[[TOC]]

Few studies investigate the impact of climate change on agriculture considering the cropping system applied in sub-Saharan Africa or make an effort to identify the least impacted cropping systems. The study of Thornton et al. (2009) is an exception, analyzing crop yield response to climate change of a maize-bean cropping sequence in East Africa under which beans grow in a separate second growing season. Analyzing different multiple cropping systems in a climate impact study requires a dataset reporting their spatial distribution in sub-Saharan Africa, which to our knowledge is not available. We identify the traditional sequential cropping systems composed of two crops following one after another in sub-Saharan African countries from a household survey containing more than 8600 households in ten African countries reporting sowing and harvest dates from 56 crops which are grown on up to three plots in up to three seasons within 12 months.

A subset of a household survey (Dinar et al., 2008) containing 8697 households in 10 sub-Saharan African countries (Burkina Faso, Cameroon, Ethiopia, Ghana, Kenya, Niger, Senegal, South Africa, Zambia, and Zimbabwe) is used to calculate the growing periods of crops grown in different cropping systems. This dataset is the product of a World Bank/Global Environmental Facility project that was coordinated by the Centre for Environmental Economics and Policy for Africa (CEEPA) at the University of Pretoria, South Africa. The household survey reports sowing and harvest dates from 56 crops which are grown on up to three plots in up to three seasons within 12 months. In the households surveyed up to six crops are grown simultaneously on a plot. For each of these countries, data from 416 to 1087 households in 17 to 61 representative sample units (district or province) were collected for only one farming season (2002/2003 or 2003/2004). Sowing and harvest dates were reported on a daily, weekly or monthly basis and were converted into a uniform date specification using the day of the year. For weekly data we assumed the first day of the week, for monthly data the 15th day of the month is assumed. The length of the growing period in days is derived from these daily sowing and harvest dates for each crop. As harvest sometimes occurs shortly after sowing but the year of sowing and harvest events is not always reported, we assume a minimum length of 2 months for the growing period (6 months for cassava).

It is not possible to simulate inter-cropping systems with overlapping growing period but only sequential cropping systems with consecutive growing periods.

We identify the sequential cropping and single cropping systems applied within one farming season in a sample unit by combining the information of the crops’ growing periods in each plot and season extracted from the household survey. As only nine out of 56 crops (cassava, cowpea, groundnut, maize, millet, rice, soybean, sunflower, and wheat) can be simulated in LPJmL we combine the remaining crops to a group of ‘‘other crops’’. We assume that sequential cropping systems if two crops are reported to be planted one after another without overlaps of more than 15 days and if their growing periods sum up to less than 365 days (Fig.1) i.e. the growing period of a crop here is restricted by the occurrence of the associated crop on the plot. In contrast, we assume that a single cropping systems exists if only one single crop is reported to grow on a plot (Fig. 1) or if more than one crop is grown on a plot but the sum of their growing periods is larger than 365 days and/or their growing periods overlap by more than 15 days (Fig. 1), i.e. the conditions for a sequential cropping system are not met. An overlap of 15 days corresponds to the maximum possible error in sowing and harvest dates owing to the conversion from monthly to daily data. If various sequential cropping systems exist within a district, we identify the most frequently applied sequential cropping system in a district and assume this system to be the traditionally applied sequential cropping system. Based on the distance between the centre coordinates of the districts and those of the 0.5° x 0.5° grid cells, the sequential cropping systems found in a district are allocated to the closest grid cell. If a district covers more than one grid cell the sequential cropping systems are distributed to all corresponding grid cells. The irrigated sequential cropping systems are negligible in most countries of sub-Saharan Africa.

Figure 1: Scheme of possible timing and length of growing periods of crops in single cropping systems (A–C) and sequential cropping systems (D–G) according to the definition used in this study. (A) Two single cropping systems with large overlap, (B) one single cropping system, (C) two single cropping systems, one spanning the turn of the year and with the sum of the growing periods exceeding 365 days, (D) sequential cropping system with small overlap, (E) sequential cropping system with long fallow period, (F) sequential cropping system with short or no fallow period, (G) sequential cropping system spanning the turn of the year with sum of growing periods below 365 days (Waha et al., 2013).

To differentiate between single and sequential cropping systems and allow to simulate them together in a grid cell we introduced the new stand type CROPSEQUENCE, added two new functions, a new parameter file and changed several other functions to cope with the additional stand type.

Accordingly there are two different PHUs per crop, one for a crop grown in a single cropping system, one for a crop grown in a sequential cropping system.

We calculate phenological heat units (PHUs) for a short-growing crop cultivar grown in sequential cropping systems (PHU_seq) and a long-growing crop cultivar grown in single cropping systems (PHU_sin) from observed growing periods and daily temperatures in sub-Saharan Africa. PHU_sin and PHU_seq are calculated by accumulating daily temperatures above a base temperature threshold summed over the growing period that is reported in the household survey. In order to estimate PHU_sin for each crop in each grid cell in sub-Saharan Africa, we use a multiple linear regression model between PHU_sin and climatic parameters in each grid cell. We found a correlation, although light for maize and groundnut, between PHU_sin , mean annual temperature and moisture conditions during the growing season:

where T is the annual mean temperature, P_gs the sum of monthly precipitation during the growing season, PET_gs the sum of monthly potential evapotranspiration during the growing season, and α, β, γ and δ are empirical parameters.

A crop´s PHU is significantly longer in a single cropping system than in a sequential cropping system (by 900°Cd on average).

The application is currently restricted to the grid cells for which a sequential systems ID is defined in multiple_cropping_districtsystem.bin.

multiple_cropping

daily_cropsequence.c, harvest_sequence.c for new stand type CROPSEQUENCE.

cft1700_2005_16cfts_multiple_cropping.bin
multiple_cropping_districtsystem.bin
cropsequence.par
pft_2layers.par

Katharina Waha, Christoph Müller, Alberte Bondeau

Input
Crop functional types
LPJmL seminar

Waha, K., Müller, C., Bondeau, A., Dietrich, J.P., Kurukulasuriya, P., Heinke, J.,Lotze-Campen, H., 2013.
Adaptation to climate change through the choice of cropping system and sowing date in sub-Saharan Africa. Global Environmental Change. 23, 130-143.

Thornton, P.K., Jones, P.G., Alagarswamy, G.,Andresen, J., 2009. Spatial variation of crop yield response to climate change in East Africa. Global Environmental Change. 19, 54-65.