Capacity building for statisticians, data scientists, public sector staff, and the general public in the Asia-Pacific region.

The enhancement of capacity for staff within government statistical agencies (NSOs) in countries across the Asia-Pacific region is of paramount importance.

Scholarships for education in the field of official statistics and data science are made available to staff members of government statistical agencies (NSOs) in countries across the Asia-Pacific region.

In the field of big data analytics and data science.

With a focus on the development of methods, algorithms, and the utilization of big data analytics and data science, particularly for the provision of official statistics and SDGs indicators.

Theme: Rice Estimation using Earth Observation Data in Indonesia

Rice data is crucial for determining food policies in Indonesia, thus the government is required to produce fast and accurate data. The rapid development of Big Data technology provides opportunities for the modernization of business processes in statistical activities, including the potential use of Big Data in agricultural statistics. Remote Sensing (RS) technology, also known as Earth Observation (EO), can be used to capture information on the Earth's surface through satellite imagery, such as land cover, crop types, land cover changes, and more. Improving methodology for rice data calculation through Earth Observation (EO) is one of the focuses of agricultural statistical modernization and digital transformation.

Initiated by One Data Rice, Statistics Indonesia (BPS), the National Research and Innovation Agency (BRIN), the Ministry of National Development Planning (Bappenas), the Ministry of Agriculture, and the Polytechnic of Statistics-STIS are committed to leveraging the technology to bolster food security. These collaborations are in the form of modernizing the current business process by incorporating the Earth Observation data, machine learning algorithm, and in-situ data collection. FAO also supports innovation with the Economic and Social Commission for Asia and the Pacific (ESCAP) for escalating the quality and nurturing the workflow.

Currently, the BPS calculates rice production data using two surveys, namely the Area Sampling Frame Survey (ASF) to identify the rice growth phases to predict harvest time, and the Crop-Cutting Survey to measure rice productivity. Data from the results of both surveys are used to calculate rice production. The opportunity of EO technology to capture surface information without visiting the field can be utilized to identify harvest times, optimizing data collection costs.

An essential part of this initiative involves the integration of satellite data with in-situ data on rice fields gathered by field enumerators, enhanced by employing a machine learning approach. This thorough method guarantees a thorough comprehension of the lifecycle of rice crops across the diverse agricultural terrains of Indonesia. These algorithms are capable of identifying the rice growth phases monthly. This information is then used to predict harvest times and the extent of harvest on a field without the need for field visits. The predicted harvest time is crucial information for estimating the timing of crop-cutting surveys.

The accuracy of the rice data produced is a critical issue. Caution is needed to generate quality data as Indonesia's agricultural landscape is highly diverse, and Earth Observation (EO) technology offers numerous approaches, and determining which approach is better must be thoroughly tested. This includes testing the model itself, validating it through Ground Truth data, and referencing various National Statistics Offices (NSOs) in countries that have implemented similar technologies. Thus, using academically tested methods will result in high-quality rice data as official statistics.

Expected Program Output

a. A validated model with Ground Truth data. A model for identifying rice growth phases that not only has high accuracy but also aligns with field conditions.

b. Harvest area, and harvest time prediction. Similar to the output of ASF surveys, this model is expected to predict potential harvest time up to 3 months in advance.

c. EO application area maps. A map indicating the areas where EO can be used as a substitute for ASF surveys. EO is only utilized in regions believed to yield high accuracy.