1.0.0b1 (October 4, 2019)¶
This is the first 1.0 beta release.
API Changes¶
- Add required argument interval_label to all model functions in
solarforecastarbiter.reference_forecasts.models
. (GH185) - Changed
resample()
keyward argument closed to label. (GH186) - Add deterministic forecast performance metrics (GH171)
- Add processed_forecast_observations argument to
template_report()
. (GH184) - Removed
rap_ghi_to_hourly_mean()
because subhourly RAP GHI data is not available. (GH186)
Enhancements¶
- Add command line interface option for reports. (GH168)
- Add classes
ProbabilisticForecastConstantValue
andProbabilisticForecast
todatamodel
to support probabilistic forecasts. (GH47) - Datamodel classes .from_dict constructors will now accept nested dicts. For example, a dict describing an Observation may now specify its site as a dict that describes a Site, rather than a Site object. (GH125)
- Documentation page for command line interface. (GH169)
- Add methods to
APISession
to support getting and posting data withProbabilisticForecastConstantValue
andProbabilisticForecast
classes. - Add probabilistic forecasts from NWP (GEFS) capability with two functions (GH54):
gefs_half_deg_to_hourly_mean()
,sort_gefs_frame()
- Add method
detect_clearsky_ghi()
to detect clear-sky periods using GHI data, for time intervals up to 15 minutes. (GH19)
Bug fixes¶
- Fix whats new documentation (GH143)
- Fix
run_persistence()
start, end time issue. (GH147) - Fix issue with negative GHI forecasts derived from GFS model due to complications with unmixing interval averages. (GH144)
- When computing reference forecast issue time of day, account for run_length when adjusting UTC to local time zone. (GH140)
- Add lower and upper limits to clearsky index calculations in
persistence_scalar_index()
function. Reduces large errors near sunrise and sunset. (GH107) - Address numpy, pandas deprecations. (GH156)
- Remove TODO, EXAMPLE text from reports template in favor of GitHub Issues. (GH167)
- Account for timezone in metrics/report generation. (GH164)
- Account for different timezones in
adjust_timeseries_for_interval_label()
with pandas >= 0.25.1. (GH173) - Bigger metrics graphics to avoid (but not yet totally prevent) label overlap. (GH163)
- Handle empty observation or forecast in current report metrics calc (GH178)
- Accept string or timelike objects to
get_forecast_values()
andget_observation_values()
(GH180) - Fix return/assignment of None in
daily_single_observation_validation()
andmake_latest_nwp_forecasts()
. (GH182) - Fix failure to pass user-specified limits to
check_irradiance_limits_QCRad()
. (GH182) - NWP reference forecasts now account for interval label. This also fixes
errors when computing forecasts with
interval_label='ending'
. (GH185) - Fix resampled/aligned observation/forecast labels in report. (GH184)
- Fix limits for temperature and wind data validation. (GH139)
- Make
Report
defaultfilter
attribute a tuple(QualityFlagFilter, )
instead of a singleQualityFlagFilter
. (GH166) - Fix bug in monthly, daily, hourly metrics calculation introduced by GH171. (GH207)
- Fix inefficient data loading for reports. (GH208)
- Fix documentation of window setting in
run_persistence()
. (GH203) - Reindex reference observation data to include NaNs when posting when appropriate (GH216)
Contributors¶
- Will Holmgren (@wholmgren)
- Leland Boeman (@lboeman)
- Cliff Hansen (@cwhanse)
- Tony Lorenzo (@alorenzo175)
- Justin Sharp (@MrWindAndSolar)
- Aidan Tuohy
- Adam Wigington (@awig)
- David Larson (@dplarson)