Categories
Uncategorized

Eadaoin’s presentation at the MÉRA workshop 2019

I gave a short presentation, “An Evaluation of Integrated Cloud Condensate in MÉRA”, about the follow-on research I’ve done on the work I presented at EMS last autumn. My talk looked at how the MÉRA reanalysis dataset needs time to spin-up for the variables: integrated cloud ice and integrated cloud water. The 09-33H forecast is best and recommended for these variables.

I compared the cloud water path (CWP) from MÉRA with satellite data which was taken to represent reality. There’s generally good agreement of CWP between both datasets. However, MÉRA tends to overestimate CWP, especially during cloudy conditions. Preliminary results indicate there may be a trend in the bias in relation to cloud type, but further analysis is needed on this. Future work also includes relating the bias to large scale weather patterns.

Feedback from the audience included a question about the accuracy of the satellite data and whether that had an impact on my results. This would be a good idea for future work. Other feedback for future work included looking to see if a similar bias exists in cloud optics or in global radiation.

Categories
Uncategorized

Eadaoin’s time at EGU 2019

On Monday I attended the session ‘Numerical weather prediction, data assimilation and ensemble forecasting’. Among other interesting talks was one titled ‘Fraction of large forecast errors in global NWP’ by Thomas Haiden, which discussed the errors in upper level variables at ECMWF and the large-scale errors. He found that the fraction of large errors is important for communicating the forecast improvements. The fraction of large errors is highly correlated with average errors. Even if the skill of an NWP model improves from day 5 to day 4 over a few years, it is still important to look at the distribution of errors and how that changes with model updates.

I also attended the session ‘Big data and machine learning geosciences’ where there was an interesting talk ‘Machine learning methods for predicting energy demand/production based on hydro-meteorological input’ by Konrad Bogner, where they looked at a number of different machine learning methods to predict the energy demand and supply in the near future. These can also be used to look at predictive uncertainties in the energy demand and supply chain. Methods included multivariate linear regression (MLR), multivariate adaptive regression splines (MARS), Quantile regression (QR), quantile regression neural networks (QRNN) quantile random forest (WRF) and deep learning quantile regression (DLQR). The demand model includes variables such as daily measurements of temperature, precipitation, global radiation and wind speed as well as details about weekdays and holidays.

On Thursday I went to an interesting session which was very relevant to my research called ‘Advances in statistical post-processing for deterministic and ensemble forecasts’. A talk ‘Statistical post-processing of dual-resolution ensemble forecasts’ by Sándor Baran, looked at the skill of post-processing a dual resolution ensemble with members with a horizontal resolution of 18km and 45km. The most popular post-processing methods include BMA and EMOS. Results found that the improvement of post-processing dual resolution ensembles is not as pronounced as using a single resolution ensemble.

The next talk was by Nina Schuhen called ‘Rapid Adjustment of Forecast Trajectories: Improving short-term forecast skill through statistical post-processing’ which was about applying post-processing techniques to older forecast runs to before the next forecast is complete. The adjustment is from short-term observations which can be applied as soon as observations are available. Results found a better forecast from RAFT applied to earlier forecast runs than newer forecast runs.

The ‘Energy meteorology’ session was on Friday. The first talk was ‘Forecasting of solar irradiance at Reunion Island using numerical weather prediction models’ by Frederik Kurzrock, who found that spatial averaging usually improves GHI forecasts.

In the session ‘Forecasting the weather and aviation meteorology’ I heard an interesting talk about ‘IMPROVER: A probabilistic, multi-model post-processing system for meteorological forecasts’ by Benjamin Ayliffe, which is an open-source probability-based post-processing system from the Met Office. It has multiple modular steps including verification, thresholds to create probability, neighbourhood processing – land and topographic aware, time lag individual models among others.

An interesting piece of information I picked up during the week was that ERA5 tends to have low wind speed, possibly due to a roughness parameter issue.

While at EGU I presented my poster – ‘Multivariate spatial post-processing for renewable energy forecasts’. I got some positive interest including interest from energy traders in industry. I also got some good feedback recommending me to extend the lead time of my forecasts out further.

Categories
Uncategorized

EGU 2019

Here are some of the interesting points I saw during EGU 2019 in Vienna.

Monday: NWP, DA and Ensembles (AS1.1)

  • Florian Pappenberger outlined plans for future work at ECMWF. Cycle 46R1 will be introduced in June 2019. Which will include the ENS being initialized from 50 member ensemble DA system. There will be web presentation on 46R1 on the 15th and 16th of May.
  • Pall August Porarinsson presented work on the use of 750m resolution runs of HARMONIE for forecasting Iceland. Compared to the operational 2.5km resolution model, the 750m forecasts offered better forecasts in areas of complex topography (such as fjords) but produced a worse temperature forecast.

Atmospheric Boundary Layer (AS2.2)

  • Domingo Munoz-Esparza presented on work using LES simulation to forecast turbulence for UAVs (aka drones). In an effort to make these simulation fast enough to be operational they have developed GPU based LES code called FastEddy.
  • Clara Garcia-Sanchez presented work using WRF with the Fitch turbine wake scheme to estimate losses from hypothetical windfarms in central USA. WRF was run at 3km as a resource assessment tool. It was found that as wind farms get larger and larger (beyond realistic scales) the energy density asymtotically reduces towards 1 W/m^2.

Convection Permitting modelling (CL5.04/AS1.28)

  • Daniel Argüeso presented work on using WRF simulations to forecast rainfall over the Maritime Continent. He ran WRF at a variety of resolutions: 32, 16, 8, 4 and 2km with and without a cu_physics scheme. It was found that explicity resolving convection allowed production of stronger rainfall events and a more accurate diurnal cycle as the convection scheme did not allow CAPE to build to large enough values.

Energy Meteorology (ERE2.1/AS1.11)

  • Frederik Kurzrock discussed using WRF to forecast incoming shortwave radiation on La Réunion. Two different WRF experiments were conducted, one with an updating cycle and WRF-DA of satellite observations and another one which was simply downscaling GFS driving data. He used spatial averaged SW to compare forecast values to observations, citing Lara-Fanego. It was found that the setup with data assimilation did not produce an improvement in overall skill although it did reduce the occurrence of large errors. More frequent cycling might be tested to see if the DA setup can be improved.
  • Adrian Acevedo presented work relating to the use of WRF downscaling and PCA analysis for forecasting for renewable at a coastal location in northern Spain. He downscaled both CFS (reanalysis) and GFS (forecast) using WRF 9km – 3km – 600m. And compared the WRF-forecast to observations and also compared the WRF-forecast to the WRF-reanalysis, treating the WRF-reanalysis as truth. He also did some post-processing stuff with PCA but I didn’t follow.
  • Jan Wohland presented work on the use of century-long reanalyses for wind resource assessments, looking at 20CR (NOAA), ERA-20C and CERA-20C (both ECMWF). It was found that these ECMWF reanalyses have spurious trends caused by the assimilation of ocean winds which can be corrupted by changes to observing system. When the trend is removed, it reveals some significant low-frequency (multi-decadal) variability in wind speed and that the period since the late seventies (the satellite-era and starting point for most global reanalyses) is during a period of faster than average wind speeds. Therefore the use of this 30-40 year period for resource assessment might not be as representative as we think. Not relevant to my work at all but was a very interesting talk.

I also presented a poster on the influence of PBL schemes in the WRF model on wind speed forecast skill over Ireland, looking at both 10m from synop stations and mast observations from wind farms.

Categories
Uncategorized

Journal Club: 17-04-2019

https://onlinelibrary.wiley.com/doi/abs/10.1002/we.2302

The WRF mode was used to investigate the prediction of wind speed ramping events which can have a large impact on renewable energy production.

Study area with WRF domains and observation stations.

5 years of forecast data was used for this analysis. 2 sets of deterministic forecasts (10km and 5km resolutions) produced by downscaling GFS 0.5° data and an ensemble forecast produced by downaling the 21 members of the GEFS 1.0° data. The area studied here are 6 coastal stations in northwest Japan. Wind speed ramps are considered changes of greater than or equal to ±5m/s in 6 hours. The ability of the forecast to predict these events as binary occurrences (“yes/no events”) was examined. A variety of skill scores were utilised to measure forecast performance such as Probability Of Detection, Success Rate, Frequency Bias, Probability Of False Detection, Critical Success Index.

Performance diagrams for the two deterministic forecasts and different probability thresholds for the ensemble forecasts.

The higher resolution deterministic forecasts outperform the coarser resolution forecasts. Ramp ups are also predicted better than ramp down events. The use of probability thresholds from the set of ensemble forecasts offered the ability to predict wind ramp events better.