Application of Seasonal Climate Forecasts for Sustainable Agricultural Production in Telangana Subdivision of Andhra Pradesh India

K. K. Singh • D. R. Reddy • S. Kaushik • L. S. Rathore • J. Hansen • G. Sreenivas

12.1

Introduction

Substantial advances in the efforts to model planetary weather systems, and resulting improvements to general circulation models (GCMs), have led to better predictability of the climate fluctuations, especially 1 to 6 months in advance (Delecluse et al. 1998). Pioneers in generation and distribution of seasonal climate forecasts include the IRI and NOAA. Wise utilization of this information by the farmers and policy makers can contribute substantially towards achieving sustainability in agricultural production. Notwithstanding constant endeavors to improve the living standards of the developing countries like India, which ranks second in the population in the world, particular challenges still remain unattended in the arena of securing sustainable food production. In this context, it is worthwhile to explore and apply climate forecasts for strategic decision-making in agriculture and related areas, especially in the semi-arid regions, which are characterized by high interannual variability in rainfall and consequent uncertainty in water availability for rainfed farming operations.

If farmers are to apply seasonal climate forecasts to improve decision-making, they must first translate forecasts into production and economic outcomes associated with alternative management strategies at the spatial scale of impacts and decisions. Locally-adapted and tested crop simulation models allow one to quickly explore the production outcomes of a range of management alternatives under a range of forecast climatic conditions (Hansen and Indeje 2004; Jones et al. 2000). However, the difference in spatial and temporal scales of dynamic seasonal climate prediction and crop simulation models presents a substantial challenge to using crop simulation to anticipate crop response to predicted climate variations. Extracting and applying information about within-season variability for crop model applications remains a more difficult challenge than downscaling in space. Several approaches for linking crop simulation models with seasonal climate forecasts have been proposed by research workers. One of the process based approaches to linking climate prediction to agricultural models is to aggregate bias-corrected climate model output into seasonal or sub-seasonal (e.g. monthly) averages, then disaggregate to produce daily time series with frequency variability that is consistent with the long-term daily record, and low-frequency variations that represent the seasonal or sub-seasonal forecasts. Temporal disaggregation involves the use of some form of stochastic weather generator approach to constrain the generated daily sequences to match predicted monthly or seasonal means or other statistical properties.

Sorghum, rice, maize and castor based cropping systems are predominant in Telangana subdivision. The total seasonal rainfall has no relevance in agricultural planning but its distribution has a major value. In the absence of advance information on the rainfall pattern, farmers plan their agricultural operations based on their experience and knowledge of the past climate. In the rainfed agricultural scenario of Telangana subdivision, dominated by the monsoon climate, the main concerns (Ramana Rao 1988) are: large variations in the dates of commencement of rainy season, variations in total seasonal rainfall received, prolonged dry spells within the rainy season, high intensity rainfall due to cyclones, depressions, etc., resulting in flood damage to the crop, and variations in the cessation date of the rainy season. Hence, early warnings based on seasonal rainfall forecasts can help farmers to adjust crop management strategies to minimize impacts of malevolent climate and maximize benefits of benevolent climate.

In addition, there is a need for multi-institutional collaboration in the region for the use of seasonal climate forecasts in analysing suitable crop management strategies, and their acceptance by the farmers and policy makers. The existing network of 107 Agro-meteorological Advisory Service Units of National Centre for Medium Range Weather Forecasting (NCMRWF), which is already working towards the dissemination of farm weather advisories in Telangana subdivision, can be used towards achieving the common goal of developing crop management strategies based on seasonal climate forecasts.

This chapter addresses the application of seasonal precipitation forecasts to the management of rainfed agricultural systems in Telengana subdivision of India with the following objectives:

a Maximize crop yield through application of seasonal climate forecast in agriculture for two selected locations, b Generate seasonal rainfall hindcasts for the two locations, c Select sowing window for selected crops, d Determining plant population density, and e Contingent planning (find alternative option when monsoon is delayed). 12.2

Methods 12.2.1

Description of Key Sites

After a detailed survey of the study area and interactions with the farmers, two contrasting sites in Telangana subdivision were selected. These sites are in two agroclimatic zones of Telengana subdivision i.e. North Telangana (assured rainfall region) agro-climatic zone and South Telengana (low rainfall region) agroclimatic zone.

North Telangana Agroclimatic Zone

North Telengana zone receives an annual rainfall of 900-1050 mm, out of which southwest monsoon contributes 780-950 mm. The maximum temperature of the zone ranges between 30-37 °C and minimum temperature ranges from 21-25 °C during southwest monsoon season. In this zone, Karimnagar district (longitude 79°09' E, latitude 18°26' N) was selected. This district has a population of 3.04 millions with total geographical area of 11800 km2. The source of irrigation is well and canal (Sri Ram Sagar project) with double cropped area and rice-rice and maize-groundnut based cropping system.

South Telangana Agroclimatic Zone

South Telangana agroclimatic zone receives an annual rainfall of 750-870 mm (southwest monsoon rainfall: 550-700 mm). The maximum temperature of the zone ranges between 28-34 °C and minimum temperature ranges from 22-23 °C during southwest monsoon season. In this agroclimatic zone, Mahabubnagar district was selected, which has a population of 3.51 millions with total geographical area of 18 432 km2. The district is drought prone and agriculture is mainly rainfed. The major crops/cropping systems are Sorghum-Fallow and Castor-Fallow.

12.2.2 Data

Weather

The historical daily weather data were collected from the Regional Agricultural Research Station, Jagtial in Karimnagar district for 1989-2002 and Palem, in Mahabubnagar district, which are nearer to the test sites. Rajendranagar center has long-term weather data (1971-2002), which are used as proxy data for Palem. Solar radiation was calculated from bright sunshine hours. District-wise historical annual and monthly rainfall data for Karimnagar and Mahabubnagar over the past 40 years were collected for analysis.

Soil

The predominant soil type of Karimnagar district is medium to deep black soils (vertisols) with clay sub soils and red sandy soils (Chalkas) with 90 cm depth. The predominant soil types of Mahabubnagar district are sandy (Dubba) and sandy loam (red chalka) soils with low water holding capacity with 80 cm depth.

GCM Predictor Selection and Rainfall Hindcasts

Climate forecast fields for rainfall were taken from the GCMs viz. ECHAM, GSCF, CCM, COLA, NCEP with approximately 2.5-3° horizontal resolution, with 18-20 vertical levels. Output from simulations that the International Research Institute for Climate Prediction (http://iri.columbia.edu) provided for the present study was used.

The coarse spatial resolution of current GCMs often leads to systematic shifts in the location of spatial rainfall patterns that reduce their prediction skill. Since the large-scale features that the GCM can predict affect local climate variations, it is possible to use this information to improve prediction of local climate variability (Benestad 2001). Model correction is necessary to account for shifts in regional rainfall anomaly patterns that result from the influence of local factors that the coarse resolution of GCMs cannot capture, such as, steep orography, vegetation contrasts and land-water contrasts. The use of statistical relationships, estimated over some past period, between observed climatic predictand fields and hindcast GCM output fields, is known as model output statistics (MOS). When the predictand is at a higher spatial resolution than the GCM output, the approach is known as MOS downscaling, or statistical downscaling. One common approach to MOS correction or downscaling uses principal component analysis applied to identify the leading modes of variability of the GCM output fields, and sometimes the predictand spatial fields (Heyen et al. 1996; Kidson and Thompson 1998). The geographical domain associated with GCM output fields for principal component (PC) analysis is 66-90° E and 5-30° N. Each PC pattern represents a predictor field with high spatial resolution and spatial coherence, yet without the risk of over-fitting the empirical model. These can then be related to the predictors by regression.

In this study, IRI provided time series of PCs, using which rainfall hindcasts for selected locations were made. After the estimation of the rainfall hindcast for different months/season for the years 1989-1998 for Jagtial and 1971-1998 for Rajendranagar, the correlation was drawn between the observed and hindcast rainfall. Correlation measures the matched variances between two time series.

Stochastic Disaggregation of Monthly Rainfall

A stochastic weather generator that is modified to allow it to generate synthetic daily weather sequences was used such that the monthly climatic means exactly match specified targets. The underlying stochastic generator is described in Hansen and Mavromatis (2001). It is an adaptation of the WGEN weather generator of Richardson (1985). For each hindcast year we generated 10 stochastic realizations of daily weather whose monthly totals match June to September monthly totals predicted from the principal components.

Crop Simulation and CERES Models

Crop yields were simulated using CERES models for crops under study. The CERES (Crop Estimation through Resource and Environment Synthesis) model is a process oriented dynamic crop growth model, which predicts status of crop on real time basis as a function of exogenous parameters. The CERES models for rice, sorghum and maize crops, used in the present study are available in DSSAT v3.5 (Hoogenboom et al. 1999). It is a daily time-step model that simulates grain yield and growth components of different varieties in a given agroclimatic condition. These models have been already validated for a wide range of climates all over the world and are independent of location and soil type encountered.

Study by Saseendran et al. (1998, 2000) using CERES-Rice V3.5 showed that the model is capable predicting of grain yield and phenological development of the crop in the climatic condition of Andhra Pradesh and Kerala in India with reasonable accuracy. The errors in grain yield prediction by the model are 7.9%, 8.3% and 5.7% respectively for Sambamahsuri, Rajavadlu and Tellahamsa in Andhra Pradesh. Reddy (1992) used CERES-Maize model to predict the silking and maturity dates and yield for cv. Ganga Safed-2 in Gujarat climatic condition. CERES-Sorghum v3.5 model was also validated for cv. CSH-1 under Maharashtra climatic condition in India for its various subroutines viz. phenology, growth, water balance and nitrogen balance by Varshneya and Karande (1999), and the growth and yield were successfully predicted by model in the rainy season.

Management Strategies Considered

The management strategies considered for the different crops are given below. These management practices are similar to those, which are followed by the farmers at study sites.

Rice

Genetic coefficients for two cultivars, which are popularly grown in the state, are required for describing the various aspects of performance of a particular genotype in the model. The rice crop varieties used in the present study are Sambamahsuri and IR-64. The Sambamahsuri is a long duration (145 to 150 days) variety having an average simulated yield level of 6 712 kg ha-1. IR-64 is a short duration (115-120 days) variety and simulated yield level is 5 623 kg ha-1. The values of the genetic coefficients for the cv. Sambamahsuri (Saseendran et al. 2000) and IR-64 are presented in Table 12.1. The same crop management practices were followed in simulation experiments with different sowing dates. The planting date considered for simulation of crop cultivars IR-64 and Sambamahsuri was 26 July. Plant population at the time of planting was 33 plants m-2 with the row spacing of 15 cm and planting depth of 5 cm. The nitrogen fertilizer was applied in three split doses of 40 kg each in the form of urea. The dates of fertilizer application were 28 July, 27 August and 1 October. The field was kept always under 2 cm of water.

Maize

Maize is generally grown as rainfed crop during rainy (Kharif) season in Andhra Pradesh. The maize cultivar used in the present study is ProAgro hybrid. The genetic coefficients for the cv. ProAgro was derived on the basis of cv. Ganga Safed-2, for which these values were available (Reddy 1992). The genetic coefficients along with values for cv. ProAgro were presented in Table 12.2. The farmers at the project site practiced sowing of the crop, when the accumulated rainfall is 75 mm after the onset of the monsoon. The planting window was taken from 2 June to 20 July with lowermost soil

Table 12.1. Genetic coefficients used in the CERES-Rice simulation model

Name

Description

Genetic coefficients

IR-64 Sambamahsuri

(PI)

Time period (expressed as growing degree days (GDD) in °C over a base temperature of 9 °C) from seedling emergence during which the rice plant is not responsive to change in photoperiod

200.0

540.0

(P20)

Critical photoperiod or the longest day length (in hours) at which the development occurs at a maximum rate

140.0

170.0

(P2R)

Extent to which phasic development leading to panicle initiation is delayed (expressed as GDD in °C) for each hour increase in photoperiod above P20

350.0

400.0

(P5)

Time period in GDD (°C) from beginning of grain filling (3 to 4 days after flowering) to physiological maturity with a base temperature of 9 °C

12.0

12.0

(G1)

Potential spikelet number coefficient as estimated from the number ofspikelets per g of main culm dry weight (less lead blades and sheaths plus spikes) at anthesis

100.0

100.0

(G2)

Single grain weight (g) under ideal growing conditions, i.e. non-limiting light, water, nutrients, and absence of pests and diseases

0.0220

0.0220

(G3)

Tillering coefficient (scalar value) relative to IR-64 cultivar under ideal conditions

1.00

1.00

(G4)

Temperature tolerance coefficient. Usually 1.0 for varieties grown in normal environments

1.00

1.00

water as 90% and uppermost soil water as 100%. Plant population at the time of emergence was maintained with 8 plants m-2 with the row spacing of 35 cm and planting depth of 6 cm. The nitrogen fertilizer was applied in three equal split doses of a 40 kg ha-1 in the form of urea i.e. at the time of sowing, 25 days after sowing (DAS), and 55 DAS.

12.2.6.3 Sorghum

Sorghum is an extensively grown rainfed crop in Andhra Pradesh, used as food, and fodder. The sorghum crop cultivar CSH-5 used in the present study is a medium duration cultivar (90-105 days), commonly grown by the farmers of Andhra Pradesh. The genetic coefficients for the cv. CSH-5 calculated by Varshneya and Karande (1999) are presented in Table 12.3. The farmers at the project site practiced sowing the crop, when the accumulated rainfall is 75 mm after the onset of the monsoon. The planting window was taken 1 June to 15 August with lowermost soil water as 70% and uppermost soil water as 100%. Plant population at the time of emergence was 18 plants m-2 with

Table 12.2. Genetic coefficients used in the CERES-Maize simulation model

Name Description Genetic coefficients for ProAgro

P1 Thermal time from seedling emergence to the 310.0 end of the juvenile phase (expressed in degree days above a base temperature of 8 °C) during which the plant is not responsive to changes in photoperiod

P2 Extent to which development (expressed as days) 0.520 is delayed for each hour increase in photoperiod above the longest photoperiod at which development proceeds at a maximum rate (which is considered to be 12.5 hours)

P5 Thermal time from silking to physiological 900.0

maturity (expressed in degree days above a base temperature of 8 °C)

G2 Maximum possible number of kernels per plant 600.0

G3 Kernel filling rate during the linear grain filling 7.90

stage and under optimum conditions (mg day-1)

PHINT Phylochron interval;the interval in thermal time 38.90 (degree days) between successive leaf tip appearances

Table 12.3. Genetic coefficients used in the CERES-Sorghum simulation model

Name Description Genetic coefficients for CSH-5

PI Thermal time from seedling emergence to the end 415.0 of the juvenile phase (expressed in degree days above a base temperature of 8 °C) during which the plant is not responsive to changes in photoperiod

P20 Critical photoperiod or the longest day length (in 13.50

hours) at which development occurs at a maximum rate

P2R Extent to which phasic development leading to 40.5

panicle initiation (expressed in degree days) is delayed for each hour increase in photoperiod above P20

P5 Thermal time (degree days above a base temperature 525.0 of 8°C) from beginning of grain filling (3-4 days after flowering) to physiological maturity

G1 Scaler for relative leaf size 10.0

G2 Scaler for partitioning of assimilates to the panicle 5.5

PHINT Phylochron interval;the interval in thermal time 49.00

(degree days) between successive leaf tip appearances a row spacing of 45 cm and a planting depth of 5 cm. The nitrogen fertilizer was applied in the form of urea in two equal split doses of 40 kg ha-1 each as basal and after 30 DAS.

12.3

Results and Discussion 12.3.1

Rainfall and Crop Yield Analysis

In order to work out the influence of rainfall variability on yield fluctuations for rice and maize in Karimnagar and sorghum in Mahabubnagar, the linear trend was fitted in the yield to remove the impact of hybrids and technological improvement. The yield deviation from the trend line was calculated for both the districts. The yield and rainfall deviations were compared and plotted in Figs. 12.1 and 12.2 for rice and maize in

-800

LtJli

Pk I"

9(^7 j \ 19?A / * 1982 V /

Fig. 12.1. Rainfall deviation and yield deviations for rice in Karimnagar district

Year

Fig. 12.1. Rainfall deviation and yield deviations for rice in Karimnagar district

Year

Fig. 12.2. Rainfall and yield deviations for maize in Karimnagar district

Year

Fig. 12.2. Rainfall and yield deviations for maize in Karimnagar district

Karimnagar district respectively and in Fig. 12.3 for sorghum in Mahabubnagar district.

In Karimnagar district, the rainfall was found to have a significant influence on yield of rice and maize. In case of rice, the trend in rainfall and yield deviation was almost similar except in few years, while in case of maize, the trend was similar up to 1992 and thereafter yield increased during the period 1993-1997 (Fig. 12.2). In Mahabubnagar district, the sorghum crop showed a similar trend except during early nineties. Though maize and sorghum crops were cultivated as rainfed, the positive yield deviation during the last decade is attributed to varietal/technological advancements and even distribution of rainfall including low rainfall years.

The rainfall data of forty years (1963-2002) were analyzed to workout the variability in mean values of decadal rainfall and its coefficient of variation at different stations. Two major periods were considered (i) thirty years (1963-1992) and (ii) recent decade (1993-2002). The results presented in Fig. 12.4 show that there was a decreasing trend in rainfall in the recent decade for the months of June, July and for the whole monsoon season. The month of July is more crucial from the agriculture point of view as most of the rainfed crops are being sown and paddy transplanting is also taken-up during this month.

Coefficient of variation of mean monthly rainfall data for 1963-1982 and 1983-2002 and presented in Fig. 12.5 shows that in the Mahabubnagar district there is an increas-

Fig. 12.4. Mean decadal (1993-2002) rainfall (mm) over three decadal mean (1963-1992) rainfall for Karimnagar district

Fig. 12.3. Rainfall and yield deviations for sorghum in Mahabubnagar district

Fig. 12.4. Mean decadal (1993-2002) rainfall (mm) over three decadal mean (1963-1992) rainfall for Karimnagar district

Fig. 12.3. Rainfall and yield deviations for sorghum in Mahabubnagar district

ing trend in variability in the months of June and July and for the rainy season. Karimnagar district showed a slight decreasing trend in the month of June whereas, in July and for the rainy season there was an increasing trend.

Hindcast of Rainfall

Time series data on X1 and X2 for all five GCMs were used to estimate rainfall hindcast for the years 1989-1998 at Jagtial and 1971-1998 at Rajendranagar. Forecasts for the individual months of June, July, August, September and October and for different combinations of months were generated keeping in view the farmer's preference for a shorter duration forecasts (Table 12.4). Of all the models tested for Rajendranagar, ECHAM was found to give a better forecast (Table 12.4, Fig. 12.6). Correlation studies revealed that the

Fig. 12.5. Coefficient of variation (%) of rainfall during June, July and for the rainy season in Karimnagar and Mahabubnagar districts
Table 12.4. Correlation coefficients between observed and predicted rainfall using different climate models for Rajendranagar

ECHAM

COLA

CCM

NCEP

GSCF

June

-0.20

-0.49

-0.30

-0.06

0.03

July

0.04

0.17

0.12

-0.02

-0.09

August

0.45

0.34

-0.20

0.13

-0.05

September

0.28

0.27

0.21

0.12

0.13

June-July

-0.36

-0.06

0.03

-0.03

-0.02

July-August

0.49

0.44

-0.07

0.13

0.00

August-September

0.59

0.47

0.15

0.24

0.16

June-July-August

0.43

0.35

-0.06

0.12

0.02

July-August-September

0.61

0.53

0.19

0.20

0.15

June-July-August-September

0.57

0.47

0.24

0.19

0.16

June-September

Fig. 12.6. Relationship between observed and predicted rainfall using ECHAM model for different months at Rajendranagar

highest significant correlation exists between observed and predicted rainfall when the August and September months put together were used with ECHAM model.

Similar work was also done for Jagtial (Karimnagar). At Jagtial the COLA model gave a better correlation for the season, whereas for the individual months (July, August, and September), the ECHAM model gave a better correlation (Table 12.5, Fig. 12.7).

Table 12.5. Correlation coefficients between observed and predicted rainfall using different climate models for Jagtial

ECHAM

COLA

CCM

NCEP

GSCF

June

0.23

-0.39

-0.19

-0.32

-0.81

July

-0.38

-0.20

-0.19

0.16

-0.15

August

-0.12

-0.09

-0.32

-0.16

-0.24

September

0.23

0.01

-0.30

-0.20

-0.17

June-July

0.00

0.15

0.03

0.08

-0.42

July-August

-0.16

0.18

0.04

0.12

0.13

August-September

0.02

0.25

-0.25

-0.06

-0.36

June-July-August

0.13

0.28

0.00

0.09

-0.28

July-August-September

0.20

0.35

0.13

0.21

0.01

June-July-August-September

0.24

0.44

0.09

0.17

Crop Yield Simulation with Actual and Hindcast Rainfall

Optimum Transplanting Time for Rice

Simulation results of grain yield of rice cv. IR-64 and Sambaamahsuri for 12 different dates of transplanting revealed that the rice yield is higher for cv. IR-64, when transplanted on 26 July as compared to other transplanting dates and for cv. Sambamahsuri, higher yield was obtained when transplanted on 19 July.

Crop Model Output with Hindcast Weather

Rice

The ten realizations of weather data conditioned on sub-seasonal (monthly) rainfall hindcasts made from each GCMs were generated and crop yield was simulated with generated weather for each realization with the same management practices as with the observed weather data for cv. IR-64 and Sambamahsuri. Further average of yield from 10 realizations for each year was worked out. Comparisons of yield based on hindcast and observed weather are shown in Figs. 12.8 and 12.9.

Maize

Comparison of the grain yield of maize simulated by the model with the hindcast and observed weather data for cv. ProAgro. Figure 12.10 shows that grain yield simulated with NCEP generated weather has the same trend as that of observed weather.

June-September

Chart Interaction Depression And Yoga
Fig. 12.7. Relationship between observed and predicted rainfall for Jagtial using; a COLA model; b ECHAM model for different months
Rice Yield Prediction
Fig. 12.8. Comparison of simulated rice yield (cv. IR-64) with observed and hindcast weather data
Rice Yield Prediction
Fig. 12.9. Comparison of simulated rice yield (cv. Sambamahsuri rice) with observed and hindcast weather data

Sorghum

The grain yield comparisons for the sorghum crop (Fig. 12.11)indicate that ECHAM model predictions were closer to the observed yield data in few years and also within the same trend.

Fig. 12.10. Comparison of simulated maize yield (cv. ProAgro maize) with observed and hindcast weather data
Fig. 12.11. Comparison of simulated sorghum yield (cv. CSH-5) with observed and hindcast weather data

Farmers Perceptions

Awareness programs were conducted periodically during monsoon season of year 2003 on seasonal climate forecasts for the farmers of both the key sites. The main aim of this exercise was to elicit farmers' views on the use of climate forecasts in their cropping strategies and the farmer's requirements. Farmers mentioned about the weekly medium range forecasts based AAS activities and long range forecasts (LRF) of India Meteorological Department given in the beginning of monsoon season. They expressed that they are unable to make use of LRF in their crop planning. Farmers were informed about the efforts being made to generate seasonal climate forecasts (SCF) for Indian region by leading international centers viz. IRI and their limitations. Interactions with the farmers brought out their following needs about weather and climate forecast:

■ Break in monsoon

■ Extreme weather events

■ Preferred monthly/fortnightly forecast

During the subsequent meetings, the farmers were educated on the use of SCFs and their limitations. In short-term planning of agriculture operations the importance of medium range forecasts was explained during these interactions. The farmers expressed satisfaction to a certain extent on the use of agro-advisory services based on medium range weather forecasts. The farmers suggested to increase the lead-time with 10-15 days. Further they felt the need to integrate the seasonal/long range climate forecasts with agro-advisory services. They suggested that this integration will help to select the right crop and the right variety based on seasonal climate forecasts and mid-season corrections like intercultural operations, supplemental irrigation, etc. using medium range forecasts.

The views of the farmers from two agroclimatic zones were also taken during extensive tours. The requirements differ between the zones. Low rainfall zone farmers are interested in correct forecast of sowing rains that is very critical. High rainfall zone farmers are interested in knowing the quantum of rainfall required to get the tanks filled up and subsequent release for paddy transplantation.

12.4

Conclusions

Results of this study showed that ECHAM model has generated a better rainfall hindcast at seasonal/sub-seasonal scale for Rajendranagar (a proxy station for Palem). For Jagtial COLA model gives better correlation between hindcast and observed rainfall at seasonal scale whereas for individual months ECHAM produced better hindcasts. Awareness was created amongst the farmers, researchers and planners about utility and limitations of seasonal climate forecast for application in agriculture through group meeting during monsoon season 2003 was created. Farmers preferred fortnightly and monthly instead of seasonal forecasts for better decision-making in agricultural operations and desired for integration of ERP along with existing AAS.

Acknowledgements

The authors are grateful for financial support from START, Washington, D.C., USA for conducting these studies.

References

Benestad RE (2001) A comparison between two empirical downscaling strategies. Int J Climatol 21:1645-1668

Delecluse O, Davey MK, Kitamura Y, Philander SGH, Suarez M, Benttson L (1998) Coupled general circulation modelling of the tropical Pacific. J Geophys Res 103(c7):14357-14373 Hansen JW, Indeje M (2004) Linking dynamic seasonal climate forecasts with crop simulation in semiarid Kenya: a comparsion of methods. Agr Forest Meteorol 125:143-157 Hansen JW, Mavromatis T (2001) Correcting low-frequency bias in stochastic weather generators. Agr

Forest Meteorol 109:297-310 Heyen H, Zorita E, Von Storch H (1996). Statistical downscaling of monthly mean North Atlantic air-

pressure to sea level anomalies in the Baltic Sea. Tellus A 48:312-323 Hoogenboom G, Wikens PW, Tusji GY (eds) (1999). DSSAT v3.5. In: DSSAT 3, vol IV. University of Hawaii, Honolulu, Hawaii, pp 95-244 Jones JW, Hansen JW, Royce FS, Messina CD (2000) Potential benefits of climate forecasting to agriculture. Agr Ecosyst Environ 82:169-184 Kidson JW, Thompson CS (1998) Comparison of statistical and model-based downscaling techniques for estimating local climate variations. J Climate 11:735-753 Ramana Rao BV (1988) Operational agricultural meteorology (problems and priorities). Indian Society of Agronomy, New Delhi

Reddy DR (1992) Crop-weather relationship in rabi maize (Zea mays L.) and testing of CERES-Maize model for the middle Gujrat agroclimatic zone. PhD dissertation, B. A. College of Agriculture, GAU, Anand Campus, Anand

Richardson CW (1985) Weather simulation for crop management models. T ASAE 28:1602-1606 Saseendran SA, Hubbard KG, Singh KK, Mendiratta N, Rathore LS, Singh SV (1998) Transplanting dates for rice in Kerala, India using both CERES and ClimProb. Agron J 90:185-190 Saseendran SA, Singh KK, Rathore LS, Singh SV, Sinha SK (2000) Effects of climate change on rice production in Kerala. Int J Clim Chang 44(4):495-514 Varshneya MC, Karande BI (1999) Genetic coefficients of Indian sorghum genotypes for CERES sorghum model. Presented at national workshop on Dynamic Crop Simulation Modeling for Agromet Advisory Services held at NCMRWF, New Delhi

This page intentionally blank

Was this article helpful?

0 0

Post a comment