-  Monitor land use/land cover changes in upper Mekong river region;
- Assess effects of land use/land cover changes in upper Mekong river on hydrological regime of Cuu Long river delta of Vietnam.

Theoretical results:
=- Inheriting the cloud-removal algorithm on multi-temporal MODIS data developed by project leader from a NAFOSTED project in 2018, this project has completed a set of automatic multi-temporal MODIS image processing tools. The whole process of cutting, merging, and interpolating to remove cloud from MODIS data is carried out automatically by modules built by project leader based on Batch language of DOS operating system.
- The study has determined that the area of forest cover in  Mekong River basin has been continuously decreased, specifically: In period 2003 - 2008 decreased by 3% (3,021,875 ha, from 60.9% to 57.9%); Period 2008 - 2013 decreased by 2.3% (2,293,950ha, from 57.9% to 55.6%); Period 2013 - 2018 decreased by 0.2% (144.375ha, from 55.6% to 55.4%).
- The study used combined machine learning models to predict water levels at 18 hydrological stations on the Mekong River, using real data measured at stations from 2000 to 2018 as training data and tested data. The results show that the experimental models all predict the water level with high confidence (R2 > 9.2), in which the Bagging (RF) model has the best reliability and 03 combined models are more reliable than single model. From this result, it can be seen that the trend of daily water level variation at the stations is still regular and predictable.
- Analyzing some indicators of hydrological regime change at Chau Doc station (the station on Mekong river and farthest from the sea) from 2000 to 2018, the results show that: The highest 2-month average water level tends to decrease, the lowest 2-month average water level tends to increase.
- The study has made statistics that the number and capacity of hydroelectric power plants increased continuously during the research period, increasing faster in quantity and capacity in the period from 2003 to 2012.
- The study analyzed the relationship between the changes of total forest area and total hydropower capacity with 2 hydrological indicators: the highest 2-month mean water level and the lowest 2-month mean water level. The univariate correlation of relational pairs is not really clear. However, the results of multivariable regression analysis showed a significant relationship, specifically: Multivariable regression correlation function between the highest 2-month mean water level (MaxTB2) with Total forest area (Sr) and Total hydropower capacity (Mtđ) have correlation coefficient R2 = 0.589; Multivariable regression correlation function between the lowest 2-month mean water level (MinTB2) with Total forest area (Sr) and Total hydropower capacity (Mtđ) has correlation coefficient R2 = 0.541.
Applied results:
- The project has built a MODIS MCD43A4 image dataset, 500m resolution, 16 days period, continuously for 19 years (from 2000 to 2018), cloud-free covering entire basin of the main rivers of Vietnam including both the Mekong River and the Red River. This is a very good quality dataset of optical remote sensing images, continuing for many years and freely available to interested researchers.
- Bagging (RF) combined machine learning model can be used to predict daily water levels at hydrological stations in the Mekong Delta with high reliability, supporting for the planning and implementation of local socio-economic activities.
- The results of analysis of the correlation between Total forest area and Total hydropower capacity with hydrological regime of the Mekong Delta can help managers have an overview of the impacts on the downstream from deforestation and hydropower exploitation activities upstream, from which there are reasonable solutions for sustainable development in the Cuu Long river delta.

-The study has conducted multi-temporal MODIS image classification for the years 2003, 2008, 2015 and 2018 using the same set of ground truth samples based on decision tree classification method - CART. This method creates a uniform classification result set between the years, minimizing the error of classification due to the difference of ground truth samples. Since then, the study has made statistics of the continuous decline of forest area from 2003 to 2018 and different from each period as described above.
- The study has used a machine learning model to predict daily water levels at hydrological stations in the Mekong Delta. The prediction results are quite reliable (R2 > 9.2), which shows that the level and water at 18 hydrological stations from 2000 to 2018 still follow the rules and can be predicted.
- The study has analyzed some indicators reflecting the hydrological regime of Mekong Delta, specifically at Chau Doc station (the station on Mekong river and farthest from the sea), the results show that: The highest 2-month mean water level tends to decrease, the lowest 2-month mean water level tends to increase.
- The study has analyzed the relationship between changes in Total forest area and Total hydropower capacity with 2 hydrological indicators: the highest 2-month mean water level and the lowest 2-month mean water level. The obtained results are: The univariate correlation between Total forest area and Total hydropower capacity with the above 2 hydrological indicators is not really clear. However, the results of multivariable regression analysis of Total forest area and Total hydropower capacity have a significant relationship with the above 2 hydrological indicators, with the correlation index R2 = 0.589 and R2 = 0.541, respectively.

-    Scientific papers in referred journals (list):
[1]    Nguyen Thanh Hoan, Nguyen Van Dung, Ho Le Thu, Hoa Thuy Quynh, Nadhir Al-Ansari, Tran Van Phong, Phan Trong Trinh, Dam Duc Nguyen, Hiep Van Le, Hanh Bich Thi Nguyen, Mahdis Amiri, Indra Prakash, Binh Thai Pham. Novel Time Series Bagging Based Hybrid Models for Predicting Historical Water Levels in the Mekong Delta Region, Vietnam. Computer Modeling in Engineering & Sciences, Vol. 131, No.3, 2022, pp.1431-1449,  doi:10.32604/cmes.2022.018699.
[2]    Ram C. Sharma1, Hoan Thanh Nguyen, Saeid Gharechelou, Xiulian Bai, Luong Viet Nguyen, Ryutaro Tateishi. Spectral Features for the Detection of Land Cover Changes. Journal of Geoscience and Environment Protection, 2019, 7, 81-93.
[3]    Nguyen Thanh Hoan, Ram C. Sharma, Nguyen Van Dung, Dang Xuan Tung. Effectiveness of Sentinel-1-2 Multi-Temporal Composite Images for Land-Cover Monitoring in the Indochinese Peninsula. Journal of Geoscience and Environment Protection, 2020, 8, 24-32.
-    Technological products (describe in details: technical characteristics, place):

1. Land use/land cover maps for the years: 2003, 2008, 2013 and 2018 of the entire Mekong river basin, scale 1:500,000
2. Land use/land cover change maps for the periods: 2003-2008, 2008-2013, 2013-2018 of the entire Mekong river basin, scale 1:500,000
3. Land use/land cover change maps in 2 key areas, scale 1:50,000
4. Database of 16-day 500m cloud-free MODIS satellite images, 16 years from 2003 to 2018 for the entire Mekong River basin and multi-temporal Landsat images of 2 key areas
5. Database collected from hydrological stations in Cuu Long river delta
6. Results of analysis and assessment of the influence of upstream land use/cover changes on hydrological regime of Cuu Long river delta
7. Report on project implementation results