Download (5MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Resource language: English
Date: 28 February 2017
Number of Pages: 220
Institution: Università degli Studi di Napoli Federico II
Department: Agraria
Dottorato: Scienze agrarie e agroalimentari
Ciclo di dottorato: 29
Coordinatore del Corso di dottorato:
Palladino, MarioUNSPECIFIED
Date: 28 February 2017
Number of Pages: 220
Keywords: drought index, soil moisture, evapotranspiration, Landsat, Uganda
Settori scientifico-disciplinari del MIUR: Area 07 - Scienze agrarie e veterinarie > AGR/08 - Idraulica agraria e sistemazioni idraulico-forestali
Additional information: The present thesis has been carried out within the activities of GULUNAP project
Date Deposited: 06 May 2017 14:54
Last Modified: 14 Mar 2018 09:30
DOI: 10.6093/UNINA/FEDOA/11771

Collection description

Drought is a natural disaster that occurs in all climatic regions, and is responsible for food insecurity in majority of developing countries. Globally, interventions to mitigate drought impacts have focused in semi-arid and arid areas. However, increased climatic variations have become more frequent in the recent past causing sporadic agricultural droughts with the end results being food shortages even in areas that receive relatively high rainfall. Many of such areas are occupied by vulnerable communities lacking strategic coping mechanisms for mitigation of drought impacts. The aim of this study was to develop a simplified approach for computing a Soil Moisture Deficit Index (SMDI) that integrates limited climatic records common in developing countries together with freely available tools for supporting soil water management decisions under rainfed agriculture. Most soil moisture based drought indices are derived from long term records of measured soil moisture time series. However, such long-term soil moisture records are scarcely available in African countries where they could be of greatest benefit in designing techniques for mitigating drought impacts. Therefore, the main objective of this research was to evaluate the performance of simulated soil moisture time series to develop a SMDI with minimal requirements of input data. To this aim, the study was organized in four consecutive objectives, namely: to identify and adapt a suitable drought indicator in relation to the data availability. Secondly, to assess the feasibility of using a calibrated agro-hydrological model for producing long time series of soil water dynamics and derive SMDI for monitoring agricultural droughts. The third objective was to upscale the SMDI through energy balance modeling using a case study in Northern Uganda. And the fourth objective was, to formulate a soil water management decision support scheme for mitigation of agricultural droughts in rain fed farming systems through application of SMDI. The study is based on agro-hydrological data collected in a dairy farm of 10 ha in Northern Uganda equipped with Mateo station and low cost commercial soil sensors to monitor soil water dynamics in the root zone during two seasons under rain fed maize crops in 2015. Because of the importance of Evapotranspiration in agro-hydrological studies and limited reported research on it at the study site, 13 different simplified reference evapotranspiration (ET0) models were compared with FAO-56 Penman-Monteith to select the best performing simplified model for application in the study area. Evaluation of the 13 ET0 models showed that the Makkink radiation model gave the best prediction of ET0 with Root Mean Squared Error (RMSE) = 0.6 mm, Mean Absolute Error (MAE) = 0.4 mm, Nash Sutcliffe Efficiency (NSE) = 0.8, Coefficient of Agreement (d) = 0.90 and Coefficient of Determination (r2) = 0.7. All temperature based models overestimated ET0 with Thornthwaite giving the worst prediction in all the test statistics. To address the first objective, the state of the art on soil moisture based drought indices were reviewed and the information gathered applied to formulate a new approach to define SMDI. The second objective was addressed through application of 1-dimensional water flow model (Hydrus 1D); firstly, in inverse mode to derive the soil hydraulic properties and secondly in direct mode to generate soil moisture time series by using the ET0 model selected in the previous step in conjunction with gridded climatic data combined with limited weather observations. In the calibration phase, Landsat 8 OLI satellite images were used to estimate crop growth variables. In the second objective, published crop coefficients where used to generate continuous boundary conditions for 21-year agro-hydrological simulations. Calibration results showed good agreement between simulations and observations of water storage in the root zone with r2 = 0.73 during calibration and r2 = 0.70 during validation. The results of the long-time series simulations were used to derive threshold parameters for SMDI definition, following the statistical approach suggested by Hunt et al. (2009). Generation of the threshold parameters; i.e.: water content at field capacity (θFC) and water content at wilting point (θWP), through agro-hydrological simulations gave good comparison with the laboratory determined values through a pressure plate apparatus on undisturbed soil core samples; with r2 = 0.95. Comparison between number of times SMDI<0, within a growing season and maize yields between 2007 to 2015, showed a negative linear correlation with r2 = 0.64. Precipitation (P) and precipitation deficit (D) were fitted on theoretical probability distributions to calculate reference drought indices i.e. Standardized Precipitation Index (SPI) and Standardized Precipitation Evapotranspiration Index (SPEI). The fitting distributions (a 2-parameter gamma distribution for P, and a 3-parameter log-logistic distribution for D) gave an acceptable Kolmogorov-Smirnov goodness of fit test at 95% level of significance in both cases. All the reference indices [i.e. SPI, SPEI and Atmospheric Water Deficit (AWD)] showed positive correlation with SMDI demonstrating the robustness of SMDI for agricultural drought monitoring in the study area. The third objective involved analysis of Landsat 8 thermal images to generate evaporative fraction (Λ) through energy balance modeling. A SMDI- Λ regression equation was developed to spatialize SMDI. A comparison between SMDI and Λ through linear regression showed good agreement with r2 = 0.84. An independent check with different sets of images were performed between the SMDI calculated using the SMDI-Λ regression equation and SMDI generated through agro-hydrological simulations provided a good agreement with r2 = 0.85. The last objective involved integration of the results obtained from objectives (i) to (iii) to formulate a decision support scheme. SMDI was found useful to delineate dry and the wet season in Northern Uganda; it showed that the dry season begins between November 25 to December 10 and the wet season begins between March 26 to April 5 of each year. A SMDI-based management decision support scheme was proposed, although it would need further investigations to verify its effectiveness. In conclusion, the approaches developed to define SMDI can easily be implemented in any developing country that experiences similar problems in rain fed farming.


Downloads per month over past year

Actions (login required)

View Item View Item