29 Jul Innovation Challenge and Akvelon
Bellevue, WA – Akvelon is teaming up with Microsoft and the United States Department of Agriculture (USDA) in their effort to help farmers, agricultural businesses, scientists and consumers with climate change and resiliency of food systems across the US.
Given the current climate changes, it is more important than ever to get more data to American farmers to create a sustainable food system for the United States. Farmers need to know what can grow well in their area, and prepare for crop changes. They need to evaluate the economics of consumer demand, in order to predict future crops.
For the first time in 100 years, USDA’s key datasets are accessible in Microsoft’s cloud, where they can be used for insights, or to create new types of end-user applications. “I am looking forward to discovering what creative ideas the community comes up with when blending together USDA and other government data sets, and novel ways of analyzing that data with access to our Microsoft Azure cloud computing platform,” says Dr. Green, a deputy managing director at Microsoft Research.
Allowing information, and researching information are two completely different tasks. Akvelon worked with Microsoft to create what is known as the “Farm Dashboard”. The Farm Dashboard allows farmers, businesses, and consumers to research and download customized queries. There are many features to the Farm Dashboard, such as searching for beet crop expenses in Washington State in 2005, or researching crop and farm financial costs, retrieving the yield of a crop, fetching the climate, and viewing where the current crops are being harvested. The Farm Dashboard is available for anyone to use, in fact, you can view it here.
You can access the following USDA databases:
- NASS (Quickstats)
The National Agricultural Statistics Service (NASS) is an on-line database containing billions of data that offers Quick Stats containing official published aggregate estimates related to U.S. agricultural production. NASS develops these estimates from data collected through hundreds of sample surveys conducted each year covering virtually every aspect of U.S. agriculture and the Census of Agriculture conducted every five years providing state- and county-level aggregates.
Agricultural Resource Management Survey (ARMS) is USDA’s primary source of information about production practices, financial conditions, and resource use of America’s farm businesses. They are also responsible for the economic well-being of America’s farm households. ARMS is a nationally acclaimed survey, administered using several phases – sample screener, field-level, and farm-level phases – targeting about 5,000 fields and 30,000 farms each year.
The geospatial data called the Cropland Categories (CropScape) is a faster, geo-referenced, crop-specific land cover data layer created annually for the continental United States, using moderate resolution satellite imagery and extensive agricultural ground truth. In general words, CropScape is a set of arrays to geographically referenced points, one array per year.
The geospatial data called the Vegetation Condition (VegScape) is a faster, geo-referenced, NDVI data layer that is available weekly. NDVI (Normalized Difference Vegetation Index) is an index, used to measure and monitor plant growth, vegetative cover, and biomass production with values ranging from 0 to 1 – with higher values indicating stronger plant vigor and high chlorophyll content. In general words, VegScape is a set of arrays to geographically referenced points, one array per week. Each point contains one value, describing NDVI at this point.
- Use Cases (Two Parts)
With data visualization, the user may get NASS data (area harvest, area planted, yield, and year), filtered by the selected state and commodity. The first chart (Production) will show the harvested area, planted area and yield over available years for the selected state and commodity. The second chart (Area Planted) will show the area planted over available years for the selected state and all non-selected commodities.
Azure Machine Learning
Azure Machine Learning (ML) offers a streamlined database for all data scientists, from setting up with only a web browser, to using drag and drop gestures and simple data flow graphs to set up experiments. The Azure ML web service on the Farm Dashboard provides an estimate for one selected crop using linear regression based on data for all crops over a period of years. The result of this algorithm is an estimated planted area for selected crop for all past years (compared to actual value) and for the current year.
Data Scientists will love the new Azure ML, as the new version incorporates Python. This capability will be powerful for data scientists, especially blending this feature in with the Farm Dashboard. “We made a lot of improvements and adding Python was part of that. Azure Machine Learning is the platform. You can copy a bit of Python code and plug it into the studio and create an API,” explained Joseph Sirosh, corporate vice president at Microsoft in an interview with TechCrunch. Additionally, data scientists will enjoy the new Azure ML supports Hadoop and Spark, allowing for a wide range of tools to process big data – including the billions of USDA data.
- Climate Data
FetchClimate exposes API, allowing arbitrary data-set requests consisting of several AJAX calls. Data scientists can visualize and grab customized clips of data, and export it to CSV, JSON, ZIP or XML file. This data consists of complex geographical information including, but not limited to, climatological information. FetchClimate will choose the best data set for your query, and perform all the necessary regridding in space and time. It will return a best guess, uncertainty, and provenance for your query and display the results on the map for visual exploration. Alternatively, the FetchClimate service can be used directly via a simple API, from within programs written in any .NET language, Python, P or MatLab.
Microsoft’s Farm Data Dashboard is one of several ways that Microsoft cloud products and machine learning capabilities are making the USDA’s data accessible to external developers to help them build applications and services. This one of a kind dashboard is available for public use.