Droughts of all kinds (e.g., meteorological, agricultural, and hydrological) are a recurring phenomenon in one or the other part of the world in any given year. The frequency of occurrence of droughts is increasing in many parts of the world under a changing climate and their magnitudes and severity are also increasing. With ballooning global population and economic activity, human-induced droughts (e.g., diverting river upstream at cost of downstream users where water is becoming increasingly precious) are as much a factor as drought by natural causes (e.g., El Niño-induced drought). NASA and NOAA recently declared that the year 2016 was the hottest year on record with 1.7° Fahrenheit above the global average temperature for the twentieth century (https://data.giss.nasa.gov/gistemp/). Going forward, based on the trends, it appears that climate variability and associated extreme events such as droughts and floods as well as greater uncertainty in timing, seasonality, and magnitude of climate (e.g., precipitation, temperature) are expected to increase across the world.
The above reality calls for advances in drought studies that enable us to better study and predict droughts. Fortunately, there are many recent possibilities to advance drought studies that include availability of improved remote sensing data (e.g., Landsat-8, Sentinel-2, Wordview-4, SMAP), cloud computing capabilities to handle massively large volumes of data (e.g., using Google Earth Engine platform), and cutting-edge methods and approaches. So, these progresses in data, computing, and methods enable us to advance our understanding, modeling, mapping, and monitoring droughts. In this context, GIScience and Remote Sensing called for a Special Issue on Advances in Remote Sensing and GIS-based Drought Monitoring. The call received many manuscripts of which we the guest editors along with the Editor-in-Chief Dr Jungho Im selected 6 papers published in this special issue based on peer-reviews and editorial scrutiny.
Paper by Tsegaye Tadesse et al. developed Vegetation Drought Response Index for Canada (VegDRI-Canada) on similar principle as the VegDRI developed for the conterminous United States earlier. They focused their study in the agricultural areas of Canadian Pairies. The strength of their approach is in using multitude of data that integrated remote sensing and non-remote sensing data and included vegetation status, climate, biophysical, land use and land cover, soil, and environmental. MODIS time-series was the main remote sensing data used. VegDRI-Canada has the potential to be made operational and used by a variety of users, especially agricultural community. However, it needs to be applied, tested, and its impact evaluated.
In drought studies, the well-known and widely used indices are Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI), Vegetation Condition Index (VCI), Temperature Condition Index (TCI), and Vegetation Health Index (VHI). Paper by Hoa Thi Tran et al. made use of these indices using long-term (1989–2016) remote sensing data from Landsat Thenmatic Mapper (TM) and Operational Land Imager (OLI) as well as MODIS to study droughts over a small area in Vietnam. They stress the advantages of Landsat time-series over MODIS time-series in study of droughts.
Agricultural cropland fallows are indicators of drought because greater cropland fallow areas mean lesser cropland areas are planted. Croplands are left fallow, overwhelmingly, because of lack of rainfall or lack of supply of irrigated water. This in turn occurs due to drought conditions. Cynthia Wallace et al. developed a fallow-land algorithm called FANTA to map croplands versus cropland fallows, year after year, using time-series MODIS data over California’s Central Valley which has about 12 million acres of croplands. How much of this is cropped and how much is left fallow depends on the water available from precipitation and irrigation. FANTA’s ability to rely on temporal and spatial anomalies by comparing a pixel to its history and its neighbors to determine automatically croplands versus cropland fallows year after year using 250-m resolution 8-day MODIS time-series data makes it an important drought monitoring method. The results help water and food security studies.
Drought has impacted the State of California, USA, profoundly many years, but especially over the last decade and a half. The year 2014 was one of the worst drought years – climatological, hydrological, agricultural, and environmental. Sensitivity of agriculture to drought is much easier to detect using remote sensing when compared with sensitivity of forests to drought which is much subtler and hence requires greater rigor to assess. In this regard, Mahesh Rao et al. showed that the well-known modified perpendicular drought index (MPDI) derived using Landsat 30-m time-series data is well correlated with climate data such as precipitation, temperature, and long-term climate water deficit. They showed this ability of MPVI to monitor drought of California’s forests during one of the worst drought years – 2014.
One of the most critical requirements for accurate drought assessment is the quality of the data itself. This requires us to understand, refine, and correct data during pre-processing, so a “clean” dataset is used in the drought analysis. This aspect was studied in detail by Saptarshi Mondal et al. by developing an algorithm for cloud noise detection and removal leading to improved time-series MODIS data for drought studies. In order to achieve this, they developed cloud possibility index (CPI) by integrating cloud flags such as cloud state, cloud shadow, cirrus cloud, and internal cloud algorithm flag. Then they use CPI along with NDVI to remove noise in MODIS time-series data. They showed, through a study in Sri Lanka, that such approach led to improved detection of growing cycle and, as a result, significantly better drought detection.
The most consistent global soil moisture data from the European Space Agency is produced by blending data from multiple sensors (e.g., ERS-1, ERS-2, METOP-A, SMMR, SSM\I, TMI, and AMSR-E). This coarse resolution data was downscaled for the 2000–2010 period and was used by Suman Kumar Padhe et al. in agricultural drought assessment of one of the worst drought prone areas of India – Bundelkhand. They showed that the time-based function derived from spatially downscaled soil moisture (FSMs) was a better predictor of crop yields than the vegetation condition index (VCI) derived from MODIS.
Thereby, this special issue contributed to advancing the drought studies using multi-sensor, multi-resolution remote sensing in several ways that included the following: (1) adopting Vegetation Drought Response Index of a country by integrating remote sensing and non-remote sensing data, (2) demonstrating the higher value of using 30-m time-series Landsat data over 250-m time-series MODIS data in drought monitoring using well-known remote sensing drought indices LST, NDVI, VCI, TCI, and VHI; (3) studying drought by developing an automated cropland fallow mapping algorithm called FANTA that looks at a pixels neighborhood and temporal anomalies; (4) establishing Landsat 30-m time-series data derived MPDI in forest drought sensitivity; (5) developing a cloud possibility index to remove noise in time-series MODIS data that enhances data by providing improved growing cycle that in turn helps better drought detection; and (6) highlighting the utility of downscaled blended soil moisture data from multiple microwave sensors that enhances drought studies relative to VCI drought index derived from MODIS.
We thank many anonymous reviewers for their insights and valuable time. Finally, we want to thank Professor Jungho Im, Editor-in-Chief, for providing us with opportunity to edit this special issue.