Remote sensing of agriculture – South/Southeast Asia research initiative special issue

South/Southeast Asian countries are growing rapidly in terms of population, industrialization, and urbanization. One of the key challenges in the region is food security. Although total food production and productivity have increased in the region because of the additional land area converted to agricultural land use during the 1960s to 2000, the growth rate of food production in recent times has slowed down, mostly due to loss of agricultural lands related to increasing urbanization and industrialization and less than optimal use of available technology. Further, the weather and climate systems in the region driven primarily by monsoon variability are resulting in droughts or flooding impacting agricultural production. Monitoring the agricultural crop production at regular intervals is essential to predict and prepare for disruptions in the food supply.

Despite the progress in remote sensing and geospatial technologies, little emphasis has been placed on developing robust methods for operational mapping/monitoring of cropped areas and forecasting crop production. Up to date, agricultural Land Cover/Land Use Change (LCLUC) information is currently limited to very few countries in South/Southeast Asia. There is an urgent need to enhance national and regional operational systems for monitoring of agricultural crops within the region. Transitioning appropriate methods developed in the research domain into operational systems is a challenge for all countries but can provide improvements in timely and valuable information for agricultural production, management and policy-making useful to address food security issues. To address the above issues and other LCLUC, the NASA LCLUC program initiated the South/Southeast Asia Research Initiative (SARI). The goal of SARI is to promote and support innovative regional research, education, and capacity building, involving state-of-the-art remote sensing, natural sciences, engineering, and social sciences to enrich LCLUC science in South/Southeast Asia. SARI has been organizing focused thematic meetings and workshops in different countries of the region with the main objective of strengthening regional science involving both developed and developing country scientists.

This special issue of papers is an output from one of the focused agricultural meetings organized in New Delhi, India during May 2–4, 2017. The papers address a variety of agriculture-related topics in South/Southeast Asia.

1. Crop mapping using very high resolution data

Maps of crop areas, the type of crops grown and the associated timing is important information, which can be used for yield prediction and production statistics. Traditional methods of obtaining crop maps and areas through census and ground surveying are being augmented, or in some cases replaced by remote sensing methods in several countries. The spectral characteristics retrieved using the remote sensing sensors are influenced by the crop leaf pigment, leaf water, and canopy structure. Daily observations from coarse resolution sensors, such as MODIS, often observe mixtures of crop types. On the other hand, Landsat 30m spatial resolution observations are very useful for crop type mapping, but have low temporal repeat cycle (16 days), with high probability of cloud contamination. Thus, crop type mapping requires high-resolution data in both space and time. Using WorldView-2 satellite imagery in Jianan Plain, Taiwan, Wan and Chang (2019) demonstrate the usefulness of support vector machines for crop type mapping with greater than 90% accuracy. Specifically, they integrate information from spectral band intensities, normalized difference vegetation index together with grey-level co-occurrence matrix (GLCM) texture measures in the classification process. The approach highlights the usefulness of grey relational analysis in locating the area of the uncertainty of target category in the thematic map, and a grey relational grade metric for detecting the regions that are harder to discriminate and in particular, mixed pixel cases. They also note a slight improvement in the classification outcome when GLCM texture measures are used.

2. Crop yield estimation using synthetic aperture radar and modeling

In several regions of the world, optical remote sensing has been used extensively for mapping and monitoring of agricultural areas. However, in the SARI region, especially outside of the dry season, persistent cloud cover limits the use of optical remote sensing. Synthetic aperture radar (SAR) is an effective and important tool useful for monitoring agricultural areas, as observations do not require clear-sky conditions. SAR is sensitive to the geometrical structures and dielectric properties of the targets and can penetrate clouds. TriSetiyono et al. (2019) showcase the potential of SAR data in crop yield forecasting in different countries of Asia including the Philippines, Vietnam, Cambodia, Thailand, and India. They show improved rice yield estimates in these countries using the SAR derived crop mapping areas integrated with the ORYZA rice crop growth model. They also demonstrate the utility of a vegetation cloud model to show good correlations between the SAR backscatter values and leaf area index and the role of incident angle effects. Parameterization of the ORYZA model with SAR data inputs is well described for each of the sites in different countries. The SAR-based yield estimation system proposed in this article is a suitable solution for operational yield forecasting of rice yields to address food security and policy issues.

3. Inversion modeling for retrieving crop biophysical parameters

Mostly, earth and plant canopies reflect radiation anisotropically which besides being a function of target surface properties is also dependent on source-target-sensor geometry. The anisotropic vegetation canopy reflectance is described by the bidirectional reflectance distribution function (BRDF), which can be derived either by carrying out measurements over a range of source-target-sensor geometries or through validated canopy reflectance models. The PROSAIL model is one of the most popular canopy reflectance model, and it describes both the spectral and directional variations of canopy reflectance as a function of leaf biochemistry and canopy architecture. The model is easily invertible and consists of leaf reflectance model PROSPECT and canopy radiative transfer model SAIL. In this special issue, Lunagaria and Patel (2019) demonstrate the use of the PROSAIL inversion model for retrieval of chlorophyll, leaf dry matter, leaf angle, and leaf area index for wheat, derived from field goniometer measurements. The PROSAIL model was inverted for all reflectance spectra for 54 different view angles and different wheat phenophases. The results indicate that the reflectance in the visible region is relatively more important for chlorophyll retrieval whereas reflectance in the near-infrared is more important for retrieving leaf inclination angle, dry matter and leaf area index. These results have implications for crop mapping and monitoring at large spatial scales. In the study, the directional sensitivity of the PROSAIL model results were also presented.

4. Mapping the impact of droughts on rice areas

In India, droughts have been occurring frequently, affecting the economy and the environment. Most of the agricultural droughts in India are due to precipitation shortages, and soil water deficits. In some regions of India, drought impacts also vary according to prevailing weather conditions, biological characteristics of the crops, and their growth stages. For large-scale studies, MODIS has immense potential considering its high revisit time of 1–2 days. A study by Gumma, Nelson, and Yamano (2019) highlight the potential of the multi-temporal MODIS normalized difference vegetation index (NDVI) integrated with field observations to identify and map drought-affected areas in India. A long-term average of NDVI during the rainy (kharif) season (June–October) was compared with NDVI from a known drought year to identify changes in rice area. Rainfall data from the Tropical Rainfall Monitoring Mission (TRMM) was used to infer satellite drought analysis. In addition, authors use spectral matching techniques to categorize drought into different classes. They report 84.7% correlation between the MODIS-derived drought-affected area and the reduction in harvested area. The methodology can be used for mapping the drought influence on crops elsewhere.

5. Salinity stress detection time series coarse resolution data

Remote sensing based time series analysis is a powerful tool for quantifying agroecosystem changes and developing precise phenological calendars. Intra-seasonal phenological patterns including crop stress can be captured using remotely sensed time series data. Despite the progress in large area crop mapping and monitoring, little emphasis has been placed on mapping soil salinity and its effects on crops. Paliwal et al. (2019) map soil salinity affecting rice crops in the coastal region of Odisha, India. In this region, soil salinity is most severe during the dry season from November to February, making soils unsuitable for rice production. The authors used MODIS 250m resolution in conjunction with a Savitzky-Golay filter and the TIMESAT package to capture seasonality in an Enhanced Vegetation Index (EVI) time series. Field measurements of soils and salinity were correlated with the EVI for different seasons, and a strong negative correlation was found. The study highlights the usefulness of the seasonal integral and amplitude of the EVI signal as promising indicators of soil salinity and intensity. The approach followed in the study is potentially useful for regional scale salinity mapping studies.

6. Oil palm plantation mapping using visual and automated methods

In several South/Southeast Asian countries, palm oil is commonly used for household cooking and cosmetics. The development of high-yield varieties has made oil palm planting a highly profitable business. The high economic value has caused extensive land-cover changes through the conversion of rich tropical forests to oil palm monoculture, most importantly in insular southeast Asia. Quantifying the positive and negative impacts of oil palm plantations on the environment and economy requires accurate mapping and monitoring efforts. Mapping of oil palm is challenging due to the age differences including smallholder versus the industrial scale of palm oil plantations. Miettinen, Gaveau, and Liew (2019) compare the results of oil palm mapping in Borneo from visual versus automated methods and highlight both the advantages and disadvantages of these methods. They infer that a visual approach detects all areas used for oil palm agriculture regardless of the phase or condition of the crop but is limited to large-scale industrial plantations whereas the automated mapping only detects existing closed canopy oil palm stands but is not limited to large-scale industrial plantations. Thus, they recommend the integration of both visual and automated methods for oil palm mapping and monitoring system. The results can be useful for deriving highly accurate oil palm maps of the region.

7. Influence of aerosols on land surface fluxes

Aerosols play a major role in climate by absorbing and scattering solar radiation. The radiation reaching the surface is influenced by the chemical characteristics of the aerosols, their size distribution, and the amount in the atmosphere. Understanding the role of aerosols in different regions of the world impacted by pollution is important in the context of climate change. In this issue, Latha, Murthy, and Vinayak (2019) highlight the impact of aerosols on surface fluxes in highly heterogeneous agricultural and industrial areas of the Indo-Ganges Basin and northeast India Plains. They use a variety of datasets to address the variations in surface fluxes from Moderate Resolution Imaging Spectroradiometer (MODIS), Global Land Data Assimilation (GLDAS-NOAH) model, National Centre for Environmental Prediction, Modern-Era Retrospective analysis for Research and Applications (MERRA) including aerosol ground observations. They show that aerosols can reduce latent heat by 60% of aerosol radiative forcing (ARF) over the permanent dense canopies, whereas the same amount of ARF reduces latent heat flux by 25% over the semi-arid regions. These results are useful to help understand land–atmosphere interactions in the region.

In summary, the seven papers published as a part of the special issue cover a variety of case studies integrating remote sensing, geospatial technologies, and simulation modelling for agricultural research in South/Southeast Asia. We hope readers will find the articles useful and will be guided to new research areas.

Disclosure statement

No potential conflict of interest was reported by the authors.