WEATHER FORECASTS AT THE WHOLE-PLANT LEVEL: IMPLICATIONS FOR ESTIMATING DISEASE RISK WITH GLOBAL CLIMATE MODELS
RC SEEM1, RD MAGAREY1, JW ZACK2 and JM RUSSO3
1New York State Agricultural Experiment Station, Cornell University, Geneva, NY 14456, USA; 2MESO Inc., Troy, NY 12180, USA; 3ZedX Inc., Boalsburg, PA 16827, USA
Background and objectives
Global climate models (GCMs) have been developed to assess the impacts of potential global climate change. However, these models do not provide specific weather information at the whole-plant level and thus provide only very gross estimates of conditions that affect plant and disease development. Over the past 4 years, we have been developing methods to hierarchically define current and forecast weather conditions down to the whole-plant level, based on nested high-resolution atmospheric numerical models. Success in this effort leads us to extend this method to GCMs where factors such as temperature, rainfall, relative humidity and surface wetness can be estimated confidently within plant and crop canopies.
Materials and methods
In the hierarchical design of weather information systems, relatively coarse-mesh models run by federal bureaux of meteorology provide the first level of the hierarchy. These models provide information about regional weather conditions. The next level of the hierarchy, high-resolution mesoscale models, provides information typically to 1 km2, or the farm level. Finally, plant canopy models of energy balance can bring the information down to the whole-plant level. Our approach is based on a two-step modelling process to bring weather information to the farm and whole-plant levels. Firstly, model output from the coarse-mesh numeric model of the US National Weather Service (NWS) was input into the Mesoscale Atmospheric Simulation System (MASS) , a high-resolution atmospheric model. Secondly, a one-dimensional energy balance model was created for a grape canopy using the approach of Tanner and Fuchs , by which output from MASS was processed to estimate temperature, relative humidity, precipitation and surface wetness duration within the grape canopy. The canopy model was parameterized from laboratory studies. Model estimates were validated under field conditions over 2 years at multiple locations.
Results and conclusions
The hierarchical weather information system was successfully linked; current and 24-h forecast weather conditions in an individual grape canopy can be obtained from the data supplied by regional-scale operational weather prediction models such as those run by the US NWS. Under validation, 24-h forecast temperature estimates had an error of 2.0°C from actual. In 1997, 14 events of surface wetness were monitored visually and with instruments. Mean duration of surface wetness was 6.9 h, while the model predicted 5.4 h (absolute error 2.1 h) and sensors measured 5.6 h (absolute error 1.4 h). Additional validation has been completed in the USA and Australia.
With the ability to model whole-plant conditions from regional-scale data, and the ability to represent that information for large areas, we can generate risk maps for the likelihood of infection by diseases, such as primary infection for grape powdery mildew or downy mildew. Further, it is possible to direct output from a GCM into our hierarchical system to develop similar risk maps, but based on estimated conditions under climate change. Such maps would allow assessment not only of changes of disease risk within regions known to harbour the pathogens, but also of the likelihood of disease moving to regions previously free of the pathogens because of unfavourable environmental conditions for disease development.
1. Kaplan ML, Zack JW, Wong VC, Tuccillo JJ, 1982. Mon. Weath. Rev. 110, 1564-1590.
2. Tanner CB, Fuchs M, 1989. Journal of Geophysical Research 73, 1299-1304.