Verification

Question 1

General things:

Answer 1

 The value of a forecast rises with lead time and reliability.

 Note: a forecast with large lead time and low performance can have more value than an accurate forecast with short lead time

 The performance of a forecast can be assessed by comparing the forecast with observed data

 There are different performance measures for continuous variables (e.g. discharge, precipitation) or for categorical/ binary variables (e.g. threshold exceeded?).

 Comparing the performance of a new model with a simple reference model allows assessment of the forecast skill of the new method

Question 2

Performance measures for continous variables:

Answer 2

• Pbias
• Root Mean Squared Error
• Correlation

Question 3

Performance measures for categorical variables:

Answer 3

• Accuracy
• Prob of detection
• Prob of false alarm