In-depth Analyses
What is China's crop cultivation potential? An application of the IIASA AEZ model.
The following discussion is based on preliminary results from the LUC AEZ study, which investigates the cultivation potential of China, Mongolia and the Former Soviet Union on the basis of recently updated soil, terrain, and climate databases (see: Fischer, G. / van Velthuizen, H.  / Nachtergaele, F., 1999) (see also the description of the AEZ methodology in this application).
Introduction
In the following sections we will discuss two basic questions:
(1) How much land in China is suitable for crop cultivation?
(2) What is China's maximal grain production potential?
We know that China's farmers are currently cultivating some 132 million hectares. But this is not all the land that is available for crop cultivation. The AEZ assessment shows that there are still some land reserves, which could be used for grain cultivation. In the following sections we will discuss how much grain China's farmers could produce, if they would use all the land that is suitable for cultivation. For that, we first have to distingush the level of agricultural input.
Not only grain yields, but also the suitability of land for crop production depend on the level of agricultural inputs. With irrigation and chemical fertilizers farmers can substantially increase grain production. However, they can also cultivate land areas, which would otherwise be to dry or would rapidly decline in fertility. With drainage systems wetlands can be cultivated. On the other hand, not all land is suitable for mechanized farming - if it is very steep or wet it often must be cultivated by hand, such as terraced rice paddies. In other words - the relationship between input level and the land area suitable for cultivation is not obvious - at a high input level the suitable area may be, in fact, significantly smaller than with low input level. This is certainly the case in China.
Scenario 1: Low agricultural input ("traditional" rainfed agriculture)
Let us start our discussion of the AEZ modeling results on China with a very hypothetical scenario - the assumption, that China's farmers would have no access to modern agricultural inputs, such as chemical fertilizers, pesticides, or machinery. The scenario also assumes that no irrigation is available. This "primitive" low input agriculture would essentially depend on hand tools, natural fertilizers, and would lack all means for large-scale irrigation. Of course, we know that China's farmers have long ago passed this stage of cultivation - they have had large-scale irrigation schemes for several hundred years and modern agricultural inputs for decades. However, as we will see below, this hypothetical scenario can teach us an important lesson.  

 
Suitable areas  
According to scenario 1, China would have more than 180 million hectares of land that could be cultivated with hand tools at low levels of agricultural input. Some 26 million hectares would be very suitable and more than 55 million hectares would be suitable for cultivation at this level. On the other hand, some 56 million hectares would be only moderately suitable, and almost 43 million hectares would be only marginally suitable for rainfed cultivation. In this calculation areas for settlements, infrastructure, industrial sites, and natural ecosystems that should be protected, such as forests or wetlands, were already taken into account by application of province-specific infrastructure correction factors. In other words, these are areas that would be really available for low-input crop cultivation (see Table 1). Scenario 1: Potential Land
Table 1
Potential grain production  
While the amount of 180 million hectares of land more or less suitable for low-input level cultivation seems to be impressive compared with the current size of China's cultivated land area, which is some 132 million hectares, the implication is actually quite dramatic. It means that China, under no circumstances, would be able to feed itself in 2025 with traditional low-input agriculture, despite the fact that the area that could be cultivated would be much larger than with a modern high-input agriculture.
This can be seen from the potential grain production results of scenario 1. With low-input rainfed grain cultivation China's farmers could - in the "best case" - only produce some 213 million tons of grain. In that case they would have to cultivate areas which are only marginally suitable (that is, areas, which produce only between 20 and 40% of the maximal yields under low-input rainfed conditions. If the farmers would use only very suitable, suitable and moderately suitable areas, they could only produce some 184 million tons of grain (see Table 2). This would be certainly not sufficient for a 1.48 billion population in 2025.		 Scenario 1: Potential Grain Production
Table 2
Scenario 2: High agricultural input / irrigation ("modern" high-input agriculture)
Now we will investigate the other extreme - the potential of modern high-input agriculture in China: Table 3 details the results of an AEZ modeling run for the assumption that the land - if necessary - would be irrigated and that the farmers would use a high level of agricultural inputs (including mechanization, chemical fertilizers, and pesticides).  

 
Suitable areas  
Under this assumptions China would have only some 118 million hectares of potentially suitable land - including 34 million hectares of very suitable land, 51 million hectares of suitable land, almost 27 million hectares of moderately suitable land, and nearly 6 million hectares of marginally suitable land. Some 702 million hectares would be completely "unsuitable" for cultivation at a high level of agricultural inputs (see Table 3). These areas have been already corrected for infrastructure demand. Scenario 2: Potential Land
Table 3
Potential grain production  
While the suitable area for crop cultivation is about 30% smaller under high-input conditions (scenario 2) as compared to the low-input scenario 1, the yields are much higher. Consequently, China's farmers would be able to produce some 615 million tons of grain - as compared to the 184 million tons under low-input conditions (see Table 4). In other words: with a high input level agriculture China's farmers could produce about  three times more grain than under low-input conditions - and this on smaller cultivated area. This grain amount would be sufficient for China's projected grain demand in 2050 (see our grain demand estimates in this application).
The modeling results demonstrate that China could feed its population by 2050 with only 118 million hectares of its best arable land (which is actually somewhat less than the current cropland area). However, it would be necessary to significantly modernize agriculture, so that average grain yields of about 9 tons per hectare could be achieved. While some crop areas have already reached this level of productivity in China, the current average is significantly lower. This is a very important result of the AEZ modeling, which underscores the importance of agricultural modernization in China.		 Scenario 2: Potential Grain Production
Table 4
Scenario 3: A realistic model
While it is certainly instructive to analyze China's resources of arable land under the extreme assumptions of a traditional, low-input versus a modern, high-input agriculture, a realistic model would certainly assume a mixture of high and low agricultural input levels. It is likely that the most suitable soils in China's flood plains are cultivated at a high input level, while more marginal soils in arid zones or on steep slopes are cultivated with more traditional low-input methods. This assumption is consistent with agronomic and economic experience, which shows that an investment into machinery, irrigation, and chemical fertilizers only pays back when very suitable soils are cultivated.
In scenario 3 a stepwise procedure was applied in assessing the suitability of China's land areas for cultivation: First, all those 5x5 km grid cells in China were identified, which are suitable or highly suitable for modern, high-input agriculture. Second, all remaining cells were checked for suitability at medium input level. Those cells that are highly suitable, suitable, or moderately suitable for cultivation at medium input level were marked. Third, what was left over from steps one and two was tested for suitability at low input levels. All cells that can be cultivated at least to some extent where taken into account. Basically, the procedure attempted to estimate China's potential arable land by assuming an optimal mix of input level for the various areas. 		 

 
Suitable areas  
Under this optimal mix of agricultural input levels China would have a potential arable land of almost 197 million hectares (see Table 5). The area is now larger than under high-input conditions, because areas where added that would be suitable under low-input conditions.
At high input levels some 34 and 51 million hectares would be very suitable or suitable, respectively. At intermediate input levels some 730 thousand hectares would be highly suitable, about 19 million hectares would be suitable and some 47 million hectares would be moderately suitable. At low input levels only some 35 million hectares of marginally suitable areas are significant. Some 709 million hectares in China would not be suitable for any kind of crop cultivation.		 Scenario 3: Potential Land
Table 5
Potential grain production  
With the optimal mix of high, medium and low input level cultivation China's farmers could produce a maximum of 675 million tons of grain. This estimate is based on the assumption that only 75% of the suitable areas are actually used for grain (cereal) production - some 25% would still be available for non-grain crops, such as vegetables and fruits. However, for a 675 million ton grain harvest the farmers would have to cultivate even marginally suitable areas (see Table 6).
If we assume that only those areas are cultivated, where the yield is at least 40% of the yield on very suitable fields - that is, if we eliminate the production on only marginally suitable land - then some 652 million tons of grain could be produced in China. It is interesting, that this amount is very close to the projected grain demand of China for the year 2020 (see our own estimates and the estimates of other authors) (see also the summary of AEZ results in Table 7).		 Scenario 3: Potential Grain Production
Table 6

 
Discussion
According to the AEZ modeling results China has some land reserves, which could be used for crop cultivation. If one takes into account all land that is more or less suitable under high, medium, and low input levels, China would have a land reserve suitable for crop cultivation of about 66 million hectares - or about 50% of the currently cultivated land area. However, this includes all kinds of areas - even those, that are only marginally suitable (which means they would have only between 20% and 40% of the yields that are possible in very suitable areas). From an economic point of view, it would hardly make sense to cultivate these areas. In other words: the 66 million hectares are China's ultimate land reserve for grain production - they could be cultivated, if the country would want to maximize grain production at all costs.
Certainly a more realistic assumption is that marginally suitable land would not be used for grain cultivation. In that case China would have a land reserve in the range of 30 million hectares, or about 23 percent of the currently cultivated land area (see Table 8). Most of these areas are in North. Heilongjiang, for instance, has some 5.8 million hectares of land which could be used for grain cultivation; Jilin has a reserve of about 5.6 million hectares and Hubei province of about 4.3 million hectares.
However, please note, that these land reserves are calculated with the assumption, that irrigation is used, wherever necessary (scenario 3). About half of the land reserves with grain cultivation potential would need irrigation. This means, that China will be able to fully exploit its land reserves for grain cultivation, if the water resources are better developed - for instance, by trans-basin water diversion from the South.
The AEZ model also shows that in some provinces - particularly in Yunnan and Inner Mongolia - the land with grain cultivation potential is actually smaller than the actually cultivated area. These are areas, where the agro-climatic conditions are so difficult, that the AEZ algorithm does not include them into the pool of land with grain cultivation potential - while in reality there is actually some cultivation. This does not indicate a failure of the model - it only shows that the AEZ algorithm is very "conservative". It tends to under-estimate the cultivation potential - particularly for marginal land.		 Summary of Results
Table 7

Actual and Potential Arable Land
Table 8
blank_3.gif (810 bytes)blank_3.gif (810 bytes)
blank_3.gif (810 bytes)blank_3.gif (810 bytes)
revision  Heilig, G.K. (2004): RAPS-China. A Regional Analysis and Planning System. Laxenburg, Austria