Prediction of above Ground Carbon Sequestration from Landuse/Landcover Changes in the Upper Northern Thailand Using CASA-Biosphere Model

Abstract

Background and Objectives :  Aboveground carbon sequestration, or Net Primary Productivity (NPP), is an important ecological indicator that reflects the potential growth of vegetation in absorbing and storing carbon dioxide from the atmosphere through photosynthesis. This process plays a critical role in offsetting carbon emissions resulting from human activities. Therefore, assessing aboveground carbon sequestration in a specific region is essential for understanding carbon balance within ecosystems and promoting ecological restoration. This study selected the Upper Northern Region of Thailand as the study area due to several unique characteristics, including high ecological diversity, the presence of various forest and agricultural landscapes, topographical variation ranging from highlands to watershed areas, and seasonal climatic fluctuations. Land use/land cover (LULC) in this region directly affects the carbon sequestration potential of each land type. However, systematic studies on the relationship between LULC types and carbon sequestration in the Upper Northern Region of Thailand remain limited, especially those employing remote sensing technology for detailed spatial analysis. This study aims to (1) Analyze the relationship between carbon sequestration and changes in land use/land cover, and (2) Predict future carbon sequestration and land use/land cover changes.

Methodology : The study begins with an analysis of the relationship between land use/land cover and aboveground carbon sequestration by classifying land use types for three time periods 2016, 2019, and 2024 using the Random Forest model. The CASA-Biosphere Model is then applied to estimate carbon sequestration, and the estimated results are validated using field data through Pearson’s correlation coefficient. Subsequently, future aboveground carbon sequestration is predicted using the hybrid Markov-CA model to simulate future LULC changes. Regression equations are then developed to predict the Normalized Difference Vegetation Index (NDVI) and Land Surface Temperature (LST). These predicted datasets, together with projected LULC data, are incorporated into the CASA-Biosphere Model to estimate future NPP. Finally, the accuracy of NPP prediction is assessed using Pearson’s correlation coefficient, Correlation Coefficient, Root Mean Square Error (RMSE), and Mean Absolute Error (MAPE).

Main Results : The analysis using the CASA-Biosphere Model revealed distinct variations in aboveground carbon sequestration across different land use and land cover (LULC) types, with forest areas exhibiting the highest sequestration potential (0.249–1.256 gC/m²) due to dense vegetation, complex canopy structures, and robust root systems, followed by perennial crops (0.245–0.836 gC/m²), while rice paddies and field crops showed moderate values (up to 0.596 gC/m² and 0.637 gC/m², respectively), reflecting differences in species composition, management practices, and growing seasons. In contrast, urban and industrial areas showed very low or negative sequestration (0.057 to –0.287 gC/m²), indicating their role as net carbon sources as a result of limited vegetation cover and high anthropogenic emissions. Model validation using three sets of field data produced Pearson’s R² values of 0.663, 0.710, and 0.994, confirming strong agreement between modeled and observed NPP. Future projections for 2033 indicate increasing carbon sequestration across most LULC categories, with forests remaining the highest (up to 1.326 gC/m²), alongside rising values for rice paddies (0.569 gC/m²) and perennial crops (0.778 gC/m²), a trend potentially influenced by climatic conditions that enhance vegetation productivity. The accuracy of NPP predictions, validated using historical data, achieved a high correlation coefficient (R² = 0.924), demonstrating the model’s strong capability to capture spatial–temporal patterns and its suitability for long-term carbon sequestration forecasting and environmental planning

Conclusions : The assessment of aboveground carbon sequestration under land use/land cover changes using the CASA-Biosphere Model reveals that forest areas possess the highest carbon sequestration potential, followed by perennial crops. Agricultural areas show moderate potential, with field crops performing similarly to rice paddies. Urban and industrial areas exhibit very low or negative sequestration, acting instead as carbon sources. The CASA-Biosphere Model demonstrates high accuracy when validated with field data. Forecasts for the year 2033 indicate increasing carbon sequestration for all LULC categories, with forests remaining the strongest contributors, followed by perennial crops and agricultural areas. The high predictive accuracy confirms the suitability of this approach for assessing carbon sequestration trends within the study region. The findings support forest conservation and sustainable agricultural practices to enhance carbon sequestration and mitigate climate change impacts. The results can guide policy-making on forest resource conservation by identifying priority areas for preservation and restoration, as well as informing national greenhouse gas reduction targets under the Paris Agreement. Moreover, the projected carbon sequestration trends can serve as a foundation for sustainable land management, low-carbon agriculture, and the development of agricultural carbon credit mechanisms to enhance regional and national carbon sequestration efficiency.