The use of detailed weather information in farm management Smart AgriMatics 2014, Paris Luka Honzak, Marko Novak, Vanja Blažica, Andrej Ceglar BO - MO, LTD. luka@bo-mo.si
Outlook Introduction Company, personnel Weather forecast How are the forecasts produced? Which data are used? What is the forecast model? Why regional models downscaling? How accurate are weather forecasts and how far ahead can the weather be predicted? Weather and agriculture The use of weather forecast in agriculture Application for FIspace trial
Company, personnel BO MO, LTD., Slovenia Established in 1994 Since 2010 working on IT, mostly on GIS and web platforms related to meteorology Group of experts in meteorology, physics and computer science Luka Honzak, BSc in Meteorology 2009 Marko Novak, MSc in Computer Science 2009 Vanja Blažica, PhD in Physics 2013 (BSc in Meteorology) Andrej Ceglar, PhD in Agronomy 2011 (BSc in Meteorology) References FP7: EuroGEOSS, FOODMETRES, FIspace Several national and international projects, collaborating with the Slovenian Environment Agency
How are the weather forecasts produced? WEATHER FORECAST MODEL OBSERVATIONS FORECAST
Observations SURFACE 69000 BUOY 9000 RADIOSONDES 600 AIRCRAFT 107000
Satellite observations POLAR-ORBITAL 600000 GEOSTATIONARY 250000 Etc. More than 35 million data collected every day, more than 95% of them are satellite data.
Weather forecast model Equations for 7 basic meteorological variables. Approximate solutions.
The computational grid Horizontal grid resolution 15 km Vertical grid 137 levels ECMWF European Centre for Medium-Range Weather Forecasts
How are the weather forecasts produced? WEATHER FORECAST MODEL ANALYSIS DATA ASSIMILATION PREVIOUS FORECAST FORECAST OBSERVATIONS
Downscaling Regional / Limited area model (LAM) Initial conditions and values at the boundaries from global model (nesting tehnique) Phenomena Convection Showers Local winds Better terrain and coast description Errors grow faster Few days ahead Nesting Usually increased resolution up to 3x Multiple nests possible (e.g. 2 nests, increase of resolution up to 9x) Increase of resolution for factor 2 -> increase of computational time for factor 8
Orography Downscaling is especially important on areas with complex orography. Resolution 120 km, 10 km, 2.5 km
Models Global ECMWF (EU) DWD GM (Germany) UKMET (UK) ARPREGE (France) GFS (USA) NOGAPS (USA) CMC (Canada) JMA (Japan) Regional/LAM ALADIN, ALARO, AROME (France, Central Europe) HIRLAM (Scandinavia) COSMO (Germany, Italy) UKMO (UK) WRF (USA)
Weather forecasts Resolution Low-resolution models (>100km) climate models Medium-resolution models (10-50 km) global models High-resolution models (<10 km) regional models Weather forecast types Nowcasting (<6h) Short range (<2-3 days) Medium range (<5-7 days) Long range (<2 weeks) Monthly, seasonal etc.
How accurate are weather forecasts and how far ahead can weather be predicted? Atmosphere is a chaotic system Small errors can lead to bigger errors in a few days Errors are not only spatial but also temporal Forecasts for 30 days in advance with 1h temporal resolution for exact locations == nonsense Predictability Theoretical limit for deterministic forecasts is 2 weeks In reality forecasts become worse even sooner: regional models loose valuable information after 3-4days, global after 8-10 days With further research and more observations these statistics improve
How accurate are weather forecasts and how far ahead can weather be predicted? Sources of uncertainty in weather forecast models Errors due to not completely known current state of the atmosphere Errors due to imperfections in the model formulation: Approximate mathematical methods to solve the equations Simplifications Incomplete theoretical knowledge Chaotic nature of the simulated processes Ensemble prediction Multiple weather forecasts using slightly different initial conditions (current state) Gives information about the stability (predictability) of the forecast
Users of weather products General public Hydrology Agriculture Transport (air, road, rail, water) Power companies Construction Engineering Civil protection Retail Defence Scientists Etc.
Weather and agriculture Weather plays an important role in agriculture. Influence on crop growth and yield development, incidence of pests and diseases, irrigation needs, fertilizer requirements and cultural field operations. Severe weather can cause physical crop damage and soil erosion. Weather during the harvest time may profoundly influence the quality of crop yield and physical operations, such as movements from the field and storage.
Weather and agriculture Weather varies spatially and temporally over an area. Time scale is important when considering the impact of weather on crop production. Weather over short time periods (daily to decadal) and year-to-year fluctuations at a place over the selected interval has to be considered for cropping. Variability of a weather parameter is crucially dependent on the time scale of interest. The effects of weather anomalies on crops usually build up slowly, but they often destabilize the agricultural production over large areas. Severe weather can lead to crop failure as well on short time scales (e.g. storms with strong wind and hail, intensive rainfall leading to flooding and water logging etc.).
Weather and agriculture Weather requirements for optimal crop growth and development depends on crop type and the growth stage. E.g. sensitivity of crop growth and development may vary between different stages of growth.
The use of weather forecast in agriculture Various agronomic practices, depending on weather forecast: Field preparation - prediction of exact time of rainfall occurence Sowing or planting temperature above and below soil surface, soil moisture, atmospheric humidity Application of agricultural chemicals air temperature, relative humidity, soil moisture, precipitation, wind speed and direction Irrigation air temperature and humidiy, solar radiation, precipitation, wind speed Weeding - rainfall Crop harvest and storage rainfall, air temperature, relative humidity and dew formation End product transportation Prevention of damage due to chilling, frost and freezes accurate and timely forecast of minimum air temperature Forestry operations, fishery operations, animal husbandry, etc.
The use of weather forecast in agriculture It is possible to adapt to or mitigate the effects of adverse weather in case when forecast of the expected weather can be provided in time. Agronomic strategies to cope with variable weather are available. Once the crop season starts, the crop-cultural practices are adopted to minimize the effects of mid-seasonal hazardous weather phenomena on the basis of expectation of their occurrences. The medium range weather forecasts enables farmers to organize and implement appropriate operations to cope with severe weather and therefore take advantage of the forecasted weather. Dissemination of weather forecasts to agricultural users needs to be quick with minimum possible temporal lag. Weather forecasts must ideally be issued for small areas In the case of well-organized weather systems the weather forecast can be generalized for larger areas. In other cases the area to which the weather forecasts will be applicable, must be unambiguously stated (e.g. specifying a location)
Formulation of weather scenario FIspace Open Call New partners will develop a set of applications demonstrating the usefulness of FIspace features. Start 1.4.2014. App: Formulation of weather scenario (Weather on Demand) A web service which delivers weather scenario for each requested location in the Netherlands. A high spatial resolution of 1 km and a temporal resolution of one hour. The time range is up to maximal one year before the actual date and 10 days after the actual date. User can request weather scenario as well as subscribe to receive them regularly. User can compare data from his own weather station to nearby interpolated fields from KNMI stations.
Formulation of weather scenarios Generation of 10-day forecast Weather forecasts for 3-4 days in advance: Dynamical downscaling of GFS predictions using WRF model. Final output on high resolution 6 km grid further interpolated to 1 km using geostatistical methodology universal kriging. Medium-range forecasts for 4-10 days in advance: ECMWF fields on 16 km resolution interpolated to 1 km. Historical meteorological observations: public available data of KNMI interpolated to 1 km grid. Variables Temperature, Relative Humidity, Rainfall, Pressure, Specific humidity, Incoming shortwave radiation - direct + diffuse, Sensible heat flux, Height of boundary layer, Wind speed at 10 m, ET0 following Penman- Monteith, ET0 following Makking
Formulation of weather scenario State of the Art weather forecast for agriculture use Working testing version available by the end of 2014 Web portal and Android App in 2015? Other Apps that will use this App: Bad weather alert (CIT DEVELOPMENT S.L.) Warns user for not foreseen unfavourable weather conditions in respect of spraying Phytopthora Advisory App Irrigation (based on water balance model)? Etc.
Forecast model (WRF) d01: 100x100 points, ~18km d02: 100x100 points, ~6km 42 vertical levels Test run: 2-day forecast computed in 7h on 4 cores
Thank you for your attention. Questions?