Subseasonal prediction for bloom dates of tart cherries in Utah and Michigan, USA: merging phenological models with CFSv2 forecast.

Temperate fruit trees require chilling for rest completion, followed by sufficient heat accumulation for onset of growth and bloom. The application of phenological models to predict bloom dates has been widely used in orchard management. Examples of such application include selecting adapted cultivars less prone to early bloom, predicting needs for frost protection, and preventing damage from late spring freezes. This study merged the Utah (chill) and ASYMCUR (forcing) phenological models by combining chill units and heat units (measured in growing degree hours) to predict bloom dates of tart cherries (Prunus cerasus L.) in Utah and Michigan, the top producing states of the USA. It was found that the modified Utah model improves the estimation of chill units compared with the original one, while the original Utah model may still be suitable for use in the colder winter of Michigan (with its later bloom dates than Utah). The combined models were applied with the temperature predicted by the Climate Forecast System v2 (CFSv2) model. The prediction was applied twice a month, starting from 1 February to 1 May. The Utah-ASYMCUR model using the forecasted temperature from CFSv2 exhibits subseasonal performance in predicting the bloom dates for 6 weeks in advance. The prediction can offer growers a way to mitigate extreme climate anomalies.

Precipitable Water Vapour Retrieval from GPS Precise Point Positioning and NCEP CFSv2 Dataset during Typhoon Events.

Radiosonde is extensively used for understanding meteorological parameters in the vertical direction. Four typhoon events, including three landfalls (MERANTI, NEPARTAK, and MEGI) and one non-landfall (MALAKAS), were chosen in analysing the precipitable water vapour (PWV) characteristics in this study. The spatial distribution of the three radiosonde stations in Zhejiang province does not meet the requirement in analysing changes in PWV during typhoon event. Global position system (GPS) observations are an alternative method for deriving the PWV. This enables improvements in the temporal⁻spatial resolution of PWV computed by the radiosonde measurements. The National Centers for Environmental Prediction (NCEP) re-analysed data were employed for interpolating temperature and atmosphere pressure at the GPS antennas height. The PWV computed from GPS observations and NCEP re-analysed data were then compared with the true PWV. The maximum difference of radiosonde and GPS PWV was not more than 30 mm at Taiz station. The Root-Mean-Square (RMS) of PWV differences between radiosonde and GPS was not more than 5 mm in January, February, March, November, and December. It was slightly greater than 5 mm in April. High RMS in May, June, July, August, September, and October implies that differences in GPS and radiosonde PWVs are evident in these months. Correlation coefficients of GPS and radiosonde PWVs were more than 0.9, indicating that the changes in GPS and radiosonde PWVs are similar. Radiosonde calculated PWVs were used for GPS PWV calibration for understanding the PWV changes during the period of a typhoon event. The results from three landfall typhoons show that the average PWV over Zhejiang province is increasing and approaching China mainland. In contrast, MALAKAS did not make landfall and shows a decreasing PWV trend, although it was heading to China mainland. Generally, the PWV change can be used to predict whether the typhoon will make landfall in these cases. PWV spatial distribution of MERANTI shows that PWV peaks change along the typhoon epicenter over Zhejiang province.

Monthly ENSO Forecast Skill and Lagged Ensemble Size.

The mean square error (MSE) of a lagged ensemble of monthly forecasts of the Niño 3.4 index from the Climate Forecast System (CFSv2) is examined with respect to ensemble size and configuration. Although the real-time forecast is initialized 4 times per day, it is possible to infer the MSE for arbitrary initialization frequency and for burst ensembles by fitting error covariances to a parametric model and then extrapolating to arbitrary ensemble size and initialization frequency. Applying this method to real-time forecasts, we find that the MSE consistently reaches a minimum for a lagged ensemble size between one and eight days, when four initializations per day are included. This ensemble size is consistent with the 8-10 day lagged ensemble configuration used operationally. Interestingly, the skill of both ensemble configurations is close to the estimated skill of the infinite ensemble. The skill of the weighted, lagged, and burst ensembles are found to be comparable. Certain unphysical features of the estimated error growth were tracked down to problems with the climatology and data discontinuities.

A new method for determining the optimal lagged ensemble.

We propose a general methodology for determining the lagged ensemble that minimizes the mean square forecast error. The MSE of a lagged ensemble is shown to depend only on a quantity called the cross-lead error covariance matrix, which can be estimated from a short hindcast data set and parameterized in terms of analytic functions of time. The resulting parameterization allows the skill of forecasts to be evaluated for an arbitrary ensemble size and initialization frequency. Remarkably, the parameterization also can estimate the MSE of a burst ensemble simply by taking the limit of an infinitely small interval between initialization times. This methodology is applied to forecasts of the Madden Julian Oscillation (MJO) from version 2 of the Climate Forecast System version 2 (CFSv2). For leads greater than a week, little improvement is found in the MJO forecast skill when ensembles larger than 5 days are used or initializations greater than 4 times per day. We find that if the initialization frequency is too infrequent, important structures of the lagged error covariance matrix are lost. Lastly, we demonstrate that the forecast error at leads ≥10 days can be reduced by optimally weighting the lagged ensemble members. The weights are shown to depend only on the cross-lead error covariance matrix. While the methodology developed here is applied to CFSv2, the technique can be easily adapted to other forecast systems.