International Research Journal of Environment Sciences________________________________ ISSN 2319–1414Vol. 2(10), 63-70, October (2013) Int. Res. J. Environment Sci. International Science Congress Association       
63
 
Comparing Maize Potential yields Predicted using Actual and Interpolated Weather data in UgandaNyombi K.1* and Balimunsi H.Dept. of Environmental management, Makerere University, Kampala, UGANDA Dept. of Agricultural and Bio-systems Engineering, Makerere University, Kampala, UGANDAAvailable online at: 
www.isca.in, www.isca.me 
Received 15th September 2013, revised 9th October 2013, accepted 19th October 2013 AbstractAcquisition of measured weather data in Uganda for crop growth modeling is a challenge due to the low number of weather stations. Often, rainfall, maximum and minimum temperatures are measured. Total solar radiation is only measured at few weather stations due to shortage of sunshine duration recorders, the time graded paper strips or the newer automated weather stations (AWOS). A number of agencies do fill this void and provide on-line interpolated daily weather data to enable long-term simulations. A dynamic crop growth model CERES within the DSSAT modeling suite was used in order to evaluate simulation results obtained using actual and interpolated weather data from Kawanda, Central and Mbarara, south-western Uganda. Generic coefficients for very short, short, medium and long season maize varieties with in DSSAT were used. Farmer planting dates for the two cropping seasons were used to start the simulation. Results showed that at Kawanda, the average actual and interpolated maximum temperature were comparable, while at Mbarara, maximum temperatures were underestimated with a deviation of 3°C. At both sites, actual and interpolated minimum temperatures were comparable. The average actual total solar radiation at Kawanda was lower, probably indicating a shift in the AWOS radiation sensors. At Mbarara, the interpolated and measured values are comparable, indicating that the solarimeter method is still very reliable. RMSEs between actual and predicted potential yields at Kawanda were larger; very short (942 kg ha1), short (1176 kg ha1), medium (1864 kg ha1) and long season maize (3055 kg ha1). Actual radiation measurements at this site were lower, which emphasizes the importance of re-calibrating radiation sensors at least every two years. At Mbarara, the RMSEs for very short (418 kg ha1), short (618 kg ha1), medium (1056 kg ha1) and long season (1896 kg ha1) were low and acceptable. Interpolated data from the NASA can be used to predict potential yields and for long-term simulations in absence of measured weather data.    Keywords: Potential yields, interpolated, weather data, radiation, modeling Introduction Solar radiation, minimum and maximum temperature data are crucial for the simulation of potential crop production. Potential yield is determined by radiation, temperature, crop physiology and canopy characteristics, with water and nutrients not limiting, and in absence of diseases, pests and weeds. This value indicates the maximum possible dry matter production and will vary with the same variety at different locations due to temperature and total solar radiation variations. The yield level may indicate the suitability of the crop to a location. Temperature determines the development rate, with higher temperature implying a faster development rate, provided it is still within the plant’s optimum temperature range. The total incoming radiation (MJ m2 d1), the radiation use efficiency (RUE; g DM MJ1) and the leaf area index (ha leaf ha1 soil) determine the radiation intercepted and the amount of dry matter produced on particular day. Climate change is likely to adversely affect communities whose livelihoods entirely depend on crop production. This is likely to result from increases in temperature above the physiological optimum for the affected crops, making the areas unsuitable and hence a production shift to new areas. According to the predictions, average rainfall in Uganda will increase with uneven distribution over the country and average temperatures are estimated to increase by 1.5ºC in the next 20 years. Coffee a major cash crop is sensitive to temperature and banana a staple food is also sensitive to drought stress and temperature. Research in the developing countries shows that the impacts of climate change on annual crop production can be reduced by adopting short and medium season varieties. However, due to the reduced growing cycle, the yields tend to be lower. In order to carry out climate change assessments, accurate weather data is required for the base year. In Uganda, it is often difficult to obtain long-term weather data to run crop growth models. This is attributed to obsolete equipment, the past political instability and the poor coverage of the weather stations. At some stations, rainfall, maximum and minimum temperatures are measured but total solar radiation is only measured at few weather stations using the traditional solarimeters. The conventional gauging is good, but is labour intensive. For solar radiation, the time graded paper stripshave 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
64
to be inserted each morning, the total sunshine hours calculated from the burnt sections and converted to daily total radiation (DTR,MJ m21). Automatic weather observation systems (AWOS) though convenient are still few due to the cost. As a country where agriculture supports over 70% of the population, accurate and timely high quality weather data must be collected and passed onto farmers to allow rational production decision making.  A number of sites or agencies such as the National Aeronautics and Space Administration (NASA) do provide long-term interpolated data to allow long-term simulations. However, there have been issues related to the accuracy of the interpolated valuesas compared with the measured or observed values. The inaccuracies in weather data may result in overestimations of yield of over 35%. The objectives of the study were to; i. compare the radiation, minimum and maximum temperatures measured from the sites and interpolated data for two sites in Uganda ii. compare the simulated potential yields and crop cycle durations using actual and interpolated weather data from NASA.  Material and Methods Weather data: Actual weather data was obtained at Kawanda Research station, central Uganda (0°25'N, 32°30'E, 1147 meters above sea level) and Mbarara experimental farm weather station, southwest Uganda (0°61'S, 30°65'E, 1405 meters above sea level). At Kawanda, measured weather data (daily total radiation - MJ m2 d1, rainfall – mmd1, minimum and maximum temperature -°C) were obtained for the years 20002006 and at Mbarara data was obtained for years 19931999. At Mbarara, measured daily sunshine hours were converted to daily total radiation using the Angstrom formula, which relates solar radiation to extraterrestrial radiation and relative sunshine duration.
Sssa
n
RabR
N


=+


Where S is the daily total solar radiation (MJ m2 d1), s + is the fraction of extraterrestrial radiation reaching the earth on clear days (n=N),n is the actual sunshine duration (sunshine hours),  is the maximum possible duration of sunshine or daylight hours (sunshine hours), n/N is the relative sunshine duration, is the extraterrestrial radiation (MJ m2 d1) and s is the regression constant, expressing the proportion of extraterrestrial radiation reaching the earth on overcast days (n=0). Since no actual direct solar radiation data are available for Mbarara and no calibration has been carried out for improved s  and  parameters, the values of 0.25 and 0.50 respectively are recommended by the Food Agricultural Organisation (FAO) and were used. For this location close to the equator, the maximum number of sunshine hours (N) is 12.  Interpolated weather data were obtained from the NASA Climatology Resource for Agroclimatology Daily Averaged http://power.larc.nasa.gov/cgibin/cgiwrap/solar/agro.cgi?email=agroclim@larc.nasa.gov.  For the NASA site, the GPS coordinates of each location were used to obtain the interpolated data. All weather files were formulated in Microsoft Excel and weather stations were created using the weatherman tool in DSSAT10. In order to complete the standard format of weather files in Weatherman, rainfall was also added but it is not required in simulating potential production.The potential yields and the crop cycle durations obtained using actual and interpolated weather data were compared using the root mean square error (RMSE). 
2
1
()
ii
ap
RMSEWhere  is the number of observations,  is the potential yield value obtained using actual data and i is the potential yield value obtained using interpolated weather data.  The DSSAT modeling suite: The CERES-Maize model was first introduced without N-supply subroutines with N assumed to be non-limiting11. It has been improved over the years and included in the Decision Support System for Agrotechnology Transfer– Crop Simulation Model - DSSAT-CSM, a collection of models that simulate growth of over 15 crops10. CERES maize simulates growth on a daily time step in response to soil, weather, environmental conditions, fertilizer rates / timing / placement, and other field management strategies. It simulates plant phenological development (emergence, end of juvenile stage, silking and physiological maturity), biomass accumulation (radiation use efficiency approach) and partitioning as a function of the development stage and final dry matter (grain yield and stovers). The CSM-CERES-Maize has been used to predict potential yields12 and study effects of climate change13. Crop coefficients: Generic parameters from the DSSAT database for the very short, short, medium varieties and a long season hybrid (PIO 3475 original) were used to fully capture the different growth durations and the possible variations in temperature during the growth period. The base or minimum temperature for maize growth is 8°C and the optimum temperature was 34°C, at which maximum development rate occurs during the vegetative and reproductive stages as used in the CERES maize model. The light extinction coefficient (ha soil ha1 leaf) was 0.85 and the radiation use efficiency was 4.2 g DM MJ1. The planting dates used by the farmers were used to initiate the simulation runs, 3rd March for season 1 and 3rdAugust for season 2. The planting method used was dry seed at a depth of 5cm. The plant spacing was 0.75×0.25m, giving a plant population of 53,333 plants ha1. The runs were done for two seasons a year for seven years. Results and DiscussionWeather data: At Kawanda, the average actual maximum temperature was 27.2°C (17.735.7°C), and interpolated 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
65
average maximum temperature was 25.74°C (19.333.3°C). This indicates a deviation of 1.46°C.The average actual minimum temperature was 16°C (6.323.5°C), and interpolated average minimum temperature was 16.72°C (8.721.2°C). The average total solar radiation was 15.10 MJ m2 d1 (1.0525.89 MJ m2 d1), and interpolated total solar radiation was 18.55 MJ 2 d1 (3.728.7 MJ m2 d1). Figure 1 shows the yearly variations in maximum temperature, minimum temperature and daily total solar radiation at Kawanda, central Uganda.  Table-1 Generic coefficients from the DSSAT v 4.5 shell used to evaluate the weather data sources. P1- Heat sum from seedling emergence to the end of the juvenile phase (°Cd); P2 – Effect of photoperiod on development; P5 – heat sum from silking to physiological maturity (°Cd); G2 - maximum possible number of kernels per plant; G3 - kernel filling rate during the linear grain filling stage and under optimum conditions (mg kernel1 day1); PHINT - Phylochron interval between successive leaf tip appearances (°Cd) Season length P1 (°Cd) P2 P5 (°Cd) G2 G3 (mg kernel1 day1) PHINT (°Cd) 
PIO 3475 orginal – Long season hybrid 220 0.70 850 907 9.90 38.90 
Medium 200 0.30 800 700 8.50 38.90 
Short 110 0.30 680 820 6.60 38.90 
Very short 5 0.30 680 820 6.60 38.90 
 
Figure-1 Actual and interpolated maximum temperature (A), actual and interpolated minimum temperature (B), and actual and interpolated total solar radiation (C) at Kawanda, central Uganda 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
66
At Mbarara, the average actual maximum temperature was 27°C (18.533.2°C), and interpolated average maximum temperature was 24°C (18.430.7°C), indicating that maximum temperatures were underestimated with a deviation of 3°C. The average actual minimum temperature was 15°C (10.719.9°C), and interpolated average minimum temperature was 16°C (10.321.8°C), implying that minimum temperatures were over estimated with a deviation of +1°C.  The average actual total solar radiation was 17.28 MJ m2 d1 (8.3526.87 MJ m2 d1), and interpolated total solar radiation was 18.47 MJ m2 d1(5.028.2 MJ m2 d1).  Figure 2 shows the yearly variations in maximum temperature, minimum temperature and daily total solar radiation at Mbarara, south-western Uganda. In comparison, the actual maximum temperatures at Kawanda were higher as compared with Mbarara due to altitude effect, but the averages were comparable. As expected, the minimum temperatures at Mbarara were lower on average as compared with Kawanda.  In general, the interpolated maximum temperatures are much lower than the observed values, but the minimum temperatures are comparable. In addition, interpolated solar radiation values are much higher than measured values, especially at Kawanda. At Kawanda, the lower average solar radiation could be attributed to reduced sensitivity or a drift in the sensitivity of the radiation sensors of the AWOS14. Often, drifts of 15% per annum are observed hence re-calibration of radiation sensors at least every two years. Though sunshine hours were converted to daily total solar radiation, actual and interpolated values were comparable, indicating that solarimeters are still very reliable15.  Simulated potential yields and days to maturity at Kawanda, central Uganda: The average predicted potential yields over 14 cropping seasons for very short season maize using actual and interpolated data were 2830kg ha1 and 3679kg ha1with a RMSE of 942kg ha1. Potential yield is under estimated by only 849 kg ha1. The average days to maturity were 82 and 85, with a RMSE of 6.69 days. For the short season maize, the average predicted potential yields using actual and interpolated data were 3916kg ha1 and 4950kg ha1 with a RMSE of 1176kg ha1. Potential yield is under estimated by 1034 kg ha1. The average days to maturity were 99 and 103, with a RMSE of 9 days. For both the very short and short duration maize varieties, interpolated data can be used with quite acceptable accuracy. The higher RMSE could be attributed to lower measured daily total radiation and increasing variation with increasing crop cycle duration. As compared with Mbarara, actual values tend to under predict yield, probably due to radiation sensor drifts14.  The average predicted potential yields for medium season maize using actual and interpolated data were 7053kg ha1 and 8384kg ha1with a RMSE of 1864kg ha1. Potential yield is under estimated by 1331 kg ha1. The average days to maturity were 122 and 126, with a RMSE of about 10 days. For the long season hybrid maize, the average predicted potential yields using actual and interpolated data were 10730kg ha1 and 13125kg ha1 with a RMSE of 3055kg ha1. Potential yield is under estimated by 2394 kg ha1. The average days to maturity were 129 and 133, with a RMSE of about 12 days. For both the medium and long season maize varieties, the increase in crop cycle duration leads to increases in the RMSEs and yield under estimations as a result of lower radiation values. The lower radiation measurements resulted in an 18% under estimation in yield. Although this value is lower than 35% reported, it is still large.   Simulated potential yields and days to maturity at Mbarara, southwestern Uganda: The average predicted potential yields for very short season maize over 14 cropping seasons using actual and interpolated data were 3676 kg ha1 and 3865 kg ha1 with a RMSE of 418 kg ha1. Potential yield is over estimated by only 189 kg ha1. The average days to maturity were 85 and 87, with a RMSE of 4 days.  For the short season maize, the average predicted potential yields using actual and interpolated data were 4784kg ha1 and 5265kg ha1with a RMSE of 618 kg ha1. Potential yield is over estimated by 481 kg ha1. The average days to maturity were 102 and 105, with a RMSE of 4 days. For both the very short and short duration maize varieties, interpolated data can be used with quite acceptable accuracy. The higher predicted potential yield values could be attributed to more solar radiation and lower maximum temperature, hence more radiation intercepted per day and slightly longer growth cycle. Higher predicted potential yields were also report for cereals, using interpolated data from NOAA16.  The average predicted potential yields for medium season maize using actual and interpolated data were 7933kg ha1 and 8665 kg ha1with a RMSE of 1056kg ha1. Potential yield is over estimated by 732 kg ha1. The average days to maturity were 126 and 129, with a RMSE of about 5 days. For the long season hybrid maize, the average predicted potential yields using actual and interpolated data were 12350kg ha1 and 14027kg ha1with a RMSE of 1896kg ha1. Potential yield is over estimated by 1677 kg ha1. The average days to maturity were 132 and 136, with a RMSE of about 5 days. For both the medium and long season maize varieties, the increase in crop cycle duration leads to increases in the RMSEs and yield over estimations.  Three gridded weather databases (GWDs), including NASA power were evaluated as compared to actual weather data (CWD) for their ability to predict potential and water limited rice yields in China, maize in USA and wheat in Germany17. Results showed that agreement between CWD and GWD results was poor, but when observed weather data from stations in the NOAA database were combined with daily total radiation from the NASA power database, simulated potential and water-limited yields were in agreement with those using CWD data with RMSE 1219% of the absolute mean. This implies that the 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
67
radiation data from NASA is quite good as noted from the data at Mbarara. Although, the agreement reported above was poor, we obtained fairly good results in this study. Overall, in absence of actual long-term weather data, interpolated from the NASA power database can be used. 
Figure-2 Actual and interpolated maximum temperature (A), actual and interpolated minimum temperature (B), and actual and interpolated total solar radiation (C) at Mbarara, southwestern Uganda 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
68
Table-2 Simulated potential yields over two cropping seasons (20002008) for very short, short, medium and long season maize varieties at Kawanda using actual and interpolated weather data Year Season Very short season (Grain yield, kg ha1) Short season (Grain yield, kg ha1) Medium season (Grain yield, kg ha1) PIO 3475 original – Long season hybrid (Grain yield, kg ha1) 
Actual Interpolated Actual Interpolated Actual Interpolated Actual Interpolated 
2000 1 3147 3652 3843 4724 7105 6792 10955 10839 
2 3033 3568 3558 4496 6201 7298 9805 11516 
2001 1 3548 4093 4055 5190 8378 8893 10915 14286 
2 2584 3375 4266 4779 9019 9202 13642 13508 
2002 1 3157 3861 4447 4639 8625 8493 13578 13591 
2 3133 3388 5100 5281 9050 8193 13886 13056 
2003 1 2326 4162 2854 4279 6174 7803 8909 12757 
2 2198 3492 3584 4836 6394 8258 10555 13141 
2004 1 2611 3304 4179 5264 6334 9572 10263 15825 
2 3315 3726 3889 4914 8311 9318 11322 13643 
2005 1 2988 4241 3478 4954 6615 8692 9815 13431 
2 2331 3310 3491 5848 5002 7948 8404 12559 
2006 1 2645 3627 4604 5115 5578 7485 8489 12127 
2 2615 3718 3485 4992 5964 9440 9687 13472 
Mean  2830 3679 3916 4950 7053 8384 10730 13125 
RMSE  942 1176 1864 3055 
Table-3 Simulated days to maturity over two cropping seasons (20002008) for very short, short, medium and long season maize varieties at Kawanda using actual and interpolated weather data Year Season Very short season (Days to maturity) Short season (Days to maturity) Medium season (Days to maturity) PIO 3475 original – Long season hybrid (Days to maturity) 
Actual Interpolated Actual Interpolated Actual Interpolated Actual Interpolated 
2000 1 81 83 94 96 117 118 124 125 
2 82 82 99 101 120 123 128 131 
2001 1 85 87 103 106 126 130 134 138 
2 91 86 114 104 139 128 148 133 
2002 1 88 85 109 101 136 126 146 132 
2 94 85 119 104 145 126 151 131 
2003 1 77 87 91 101 113 122 121 131 
2 82 85 97 101 120 122 126 130 
2004 1 78 82 91 103 113 126 119 135 
2 81 89 99 107 120 130 125 138 
2005 1 79 86 96 101 118 127 126 134 
2 83 88 99 105 119 128 124 136 
2006 1 78 89 95 107 117 132 123 138 
2 78 89 93 107 114 131 123 137 
Mean  82 85 100 103 122 126 129 133 
RMSE  6.69 9.0 10.8 12 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
69
Table-4 Simulated potential yields over two cropping seasons (19931998) for very short, short, medium and long season maize varieties at Mbarara using actual and interpolated weather data Year Season Very short season (Grain yield, kg ha1) Short season (Grain yield, kg ha1) Medium season (Grain yield, kg ha1) PIO 3475 original – Long season hybrid (Grain yield, kg ha1) 
Actual Interpolated Actual Interpolated Actual Interpolated Actual Interpolated 
1993 1 3660 3475 4790 4851 7365 7683 12178 12416 
2 4272 3973 5621 6248 8034 9663 12662 15441 
1994 1 4198 3624 4569 4361 8179 8303 13400 13564 
2 3363 4029 4486 5372 8873 9353 12950 14319 
1995 1 3640 3945 4815 5141 8691 9073 11903 14316 
2 3040 3712 5097 5569 7962 9456 12728 15517 
1996 1 3427 3397 4926 5050 8830 8138 12049 12697 
2 3903 4414 5137 5807 8050 9270 12894 14583 
1997 1 3702 3889 4381 4651 7757 8374 12612 14406 
2 3050 3655 4721 5819 7196 9048 10762 13283 
1998 1 4045 4130 4580 4921 6381 6874 10431 12126 
2 3882 3898 4437 5664 8976 8521 13207 14817 
1999 1 3692 3815 4749 4972 7014 8662 12103 14897 
2 3589 4154 4673 5291 7757 8897 13027 14006 
Mean  3676 3865 4784 5265 7933 8665 12350 14027 
RMSE  418 618 1056 1896 
Table-5 Simulated days to maturity over two cropping seasons (1993-1998) for very short, short, medium and long season maize varieties at Mbarara using actual and interpolated weather data Year Season Very short season (Days to maturity) Short season (Days to maturity) Medium season (Days to maturity) PIO 3475 original – Long season hybrid (Days to maturity) 
Actual Interpolated Actual Interpolated Actual Interpolated Actual Interpolated 
1993 1 84 84 100 100 126 121 132 129 
2 84 79 101 100 123 123 129 132 
1994 1 87 86 101 101 126 125 133 133 
2 85 86 105 105 130 131 137 141 
1995 1 84 91 104 108 126 132 132 138 
2 84 87 104 106 126 132 134 139 
1996 1 86 89 104 108 127 131 133 139 
2 86 91 102 108 128 134 137 141 
1997 1 90 93 104 109 127 132 134 139 
2 81 83 97 102 123 129 129 135 
1998 1 84 88 96 103 120 122 126 130 
2 89 83 103 104 125 125 130 133 
1999 1 83 91 102 109 123 133 129 140 
2 86 90 106 109 129 133 135 142 
Mean  85 87 102 105 126 129 132 136 
RMSE  4.34 4.06 4.87 5.36 
Conclusion At Kawanda, the average actual and interpolated maximum temperature were comparable with a deviation of 1.46°C, while at Mbarara, maximum temperatures were underestimated with a deviation of 3°C. At both sites, actual and interpolated minimum temperatures were comparable. This indicates that maximum temperatures were not captured well at Mbarara. The average actual total solar radiation at Kawanda was 15.10 MJ 2 d1 and interpolated total solar radiation was 18.55 MJ m21, indicating that actual values were lower. At this site, an AWOS was in use, probably the shift in radiation sensors could 
International Research Journal of Environment Sciences______________________________________________ ISSN 2319–1414 Vol. 2(10), 63-70, October (2013)     Int. Res. J. Environment Sci. International Science Congress Association             
70
be the factor.  At Mbarara, sunshine hours from the solarimeter were converted to daily total radiation. The interpolated and measured values are comparable, indicating that this old method is still very reliable.  The larger RMSEs between actual and predicted potential yields at Kawanda may be attributed to the lower daily total radiation measurement. This indicates the importance of checking the radiation sensors or sending them back to the manufacturer every two years for re-calibration. At Mbarara, the RMSEs for very short (418 kg ha1), short (618 kg ha1), medium (1056 kg ha1) and long season (1896 kg ha1) were low and acceptable. We conclude that interpolated data from the NASA Climatology Resource for Agroclimatology Daily Averaged can be used to predict potential yields and for long-term simulations in absence of measured data. References 
1.Wassmann R., Jagadish S.V. K., Heuer S., IsmailA., Redona E., Serraj R., Singh R.K., Howell G.,Pathak H. and Sumeth K., Climate ChangeAecting Rice Production: The Physiological andAgronomic Basis for Possible Adaptation Strategies, Adv. Agron., 101, 59–122 (2009)
2.Lövenstein H., LantingaE.A., Rabbinge R. and van Keulen H., Principles of production ecology. Department of Theoretical production ecology and Centre for Agrobiological Research (CABO-DLO), Wageningen, The Netherlands,  7 (1995)3.Adger W.N., Huq S., Brown K., Conway D. and Hulme M., Adaptation to climate change in the developing world, Progress Develop. Stud., 3(3), 179195 (2003) 4.Hepworth N. and Goulden M., Climate Change in Uganda: Understanding the implications and upraising response, LTS International Edinburg (2008)5.van Asten P.J.A., Fermont A.M. and Taulya G., Drought is a major yield loss factor for rain-fed East Africa highland banana, Agric. Water Manag., 98(4), 541552 (2011)6.Traerup S.L.M. and Mertz O., Rainfall variability and household coping strategies in northern Tanzania: a motivation for district-level strategies, Reg. Environ. Change, 11(3), 471481 (2011)7.Stampone M.D., Hartter J., Chapman C.A. and Ryan S.J., Trends and variability in localized precipitation around Kibale National Park, western Uganda, Africa, Res. J Environ. Earth Scie.3(1), 1423 (2011)8.Nonhebel S., Inaccuracies in weather data and their effects on crop growth simulation results, 1. Potential production, Climate Res. , 4760 (1994) 9.Angstrom, A. Solar and terrestrial radiation, Q.J.R. Meteorol.Soc., 50,121125 (1924) 10.Hoogenboom G., Jones J.W., Porter C.H., Wilkens P.W., Boote K.J., Batchelor W.D., Hunt L.A. and Tsuji G.Y. (Eds). Decision Support System for Agrotechnology Transfer Version 4.0. Vol. 1. University of Hawaii, Honolulu, HI (2003)11.Jones C.A. and Kiniry J.R., CERES-Maize, a simulation model of maize growth and development, Texas A&M University Press, College Station (1986)12.Walker N.J. and Schulze R.E., An assessment of sustainable maize productdifferent management and climate scenarios for smallholder agro-ecosystemNatal, South Africa, Phys. Chem. Earth, 31, 9951002 (2006)13.Wasige J.E., Assessment of the impact of climate change and climate variability on crop production in Uganda. Project report submitted to the Global change system for analysis, research and training (START) and the National science foundation (NFS) (2009)14.Davis Solar radiation sensor, standard, industrial and vantage versions. Rev C manual (1/12/01), Davis instruments corporation, Hayward, CA, USA (2000)15.Jennings S.B., Brown N.D. and Sheil D., Assessing forest canopies and understory illumination: canopy closure, canopy cover and other measures, Forestry, 72(1), 6073 (1999)16.Van Wart J., Kersebaumb K.C., Peng S., Milnera M. and Cassman K.G., Estimating crop yield potential at regional to national scales, Field Crops Res., 143, 3443 (2013)17.Van Wart J., Grassini P., and Cassman K.G., Impact of derived global weather data on simulated crop yields, Global Chang. Biol.doi: 10.1111/gcb.12302 (2013)
<end>