Agricultural drought mitigating indices derived from the changes in drought characteristics

Highlights

• Assessing drought mitigation performance with changes in drought characteristics

• Two drought mitigation indices (ADFC and ADAC) are proposed and tested.

• ADAC has the ability to assess mitigation performances to extreme drought event.

• PDSI and VHI detect natural and actual agricultural drought separately.

Abstract

Droughts can cause tremendous losses to agricultural and economic development; humans have explored many anti-drought measures to mitigate the influence of drought and accordingly altering the characteristics of actual agricultural drought. In this study, a method is proposed to detect the spatiotemporal changes in drought characteristics, which show the effects of anti-drought measures on drought mitigation. Two agricultural drought mitigation evaluation indices are proposed, the agricultural drought frequency change (ADFC) and agricultural drought area change (ADAC), which are calculated by combining the Palmer drought severity index (PDSI) and vegetation health index (VHI), two widely used drought monitoring indices. The PDSI and VHI represent the natural and actual agricultural drought severity under natural and actual conditions respectively, and their differences in drought frequency and affected area reflect the level of anti-drought measures in mitigating agricultural drought. The feasibility of using ADFC and ADAC to quantify the effects of anti-drought measures for agriculture is explored using data from six typical agricultural provinces in the North China Plain and Northeast China. The results show that ADFC and ADAC could reflect both the spatiotemporal changes in agricultural drought characteristics and the influence of anti-drought measures on agricultural drought. The trend of the drought mitigation index is consistent with agricultural activity statistics. These two indices could be further used to evaluate the effects of different anti-drought methods and aid in defeating agricultural drought across many countries.

Introduction

Drought, representing a water shortage relative to normal conditions(Wardlow et al., 2017; West et al., 2019), is among the disasters that cause heavy agricultural losses (Lesk et al., 2016). Drought can directly affect agricultural production through its effects on photosynthesis (Yan et al., 2016a; Yan et al., 2016b) by reducing both chlorophyll and the water content in vegetation (Kogan et al., 2019) and by changing the frequency and intensity of other disturbance factors (Anderegg et al., 2012; Sangüesa-Barreda et al., 2015; Tian et al., 2014; Tian et al., 2015). The annual average losses in the Asia Pacific region due to agricultural drought reach $404 billion, which is approximately 1.4% of the region's GDP (UNESCAP, 2019). The annual economic losses caused by drought reached $7 billion per year from 1984 to 2017 in China (Su et al., 2018). Harvested areas have been found to decline by a global average of 4.1% during droughts (Lesk et al., 2016). Drought impacts can include conflict, migration, fires, food shortages and famine (Patrick, 2017; Trnka et al., 2019).

Drought has been defined in a number of ways (Trenberth et al., 2014; Van Loon and Laaha, 2015; West et al., 2019), including meteorological drought, which is defined as precipitation deficiency, agricultural drought, which is defined as deficiency in soil moisture, hydrological drought, which is defined as low streamflow and ground water reduction, and socioeconomic drought, which is defined as the gap between water demand and supply. The development of droughts consists of multiple climate process interactions and land-atmosphere responses, as well as human activities that influence the cause/linkages of droughts (Wang et al., 2016b). Factors that influence drought propagation include regional climate and local land surface characteristics, such as landcover, the vegetation community, soil context, topography and human interventions (Barker et al., 2015; Van Loon and Laaha, 2015). Drought propagation occurs not only among different drought phases (e.g., from meteorological to agricultural drought) but also within a certain drought phase (Parry et al., 2016).

Drought can be characterized by the degree of influence, temporal properties, spatial properties (Dai, 2011), the percentage of the area affected, the frequency of occurrence in certain periods, long-term trends (Yu et al., 2014), and the peak time of drought occurrence (Yao et al., 2020). Various drought indices have been developed to describe the characteristics of drought at different phases, including severity, extent, frequency and duration (Hayes et al., 2012; Yao et al., 2020), and these indices usually measure the departure of a variable from the local normal condition based on its historical distribution. Drought indices are also used to quantify drought impacts by establishing the relationship between the temporal variability of different drought indices and response variables (Vicenteserrano et al., 2012), such as crop yields or crop productivity (Hornbeck and Keskin, 2014; Li and Ren, 2019; Mavromatis, 2007; Mohtashami et al., 2020; Potop, 2011; Quiring and Papakryiakou, 2003; Vergni and Todisco, 2011; Wang et al., 2018), soil moisture (Sims et al., 2002), groundwater level (Khan et al., 2008), river discharges (Wu et al., 2018), and ecosystem services(Jones et al., 2019; Seneviratne and Ciais, 2017), etc.

For meteorological drought, precipitation is the primary variable used in indices such as the standardized precipitation index (SPI) (McKee et al., 1993), and surface air temperature is used as a secondary variable to consider the effect of evapotranspiration, such as in the standardized precipitation evapotranspiration index (SPEI) (Vicenteserrano et al., 2010). The soil water balance is further used for water supply and demand as in the Palmer drought severity index (PDSI) (Palmer, 1965). The PDSI was found to be highly correlated with the soil moisture differences within the uppermost 1 m according to site observations; thus, site observation are usually adopted to quantify agricultural droughts (Dai, 2011) but can reflect only the changes in soil moisture caused by climate change (Mu et al., 2013) and do not consider the impacts of anti-drought measures. The SPI, SPEI and PDSI have been used to describe the spatiotemporal variations in drought characteristics, drought extent, drought severity and tendency (Trenberth et al., 2014; Yao et al., 2020; Zhai et al., 2010). Drought-affected areas significantly increased in northern China from 1951 to 2003 based on the PDSI (Zou et al., 2005) at a rate of approximately ~3.72% per decade (Yu et al., 2014). The SPEI revealed decadal variations over 1961–2012 in China with the most frequent and severe droughts before the 1980s and in the 2000s (Chen and Sun, 2015; Troy et al., 2015; Yu et al., 2014). In particular, North China experienced significant drying trends and had the longest drought durations during the 1990s and 2000s (Yang et al., 2016; Yu et al., 2014; Zhai et al., 2010). Although meteorological drought indices are highly correlated with crop yield, crop yield is influenced by many factors more than just climate variability (Quiring and Papakryiakou, 2003; Tian et al., 2018; Wang et al., 2016a). However, these indices include no information about agricultural irrigation (Yu et al., 2019), varieties, tillage practices and any other anti-drought measures, therefore are capable to describe drought characteristics under natural condition.

Agricultural drought is related to moisture deficits in vegetation roots, which lead to crop growth stress, crop yield reduction or failure driven by low precipitation over a sustained period (Narasimhan and Srinivasan, 2005). The characteristics of agricultural drought can be directly estimated from in situ soil moisture measurements (Yang et al., 2018) or satellites with ground-calibrated soil moisture models such as the Soil Moisture Active Passive (SMAP) mission (Entekhabi et al., 2008; Yang et al., 2018) or indirectly estimated using the amount of chlorophyll and water content to represent vegetation health (Kogan et al., 2019; Wang et al., 2016a; Wardlow et al., 2017), as well as the passive and active microwave sensors (Eswar et al., 2018; West et al., 2019), such as Soil Moisture Agricultural Drought Index (SMADI) combining soil moisture with vegetation indices(Sánchez et al., 2016). Vegetation is considered unhealthy if it is not well developed or vigorous, wilting and not very green (or brown). The linkage between satellite-derived vegetation growth conditions and soil moisture deficits has been shown in agricultural drought monitoring (Gu et al., 2008; Wardlow et al., 2017; Weiss et al., 2020; West et al., 2018). Satellite-based vegetation indices (VIs) such as the anomaly of normalized difference vegetation index (NDVI), vegetation conditions (VCI) and normalized difference water index(NDWI) (Gu et al., 2008; Hayes et al., 2012; Qin et al., 2015; Wardlow et al., 2017; West et al., 2019) can provide valuable information for the characterization of agricultural drought, its spatial extent, and the severity of vegetation stress. However, it is difficult to detect the ending time of an agricultural drought event since stressed vegetation cannot recover promptly after drought (Zhang et al., 2019). Among all satellite-based indices, the vegetation health index (the VHI) is one of the most popular indicators (Bento et al., 2018; Gomes et al., 2017; Rahman et al., 2009; Rojas et al., 2011) for monitoring drought extent (Sholihah et al., 2016). The VHI inherently considers the local biophysical (soil, slope) and climate conditions (García-León et al., 2019). The VHI is a stable and robust remote sensing index for actual agricultural drought monitoring in various agro-meteorological zones of China (Yan et al., 2016a). The VHI exhibits a high correlation with crop yield, especially in the critical stage of crop growth (Kogan et al., 2012; Prasad et al., 2006; Rahman et al., 2009; Yang et al., 2018).

Meteorological and agricultural indicators exhibit differences in their ability to capture drought characteristics (Hayes et al., 2012; Qin et al., 2015; Wang et al., 2016a; Wang et al., 2018). The performances of the soil moisture deficit, meteorological drought indices and satellite-based VIs were compared in North China (Qin et al., 2015; Yang et al., 2018). The drought-affected area exhibited a significant increasing trend, and the drought event that occurred in 1999 had the largest extent of all years with data according to the soil moisture, but the same events were shown to have shorter durations and even larger extents on average when derived from the SPI, and annual improvements in the vegetation condition were generally observed in North China (Yang et al., 2018). The declining periods of NDVI were in fairly good agreement with the drought events identified in the same periods. The correlation between the annual soil moisture deficit and the NDVI anomaly was higher than that between the annual soil moisture deficit and the SPI (Qin et al., 2015). The correlation between the vegetation index and meteorological drought index was found to be much lower over cropland areas than over grassland, shrub, and forest areas because human interventions, including tillage practices, irrigation, fertility and disease control, will also affect crop growth (Yang et al., 2018). It was also found that irrigation plays a significant role in establishing the relationship between drought indices and crop yield loss (Wang et al., 2018). Differences in drought characteristics are derived from different indices, particularly in terms of the drought-affected areas and occurrence frequencies. However, few studies have explored the value and significance of these differences and what they stand for.

When a drought occurs, many anti-drought measures are often explored to mitigate the detrimental impacts of drought, accordingly altering the characteristics of actual agricultural drought, such as drought affected extent, occurring frequency. These measures include irrigation practices (Hornbeck and Keskin, 2014; Troy et al., 2015; Uwizeyimana et al., 2018), the use of drought-tolerant seed varieties (Simtowe et al., 2019), crop intensity adjustments (Solh and van Ginkel, 2014), and agricultural water conservation methods (mulching and ridges) (Uwizeyimana et al., 2018). Actual agricultural drought may not occur if anti-drought measures are appropriately adopted in time to preserve soil moisture and fulfill crop growth requirements (Van Loon et al., 2016), even if meteorological drought occurs with a significant reduction in precipitation (Hayes et al., 2012; Pablos et al., 2017).

Irrigation alleviates drought impacts, improves crop yield, and increases crop water productivity (Hornbeck and Keskin, 2014; Kresovic et al., 2014; Suarez et al., 2019; Troy et al., 2015; Yan et al., 2016a) since irrigation stabilized the climate extremes and variability, which led to yield reductions in rainfed crops but not irrigated crops (Troy et al., 2015). Irrigation changed the effects of drought by shifting the threshold to beyond the level at which crops will be negatively affected (Araujo et al., 2016; Li et al., 2018a; Schauberger et al., 2017; Zipper et al., 2016) and by mitigating dry spells (Uwizeyimana et al., 2018), therefore leading the changes of agricultural drought characteristics. In addition, stress-tolerant crops can potentially mitigate the negative impacts of drought (Fisher et al., 2015). Farmers with drought-tolerant maize varieties did not experience the impact of a drought, such as reductions in agricultural productivity, when the most severe drought occurred in 2015 in Uganda (Simtowe et al., 2019). A trend analysis of the NDVI residuals revealed that new grain varieties and farming practices, not irrigation increases, generated the apparent departure of the NDVI tendency from rainfall in West Australia (Burrell et al., 2017), which means that new varieties and tillage practices have changed the crop response (growth stress, yield reduction or failure) to drought or drought characteristics.

Therefore, by mitigating the impact of droughts on crop conditions, anti-drought measures have significantly changed the manifestation of actual agricultural drought characteristics such as extent, intensity, duration and frequency, altered the propagation process of drought development (Wang et al., 2016b) and affected drought vulnerability (Wilhelmi and Wilhite, 2002). However, a method to quantify the changes in drought characteristics is lacking, which in fact are the effects of anti-drought measures on drought mitigation.

Although drought is a relatively foreseeable climatic phenomenon with biophysical mechanisms, reactions to drought from the individual to the community and even at the state level are different (O'farrell et al., 2009). Most countries are still in the early stages of developing and implementing mitigation schemes to offset the influences of agricultural droughts, including irrigation modernization, water flow meter installment, surface water storage, and groundwater pricing (Patrick, 2017). It is necessary to investigate the overall role of anti-drought measures, which will improve our understanding of the effect of drought mitigation measures, since a “lack of information” on drought mitigation performance is the greatest barrier to investing in drought mitigation strategies (Wilhite et al., 2007). A drought mitigation effect analysis and conclusion can support climate change research and provide information for making decisions related to food security.

The objective of this paper is to propose a method to assess the changes in drought characteristics to quantify the effects of anti-drought measures on drought mitigation. Spatial-temporal analyses of the changes in drought characteristics in six provinces over the North China Plain and Northeast China over 37 years were conducted for agricultural droughts under both natural and actual conditions.