Release Notes

Solargis TMY data is constructed using multi-year time series (TS) by selecting the most representative months which are finally concatenated into one artificial and representative single year.

We introduced a new methodology for generation of different probability scenarios for TMY PXX (P99, P95, P90, P80, P75, P25, P20, P10, P5, P1) into the data delivery process.

The new methodology is based on increasing the number of months used to generate the TMY from TS. For example: if we have 27 years at the input, we create 729 synthesized candidates. This allows increasing the data points of constructing TMY in a simpler way but also increases the accuracy.

The calculation for the interannual variability of DNI for TMY Pxx (P90/95/99 - excluding P50) has been improved. The interannual variability affects the combined uncertainty, as such the expected DNI Pxx value is used for TMY generation. The combined uncertainty is now higher.

We extended the nowcast availability provided in Solargis. The nowcasting is accessible in the timeframe of up to 6 hours ahead thanks to the Solargis Cloud Motion Vector (CMV) model.

We have extended the nowcast in MSG/MFG PRIME (Europe and Africa) region to Azores, Egypt and Zambia.

Solargis' metadata was updated in all CSV file header of products. The updates include information on whether horizon shading is applied for GHI, DNI, DIF and GTI data, e.g. #GHI - Global horizontal irradiation [Wh/m2], no data value -9, with terrain shading.

Key features:

• All charts are now exportable and with more additional info added to exported images.
• Recent projects on Dashboard and Prospect application now show the same projects.
• Unused language selector was removed from Data exports screen.
• Improved Compare links - Projects in Compare stay selected after the page reload.
• Default time format for new users based on their local settings.
• UI improvements

Bug fixes:

• Fixed issues with user registration and trial requests.
• Fixed issues in company invitation workflow.
• Deleted users no longer block subscription slots.
• SNOWD parameter is now visible in Compare.
• Displaying charts in Compare tool.
• Corrected texts in CSV exports.

We have extended the nowcast data access to New Caledonia. The nowcasting model, based on the calculation of cloud motion vectors from satellite data, provides forecasts within the next 6 hours timeframe. 

The processing of Solargis historical time series (TS) and long-term average (LTA) data were updated and related bugs have been fixed.

Key features:

• The monthly, yearly and LTA irradiation data were updated from daily average to sums
• We fixed the yearly PVOUT aggregation previously obtained from daily average, to yearly sums
• We fixed the UVA, UVA aggregation for monthly and yearly time period
• We increased the number of decimals in exports for the UVA and UVB as it was insufficient for the winter season
• We fixed the WOUT aggregation for monthly, yearly sums and LTA
• We fixed the monthly and yearly experimental uncertainty calculation and changed representation from daily average to sums as well

Wind power (WOUT) has been developed and integrated into the standard delivery method of Solargis' forecasts.

This is a preliminary version and we are currently working to release a more developed one.

Key features:

• Push API enablement for FTP delivery type
• Access to a global coverage

Forecast services are now providing predictions up to 15 days ahead. This is related to the recent integration of input parameters from the GFS model, which have been extended to day +15.

Within the release of this new version, improvement on data post-processing methods affecting forecasting of atmospheric pressure (AP) has been implemented too.

GOES-R satellite located at 75º West longitude (East location) covering mostly the West coast of continental US and Canada was updated in our historical time series for the period of May to December 2019.

Bug fixes:

Previously, time series data were incorrectly calculated for late afternoon hours for the 1st and last day of the month for the whole period of May to December 2019. It resulted in an underestimation of the cloud coverage for corresponding time series data.

This issue is now fixed in the updated version of our model v2.2.12. All data previously impacted for the west coast of continental US and Canada (approximately 0.5-1%) were correctly recalculated.