Using MODAWEC to generate daily weather data for the EPIC model
Introduction
Over the past two decades, the Environmental Policy Integrated Climate (EPIC, originally known as Erosion Productivity Impact Calculator) model has played an important role in the agricultural and environmental studies in the U.S. and in the other regions of the world. The EPIC model is a field-scale model that is designed to simulate drainage areas characterized by homogeneous weather, soil, landscape, crop rotation, and management system parameters. It was first developed in 1981 to support assessments of soil erosion impacts on soil productivity in the U.S. (Williams et al., 1984). Since then, it has continuously been developed by integrating and improving a number of additional functions including water quality, atmospheric CO2 change, and carbon cycling routines. The model has been applied in a wide range of studies in agriculture, meteorology, and environment, e.g. crop growth and yield (Williams et al., 1989, Easterling et al., 1996), impacts of climate change (Easterling et al., 1992, Brown and Rosenberg, 1997), nutrient cycling and nutrient loss (Jackson et al., 1994, Pierson et al., 2001), wind and water erosion (Potter et al., 1998, Bhuyan et al., 2002), pesticide losses (Sabbagh et al., 1991, Williams et al., 1992), impacts of irrigation on crop yields (Cabelguenne et al., 1995, Rinaldi, 2001), soil temperature (Potter and Williams, 1994, Roloff et al., 1998), soil carbon sequestration (Lee et al., 1996, Potter et al., 2004), and economic–environmental analysis (Bernardo, 1993a, Bernardo, 1993b, Kurkalova et al., 2004). Partly due to its good performance, the EPIC model has been applied in several regional, national and even global assessments. For example, a “spatial EPIC” system (Priya and Shibasaki, 2001) was developed to assess the national crop productivity in India. GIS-based EPIC models, which integrate EPIC with a geographic information system (GIS), were also used to study crop yield with high spatial resolutions for China (Liu et al., 2007a), for Africa (Liu et al., in press) and for the entire world (Tan and Shibasaki, 2003, Liu et al., 2007b, Liu et al., 2008, Liu, 2009).
Daily weather data are needed for the simulation of most processes in the EPIC model, but such data are often not available or not complete in many parts of the world. For example, so far, one of the most comprehensive daily weather data products, the Global Surface Summary of the Day produced by the National Climatic Data Center (NCDC), covers historical data of over 10,000 stations from 1929 to the present, with data from 1973 to the present being the most complete (http://www.ncdc.noaa.gov). For this dataset, there is an uneven distribution of meteorological stations among countries with sparse stations in many underdeveloped countries. Furthermore, the daily data are often not complete with many missing data in individual stations. In addition, NCDC does not provide projected future weather data for these stations.
When not available or not complete, the daily weather data can be generated with EPIC's built-in “weather generator” (WXGEN). WXGEN incorporates a first-order Markov chain technique for a wet or dry day decision. When a wet day is generated, a skewed normal distribution is used to generate the amount of daily precipitation. WXGEN first independently generates precipitation for a day. Maximum temperature, minimum temperature, solar radiation and relative humidity are generated on the presence or absence of rain for the day. Daily wind speed is generated independently. Detailed description of WXGEN can be found in Sharpley and Williams (1990). The inputs to WXGEN are several monthly statistics taken from long-term daily weather records. Monthly statistics such as monthly skew coefficient and monthly probability of wet day after dry day or wet day are difficult to obtain without daily weather data. When the necessary monthly statistics are available, WXGEN is very useful in simulating daily weather sequences that have statistical properties similar to those of measured weather in the same region. It can provide any number of equally likely weather sequences for use in evaluating management strategies under varying climatic conditions. Also, it can repeat the same weather sequence of any length (hundreds of years) as many times as needed in evaluating various management strategies under the same climatic conditions. However, since WXGEN is a stochastic model, the generated weather sequences do not resemble measured weather records year to year, although their long-term statistical properties are similar. Daily time step models like EPIC require daily weather data, but these data cannot be generated accurately for individual years by WXGEN.
Monthly weather data are easier to obtain than daily weather data. For example, monthly precipitation, maximum and minimum temperature, and wet days are available on a global scale with a spatial resolution of 30 arc-min (about 50 × 50 km in each grid cell near the equator) for 1901–2000 through the Climatic Research Unit (CRU) at the University of East Anglia (Mitchell et al., 2004). The Tyndall Centre for Climate Change Research (TYN) from the same university provides monthly variations of the above climate data for 16 different climate scenarios for 2001–2100 (Mitchell et al., 2004). These monthly data are valuable for conducting past, current and future global environmental assessments. However, they cannot be used directly by EPIC because the model operates on a daily time step. Since only monthly weather is available in many locations, there is a need for a method for converting monthly data to daily data.
The purpose of this study is to develop a stand-alone weather generator MODAWEC (MOnthly to DAily WEather Converter) for the EPIC model (EPIC0509). As our main interests are crop yield and crop water use, one important objective of model development is to generate reliable daily weather data for the EPIC model to simulate crop yield and crop evapotranspiration (ET). A case study is provided to test the reliability.
Section snippets
The MODAWEC model
The MODAWEC model converts monthly precipitation (in mm) and maximum and minimum temperature (in °C) to daily values while preserving the monthly totals and averages. The main inputs of the MODAWEC model include monthly precipitation, monthly wet days, and monthly maximum and minimum temperature in each year. The outputs are daily precipitation, daily maximum temperature, and daily minimum temperature. The flowchart of the MODAWEC model is depicted in Fig. 1. According to the classification by
Site description and model parameterization
The weather, soil, and management data used in this test case were from a long-term experiment conducted at the Arlington Agricultural Research Station of the University of Wisconsin in the south central Wisconsin (43° 18′ N, 89° 21′ W). The station is located on an extended plain with 1%–2% slope on a Plano silt loam soil (fine-silty, mixed, mesic, Typic Argiudoll). The long-term experiment was established in 1958 in order to evaluate the response of continuous corn (Zea mays L.) to nitrogen
Conclusions
A MODAWEC model has been developed to generate daily precipitation and maximum and minimum temperature with monthly weather data. The results show that the simulated crop yield or ET does not differ significantly when the measured daily weather data or the generated daily weather data by the MODAWEC model are used. The MODAWEC model enables the application of the EPIC model in regions where only monthly data are available. Particularly, for global level studies, daily weather data are often
Acknowledgements
This study was supported by the European Commission within the GEO-BENE project framework (Global Earth Observation–Benefit Estimation: Now, Next and Emerging, Proposal No. 037063). Special thanks are given to Monireh Faramarzi and Jafet Andersson from Eawag for their advice on sensitivity analysis in this paper.
References (48)
- et al.
Weather analogue: a tool for real-time prediction of daily weather data realizations based on a modified k-nearest neighbor approach
Environmental Modelling and Software
(2008) - et al.
Sensitivity of crop yield and water use to change in a range of climatic factors and CO2 concentrations: a simulation study applying EPIC to the central United States
Agricultural and Forest Meteorology
(1997) - et al.
Simulations of crop response to climate change: effects with present technology and no adjustments (the “dumb farmer” scenario)
Agricultural and Forest Meteorology
(1992) - et al.
A daily weather generator for use in climate change studies
Environmental Modelling and Software
(2007) - et al.
Sensitivity of the US corn belt to climate change and elevated CO2: II. Soil erosion and organic carbon
Agricultural Systems
(1996) A GIS-based tool for modelling large-scale crop–water relations
Environmental Modeling & Software
(2009)- et al.
GEPIC – modelling wheat yield and crop water productivity with high resolution on a global scale
Agricultural Systems
(2007) - et al.
National spatial crop yield simulation using GIS-based crop production model
Ecological Modelling
(2001) Application of EPIC model for irrigation scheduling of sunflower in Southern Italy
Agricultural Water Management
(2001)- et al.
First-order Markov chain approach to wind speed modelling
Journal of Wind Engineering and Industrial Aerodynamics
(2001)