A framework was developed to elucidate (1) the drivers of land degradation, (2) pressures, (3) local impacts and vulnerabilities and (4) adaptation strategies. The combination of participatory approaches, statistical data analysis, time series Landsat imagery and spatial data mining was tested in southeast Vietnam where the impacts of land degradation on the environment and economy are considerable. The major drivers of land degradation are climate, notably drought, and population density. The pressures include natural resource management and land use/cover change. A Landsat archive analysis showed an increase in agricultural land use from 31% to 50%, mostly at the expense of forests, from 1990 to 2019. Farmers adapted by investing in the irrigation of rice and dragon fruit, and by selecting their rainfed crops in line with the changing environment. The most vulnerable were the rural poor and farmers without access to land and water resources. The best protection against land degradation was prosperity, which is enhanced by the region’s location along Vietnam’s major national route, connecting major cities along a north–south axis. Our analysis shows that southeast Vietnam emerged as a region with an important human ecological resilience strengthened by increased prosperity. The current adaptation options and limitations warrant further research.

Soil salinity has a major impact on agricultural production. In a changing climate with rising sea-levels, low-lying coastal areas are increasingly inundated whereby saltwater gradually contaminates the soil. Drought prone areas may suffer from salinity due to high evapotranspiration rates in combination with the use of saline irrigation water. Salinity is difficult to monitor because soil moisture affects the soil’s spectral signature. We conducted Fourier-transform infrared spectroscopy on alluvial and sandy soil samples in the coastal estuary of the Red River Delta. The soils are contaminated with NaCl, Na₂CO₃ and Na₂SO₄ salts. In an experiment of salt contamination, we established that three ranges of the spectrum were strongly influenced by both salt and moisture content in the soil, at wavenumbers 3200–3400 cm⁻¹ (2.9–3.1 µm); 1600–1700 cm⁻¹ (5.9–6.3 µm); 900–1100 cm⁻¹ (9.1–11.1 µm). The Na₂CO₃ contaminated soil and the spectral value had a linear relationship between wavelengths 6.9 and 7.4 µm. At wavelength 6.99 µm, there was no relationship between absorbance and soil moisture, but the absorbance was proportional to the salt content (R² = 0.85; RMSE = 0.68 g) and electrical conductivity (R² = 0.50; RMSE = 3.8 dS/m). The relationship between soil moisture and spectral absorbance value was high at wavelengths below 6.7 µm, resulting in a quadratic relation between soil moisture and absorbance at wavelength 6.13 µm (R² = 0.80; RMSE = 5.2%). The spectral signatures and equations might be useful for mapping salt-affected soils, particularly in difficult to access locations. Technological advances in thermal satellite sensors may offer possibilities for monitoring soil salinity.

Chlorophyll-a (Chl-a) concentration, a crucial indicator of phytoplankton biomass, is sensitive to seasonality. The variations in trophic states regarding seasonality and the changes of spectral properties of water bodies pose uncertainties to the accuracy of remote sensing semiempirical models. In particular, lakes in subtropical regions generally experience different trophic states in dry and wet seasons. In this study, a season-insensitive Chl-a retrieval model using multitask convolution neural network with multiple output layers (MCNN) is proposed. A layer-sharing network combined with data augmentation is adopted to alleviate the issue of insufficient quantity of in situ samples. In addition, a hyperparameter optimization is performed to automatically refine the MCNN architecture. To evaluate the accuracy of proposed method, Laguna Lake, one of the largest lakes in Southeast Asia, is selected as the validation target. The lake is characterized by oligotrophic and mesotrophic states in wet season, whereas the states change to mesotrophic and low-level eutrophic states in dry season. A collection of Sentinel-3 Ocean and Land Colour Instrument Level-2 images and 409 in situ samples with the Chl-a concentration range 1.24–22.30 mg $\cdot$ m $^{-3}$ were used for model calibration and evaluation. Experimental results showed that MCNN with the performance of average $\boldsymbol{R^{2}}$ = 0.74, RMSE = 2.06 mg $\cdot$ m $^{-3}$ , Pearson's $\boldsymbol {r}$ = 0.86 outperforms related semiempirical models, including normalized difference chlorophyll index, two-band and three-band models, and WaterNet. The Chl-a prediction accuracy was improved by 7.19–14.6%, in terms of RMSE, compared with WaterNet.