338 
-methods to be used for logging 
-enrichment plantings 
-regeneration of logged areas. 
The monitoring capability would also make it 
possible to watch closely the different phases of the 
exploitation of the forest like: 
-the preparations (construction of roads, and 
infrastructure) 
-the logging (selective or clear cutting) and 
transport 
-the regeneration (planting of open parts, enrichment 
plantings, reforestation) 
It is also possible to monitor the extent and 
results of other factors related to forestry. The 
population growth has an impact on forestry; shifting 
cultivation and the related forest fires as well as 
overgrazing are the main causes of damage and 
degradation of the forests. Already an area of about 
15 million ha consists of deteriorated forest area. 
With timely information on new (illegal) settlements 
it will be possible to take precautionary actions and 
to avoid forest fires and degradation of valuable 
forests. 
- the determination of yield for the final production 
estimates is based on sampling with the problem of 
having a real aselect and representative distribution 
over the different conditions and rice varieties . 
- the conversion factor of wet stalk paddy to grain 
and to milled rice is not a fixed one but is among 
others dependent on variety of rice, season, as well 
as on the milling process. 
Bulog maintains large buffer stocks, which are used 
to maintain floor prices for the farmers and ceiling 
prices for the consumers. The buffer stocks are 
replenished by domestic purchases and in the past 
also through the import of rice. The cost of storing 
the large amounts of rice and shipping these between 
the various islands is considerable. Before the 
formal estimates are available, estimates are already 
required for reasons of budget allocation. Apart from 
the forecasts and estimates by BPS, Bulog prepares 
informal estimates for planning imports and national 
purchases of rice and relies on its own experts 
visiting the main rice producing regions and 
reporting directly to the Jakarta Headquarters of 
Bulog. The evaluation of the crops is qualitative and 
comparative and no figures are available on the 
accuracies of the Bulog estimates. 
SP 
M( 
4 OUTLINES OF A RICE PRODUCTION MONITORING SYSTEM 
The Central Bureau of Statistics (Buro Pusat 
Statistik, BPS), in cooperation with the Department 
of Agriculture, is responsible for producing fore 
casts and estimates of the rice production of 
Indonesia. The National Food Stock Authority (Badan 
Urusan Logistik, Bulog) is assigned the task to 
ensure an adequate supply of rice throughout the 
country at stable prices. 
4.1 Present system for rice crop statistics 
Three forecasts of the production of the whole year 
are produced, mid February, June and October. The 
next year in February a preliminary estimate of the 
production of the whole past year is made, after 
which in July the final estimate follows. The current 
method to determine the acreages for both the 
forecasts and the estimates is based on monthly 
reports of the village heads to the local staff of 
BPS and Dept, of Agriculture. 
These reports include: 
- standing area under production at the start of the 
month 
- area harvested during the month 
- area damaged or destroyed during the month 
- area planted during the month 
- standing area under production at the end of the 
month. 
The first forecast of mid February is made from the 
reported planted area in December and a yield estima 
tion based on the actual yield in preceding years. 
The forecasts of June and August are made, taking 
into account the reported area and the sampled yield 
data of the already harvested crops. The yearly 
production estimates are based on the reported 
harvested areas and the sampled yield data. 
As the number of potential rice producing villages 
is very large it is clear, that there are quite many 
sources of errors and delays in the process of 
collecting and transfer of the data, so it is very 
difficult to meet the deadlines for the forecasts. 
BPS has developed statistical techniques to cope with 
missing data, but these do not remove the uncertainty 
in the forecasts. 
The relation between the official estimates and the 
actual production is not accurately known and 
difficult to assess because: 
- the acreage as used in the estimation of the 
production is based on the same source of information 
as used for the forecasts i.e. the monthly reports 
with their inherent inaccuracies 
4.2 Multitemporal/multispectral identification of wet 
rice areas 
The use of Landsat MSS and aerial photography for 
these forecasts has been studied by the Central 
Bureau of Statistics. These sensors have their 
drawbacks and it was no wonder that the conclusion 
arrived at, was not very encouraging. This is caused 
by the coarse geometrical resolution of the MSS 
scanner and the cloud conditions in Indonesia. 
The new satellites, like Landsat 5, Spot and 
eventually a TERS, will improve this situation. In 
the next years it will be demonstrated what can be 
obtained by earth observation from space. Actual 
measurements will prove the feasibility of the 
methods and in parallel scenarios will be worked out 
on how such new data and methods can be integrated in 
the existing organisations. 
Based on the time-evolution of the spectral 
features of irrigated and rain-fed rice during its 
growth, it is possible to identify rice areas with 
high accuracy. The proposed method starts from the a 
priori selection of areas where wet rice cultivation 
is possible on the basis of hydrological, soil and 
terrain conditions. The next step is that in the 
course of a continuous monitoring of these potential 
wet rice areas, the irrigated areas will be 
identified. These irrigated areas are then monitored 
with respect to the development of apparent 
greenness. The sequence of the spectral change, 
related to the change of submersed irrigated terrain 
into green vegetation cover, followed by crop 
yellowing prior to the harvest, will be used as a 
model in order to identify wet rice. 
The data will be used to estimate the harvested 
area per subdistrict. It is proposed to determine the 
rice area by means of supervised machine processing 
of the satellite data. Besides the monitoring of the 
total area of planted, growing and harvested rice per 
subdistrict, imagery suitable for visual analysis is 
generated. By means of a multistage approach and 
subarea selection, the area sample frame method can 
be applied for the selection of sampling units for 
field collection of yield samples. Consequently rice 
yield and harvested rice area per subdistrict are 
used for the calculation of the rice production for 
each subdistrict. 
The acquisition of cloudfree imagery is required at 
intervals of at least once a month, but preferably Figure 
each two to three weeks, to identify and to follow 
the main stages of the cultivation of rice. These 
stages are: dry soil after harvest or growth of other 
crops than rice; irrigation; increase of green vege 
tation; green canopy and start of maturing; crop