TL;DR
This study leverages Google Earth Engine and multi-source satellite data to monitor crop growth across China during the initial rapid spread of COVID-19. Surprisingly, despite widespread restrictions, crop growth was generally better than in previous years, suggesting minimal direct impact from the pandemic on overall agricultural output during that specific period.
What Problem Does This Paper Address?
The COVID-19 pandemic presented unprecedented challenges, including widespread lockdowns and restrictions on movement, which raised concerns about potential impacts on global food security. Specifically, understanding how these measures affected agricultural activities and crop growth in major food-producing regions like China was crucial for timely policy formulation and ensuring food supply. Traditional methods of monitoring crop growth over large areas are often time-consuming and spatially limited. This paper addresses the need for a rapid, large-scale assessment of crop health during a critical period of the pandemic, using remote sensing to overcome the limitations of ground-based observations.
Methods
The researchers employed Google Earth Engine (GEE) as their primary platform for accessing and processing vast amounts of satellite imagery and meteorological data.
- Crop Growth Monitoring: The Enhanced Vegetation Index (EVI) was chosen as the key indicator for crop growth due to its sensitivity in dense vegetation and reduced influence from soil and atmosphere. EVI was calculated from MODIS (MOD09GA) and Sentinel-2 (MSI) reflectance data.
- Climate Analogous Year Selection: To compare crop growth in 2020 with previous years under similar climatic conditions, a method developed by the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) was adapted. This method identifies the year with the most analogous temperature and precipitation patterns to the target period (January 23 - March 11, 2020).
- Relative Change Rate (CR): The difference in EVI between 2020 and the analogous year was calculated as a relative change rate (CR) to quantify crop growth status (better, similar, or worse).
- Spatial Analysis: The proportion of area (PA) for each growth category (better, similar, worse) was calculated at provincial and city scales.
- Correlation Analysis: To investigate potential indirect impacts of the pandemic, the correlation between crop growth (PA and CR) and the cumulative number of confirmed COVID-19 cases at the city level was analyzed.
Datasets
- Remote Sensing Data:
- MODIS MOD09GA (daily, 463m resolution)
- Sentinel-2 MSI (10-day, 10m resolution)
- Landsat land cover data (30m resolution) for cropland identification.
- MCD12Q1 (IGBP classification, 463m resolution) for cropland identification.
- Meteorological Data:
- ERA5 daily temperature and precipitation data (0.25° resolution) from ECMWF.
- COVID-19 Data:
- City-scale cumulative confirmed cases released by the National Health Commission (NHC) of China.
Key Insights and Ideas
- Unexpected Resilience: The most striking finding is that crop growth in China during the rapid spread of COVID-19 (Jan-Mar 2020) was generally better than in the preceding four years under similar climatic conditions. This suggests that the initial phase of the pandemic, despite severe restrictions, did not significantly hinder overall crop development.
- Geographic Distribution of Growth: Better crop growth was predominantly observed in the North China Plain (a key winter wheat region) and parts of South China. Conversely, some areas in southern China experienced worse growth.
- Limited Correlation with Pandemic Severity: While a correlation between worse crop growth and higher cumulative COVID-19 cases was found in some cities (particularly those with over 500 cases), this relationship was not widespread. This indicates that the pandemic's direct impact on crop growth was localized and not a universal consequence of case numbers.
- Data Consistency: The study validated the consistency of crop growth monitoring results derived from different spatial resolution land cover products (30m LUCC vs. 463m MCD12Q1) and different sensor data (MODIS vs. Sentinel-2), bolstering the reliability of their findings.
Key Results
- Nationally, 51% of crops showed better growth, 27% similar growth, and 22% worse growth during the study period.
- The average relative change rate (CR) for better growth was 17%, while for worse growth it was -14%.
- In Hubei province, 61% of crops showed better growth and 27% showed worse growth.
- A significant correlation (p < 0.05) between worse crop growth and cumulative confirmed cases was observed in cities with over 100 cases, particularly those exceeding 500 cases.
- Comparisons between MODIS and Sentinel-2 data showed significant correlations for relative CRs of better (r=0.87) and similar (r=0.71) growth, indicating good agreement despite resolution differences.
Limitations and Future Work
- Data Uncertainties: The authors acknowledge potential uncertainties in land cover data, particularly in distinguishing small croplands from natural vegetation in some classifications.
- Scale Effects: While Sentinel-2 and MODIS showed good agreement, scale effects might exist, with MODIS potentially reflecting smaller CRs due to its coarser resolution.
- Indirect Influences: The study acknowledges that changes in rotation systems, seed improvements, and other factors over the past five years could influence EVI values, and these were not fully isolated.
- Other Factors: Photosynthetically active radiation (PAR) was not included due to data limitations, but it could be a significant factor affecting crop growth in some regions.
- Continuous Process: Crop growth is a continuous process, and the study notes that the impact of events like the pandemic can extend beyond the immediate period. Future work could involve monitoring crop growth over longer periods (e.g., March-May) to capture these extended effects. The paper also suggests that real-time monitoring is essential for ensuring food security.
Summary updated