Tempocampo: A System for Operational Forecasting of Brazilian Sugarcane and Soybean YIELD.

Crop modelling efforts directed at yield prediction have only rarely progressed beyond model development to successful model implementation and operational use. For the Brazilian sugarcane and soybean, weather variations cause large inter-annual yield fluctuations and pre-harvest yield estimates which take climatology into account are essential in planning sales and management from the farms. Since weather (represented by air temperature, rainfall, solar radiation, wind and air humidity) is the most variable driver for the next harvest season, once monitored (consisted and gap filled) can be injected in a previously calibrated process based crop model to produce yield scenarios for coming harvest assuming weather as a key factor. Simulations from TEMPOCAMPO System (TC) were first released late in 2016 (www.tempocampo.org), after few years working on the support software for the whole system to automatically operate. Since them it has been release its reposts in the internet to give some advice to decision makers on how climate could affect the crop on the season. It has been using the SAMUCA and DSSAT/CROPGRO-SOYBEAN as the crop models for sugarcane and soybean, respectively. Yields and TC simulations for 1980–2016 were well correlated (r=0.67 and r=0.58 for sugarcane and soybean, respectively) with a standard error of estimate (4% and 6% of the mean yield for sugarcane and soybean, respectively). Results were better when comparing the year variation trend regarding the average, when correlation exceeded r=0,90 for both crops. Although already in operation, further research and software improvements for decreasing the system uncertainty still being needed.

KEY-WORDS: weather, agrometeorology, crop modeling, SAMUCA, DSSAT