more descriptions on leaderboard

website/leaderboard.md

@@ -2,12 +2,16 @@
 
 We provide a leaderboard for both deterministic and probabilistic models/metrics.
 This interactive page is inspired by the [WeatherBench 2 design](https://sites.research.google/weatherbench/).
-All results are evaluated on the year 2022.
+
+We showcase select variables (`t-850`, `z-500`, `q-700`) on select metrics e.g., `RMSE`, `ACC` for deterministic models or `RMSE`, `CRPSS` for ensemble models. 
+Learn more about the metrics [here](https://leap-stc.github.io/ChaosBench/metrics.html) and access the complete metrics across variables [here](https://huggingface.co/datasets/LEAP/ChaosBench/tree/main/logs). Unless stated, all results are evaluated for the year 2022.
 
 ## Deterministic Models
+Deterministic models make a single (control) forecast. This method is particularly sensitive to intiial condition, and is therefore not recommended for long-term chaotic prediction such as in the S2S case. 
 
 <iframe src="https://htmlpreview.github.io/?https://raw.githubusercontent.com/leap-stc/ChaosBench/main/website/html/control.html" width="100%" height="600px" frameborder="0"></iframe>
 
 ## Ensemble Models
+Ensemble models make multiple forecasts (members) at a time to give us a range of likely future outcomes. This (plus a well-dispersed ensemble) is particularly important for S2S forecasting. 
 
 <iframe src="https://htmlpreview.github.io/?https://raw.githubusercontent.com/leap-stc/ChaosBench/main/website/html/ensemble.html" width="100%" height="600px" frameborder="0"></iframe>