aWhere’s Ag Intelligence - Foundation ‘big data’ Asset

06/28/17
0

aWhere’s Agricultural Intelligence products provide users with accurate – and localized - agricultural World-Cloud-Information-Big-Data-Data-Global-1667184.jpginformation for the entire agricultural earth. In order to accomplish this, aWhere integrates information from ground weather stations, Doppler weather stations, satellites, and other sources from around the world in order to create a continuous global ‘weather surfaces’(continuous, accurate, geographic layers of weather information). By collecting and organizing this information, aWhere is able to create a global network of more than 1.5 million virtual weather stations that provide daily observed and hourly forecast information. With quality data since 2006, recent increases in weather variability are captured at a hyper-local resolution.

Our innovation:
aWhere consumes more than 13,000 ground stations globally every day. Each daily weather observation is statistically examined by a suite of meteorologically tuned algorithms to ensure validity. To create aWhere’s innovative weather surfaces our proprietary 3D interpolation algorithms come into play. “3D” leverages location (latitude and longitude) as well as the elevation of the observation. Weather surfaces are accurate because the elevation is known between the irregularly spaced ground station observations.

 Leveraging elevation connects our interpolation method to the physics of the atmosphere: temperature is absolutely related to the elevation of the location in question relative to nearby observations: if higher in elevation, the temperature will be cooler and therefore humidity will be higher – this is adiabatics.

 This suite of elegant 3D interpolation techniques delivers accurate local temperatures and humidity. The aWhere system is sufficiently agile and localized that each and every temperature observation is tested against an interpolated surface using all the other stations (an “out of sample test”) every single day. Observations deemed to have too great an error (between the observation and the geographic neighbors interpolated ‘prediction’) are dropped for that day – because the physics of the atmosphere dictate the change in temperature associated with elevation. Erroneous observations – like in the case of a bad sensor or even an odd event (a parked car too close to the sensor!!), that specific observation is dropped for that day. This is ‘big data’ at its core: a huge number of calculations are done over and over each day against 13,000 observations of each variable to ensure accurate daily temperature and humidity weather data every 9km.

 For rainfall, the situation is more complex. Rainfall – and particularly daily rainfall - cannot be accurately interpolated: it is a discontinuous variable and rainfall in any one location may have no relation to the absence – or magnitude - of rainfall nearby. To create a daily total rainfall everywhere across the agricultural earth at this 9km resolution, aWhere processes a huge amount of data, mostly from satellites and active radar (where available). It is the rainfall observations that set the resolution of the weather surfaces. The 9km spacing (technically 5 arc-min) pushes the satellite observations to their most precise with good accuracy and low error - and it is the highest resolution our partners (primarily CIRA – the Cooperative Institute for Research in the Atmosphere) can produce. We process hourly, global, rainfall at this 9km resolution (aWhere has exclusive access to this product), merge this with publicly available satellite derived data (typically 3 hour aggregate); integrate this with the 13,000 ground station observations (we calibrate the satellite data – much stronger calibration than the 100 or so utilized by the publicly available satellite product); we then leverage active radar where available (we adjust active radar as the distance from source increases - noting that active radar becomes less and less accurate as you move more than ~100km from the source) to produce the most accurate daily observed weather data possible.  

 aWhere’s weather surfaces, covering the agricultural earth, are the most accurate agriculturally optimized information available today. aWhere creates more than 7 billion data points per day from a variety of information sources in order to build aWhere's 1.5 million virtual weather stations which cover the entire agricultural earth at 5 Arc-minutes or approximately 9km at the equator.

vws.png

Agronomic Attributes
aWhere provides insights for the agriculture industry through derived agronomic attributes built on the foundational ag-weather surfaces. These attributes include the straight forward accumulation (Growing Degree Days, rainfall) but importantly also include the calculation of Potential EvapoTranspiration or PET (see http://blog.awhere.com/how-thirsty-is-the-atmosphere ).   PET is the evaporative demand of the environment – a critical component of the environment to understand crop growth. It is not enough to have precipitation – if temperatures are warmer than normal (and with warmth, typically humidity is lower) then crop water demand increases. The ratio of P/PET is an excellent indicator of water availability for the plant. As this ratio drops below 1.0 – there is less water available than the environment demands and water stress starts to impact the plant. Buffered by the water stored in the soil, aWhere monitors crop development leveraging the science of crop-water demand (a young crop after germination needs much less water than the same crop during reproduction and grain fill when the plant is full grown). These crop specific water balance monitoring are one of the the keys to interpreting agricultural weather information into yield and production forecasts. 

Base, advanced, and agronomic information are all accessible through an industry standard, robust, API. aWhere also generates considerable insight from these data provisioning actionable intelligence for input providers, farmers and those who track the market (i.e., supply chain managers and commodity traders).

Topics: Big Data

Recent Posts

Search

Subscribe to Email Updates