EarthSat GeoTools is a suite of tools that utilize sampling methodology to facilitate performing agricultural assessments with the primary inputs being multiple resolutions of imagery.
Several techniques are utilized, using either a single level or multiple nested levels of sampling.
Two characteristics distinguish the GeoTools approaches:
- High resolution imagery are used to accurately discriminate areas of a crop or cover of interest. This imagery takes the place of ground truth, and makes the procedure particularly useful when ground surveys are prohibitively expensive or otherwise impractical.
- Stratification by spectral means is used to extend the results from the high resolution sample areas to broader areas. When double sampling is utilized, the second level of sampling serves to improve the estimate made from the first level of samples.
By a combination of these two approaches, the areal extent of a crop or cover of interest can be estimated with a high, measurable degree of accuracy, at a lower cost than would be possible by doing a full enumeration of the entire area. The advent of the availability of high-quality commercial imagery of high spatial resolution makes this a technology whose time has come.
Louis Dreyfus Corporation
MDA was contracted by the Louis Dreyfus Corporation to estimate the area of coffee production in the Mogiana region of Sao Paulo, Brazil. MDA applied the GeoTools approach to calculate a coffee estimate for this region. The terrain is very rugged and results of typical ground surveys inaccurate. A combination of LandSat 5 imagery and aerial photography was used to estimate the hectares of coffee production in this region. There is no clear spectral signature for coffee because the coffee plant is typically less than 3-meters high and retains greenness except for pruning. Coffee is very closely mixed with the "scrub shrub" type of land cover in this region.
Souther States Cooperative
A major east coast agricultural chemical provider/applier wanted to be able to monitor crop health in near real-time and make management decisions based on the monitoring data. MDA was contracted to create and ESRI ARC Internet Mapping Service (ArcIMS) to post Landsat 7 images, digital soils data, and NDVIs (Normalized Difference Vegetative Index) for corn and soybeans during the 2000 growing season. The system worked well and allowed the client to monitor crop greenness throughout the growing season.
The project continues with the overlay of yield data on NDVIs from various times to determine if NDVI is a good predictor of the magnitude of yield. The capability to extract and email small portions of the data "clip & ship a farm field" for import to the tractor, using the AGRIS GIS system is underway. The project is continuing in the 2001 season with the goal of east coast coverage by 2003.