Government agencies and agricultural managers require information on the spatial distribution and area of cultivated crops for planning purposes. Agencies can more adequately plan the import and export of food products based on such information. Although some ministries of agriculture and food security annually commission their staff to map different crop types, these ground surveys are expensive and yet cover only a sample of farms. Remote sensing data, together with ancillary information, enable the determination of the spatial distribution of crops at varying spatial scales with relatively little financial resources. This can, however, be achieved with various degrees of uncertainty, as some crops are spectrally similar and pixel sizes sometimes bias the estimation of crop acreages.
Knowledge of the growth cycle or calendar of crops is essential for accurate interpretation of RS data (Forkuor et al., 2014; Peña-Barragán et al., 2011; Son et al., 2013). The cropping calendar covers the period from land preparation, planting, growth, pollination, senescence to harvesting. The spectral reflectance of crops at each of these growth stages is called the temporal profile of the crop. This information is extracted from satellite images based on surveyed locations (boundaries) of dominant crop types in the area of interest. Dominant crop types are surveyed during field campaigns that ideally coincide with the timing of satellite image acquisition.
For example in the STARS project, the boundaries of 48 fields representing six dominant crops in Mali were mapped during an extensive field campaign, from which the temporal profiles of each crop were extracted. Figure 4.8 shows typical temporal profiles of dominant crops in Mali based on NDVI data derived from Digital Globe WorldView images. It shows the typical growth cycle of each crop during the cropping season (May – October).
- Figure 4.8 NDVI development of dominant crops in Mali derived from WorldView images (Source: STARS Mali team)
Based on the potential uniqueness of these temporal profiles, UAV and satellite images can be used to reveal the spatial distribution of crops in the area of interest. NDVI is a simple metric based on spectral information; other spectral indices exist and may be useful in certain settings, dependent on context.
Apart from spectral information, other information layers can be added to improve the accuracy with which crops can be identified. An example is textural information (Haack and Bechdol, 2000; Sheoran and Haack, 2013).
Texture represents the degree of local spatial variation in an image. A crop in combination with a planting practice, by virtue of its spatial arrangement, displays textural properties. Derivation and addition of textural measures to the spectral information can therefore improve crop identification accuracy.
Both texture and context are adding important information for the classification of image segments. Context refers to the relation between coarse and fine image segments. Texture serves as a valuable parameter in addition to spectral reflectance to identify the crop per segment. The textural parameters that worked best, were those in GLCM (Grey Level Co-occurrence Matrix) and GLDV (gray-level difference vector) (Conrad et al., 2010; Novack et al., 2011).
Despite acquiring sufficient UAV/satellite and field data, crop identification can be challenging and may result in low accuracies. At the core of this is high variability in the spectral characteristics of the crops under study. In other words, the temporal profiles of the crops are ideally unique for each crop, but this is often not the case.
Figure 4.9, for example, shows the monthly temporal profiles of the dominant crops in the STARS Mali site. Each column (labelled by month) depicts the spectral patterns extracted from five quadrats within a field. The figure shows high similarity in the profiles, which challenges crop identification and separation in satellite/UAV images.
This high spectral and spatial variability can have various causes. These include:
- Overlaps in cropping calendar, especially in predominantly rainfed agricultural areas where different crops are planted and harvested around the same time, leading to similarity in their temporal profiles.
- Differences in management practices (e.g., tillage, weeding, fertilization) between and within fields
- Variations in soil type, depth and fertility
- Intercropping, i.e., cultivation of different crops in the same field
- Proximity of natural/semi-natural vegetation to cultivated areas
- Occurrence of many or large trees in farm fields
- Water accumulation
- Occurrence of pests and diseases in portions of a field
Figure 4.10: Farm field in Njombe District before and after weeding, Tanzania (Source: STARS AgriSense team)
Figure 4.11: Intercropping of pumpkin with maize (Source: STARS AgriSense team)
To reduce the effects of the above-mentioned factors on crop identification, a number of measures can be taken. These include:
- Inclusion of additional RS data, e.g., Synthetic Aperture Radar (SAR) data (McNairn et al., 2009; Forkuor et al., 2014)
- Landscape stratification based on soil, topography, climate, or other characteristics
- Performing object-based, instead of pixel-based, image analysis (Peña-Barragán et al., 2011)
- The use of other identification approaches such as the sequential masking classification algorithm (Van Niel and McVicar, 2004; Forkuor et al., 2015)