Yield Prediction

In March of 2004, the Institute for Technology Development (ITD) wrote a proposal to the NASA Earth Science Application Directorate (ESAD) to use the experience ITD has developed in agricultural remote sensing applications via efforts such as the Ag2020 project to work with partnering government entities. The United States Department of Agriculture (USDA) was targeted as a primary partner to benefit from the use of NASA Earth Science Enterprise’s (ESE) suite of remote sensing tools in their daily tasks, and to specifically enhance the use of their current decision support systems. One agency within the USDA that was identified was the National Agriculture Statistics Service (NASS). The mission of NASS is to provide timely, accurate and useful statistics to the U.S. agricultural industry including information on crop acreage and yield. After discussing possibilities with members of NASS’ research and development team, ITD proposed to NASA that imagery from ESE’s suite of satellites could offer opportunities for NASS to improve yield prediction models.

The challenge for the project was to determine a use for remotely sensed imagery that NASS had not already investigated. NASS has a long history of working with remotely sensed imagery and was one of the first government agencies to use satellite imagery in production mode. Since the early 1970’s, the agency has been utilizing Landsat satellite imagery for crop acreage estimates and crop condition monitoring.

ITD had two primary objectives which would lead to the completion of this project.

· The collection and processing of hyperspectral imagery over areas identified in central Iowa by NASS and ARS in late July and late August of 2004.
· The development of a database of five years of MODIS, soils and weather data over the states of Illinois and Indiana.

To achieve the first objective, plans had to be made for the collection of the imagery using ITD’s hyperspectral sensor coordinated with ground data collection, then the imagery had to be referenced to real-world coordinates and calibrated to radiance.

Figure 1. Location and flight lines provided by USDA ARS personnel to ITD for collection of hyperspectral imagery.

Figure 2. Steps for referencing RDACSH imagery to real-world coordinates.

For the second objective, weather and soils data would have to be purchased and converted to shapefiles and MODIS imagery would have to be ordered from NASA and downloaded or shipped.

Figure 3. Information derived from the Midwest Regional Climate Center.

Figure 4. Information derived and station locations for the TD3200 Cooperative Network.

Figure 5. Information derived and station locations for the Global Surface Summary of the Day.

The complete success of the research still will be seen in the results from the analysis being performed by ARS and NASS. It is hoped that the information and documentation provided by ITD to NASS and ARS will prove useful to the Services and lead to the eventual production use of ESE sensor data.

For more information on this or any other past research, contact us.