Welcome to the Forecasting Rodeo 2 Challenge Series!
Why This Matters
Water managers need more skillful information on weather and climate conditions in order to help efficiently utilize water resources to reduce the impact of hydrologic regimes. Examples of hydrologic regimes include the onset of drought or a wet weather extremes from high levels of precipitation. Lacking skillful sub-seasonal information limits water managers’ ability to prepare for shifts in hydrologic regimes and can pose major threats to their ability to manage the valuable resource.
The challenge of sub-seasonal forecasting is that it encompasses the time frame where initial state information (e.g., coupled land-atmosphere processes) becomes less important, and slowly varying long term states (e.g., sea surface temperatures, soil moisture, snow pack) become more important to prediction skill. Additionally, this information is needed 15 days to 45 days ahead of time, or even longer.
The Rodeo II Challenge series will create algorithms to enhance sub-seasonal forecasting, reducing risks to our water systems, and helping water managers efficiently manage hydrological regimes.
Challenges and Prizes
This challenge series will first have four traditional marathon match challenges to establish predictive algorithms on historical weather data, and then recurring data science challenges over the next full year to refine those solutions on live weather data.
The first challenges kick off July 19th, 2019.
Each of the four concurrent marathon matches will have the following prize structure with 10 placements.
In addition, if any of the top ten placements are new submitters*, they will earn an additional $100 bonus! All new participants who submit passing submissions will earn a $25 bonus.
*A new submitter is someone who has never submitted to a Topcoder Marathon Match and who registered for Topcoder as a new member as of June 27, 2019.
You need to be a Topcoder member to participate in this series.
The contest is extended till Sunday (25th) 10:00am EDT, feel free to submit now to all 4 challenges.
We think now that the difference in scores had nothing to do with architectural differences between the CPU and GPU setups, nor were our test servers hacked, as some of you suggested. It seems that at the end of the CPU-based testing period the set of target test days had one extra day compared to the set that we planned to use, that's why you saw different scores in identical submissions. We are still investigating how this could have happened. We also think that the current setup is correct in all 4 systems. To compensate for the downtime and the submission issues we extended the contest with ~2 days. Apologies for the inconvenience caused by this.