Climate trends and extremes in the Indus river basin, Pakistan: Implications for agricultural production

  Citation: Heureux, A.M.C.; Alvar-Beltrán, J.; Manzanas, R.; Ali, M.; Wahaj, R.; Dowlatchahi, M.; Afzaal, M.; Kazmi, D.; Ahmed, B.; Salehnia, N.; et al. Climate Trends and Extremes in the Indus River Basin, Pakistan: Implications for Agricultural Production. Atmosphere 2022,13, 378. https://doi.org/ 10.3390/atmos13030378 Academic Editors: Liming Ye and Waqar Ahmad Received: 19 January 2022 Accepted: 16 February 2022 Published: 24 February 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). atmosphere Article Climate Trends and Extremes in the Indus River Basin, Pakistan: Implications for Agricultural Production Ana Magali Carrera Heureux 1,† , Jorge Alvar-Beltrán 1,† , Rodrigo Manzanas 2,*,† , Mehwish Ali 3, Robina Wahaj 1, Mina Dowlatchahi 1, Muhammad Afzaal 4, Dildar Kazmi 4, Burhan Ahmed 4, Nasrin Salehnia 5, Mariko Fujisawa 1, Maria Raffaella Vuolo 1, Hideki Kanamaru 6 and Jose Manuel Gutiérrez 7 1Food and Agriculture Organization (FAO) of the United Nations, Headquarters, 00153 Rome, Italy; [email protected]g (A.M.C.H.); jor[email protected] (J.A.-B.); r[email protected]g (R.W.); [email protected] (M.D.); [email protected] (M.F.); mariar[email protected] (M.R.V.) 2Meteorology Group, Departamento de Matemática Aplicada y Ciencias de la Computación, Universidad de Cantabria, 39005 Santander, Spain 3Food and Agriculture Organization (FAO) of the United Nations, Pakistan Office, Islamabad 1476, Pakistan; [email protected] 4Pakistan Meteorological Department, Islamabad 1214, Pakistan; afzaalkar[email protected] (M.A.); [email protected] (D.K.); [email protected] (B.A.) 5School of Earth and Environmental Science, Seoul National University, Seoul 08826, Korea; [email protected] 6Food and Agriculture Organization (FAO) of the United Nations, Regional Office for Asia and the Pacific, Bangkok 10200, Thailand; [email protected]g 7 Meteorology Group, Instituto de Física de Cantabria, CSIC-Universidad de Cantabria, 39005 Santander, Spain; [email protected] *Correspondence: r[email protected] † These authors contributed equally to this work. Abstract: Historical and future projected changes in climatic patterns over the largest irrigated basin in the world, the Indus River Basin (IRB), threaten agricultural production and food security in Pakistan, in particular for vulnerable farming communities. To build a more detailed understanding of the impacts of climate change on agriculture s in the IRB, the present study analyzes (1) observed trends in average temperature, precipitation and related extreme indicators, as well as seasonal shifts over a recent historical period (1997–2016); and (2) statistically downscaled future projections (up to 2100) from a set of climate models in conjunction with crop-specific information for the four main crops of the IRB: wheat, cotton, rice and sugarcane. Key findings show an increasing trend of about over 0.1 ◦ C/year in observed minimum temperature across the study area over the historical period, but no significant trend in maximum temperature. Historical precipitation shows a positive annual increase driven mainly by changes in August and September. Future projections highlight continued warming resulting in critical heat thresholds for the four crops analyzed being increasingly exceeded into the future, in particular in the Kharif season. Concurrently, inter-annual rainfall variability is projected to increase up to 10–20% by the end of the 21st century, augmenting uncertainty of water availability in the basin. These findings provide insight into the nature of recent climatic shifts in the IRB and emphasize the importance of using climate impact assessments to develop targeted investments and efficient adaptation measures to ensure resilience of agriculture in Pakistan into the future. Keywords: climate change impacts; extremes; trend analysis; food security; agriculture; Pakistan 1. Introduction The agricultural sector in Pakistan accounts for 18.5% of the Gross Domestic Product and employs 38.5% of the workforce nationally. Within the country, the Indus River Basin Atmosphere 2022,13, 378. https://doi.org/10.3390/atmos13030378 https://www.mdpi.com/journal/atmosphere Atmosphere 2022,13, 378 2 of 15 (IRB) is the primary agricultural region, contributing to more than 90% of the country’s food production [ 1 ]. Pakistan is among the countries most affected by climate change and extreme weather events over the last two centuries [ 2 ], having documented a succession of climate related disasters including glacial lake outburst floods, droughts, heat waves, pest infestations and changing monsoon patterns [ 3 ]. The impacts of climate on agricultural production are often compounded by the high degree of vulnerability of small-scale farmers and the limited adaptive capacity of livelihoods and institutions [ 4 – 6 ]. As a result, changes in climate pose a serious threat to Pakistan’s food security and the sustainability of future agricultural production [7–9]. The IRB hosts the world’s largest contiguous irrigation system [ 10 ] that sustains approximately 90% of agricultural production in the country [ 11 ], which in turn puts significant pressure on water resources. Changing climatic conditions threaten to exacerbate future pressure on water availability, in particular increasing temperature, associated evaporation rates and more irregular precipitation patterns. According to national surveys, one of the main factors driving the recent decline in crop productivity along the IRB is the more frequent extreme weather events and shifts in seasonal rainfall [12]. Uncertainty in the timing and extent of water availability and extreme events in the IRB makes the adaptation of the vulnerable agriculture sector essential for the future [ 13 ]. Previous studies found that water sources to the IRB are expected to shift, including increases in the upstream sub-basins of Hunza, Shigar and Shyok, and decreases in the lower altitude sub-basins [ 13 , 14 ]. Likewise, increased precipitation and intensification of heavy rainfall events is likely to exacerbate river flow variability [ 15 ]. In conjunction with increases in glacial and snow-melt runoff, this suggests a higher risk of flooding and damage to cropping systems during this season [ 13 ]. For instance, in 2010, floods in the Sindh province resulted in a decline of almost 30% in rice production [ 16 ]. Recently, August 2020 has been documented as the wettest month on record for Pakistan, affecting 77,000 hectares of agricultural land predominantly along the Sindh province [17]. Previous works have reported that maximum and minimum temperatures have increased between 0.5 and 1 ◦ C, on average for Pakistan, over the period 1960–2007 [ 7 , 18 , 19 ]. However, these studies have also demonstrated a high degree of heterogeneity both spatially and temporally. With respect to precipitation, historical trends highlight an average annual increase, with the frequency of extreme precipitation events also increasing across the country over the period 1965 to 2009 [15]. Future projections of temperature in Pakistan show an continued increase in average temperature, ranging between 3 and 9 ◦ C by 2100 [ 11 ]. Overall annual rainfall in the IRB is projected to increase, however, rainfall projections exhibit significant temporal and spatial variability over the 21st century [ 20 ], with a decrease in the number of rainy days accompanied by an increase in rainfall intensity over shorter periods [20–22]. From the crop perspective, previous studies have investigated the impacts of climate on crop production in the region. Ali et al. [23] found that net production for cotton and wheat increased in Southern Punjab, while yield declined in correlation with temperature extremes. This study, however, was limited to a small region in the IRB over the historical period and did not investigate trends into the future using climate projections. To build a more detailed analysis of the past and future climate and its related impacts on the agriculture sector in the IRB, the present study employs national records of daily temperature and precipitation to assess the trends of observed extreme indicators during a recent historical period (1997–2016) and statistically downscaled future projections from climate models to identify some of the key risks (related to heat stress) that are expected to limit crop productivity in the region along the entire 21st century. We focus on the two main provinces within the IRB, Punjab and Sindh—which account for over 90% of the country’s food production and 75% of the country’s export revenues [ 1 ]—and analyze four key crops: wheat, cotton, rice and sugarcane. The paper is organized as follows: Section 2describes the area of study, data and methods used for the analyses performed. The results obtained are presented and discussed Atmosphere 2022,13, 378 3 of 15 in Section 3. Finally, Section 4summarizes the implications of our key findings for the agricultural sector. 2. Materials and Methods 2.1. Area of Study The climate in Pakistan varies from arid to semiarid with a range of annual rainfall from 250 mm/year to up to 2000 mm/year in the southern slopes of the Himalaya and submountain northern regions. Heavy monsoons in summer months account for approximately 60% of the total annual precipitation nationally and define precipitation patterns in the south-eastern regions. The north-western regions of the country receive rains mainly during winter months (December to March) through western weather disturbances [11]. The alluvial plain of the Indus River has an area of approximately 207,200 km 2 , covering nearly 65% of Pakistan’s territory. The basin is the country’s main center of agricultural production, the source of over 90% of the country’s food and agriculture commodities [ 1 ]. Of the total cropped area in Pakistan (23.4 million ha), 77% is located in the Punjab province and 14% in the Sindh province [ 24 ], which are shown in Figure 1. The present study targets these two provinces central to agricultural production. Punjab Sindh Figure 1. Study area and location of the 15 weather stations used in this work (red triangles), together with a land-use map for the Punjab and Sindh provinces. Note that the Jammu and Kashmir regions, whose current status has not yet been agreed upon the different parties involved, are not shown in the map. Sources: PMD and FAO-Pakistan. 2.2. Observational Data Daily precipitation, maximum and minimum temperature records for 15 weather stations across Punjab and Sindh provinces (red triangles in Figure 1; details in Table 1) were provided by the Pakistan Meteorological Department (PMD) for the period 1997–2016. PMD data was used (i) to compute the climate indicators described in Section 2.2 and their Atmosphere 2022,13, 378 4 of 15 corresponding trends (Section 2.4), and (ii) to calibrate the statistical models used to obtain the local projections of climate change (Section 3.2.1). Table 1. Details of the 15 PMD stations used in this study. Station Name Longitude (◦) Latitude (◦) BADIN 68.90 24.63 BAHAWAL-NAGAR 73.25 29.95 BAHAWAL-PUR 71.78 29.33 DERA ISMAIL KHAN 70.92 31.82 FAISALABAD 73.10 31.43 HYDERABAD 68.42 25.38 JACOBABAD 68.47 28.30 JHELUM 73.72 32.93 KARACHI (AIRPORT) 67.13 24.90 KHANPUR 70.68 28.65 LAHORE 74.33 31.55 MULTAN 71.43 30.20 MURREE 73.38 33.92 NAWABSHAH 68.37 26.25 SIALKOT 74.53 32.50 Figure 2shows the observed mean climatology for precipitation, maximum and minimum temperature (in rows) for the 15 PMD stations considered for the period 1997–2016 , for the entire year and two seasons of particular interest, Kharif (May–October) and Rabi (November–April), which cover the growth cycles of the four main crops in the IRB: wheat, cotton, rice and sugarcane (see Table 2). 1 0 2 4 20 30 40 10 20 30 Year Kharif (May-Oct) Rabi (Nov-Apr) Precipitation Maximum temperature Minimum temperature ºC ºC 25 35 15 25 3 5 mm/day Figure 2. Observed mean climatology for precipitation, maximum and minimum temperature (in rows) for the 15 PMD stations over the period 1997–2016, for the entire year and the Kharif and Rabi seasons (in columns). Atmosphere 2022,13, 378 5 of 15 Table 2. Cropping calendar for wheat, cotton, rice and sugarcane (in rows). The blue/green/yellow boxes, labelled as P/G/H correspond to the planting/growing/harvesting phase. Brown boxes identify those periods of the year which can correspond to more than one phase, depending on the region. Source: Personal communication from Jam Khali (FAO-Pakistan). Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Wheat (Rabi) G G H H P P G Rice (Kharif) P P P G G G/H H H Cotton (Kharif) P P P G G G H H H Sugarcane (Kharif) G/H P/H P P/G G G G G G G G G Sugarcane (Rabi) G G G G G G G G P/H P/H P/H G 2.3. Extreme Climate and Crop-Specific Indicators To assess the implications of climate variability on crops during the recent historical period, this work uses the precipitationand temperature-based indicators described in Table 3. Table 3. Precipitation and temperature-based extreme indicators used. Related Variable Indicator Description Units CDD Largest number of Consecutive Dry Days (precipitation < 1 mm) days CWD Largest number of Consecutive Wet Days (precipitation ≥1 mm) days SDII Simple Daily Intensity Index mm/wet day Precipitation R20 Number of very heavy precipitation days (precipitation ≥20 mm) days TX90p 90th percentile of maximum temperature ◦C SU Summer days: Number of days with maximum temperature above 35 ◦Cdays TR Tropical nights: Number of days with minimum temperature above 20 ◦Cdays Temperatures ETR Extreme diurnal temperature range (difference between the highest maximum temperature and the lowest minimum temperature) ◦C With the exception of ETR, all these indicators have been defined by the Expert Team for Climate Change Detection and Indices (ETCCDI: http://etccdi.pacificclimate.org/ list_27_indices.shtml (accessed on 14 February 2021). In addition, the indicator summer days (SU) has been modified for this study to represent the agro-climatic conditions of Pakistan. In particular, we have increased the 25 ◦ C threshold defined in ETCCDI to 35 ◦ C to account for the optimal growth of wheat, cotton, rice and sugarcane have been empirically determined to occur at 30, 32, 36 and 34 ◦C, respectively, [25–29]. 2.4. Trend Analysis In this study, we analyze the observed annual trends over the 1997–2016 period. Note that, although a longer period would be more appropriate for a robust computation of trends, we are using for this study the longest observational records available from the national meteorological authority of the country. For the trend analysis the following climatic variables were selected: (i) precipitation, maximum and minimum temperature and (ii) precipitationand temperature-based extreme indicators as shown in Table 3. In all cases, trends were calculated as the slope of the straight line that best fit (based on least squares) to a given set of data. This line responds to the equation y=ax +b , where the slope acan be computed as: Atmosphere 2022,13, 378 6 of 15 a= n n ∑ i=1 xiyi− n ∑ i=1 xi n ∑ i=1 yi n n ∑ i=1 x2 i− n ∑ i=1 xi!2 Moreover, the non-parametric Mann–Kendall (MK) test [ 30 , 31 ] was applied to assess whether or not the trends were statistically significant. MK is based on the number of positive differences minus the number of negative differences that take place in a time-series, S: S= n−1 ∑ k−1 n ∑ j−k+1 sign(xj−xk) where j and k denote different times (from 1 to n ). For n> 10, the variance of S can be obtained according to the following Equation [32]: Var(S) = 1 18"n(n−1)(2n+5)−∑ p Tp(Tp−1)(2Tp+5)# where p varies over the set of tied groups and Tp is the number of observations in the pth group. Then the statistic of the test, ZMK, can be computed as ZMK =       S−1 √Var(S),i f S >0 0 , i f S =0 S+1 √Var(S),i f S <0 The null hypothesis of no trend in the data can be rejected if ZMK >=Z1−α , where Z1−α is the 100 ( 1 −α)th percentile of the standard normal distribution. In this study, a confidence level of 95% (α= 0.05) was considered. Note that previous works targeting the assessment of trends from gauge data across the globe have used the same approach followed here [33–38]. 2.5. Statistical Downscaling Local daily projections (up to 2100) of precipitation, maximum and minimum temperature for the 15 PMD stations analyzed were obtained by means of statistical downscaling as explained in Section 2.2.1 of [39] (the interested reader is referred to the latter reference for further technical details on the downscaling process). As described therein, the four Earth System Models (ESMs) from the Coupled Model Intercomparison Project–Phase 5 (CMIP5) listed in Table 4were downscaled under two Representative Concentration Pathways (RCPs) representing different socio-economic and emission scenarios, the moderate RCP4.5 [40] and the extreme RCP8.5 [41]. Table 4. The four CMIP5 Earth System Models (ESMs) used in this study. ESM Institution Acronym and Country Horizontal Resolution Reference CAN-ESM2 CCCMA (Canada) 2.8◦×2.8◦[42] CNRM-CM5 CNRM-CERFACS (France) 1.4◦×1.4◦[43] MPI-ESM-MR MPI (Germany) 1.8◦×1.8◦[44] NOR-ESM1 NCC (Norway) 1.5◦×1.9◦[45] Atmosphere 2022,13, 378 7 of 15 3. Results and Discussion 3.1. Observed Historical Climate 3.1.1. Trends in Precipitation and Temperatures Precipitation exhibits high spatial variability, with the northernmost parts of the country recording the largest values, especially in the summer season. Very low precipitation is received across the basin during winter. On the other hand, maximum and minimum temperatures vary up to 15 ◦ C and 10 ◦ C across the basin, respectively. The present study presents results over the entire year as well as disaggregated by the two cropping seasons in the IRB, Kharif season (May–October) and Rabi season (November–April). Figure 3shows the observed trends in annual precipitation, maximum and minimum temperature (in rows) over the 15 PMD stations considered, for the entire year, the Kharif and the Rabi seasons (in columns). Significant (95% confidence level) trends are marked with a white dot. The results in Figure 3show a general increasing trend for precipitation (with values above 10 mm/year in some locations), supporting trends identified by [ 7 ] for the Punjab province. However, the latter study finds a decreasing trend in precipitation over the Sindh province for the period 1910–2007, while the present work shows an increasing trend, although with a smaller magnitude and without statistical significance. The different signal found in [ 7 ] can be explained by the difference in time periods analyzed. Here, trends are presented for the two main cropping seasons elucidating the fact that increased precipitation is driven almost entirely by an increase in the Kharif season (May–October) while near to zero trends are found over the Rabi season (November–April). With respect to temperatures, there is not a clear spatial pattern of observed trends, neither for maximum nor for minimum values, while no significant changes are found in general for maximum temperature, significant increasing trends of more than 0.1 ◦ C/year are encountered for minimum temperature in many locations, with both Kharif and Rabi seasons contributing to a similar extent to this warming. These results are in partial agreement with those from previous studies (see, e.g., [ 7 ]), which identified non-significant changes of up to 0.35 ◦ C in maximum temperature in Sindh and slightly decreasing (almost zero) change in maximum temperature in Punjab over the period 1910–2007. With regards to annual minimum temperature, an increase over the country, including the Punjab and Sindh provinces, has been found in previous works [ 7 ]. Although the period analyzed in the present study differs, the warming tendency is consistent between the two studies. To give further insight into the temporal distribution of precipitation, maximum and minimum temperature throughout the year of the trends shown in Figures 3and 4presents a detailed analysis of annual trends, separately for each calendar month, over the period 1997–2016. Figure 4reveals that the positive trends observed in yearly precipitation are mainly driven by increases in August and September, and also in March (to a lesser degree), particularly in the Punjab province. It also shows a slight decreasing trend in October and no substantial changes in the other months. These results suggest a shift in the wet season to end slightly earlier, and therefore receive increased total rainfall over a shorter period. Regarding maximum temperature, monthly trends exhibit great variability, both temporally and spatially. Particularly in the Sindh province, Figure 4shows a weak signal of warming in June, July and October. To the contrary, a cooling trend emerges from January to April particularly in the Punjab province. Minimum temperatures in the region, however, are increasing throughout the entire year, with the highest warming in October and March, which suggests the elongation of the Kharif season and general shifts in the warm periods throughout the year. Atmosphere 2022,13, 378 8 of 15 Year Kharif (May-Oct) Rabi (Nov-Apr) Precipitation Maximum temperature Minimum temperature -0.250 0.000 0.250 -0.125 0.125 ºC/year -20 0 20 -10 10 mm/year Figure 3. Observed annual trends in precipitation, maximum and minimum temperature (from top to bottom ) over the period 1997–2016 for the whole year and the Kharif and Rabi seasons (in columns). Significant (95% confidence level) trends are marked with a white dot. J an Feb Mar Apr May J un J ul Aug Sep Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep Oct Nov Dec Precipitation Maximum temperature Minimum temperature 5 0 -5 -2 -3 -4 -1 2 1 3 4 mm/year -0.20 -0.10 0.00 -0.15 -0.05 0.05 0.15 0.10 0.20 ºC/year -0.20 -0.10 0.00 -0.15 -0.05 0.05 0.15 0.10 0.20 ºC/year Figure 4. Observed annual trends in precipitation, maximum and minimum temperature (panels from left to right ) over the period 1997–2016 for each of the twelve calendar months. Significant (95% confidence level) trends are marked with a white dot. 3.1.2. Trends in Extreme Climate Indicators Figure 5shows the observed annual trends for the precipitation and temperaturebased extreme indicators listed in Table 3(left and right panel, respectively). In agreement Atmosphere 2022,13, 378 9 of 15 with the results from Figures 3and 4, the analysis of precipitation-based indicators reveals an overall decrease in the number of consecutive dry days (CDD), driven by a negative shift in the Kharif season. During the Rabi season, the number of CDD is increasing in some parts of the basin, which points out to increased periods without rain during the winter months. Over the basin, there is a slight increase in the number of consecutive wet days (CWD) during the Kharif season, however no change during Rabi. The Simple Daily Intensity Index (SDII) shows a slight increase in Kharif and no change during the Rabi season. The analysis also finds an increase in the number of very heavy precipitation days (R20) over the period analyzed driven by changes in the Kharif season. The results support previous discussion of increasing intensity of rain during the wet summer season, however extended dry periods during the already dry winter season. This decrease in dry days may impact non-irrigated Rabi crops, in particular wheat and sugarcane, which will be in critical growth phases during this period [7,39]. Year Kharif (May-Oct) Rabi (Nov-Apr) SUTRETR -2 0 2 -1 1 -1.0 0.0 1.0 -0.5 0.5 -1.0 0.0 1.0 -0.5 0.5 days/year days/year ºC/year Year Kharif (May-Oct) Rabi (Nov-Apr) CDDCWDSDIIR20 -2 0 2 -1 1 days/year -2 0 2 -1 1 mm/day/year -0.50 0.00 0.50 -0.25 0.25 days/year -0.50 0.00 0.50 -0.25 0.25 days/year TX90p -0.50 0.00 0.50 -0.25 0.25 ºC/year Figure 5. Observed annual trends for the temperatureand precipitation-based extreme indicators listed in Table 3( left and right panel, respectively) over the period 1997–2016. Significant (95% confidence level) trends are marked with a white dot. Temperature-based indicators show overall large spatial variability. The indicator exhibiting the most significant trend is TR, defined as the number of days with minimum temperatures above 20 ◦ C. Analysis of seasonal trends show that the increase is driven by changes mainly during the Kharif season in the basin. The temperature indicators such as the 90th percentile of maximum temperature (TX90p) and the number of days with temperature above 35 ◦ C (SU), both indicate decreases in maximum temperature driven by a decrease during the Rabi or winter season. SU shows increasing values in some stations in the Kharif season, suggesting increases in high temperature in the hot summer season. The extreme diurnal temperature range (ETR) exhibit mostly significant decreasing trends in the north during Rabi and in the central regions during Kharif. This finding highlights that different regions experience increased night-time temperatures during different cropping seasons, and therefore crop’s exposure to heat-stress conditions will vary spatially. Changes in the diurnal temperature range (ETR) exhibit a significant decrease driven by changes in the Rabi season. These results suggest that Kharif crops (sugarcane, rice and cotton) might likely be more negatively impacted than Rabi ones due to the compounding impacts of heat stress and increased water-loss. This finding is supported by the results [ 39 ], who found