote Sensing, 6, 7952-7970. http://dx.doi.org/10.3390/rs6097952.
PublicationFlood, N., Danaher, T., Gill, T. and Gillingham, S. (2013). An Operational Scheme for Deriving Standardised Surface Reflectance from Landsat TM/ETM+ and SPOT HRG Imagery for Eastern Australia. Remote Sens., 5(1), 83-109. http://dx.doi.org/10.3390/rs5010083
PublicationSchmidt, M., Dedadas, M.,  Pringle, M. and Denham, R. (2014). Large-scale, operational crop-frequency mapping to inform state resource management, Sentinel-2 for Science Workshop, ESA-ESRIN, 20-22 May 2014, Frascati (Rome), Italy: http://seom.esa.int/S2forScience2014/files/05_S2forScience-LandCoverI_SCHMIDT.pdf
PublicationSchmidt, M., Pringle, M., Devadas, R., Denham, J., and Tindall, D. (2016). A Framework for Large-Area Mapping of Past and Present Cropping Activity Using Seasonal Landsat Images and Time Series Metrics. Remote Sensing, 8, 312; http://dx.doi.org/doi:10.3390/rs8040312
 PublicationSchmidt, M., and Tindall, D. (2016): Crop frequency mapping for land use intensity estimation during three decades. Proceedings SP-740 of the Living Planet Symposium 2016, Prague, Czech Republic 9-13 May.

Metadata history

DateDetail
2015-04-30Metadata creation date
2015-05-19Added DOI and minor format changes
2015-05-22Minor changes
Tags:
Created by Bec Trevithick on 2017/11/30 12:38

This wiki is licensed under a Creative Commons 2.0 license
XWiki Enterprise 4.5.3 - Documentation