Interactions between meteorological and agricultural droughts at different temporal and spatial scales

The Tensift basin is prone to drought and, with the increasing frequency of extreme events, their forecasting and monitoring are becoming more complex. The present work aims to shed light on the interactions between meteorological and agricultural droughts while using multiple drought indices, and analyzing its temporal and spatial patterns over the Tensift basin in Morocco. To this purpose, we initially performed a trend analysis of the main parameters used in this study namely precipitation, temperature, NDVI, and soil moisture using the Mann-Kendall test. Moreover, a data-driven approach was adopted here to reveal the impact of lack of precipitation on the soil and vegetation cycles. Remote sensing data of precipitation from ERA5Land and soil moisture data from ESA-CCI as well as land surface temperature and NDVI from MODIS are used to calculate the standardized precipitation index (SPI), the vegetation condition index (VCI), the temperature condition index (TCI), and the soil moisture condition index (SMCI) for the period 2001–2021. A comparison analysis was conducted to test the performance and concordance of the indices. Then, to analyze the propagation of meteorological drought to the other components we conducted a cross-correlation study between drought indices.  The results reveal an upward trend of NDVI which is noticeable from the first decade (2009) and is attributed to the development of irrigated areas in this period. In contrast, the basin has shown a significant decline in monthly soil moisture for the period extending from 2001 to 2021, which could be explained by the way how soil moisture is retrieved in the ESA CCI product, and the trend in vegetation. On the other hand, the monthly precipitation and land surface temperature time series show no significant trend. The comparison between the indices showed moderate to low agreement. Correlations between TCI and SPI were eventually negative and significant at small time scales. A moderate correlation was observed between SPI1, SPI3, and TCI (0.45). The strongest correlations between SMCI and SPI were found at the 3 and 6-month time scales. Furthermore, the concordance between VCI and SPI is stronger at larger SPI time scales, the best correlation was observed between the indices VCI and SPI at 12 months with a correlation coefficient of 0.44. The correlations in the Tensift basin reflect spatial heterogeneities where some indices are more prevailing than others. Lag analysis results demonstrate valuable insights into the leading and preceding behavior of different variables regarding SPI. Relevant responses were identified at short, mid, and long-term influence of precipitation deficits on soil moisture, vegetation, and temperature. The results of this study highlight the interest in analyzing drought with different indices dedicated to each type of drought in order to improve early warning systems and risk management strategies for semi-arid areas.