AgroNum's I/O source/sink data is robust

Precise grain yield & nitrogen fertilization recommendations require precise input data. This said, at the field level, very few such data are easily available. In fact only grain-yield dry-matter & nitrogen-content, along with kg-N_fertilizer/ha (and other fertilizer & crop protection inputs, of course) are known precisely. The rest - soil organic matter stocks, soil physico-chemical analyses, clay contents, texture, etc., are at best error prone averages.

Ladha et al. 2016 recognized this when stating that " ... [their] scaling-down approach to estimate the contribution of non-symbiotic N2 fixation is robust because it focuses on global quantities of N in sources & sinks that are easier to estimate, in contrast to estimating N losses per se, because losses are highly soil-, climate-, and crop-specific ..." (sic).

"Sources & sinks", i.e. N-fertilizers application rates (TUN) & grain N-yields (RDN), are also at the core of Polyor SAS's AgroNum approach to sustainable agriculture (https://lnkd.in/eBmfT3NM). Farmers need only indicate the centermost GPS coordinates of the field-plot. The AgroNum AI algorithme then generates a unique RDN/TUN response curve for that particular field. A vast ensemble of soil organic & mineral characterestics along with climate data pre-existes and is processed by AgroNum's AI. Minimal data uploading and no soil sampling, no sensors or drones, etc. Strictly I/O, sources & sinks. The rest is data-science &  AI, unseen & unheard of by the farmer.

AgroNum is thus precise, ergonomic & ... "robust". It is also interoperable with azotobacterial fertilziation as a means of non-symbiotic N2 fixation as anticipated by some of Ladha et al. 2016's data.

#sustainable #agriculture #ai #artificalintelligence #sustainableagriculture #durabilité #azotobacterial_fertilization