LAUSR

Many models are designed to generate future predictions under climate-change scenarios. Such models are typically calibrated for a study area with climate data that represent historical conditions. However, future projections of the model may include outputs for which the model has not been calibrated. Ideally, a climate-change-impacts model would be calibrated for recent conditions and also for possible future climate conditions. We demonstrate an approach, where a vegetation model is subjected to two calibrations: conventionally to the study area and separately to the study area plus additional areas representing analogues of potential future climate. We apply the dynamic vegetation model MC2 to a mountainous ecosystem in the Pacific Northwest, USA. We compare the conventional model calibration with the extra-study-area calibration and future projections. The two calibrations produce different outputs in key ecosystem variables, where some differences vary with time. Some model output trends for net primary productivity and plant functional type are more influenced by climatic input, while for others, the calibration area has greater consequence. Excluding areas representing potential future climate may be an important omission in model calibration, making the inclusion of such areas a decisive consideration in climate-change-impact simulations.